application of physiology in wheat breeding

Appl i cati o n of Phy s iol ogy i n Wheat Br eedi n g

M. P. Re y nold s , J. I . Or tiz -M

o n aste r io, a nd A. McNa b, E d itor s

M.P. Reynolds,J.I. Ortiz-Monasterio,and A. McNab,Editors

ISBN: 970-648-077-3

International Maize and Wheat Improvement CenterApdo. Postal 6-641, 06600 Mexico, D.F., Mexico

www.cimmyt.org

Applica tion ofPhysiology

in W hea t Breeding

M.P. Reynolds, J.I. Ortiz-Monasterio,

and A. McNab, Editors

CIMMYT® (www.cimmyt.org) is an internationally funded, nonprofit, scientific research and training organization.Headquartered in Mexico, CIMMYT works with agricultural research institutions worldwide to improve the productivity,profitability, and sustainability of maize and wheat systems for poor farmers in developing countries. It is one of 16 food andenvironmental organizations known as the Future Harvest Centers. Located around the world, the Future Harvest Centersconduct research in partnership with farmers, scientists, and policymakers to help alleviate poverty and increase food securitywhile protecting natural resources. The centers are supported by the Consultative Group on International Agricultural Research(CGIAR) (www.cgiar.org), whose members include nearly 60 countries, private foundations, and regional and internationalorganizations. Financial support for CIMMYT’s research agenda also comes from many other sources, including foundations,development banks, and public and private agencies.

Future Harvest® builds awareness and support for food and environmental research for a world with lesspoverty, a healthier human family, well-nourished children, and a better environment. It supports research,

promotes partnerships, and sponsors projects that bring the results of research to rural communities, farmers, and families inAfrica, Asia, and Latin America (www.futureharvest.org).

© International Maize and Wheat Improvement Center (CIMMYT) 2001. All rights reserved. The opinions expressed in thispublication are the sole responsibility of the authors. The designations employed in the presentation of materials in thispublication do not imply the expression of any opinion whatsoever on the part of CIMMYT or its contributory organizationsconcerning the legal status of any country, territory, city, or area, or of its authorities, or concerning the delimitation of itsfrontiers or boundaries. CIMMYT encourages fair use of this material. Proper citation is requested.

Printed in Mexico.

Correct citation:Reynolds, M.P., J.I. Ortiz-Monasterio, and A. McNab (eds.). 2001. Application of Physiology in WheatBreeding. Mexico, D.F.: CIMMYT.

AGROVOC Descriptors:Plant breeding; Breeding methods; Plant physiology; Wheats; Germplasm; Selection;Environmental factors; Resistance to injurious factors; Yields

Additional Keywords: CIMMYT

AGRIS Category Codes: F30 Plant Genetics and BreedingF60 Plant Physiology and Biochemistry

Dewey Decimal Classif.: 631.53

ISBN: 970-648-077-3

Design and layout: Marcelo Ortiz S., Eliot Sánchez P., and Miguel Mellado.

Cover photo credits: Demonstration of the infrared thermometer, Poza Rica, Mexico, G.P. Hettel

Waterlogging treatment, Ciudad Obregon, Mexico, K.D. Sayre

Wheat on the edge of the Sahara Desert, Egypt, M.P. Reynolds

C O N TEN TSiv Preface

General Considerations in Physiological Breeding2 Introduction. Application of Physiology in Wheat Breeding—M.P. Reynolds, R.M. Trethowan,

M. van Ginkel, and S. Rajaram

11 Chapter 1. Directions for Physiological Research in Breeding: Issues from a Breeding Perspective—P.A. Jackson

17 Chapter 2. Searching Genetic Resources for Physiological Traits with Potential for Increasing Yield—B. Skovmand, M.P. Reynolds, and I.H. Delacy

29 Chapter 3. Genetic Basis of Physiological Traits—J.-M. Ribaut, H.M. William, M. Khairallah,A.J. Worland, and D. Hoisington

48 Chapter 4. Managing Experimental Breeding Trials—P.R. Hobbs and K.D. Sayre

59 Chapter 5. Recent Tools for the Screening of Physiological Traits Determining Yield—J.L. Araus,J. Casadesus, and J. Bort

78 Chapter 6. Economic Issues in Assessing the Role of Physiology in WheatBreeding Programs—J.P. Brennan and M.L. Morris

Breeding for Adaptation to Environmental Factors88 Chapter 7. Traits to Improve Yield in Dry Environments—R.A. Richards, A.G. Condon,

and G.J. Rebetzke

101 Chapter 8. Salinity Tolerance—K.N. Singh and R. Chatrath

111 Chapter 9. Cold Tolerance—N.N. Sãulescu and H.-J. Braun

124 Chapter 10. Heat Tolerance—M.P. Reynolds, S. Nagarajan, M.A. Razzaque, and O.A.A. Ageeb

136 Chapter 11. Waterlogging Tolerance—A. Samad, C.A. Meisner, M. Saifuzzaman, and M. van Ginkel

145 Chapter 12. Preharvest Sprouting Tolerance—R.M. Trethowan

148 Chapter 13. Selection Traits for Improving Yield Potential—R.A. Fischer

160 Chapter 14. Manipulating Wheat Development to Improve Adaptation—G.A. Slaferand E.M. Whitechurch

Breeding for Nutritional and Soil Factors172 Chapter 15. Acid Soils and Aluminum Toxicity—A.R. Hede, B. Skovmand, and J. López-Cesati

183 Chapter 16. Genotypic Variation for Zinc Efficiency —I. Cakmak and H.-J. Braun

200 Chapter 17. Nitrogen and Phosphorus Use Efficiency—J.I. Ortiz-Monasterio,G.G.B. Manske, and M. van Ginkel

208 Chapter 18. Techniques for Measuring Genetic Diversity in Roots—G.G.B. Manske, J.I.Ortiz-Monasterio, and P.L.G. Vlek

219 Chapter 19. Micronutrients—J.S. Ascher-Ellis, R.D. Graham, G.J. Hollamby, J. Paull, P. Davies, C. Huang,M.A. Pallotta, N. Howes, H. Khabaz-Saberi, S.P. Jefferies, and M. Moussavi-Nik

iii

We applaud this practical guide to the application of physiology in wheat breeding, whichbrings together in one volume the working knowledge of a broad range of experts in salinity,drought, cold, waterlogging, micronutrients, and other key topics.

The more understanding plant breeders have of the physiological processes that underlie plantperformance, the more efficiently they can exploit relevant physiological mechanisms toimprove crop performance. Wheat breeders have become increasingly able to use physiologicaltraits directly as selection criteria, as their knowledge of physiological processes has expandedand as traits have been identified that can be used as selection criteria to achieve results morequickly and efficiently than selecting for yield performance alone.

Nonetheless, there are still major gaps in our understanding of how crops adapt to theenvironment, and this calls for further physiological research. Indeed, a more completeunderstanding of crop physiology will be a prerequisite to the effective application of newtechniques such as genetic transformation, functional genomics, and marker-assisted selectionin wheat breeding.

The improved varieties developed though wheat breeding are important catalysts for increasingcrop performance at the farm level, where a range of biotic and abiotic stresses impinge onyields. However, for the maximum genetic yield potential of improved varieties to be fullyexpressed, scientists must also pay due attention to crop management practices. Withoutadequate soil fertility, appropriate planting methods, effective control of weeds and pests, andefficient water management, the full economic benefits of genetic improvement can never berealized.

Brief theoretical explanations are provided throughout this book, but the main focus is onpractical procedures breeders can readily apply. Such topics as economic issues related to therole of physiology in wheat breeding and the search for genetic diversity that could contributeto increasing yield will help breeders take full advantage of existing methodologies andresources to do their work more efficiently. The chapter on the genetic basis of physiologicaltraits brings out the point that though field testing is indispensable, proper combination withmolecular data could lead to more efficient use of limited resources.

The collected wisdom contained in this book was generously contributed by the authors, and wethank them for sharing the fruits of their varied experience. Through this book, their expertisewill be accessible to breeders everywhere, but especially in developing countries, whereinformation on this newly emerging field is rarely available.

____________________________ ____________________________

Sanjaya Rajaram Norman E. Borlaug

Director Senior ConsultantCIMMYT Wheat Program CIMMYT

PREFA CE

iv

1

GENERAL CONSIDERATIONS

IN PHYSIOLOGICAL

BREEDING

2

Application of Physiologyin Wheat BreedingM.P. Reynolds, R.M. Trethowan, M. van Ginkel, and S. Rajaram1

How can disciplinary research inphysiology complement wheatbreeding? This introductory chapter isintended to provide broad guidelines tohelp breeding programs: 1) assesswhether physiological criteria should beincluded in a breeding strategy; 2)evaluate specific physiological selectiontraits and determine their usefulness inbreeding. The other chapters in thisbook provide more explicit informationon how physiological approaches can beused in breeding work for a variety ofenvironmental conditions.

Physiological criteria are commonlythough not explicitly used in breedingprograms. A good example is selectionfor reduced height, which improveslodging resistance, partitioning of totalbiomass to grain yield, andresponsiveness to management. Anotheris differential sensitivity to photoperiodand vernalizing cold, which permitadaptation of varieties to a wide rangeof latitudes, as well as to winter- andspring-sown habitats. Despite a lack ofdetailed understanding of howphotoperiod and vernalizationsensitivity interact with each other andthe environment, the relatively simpleinheritance of photoperiod (Ppd) andvernalization (Vrn) sensitivity genes and

their obvious phenotypic expression (i.e.earliness versus lateness) has permittedthem to be modified in many breedingprograms. The same is true for theheight reduction (Rht) gene. In the futurean increased understanding of thegenetic basis of these traits may enablebreeding programs to exploit themfurther.

Selection for reduced height andimproved adaptation to environment hashad a profound impact on modern plantbreeding, and the improvement in yieldpotential of spring wheat since theGreen Revolution has been shown to beassociated with a number of otherphysiological factors (Reynolds et al.,1999). Nonetheless, most breedingprograms do not put much emphasis onselecting physiological traits per se(Rajaram and van Ginkel, 1996).Exceptions would include: 1) the stay-green character, which has been selectedfor in relation to improved diseaseresistance and is associated with highchlorophyll content and photosyntheticrate in Veery wheats, for example Seri-82 (Fischer et al., 1998), and 2) moreerect leaf angle, a common trait in manyhigh yielding bread and durum wheatplant types that was introgressed into theCIMMYT germplasm pool in the early1970s (Fischer, 1996).

A recent survey of plant breeders andphysiologists addressed the question ofhow physiological approaches in plantbreeding could have greater impact(Jackson et al., 1996). According to thesurvey, while the impacts ofphysiological research on breedingprograms have been limited in the past,future impacts may arise through:

• Focusing physiological work on anappropriate range of germplasm(which will depend on the specificbreeding objectives);

• Working with larger populations toenable extrapolation of findings tobreeding methods;

• Identifying traits for use as indirectselection criteria, in addition to thosealready used in core breedingprograms;

• Identifying traits for use as selectioncriteria in introgression programs;

• Conducting selection trials in morerepresentative environments, and

• Developing tools that could bequickly and easily applied to largenumbers of segregating lines.

In this and the following chapters, manyof these suggestions are incorporatedinto a research framework for assessingthe value of physiological selection traitsin a breeding context.

1 CIMMYT Wheat Program, Apdo. Postal 6-641, Mexico, D.F., Mexico, 06600.

IN TR O DUCTI O N

3

Assessing thePotential ofPhysiological SelectionCriteria to ComplementBreeding Strategies

The Art, Science, andEmpiricism of BreedingBreeding is frequently referred to as ablend of science and art, as well as anempirical process. The science refers tothe routine application of establishedfacts, such as the documented role ofspecific genes in conferring diseaseresistance or environmental adaptation.The art of breeding refers to the intuitiongained by working with germplasm andthe integration of that experience withestablished knowledge. In other words,intuition enables good breeding decisionsto be made based on our incompleteknowledge of the biology and ecology ofplants. Empiricism refers to the use ofmultiple crossing and selection strategiesto achieve a single objective, sometimesloosely referred to as “the numbersgame.” Physiological understanding addsto the science, and as such complementsthe intuitive knowledge required toconduct good breeding. Use ofphysiological selection criteria canimprove the probability of success bymaking empirical selection moreefficient.

Theoretical Basis for UsingPhysiological TraitsAssuming significant genetic diversity fora trait is established, the question of howits use as a selection criterion wouldimprove breeding efficiency can beaddressed. Without experimentation, thiscannot be predicted with any certainty,any more than a breeder can know inadvance which of many crosses willproduce the desired variety. An essentialquestion is whether selection for a givenphysiological trait as part of an integrated

breeding approach will achieve resultsmore quickly and efficiently thanselecting parents and/or progeny forperformance alone.

Many traits may appear to be of potentialbenefit to yield. To assess which trait(s)should be prioritized, alternatehypotheses may be tested empirically,based on a conceptual model thatincorporates current understanding ofphysiological and biochemicalconstraints to performance (Figure 1).For example, if the principal yield-limiting factor is water stress,physiological understanding wouldsuggest that genotypes with deeper rootswill have an advantage over others,assuming moisture is available deeper inthe soil profile. However, selection foryield alone will not guarantee that goodlines have the deepest roots, becausedrought tolerance may be conferredthrough genetic superiority of othermechanisms, such as osmotic adjustment,accumulation and remobilization of stemreserves, superior spike photosynthesis,heat tolerant metabolism, and goodemergence and establishment undermoisture stress (Figure 1).

Genetic advance in dry environments hasbeen quite limited in most breedingprograms. Slower progress in moisturestressed environments compared toirrigated environments is usually ascribedto the heterogeneity of selection nurseriesunder dry conditions, which rendersperformance-based selection unreliable.Selection for specific traits in morecontrolled environments is likely to bemore effective. In addition, if more thanone mechanism is involved in droughttolerance, deliberate selection with aview to combining synergistic traits islikely to achieve results sooner thanadopting a strategy whereby parents andprogeny are selected on the basis ofperformance alone (see chapter byRichards et al.).

Figure 1. A conceptual model for droughttolerance in wheat. The theoretical ideotype hashigh expression of the following traits (not all of whichwould be useful in all drought environments): seed size& coleoptile length (improve early crop establishment),early ground cover & pre-anthesis biomass (reduceevaporation of soil moisture), stem reserves /remobilization & spike photosynthesis (help grainfillingduring severe post-anthesis stress), stomatalconductance (indicative of roots which are able toextract soil water at depth), osmotic adjustment(maintains cell functions at low water potential),accumulation of abscisic acid (pre-adapts cells tostress), heat tolerance (heat stress may be caused bylow leaf transpiration rates under drought), leafanatomical traits e.g. waxiness, pubescence, rolling,thickness (reduce risk of photo-inhibition), high tillersurvival and stay-green (easily observed integrativetraits indicative of good drought tolerance).

High spike photosynthesis

Stem reserves

Cellular traits: osmotic adjustment,heat tolerance, ABA, etc.

Leaf traits: wax, rolling, thickness, etc.

High pre-anthesis biomass

Early ground cover

Long coleoptile

Large seed

Water relations traits:stomatal conductance, etc.

APPLICATION OF PHYSIOLOGY IN WHEAT BREEDING

H2O

4

Establishing the GeneticBases of Physiological TraitsOnce the value of a physiological traithas been established, it may be useful todetermine its genetic basis, such as thenumber and location of genes involvedin its expression. (Genetic studies,including the identification of molecularmarkers, can be conducted on the samekinds of populations developed toestablish genetic gains associated withselection of phenotypic traits.) In theory,such an investment would enablefingerprinting for stress tolerance, orother desirable traits, on fixed lines inany breeding program worldwide. Thisinformation would allow strategiccrossing programs to improve thelikelihood of pyramiding droughttolerance traits, without necessarilyhaving to measure phenotypicexpression in the parents or progeny. If atrait is genetically complex and itsexpression a function of epistatic andother interactions among genes, thengenetic markers would need to beidentified in several different geneticbackgrounds to gain comprehensiveinformation about which loci may beinvolved.

Although identifying adaptivephysiological mechanisms and theirgenetic markers may be time-consumingand costly, once the initial investment ismade, the information is permanentlyavailable. The information can be used atdifferent stages of the breeding process,depending on the resources available.

In a relatively low investment scenario,information on important physiologicaltraits can be collected on potentialparental lines. For example, it might beworth screening an entire crossing blockor a subset of commonly used parents toproduce a catalogue of usefulphysiological traits or their geneticmarkers. The information can be usedstrategically in designing crosses,thereby increasing the likelihood of

transgressive segregation events thatbring together desirable traits. In ascenario where more resources areavailable to screen for physiologicaltraits, the same selection criteria could beapplied to segregating generations, inyield trials, or any intermediate stage,depending on where genetic gains fromselection are optimal.

The rest of this chapter will describegeneralized procedures for evaluatingphysiological criteria within a breedingprogram, and how they might be applied.

Standard Proceduresfor IncorporatingPhysiological Criteriainto a BreedingStrategy

The procedure for incorporatingphysiological criteria into a breedingprogram has two phases, each of whichconsists of a number of experimentalsteps (see boxes).

Phase 1. Identifying TraitsAssociated with PerformanceDefine the target wheat-growingenvironmentBefore designing a research program, itis crucial to define the physical andagricultural characteristics of the targetenvironment (Table 1), for a number ofreasons. First, this information enablesselection traits to be chosen which aremost relevant to the environmentalfactors limiting performance. Forexample, when aiming to improveperformance under drought, there is nopoint in selecting for deep rootingcapacity if target environments arecharacterized by soil profiles that lackadditional water at depth. Similarly, itwould not be beneficial to select for earlyground cover with a view to conservingsoil moisture if farmers practice residueretention or soil mulching.

Second, an accurate description of thetarget environment is important tofacilitate experimental design bypermitting identification of appropriateresearch sites (based on temperatureprofile, latitude, soil type, etc.) as well aschoice of experimental treatments (i.e.sowing dates, crop management, etc.).By matching the conditions of theexperimental environment with those ofthe target environment as far as possible,results are more likely to berepresentative of the target environmentas a whole. In many instances it isadvisable to replicate trials across anumber of locations within the targetenvironment. It may not be possible tomimic the target environment precisely atany experimental site, and for strategicreasons these sites may be manageddifferently. Nonetheless, decisions of thistype should be based on as complete aknowledge base as possible of the targetenvironment (see chapter by Hobbs andSayre).

Identify physiological traits/selection criteriaWhen considering the incorporation ofphysiological criteria into a breedingstrategy, previously published work ontraits and methodologies can be

Steps for IncorporatingPhysiological Criteria into aBreeding Strategy

Phase 1: Identifying TraitsAssociated with Performance

• Define the target wheat-growingenvironment

• Identify physiological traits/selectioncriteria

• Choose genotypes appropriate forevaluating trait expression

• Design the experimental environment• Develop protocols to optimize trait

expression• Measure trait expression and its

association with performance

M.P. REYNOLDS ET AL.

5

Table 2. Desirable characteristics forgermplasm used in physiological breedingexperiments.

• Generally adapted to the target environment• Acceptable range of maturity• Acceptable agronomic type• Resistant to prevalent diseases and pests• Broad genetic background• Not contrasting in height class†

• Similar response to photoperiod andvernalization†

† Unless characteristics are those under investigation.

consulted. Much of this literature hasbeen reviewed—for example, by Blum(1988) for abiotic stresses, by Evans(1993) for yield potential, and by Lossand Siddique (1994) for wheat underdrought. In addition, many traits andreferences will be included in subsequentchapters related to yield potential, heatstress, drought stress, etc. To be ofpotential use in breeding, a trait mustmeet two broad criteria 1) evidence ofsignificant genetic variability for the traitmust exist, and 2) selection for the traitmust be economically advantageousbased on relative costs and benefits (seechapter by Brennan and Morris).

In choosing traits it may be helpful tothink in terms of their level of integrationat the whole plant level. Traits can beclassified into two broad categories:simple traits associated with a particularmorpho-physiological attribute, such asleaf waxiness, and integrative traitsproduced by the net effect of a number of

Table 1. Key parameters defining target and experimental environments.

Parameter Units

ClimateMaximum daily temperature ºC (monthly mean)Minimum daily temperature ºC (monthly mean)Sunhours or incident radiation h/day or joules/m2/dAnnual rainfall mm/monthRelative humidity (% max/min)

Crop environmentAverage yield t/haPreceding crop(s) e.g. summer crop speciesBiotic stresses Typical % yield loss

SoilSoil type e.g. clay/ loam/sandSoil pH pHPhysical properties e.g. compaction zonesOrganic matter %Rooting depth approx in cm

ManagementTypical fertilizer rates NPK etc. kg/haTypical irrigation schedule Frequency/mm applied (if available)Disease, pest, and weed control Frequency

simpler traits—for example, canopytemperature. Being a function of severalsimpler traits, integrative traits arepotentially powerful selection criteria forevaluating breeding progeny; however,the heritability (measured as the ratio ofgenetic variance σ2

g to phenotypicvariance σ2

p) of such traits is generallylower than that of simple traits. Clearlythe heritability of a target trait willinfluence the ease with which it can bemeasured and employed in a breedingprogram.

Choose genotypes appropriate forevaluating trait expressionThe initial choice of germplasm iscritical since conclusions will hinge on itbeing representative of current breedingobjectives. Crossing block materials andadvanced breeding lines are goodsources. Germplasm collections mayprovide useful sources of geneticdiversity, especially if the accessionsoriginate from environments where the

yield constraints are similar to those inthe target environment. Ideallygermplasm should have a number ofcommon characteristics that will facilitateexperimental work and interpretation ofthe results (Table 2). Differences inheight, maturity, adaptation, diseasesusceptibility, etc., are all factorspotentially confounding to the trait understudy, and variation may increaseexperimental error. It is unlikely that allmaterial will meet all of the criteria;however, research findings will be greatlyenhanced if most of them are met.

If significant genetic diversity for the traitcannot be found in agronomicallyacceptable backgrounds, it may benecessary to introgress the trait intocommonly used parental lines before traitevaluation can begin. Since this requiresconsiderable resources, there has to begood reason to believe that the trait iseconomically advantageous, based on, forexample, studies in related species orpreliminary studies on lines in whichmodest genetic diversity for the traitindicated a strong association withperformance.

An important additional point to checkbefore proceeding is the current status ofgermplasm coming out of ongoingselection programs with respect to thetrait of interest. If that material alreadyshows high expression for the trait ofinterest with little significant genetic


6

diversity, then the methodologies beingused are effective in making geneticgains for the trait. Developingphysiological selection methodologiesis only worthwhile if they can achievegreater efficiency than existingapproaches. Otherwise, the only role ofphysiological measurements may be toidentify new and better genetic sourcesof the trait.

The number of lines studied must besufficient to ensure a range of geneticdiversity for the trait of interest,preferably in several diverse but locallyadapted genetic backgrounds. It may besensible to start with a large number oflines for preliminary observations. Thiscould run into a hundred or so,depending on the complexity of thetrait; the precise number will depend onthe likelihood of finding geneticdiversity for the trait of interest. Oncediversity has been established frompreliminary observations in a controlledtest cycle, numbers can be reduced (e.g.20-50) to include the best germplasm,encompassing the full range of geneticdiversity, for more detailed observationsin subsequent cycles.

Design the experimentalenvironmentTrials should be managed optimallybecause factors such as disease pressureand irregular irrigation can introduceerrors into the expression ofphysiological traits. Besides choosingan appropriate test site that isrepresentative of the target environment,a number of management factors needto be considered (Table 3). Somefactors, such as sowing dates that mimictemperature and photoperiod regimes,should be as representative as possibleof the target environment to avoidexpression of traits not relevant to theexperimental objectives. Others, such asland preparation and pest control,should be managed optimally to reduce

experimental error. Seed quality isanother factor which should be managedoptimally to ensure good standestablishment. In addition, seed shouldcome from the same growingenvironment. Differing seed sources mayconstitute a serious confounding factor,for example when studying responses tomicro-elements known to be stored tovarying degrees in the seed, dependingon local growing conditions (see chapterby Ascher-Ellis et al.). More details onmanagement of trials is given in thechapter by Hobbs and Sayre.

Develop protocols to optimize traitexpressionThe efficiency of physiological traitselection will be related to how well atrait is expressed and measured.Therefore, for any trait, experimentationmust take place to establish how andwhen measurements should be made tomaximize genetic resolution for itsexpression. Three groups of factors mayinteract with the expression of a trait: 1)macro-environment, i.e. temperature,radiation, irrigation status, nutritionalstatus, and soil type; 2) micro-environment, i.e. small daily fluctuations

Table 3. Management factors in traitevaluation work.

Factors that should be representative oftarget environment:• Crop rotation• Sowing date (to mimic target photoperiod

and temperature regime)• Planting method• Seed rate• Fertilizer regime• Irrigation regime

Factors that should be managed optimally:• Seed quality and source• Land preparation• Pest and disease control• Weed control• Appropriate statistical design

in temperature and radiation, etc., aswell as small environmental differencesamong plots or between plants caused,for example, by soil heterogeneity,weeds, pests, etc.; and 3) physiologicalfactors, i.e., age of plant or its organs,diurnal rhythms of plants, and smallamounts of genetic diversity that mayexist within so-called fixed lines.

For example, leaf chlorophyll isrelatively simple in its expression in thatit is not affected by diurnal changes.However, its expression may vary underdifferent nutritional regimes.Chlorophyll is also a function of leafage, and some standardization inmeasurement will be necessary to takethis into account. One might choose tomeasure chlorophyll in the flag-leaf atflowering or in the youngest fullyexpanded leaf at regular intervals aftersowing. On the other hand, traits likeleaf conductance or canopy temperaturedepression (CTD) are strongly affectedby temperature and relative humidity(i.e. vapor pressure deficit) and have adiurnal function. Studies in CIMMYT(Amani et al., 1996) have shown thatCTD is best expressed on warm, sunny,cloudless afternoons, in well-wateredplots. The trait is also affected byphenology and, while pre-headingreadings are usually higher, readingsmade during grainfilling are bestassociated with yield potential.

Avoiding confounding factors and useof experimental design. While it isimpossible to completely avoidconfounding factors in fieldexperiments, much can be done tominimize their effect through planning,good crop management, and appropriateexperimental design. For example, it hasalready been mentioned that germplasmshould be chosen to avoid excessivecontrasts in maturity date and height(Table 2). In addition, if test sites arecarefully chosen and managed


7

optimally, plot-to-plot variability can bereduced. For traits affected by weatherconditions, investigators may need to beflexible in the timing of datameasurement, collecting readings onlyunder relatively favorable conditions.

Fortunately, appropriate statisticalprocedures can help attenuate some ofthese problems. The use of replication incombination with appropriate blockingstructures is very powerful and will helpcontrol the effects of heterogeneityinherent in all field experimental sites.Covariance analysis can help reduce theconfounding effect of plant age andphenology, provided the trait andmaturity class are not genetically linked.Multiple sampling will reduce errorsassociated with measurement (humanand instrument error). These strategies,in combination with advanced dataanalysis procedures (i.e., spatialanalysis), will reduce error associatedwith environmental variability.

Genotype by environment interaction.The factors that can affect the expressionof a trait (i.e., macro-environment,micro-environment, and physiologicalfactors) may show interaction withgenotype, accounting for what iscollectively called genotype byenvironment interaction (G×E). Sometraits demonstrate little G×E; that is tosay that genotypes ranked based onthese traits will largely maintain thisrank across different environments,regardless of the absolute expression ofthe trait. These traits are highlyheritable, as environment has littleinfluence on their expression. Therefore,selection for these traits will be effectiveacross locations and years. In general,the greater the genetic complexity of atrait, the greater is the probability ofobtaining significant G×E.

Measure trait expression and itsassociation with performanceOnce experimental protocols have beenrefined, data must be collected in at leasttwo or three environments (these may bedifferent representative sites and/oryears), and assessed for 1) significantand consistent expression of the trait ofinterest, and 2) association of the traitwith performance among genotypes. Forthe latter, correlation between the traitand performance should be tested usingthe mean values for both, averagedacross replications within environments.(Correlations should not be interpretedfrom individual replication data, sincethese may be highly confounded byenvironmental differences amongreplications; ideally genetic correlationsshould be calculated.)

Any interpretation of data from unrelatedfixed lines is speculative, since theassociation between traits andperformance may be confounded byother genetic factors, such as differencesin phenology and plant type. For thisreason, assuming the above criteria aremet, a second phase of experimentationis needed to demonstrate a definitivegenetic linkage between the trait andperformance in more closely relatedmaterials such as homozygous sisterlines. Genetic gains resulting fromselection and measured as improvedperformance can then be estimated.

Phase 2. EstimatingHeritability of Traits andResponse to SelectionMake experimental crosses withparents contrasting in traitThe initial objective of producingexperimental germplasm is thegeneration of homozygous sister lines,i.e., recombinant inbred lines (RILs),which may be F4 (or later) generationderived so as to be reasonablyhomozygous and homogenous.

Experimental germplasm can, in theory,be derived from any cross. In practicethe parents should be sufficiently welladapted to the target environment topermit field experimentation, and showgenetic diversity for the selection traitunder investigation; preferably, theyshould also meet the additional criterialisted in Table 2. Lines on which initialstudies were made are a good source,since they will have been wellcharacterized in Phase 1. Two or threecrosses would probably be a minimumsince traits of interest may interactwithin different genetic backgrounds.

Develop randomly-derivedhomozygous sister linesTo demonstrate a genetic linkagebetween traits in homozygous lines,ideally there should have been noselection pressure applied during theirdevelopment, thereby ensuring that alllines are randomly derived. One way toachieve this is through using the singleseed descent (SSD) method. However,SSD is a resource-intensive way ofproducing germplasm. The cheapestalternative is to simply advance eachgeneration in bulk without deliberateselection. However, the disadvantage ofbulking is that natural selection and

Steps for Incorporating PhysiologicalCriteria into a Breeding Strategy

Phase 2: Estimating Heritability of Traitsand Response to Selection

• Make experimental crosses with parentscontrasting in trait

• Develop randomly-derived homozygoussister lines

• Test genetic links between quantitativetraits and performance

• Estimate heritability and genetic gains fromselection

• Apply physiological traits to complementbreeding


8

genotypic competitivity (e.g., tilleringability, height) will influence genefrequencies. A reasonable compromise isto grow bulks at low density startingwith F2 seed, such that individual plantscan be recognized. To minimize theeffects of interplant competition, thebulk is maintained by harvesting a singlespike from each plant, rather than thewhole plant.

An alternative to the above is to producedoubled haploid populations. Thisstrategy allows the researcher to producea large number of totally unselected andgenetically fixed lines in a relativelyshort space of time; however, thetechnique is quite costly and resourcedependent (Snape, 1989).

Relevance of “unselected” germplasmto breeding objectives. There are anumber of practical problems associatedwith working on “unselected” materials.The most obvious relates to relevance toa breeding program, in which unsuitablematerials are normally discarded as earlyas possible. Conclusions based onstudies with unselected material maythus not be relevant to a practicalbreeding program.

A second problem relates to potentiallyconfounding factors inherent inunselected material with respect tomeasuring physiological traits. Forexample, expression of manyphysiological traits may be seriouslyconfounded by phenotypic variability inmaturity date or height. Unless thechoice of experimental germplasm hasavoided all potentially confoundingfactors (Tables 2 and 4), segregatinglines are likely to show considerablevariation for such characteristics.

As a compromise between maintainingmaximum genetic diversity on the onehand and, on the other, obtaining resultsthat are relevant to breeding objectivesand still experimentally valid, negative

selection can be practiced onexperimental germplasm to removetotally unsuitable material. While theprecise nature of such characteristics willvary with the target environment, somecommon ones are listed (Table 4),bearing in mind that lines withundesirable characteristics should neverbe removed if they are in any wayrelated or linked to the trait under study.Undesirable traits are the same ones thatwould normally be discarded during thebreeding process.

Test genetic links betweenquantitative traits andperformanceThe relevant generation at which to lookfor a genetic link between a trait andperformance will depend on howcomplex the trait is. Due to the nature ofgenetic segregation, the number ofheterozygous loci in a genotype isapproximately halved after eachgeneration of self-fertilization. Thus,based on probability, any given genotypewill be approximately 50% homozygousat all loci in the F2, 75% in the F3, 87.5%in the F4, etc., such that the probabilityof being genetically stable increases witheach subsequent generation. As aconsequence, traits controlled by fewgenes are more likely to become fixed inearlier generations than more complexones. In addition, the more genes

Table 4. Undesirable agronomic charac-teristics that may confound results inexperimental breeding.

• Extremes of height• Unsuitable phenological development• Strong lodging tendency• Very poor tillering ability• Shriveled kernels• Very low yield potential• Chlorotic and necrotic symptoms• Susceptibility to diseases that are widespread and

difficult to control

involved in the expression of a trait, thegreater the number of different genotypespossible.

Since in practice it is difficult to establishthe number of genes involved for any butthe most simple of traits, we mustassume that a trait is relatively complexor relatively simple based on segregationratios for qualitative (or Mendelian) traitsor, in the case of quantitatively inheritedcharacters, the degree of G×E observedin the trait’s expression. The heritabilityof a trait in early generations will alsoindicate its genetic complexity. Forsimply inherited characters such as thepresence or absence of awns, theheritability of expression is 100%;however, the expression of quantitativelyinherited characters such as grain yieldwill be greatly influence by G×E.

Based on the above considerations, onemight harvest a number of (relatively)randomly derived lines anywhere fromF4 to F8 by taking the seed of individualplants in the requisite generation andmultiplying it for yield testing. Theprocedure is then the same as describedin Phase 1, when looking for theassociation between a trait andperformance in fixed lines. The geneticlink will be validated when tested in anadequate number of environments.

Evaluating physiological traitsindependently of biotic stress. A majorconfounding factor in evaluatingphysiological traits is disease incidence.Generally speaking diseases should becontrolled chemically, if not genetically,so as not to confound conclusions. Thisis reasonable considering the verydifferent nature of the geneticmechanisms involved. For example,when comparing resistance to bioticversus abiotic stresses, it is clear thatgermplasm with improved abiotic stresstolerance will maintain its superiorperformance indefinitely under a definedphysical environment. Resistance to


9

pests and diseases, on the other hand,may break down very rapidly. Hence,within the context of experimentalgermplasm at least, it makes no sense torestrict genetic variability for aphysiological trait based on diseaseresistance criteria that may becomeobsolete in time.

One exception to the above would beparticularly intractable disease problemsthat are of major economic importance inthe target area, where maintenancebreeding is costly. In that case, not onlywould it be necessary for germplasm tobe derived from at least one resistantparent, but susceptible progeny wouldneed to be eliminated systematically.Although it may be decided to controldiseases in experimental work, diseaseexpression should be encouraged inongoing breeding work so thatsusceptible lines can be eliminatedbefore they are evaluated forphysiological criteria.

Estimate heritability andgenetic gains from selectionIf a trait is highly heritable, it is moreefficient from a breeding point of view toselect lines as early as possible in thebreeding process. If shuttle breeding isbeing applied, the environment mostsuitable for expressing a specific traitmay coincide with a particular generationor generations. In either case, it isnecessary to establish the heritability ofthe trait at that generation, so that geneticgains can be evaluated using the formula

R = ih2σp (1)

where i = intensity of selection or theproportion of the population included inthe selected group and σp is thephenotypic standard deviation.

In this case, realized heritability(estimated as σ2

g/σ2p) is perhaps the most

relevant parameter to calculate. This isdone by measuring trait expression in apopulation of lines or bulks from a crossand, using an arbitrary selection intensity,dividing the population into high and lowgroups. These are advanced onegeneration, and then the trait is measuredagain. The smaller the difference betweenthe high and low groups in subsequentgenerations, the lower the heritability.This can be represented by the followingequation:

hr = (Fn+1 high – Fn+1 low)/ (Fn high – Fn low) (2)

Where Fn high is the mean trait value ofthe plants selected for high expression inthe nth generation; Fn low is the meantrait value of plants selected for lowexpression in the nth generation; Fn+1high is the mean value in Fn+1 of thesame high lines from Fn; and Fn+1 low isthe mean value in Fn+1 of the same lowlines in Fn (Falconer, 1990).

Apply physiological traits tocomplement breedingWhen a trait shows a strong associationwith performance in unrelated fixed lines(Phase 1) as well as in homozygous

sister lines (Phase 2), it is probably worthapplying selection pressure for the trait inpreliminary trials (PTs). However, thenature of the association should also beexamined, i.e., the distribution of thevalues of the trait in relation toperformance of lines. For example, whenthe distribution is linear at first butflattens off at higher yields (Figure 2),selection for the trait may only beeffective in eliminating the poorestmaterial.

When a trait shows high heritability andgood association with performance insister lines, it may lend itself to earlygeneration selection (EGS), instead of orin addition to selection in PTs (Table 5).Selection pressure might be applied in F3plants or F3:4 plots etc., depending on thesensitivity of trait expression to plantingmethod. Even for a trait that is relativelyweakly associated with performance, buthighly heritable, early generationselection may be a useful tool foreliminating the poorest material (Table5). Assuming that the data pointconvincingly to the use of a physiologicaltrait in breeding, it must be selected forwithin the overall framework of thebreeding program. Disease resistance,agronomic type, and industrial quality areamong economically important traits thatare essential if a new variety is tosucceed. A theoretical scheme forincorporating physiological traits into aconventional breeding program ispresented below (Table 6).

NDVI0.90

0.80

0.70

0.60

0.50

0.4010 12 14 16 18 20 22 24

Biomass (t ha-1)

Figure 2. Relationship between spectralreflectance index (NDVI, normalizeddifference vegetation index) measuredduring grainfilling and biomass of irrigatedspring wheat advanced lines, Obregon,northwestern Mexico, 1996-97.Source: Reynolds et al. (1999).

y = 3E-09x2 + 0.0001x – 0.4153r2 = 0.56


Table 5. Criteria for applying physiologicaltraits in a breeding program.

Association of trait withTrait performanceheritability Strong Weak

Low Selection in PYTs No application

High Early and/or Negative selectionlate generation in earlyselection generations

10

References

Amani, I., R.A. Fischer, and M.P. Reynolds. 1996.Canopy temperature depression associationwith yield of irrigated spring wheat cultivarsin hot climates. Journal of Agronomy andCrop Science 176:119-129.

Blum, A. 1988. Plant Breeding for StressEnvironments. CRC Press, Inc., Boca Raton,Florida.

Evans, L.T. 1993. Crop Evolution, Adaptation andYield. Cambridge University Press,Cambridge. 500 pp.

Falconer, D.S. 1990. Introduction to QuantitativeGenetics. Third edition. Longman Scientificand Technical, New York. pp. 313-335.

Jackson, P., M. Robertson, M. Cooper, and G.Hammer. 1996. The role of physiologicalunderstanding in plant breeding; from abreeding perspective. Field Crops Research49:11-37.

Loss, S.P., and K.H.M. Siddique. 1994.Morphological and physiological traitsassociated with wheat yield increases inMediterranean environments. Advances inAgronomy 52:229-276.

Rajaram, S., and M. van Ginkel. 1996. Yieldpotential debate: Germplasm vs.methodology, or both. In M.P. Reynolds, S.Rajaram, A. McNab (eds.). Increasing YieldPotential in Wheat: Breaking the Barriers.Mexico, D.F.: CIMMYT.

Reynolds, M.P., S. Rajaram, and K.D. Sayre.1999. Physiological and genetic changes ofirrigated wheat in the post green revolutionperiod and approaches for meeting projectedglobal demand. Crop Science 39:1611-1621.

Snape, J.W. 1989. Double haploid breeding:theoretical basis and practical applications.In A. Mujeeb-Kazi and L.A. Sitch (eds.).Review of Advances in Plant Biotechnology,1985-88: 2nd International Symposium onGenetic Manipulation in Crops. Mexico,D.F., Mexico, and Manila, Philippines:CIMMYT and IRRI.

Table 6. Theoretical scheme for incorporating some physiological selection criteria into aconventional breeding program showing different alternatives for when traits could bemeasured, depending on resources available.

Breeding generation when selection should be conducted

All generations F3 F4-F6 PYTs / Advanced lines

Simple traitsDisease visualHeight visualMaturity visualCanopy type visual †

Complex traitsYield visual yield plotsIndustrial quality grain grainLodging small plots small plots yield plotsCanopy temperature depression small plots yield plotsStomatal conductance plants small plots yield plotsLeaf chlorophyll plants small plots †Spectral reflectance small plots yield plots

† Selection in early/ intermediate generations is probably sufficient, as GxE is low.

In summary, the advantages of EGS ofphysiological traits over later generationselection are 1) resources may be savedby eliminating physiologically inferiormaterial from the program, and 2) thelikelihood of discarding favorablegenetic diversity is decreased. Thepotential disadvantages are 1) withoutclose interdisciplinary collaboration,time may be wasted by measuring traitson agronomically unsuitable material, oreven promoting it, and 2) in earlygenerations large numbers of plants mustbe tested, and some currently availablephysiological tools are either tooexpensive or cannot be applied quicklyenough.

Conclusions

This chapter attempts to provide basicguidelines for evaluating and applyingphysiological selection traits in breedingwork. If adopted, physiological criteriarepresent a refinement of a breedingprogram and in no way replace traditionalmethods of selection. Nonetheless,breeding efforts aimed at meetingongoing challenges, such as breakingyield barriers and improving performanceunder abiotic stresses, are more likely toachieve success if physiologicalunderstanding is used to complementtraditional approaches.


11

C HAPTER 1

1 CSIRO Tropical Agriculture, Davies Laboratory, Queensland, Australia.

Directions for PhysiologicalResearch in Breeding: Issues froma Breeding PerspectiveP.A. Jackson1

Plant breeding has traditionally applied atrial-and-error approach in which largenumbers of crosses are made from manysources of parental germplasm.Progenies are evaluated for characters ofdirect economic interest (e.g., grain yieldand grain quality) in targetenvironments. Good performing parentalgermplasm, crosses, and progenies areselected for further use or testing. Inmany programs “breakthroughs” inimprovement are made simply byfinding superior sources of parentalgermplasm among the numerous sourcestested. This conceptually simpleapproach has been highly successful inmany crop species and numerousbreeding programs.

The approach has often succeeded in theabsence of in-depth knowledge about thephysiological basis for superiorperformance. In some crops suchknowledge has been obtained by doingretrospective analyses of prior geneticgains. Breeders have not applied thisknowledge to a significant extent as aguide to further improvements, butinstead have taken any avenue ofimprovement that happens to arise fromdirect selection for yield and economicperformance.

Given the success of such approaches todate, to what extent can plant breedingprograms benefit from physiologicalresearch? A recent survey of plantbreeders and physiologists (reported inJackson et al., 1996) indicated a generalview among both groups that in thefuture physiological research would havean increasing role in plant breedingprograms. However, the same surveyalso indicated that many respondents feltoutputs from physiological research todate had not developed into practicalimprovements to the extent they hadexpected or thought was desirable.

There has been considerable discussionin the literature about the potential roleof physiological research in plantbreeding. Much of this discussion hasbeen from a physiological perspective—i.e., examining the potential merits ofdifferent plant traits for improving yieldunder different environmentalconditions. The aim of this chapter is totake a breeding perspective of ways inwhich physiological understanding maybe applied in traditional breedingapproaches. The assumption is that plantbreeding programs will continue to relyheavily on large scale evaluation ofparental and progeny populations.Physiological understanding couldenhance and refine this approach.

Applying PhysiologicalUnderstanding inBreeding

Illustrated in Figure 1 are the ways inwhich physiological understanding maybe applied in a breeding program; thisprovides the basis for the subsequentdiscussion. The figure also showspossible ways to enhance this process,all of them based on physiologicalresearch or understanding.

Understanding yield-limiting factorsUnderstanding the biological factorslimiting the performance of genotypesacross target environments is essentialfor improving breeding programsthrough physiological research.Examples of such factors are moisturestress during different phenologicalperiods (Fischer, 1979; Woodruffe andTonks, 1983), soil fertility constraints(Carver and Ownby, 1995), productionof sink capacity (grain number) andsubsequent partitioning of dry matter tograin (Gallagher et al., 1975), canopylight interception during reproductivegrowth (Lawn and Byth, 1974), andpresence of a plant disease. The definingfeature of a limiting factor in this contextis that improving genetic response to thatspecific factor would result in higheryields. In all physiological researchtargeting crop improvement, knowledgeof what these limiting factors are for aparticular crop species x targetenvironmental domain combination is

12

either taken from previous experience orresearch, or established at the outset asan objective of the research.

Knowledge of the important limits tobetter performance may lead to moreeffective approaches to cropimprovement, which will be discussedlater. An example of the application ofsuch knowledge is the use of diseasescreening trials in breeding programs. Insuch cases, the breeder has previouslyrecognized that the disease in question isan important limiting factor in at leastsome target environments. Genotypeswill be deliberately evaluated in thepresence of that disease at some stage ofthe selection process, and an appropriateweighting given to the results.Furthermore, if necessary, sources ofparental germplasm that exhibitfavorable responses to the disease mayalso be sought outside the breedingprogram. Thus breeding is more focused

and effective than when such a limitingfactor is not known.

While a breeder may have someknowledge of the major constraints tohigher yields in his or her targetenvironments, often it may besuperficial or deficient. Physiologicalresearch can help to fill the gap.Different approaches have been used,including focused agronomic trials inwhich factors suspected of limitingperformance are manipulated to verifyand quantify their effect (Nix, 1980). Ata slightly more sophisticated level,simple but very informative conceptualmodels of yield accumulation processes(e.g., Fischer, 1979) may be developed.Finally, crop growth models toquantitatively assess constraints havebeen advocated (e.g., Muchow andCarberry, 1993) and may have particularvalue where highly variable seasonalconditions exist. However, it should be

noted that development of crop growthmodels is expensive and, in many cases,may not be necessary to gain anadequate level of knowledge.

Identification of the environmental orphysiological constraints to higher yielddoes not automatically lead to moreeffective breeding and selectionapproaches. However, it is a startingpoint for developing the approachesdiscussed below.

Choosing environments forselection trialsMost resources in large plant breedingprograms are devoted to conductingselection trials for progeny lines. Theaim is to select superior lines that willbe more precisely evaluated insubsequent trials, and eventually torelease worthwhile lines. Selection trialsare usually conducted in environmentsconsidered representative of targetproduction areas.

Limiting factors are significant enoughto be the object of selection trials if: 1)they are economically important, i.e.,they have a widespread or large impacton yield in the target region, and 2)there is genetic variation for response tothem, i.e., they allow discriminatingamong the genetic materials beingevaluated. Enhanced knowledge of thesefactors will help the breeder choose sitesfor selection trials and decide how tomanage them. The breeder evaluatesgenetic materials so as to be able toobserve genotypic response to thelimiting factors. Based on theseobservations, he can select genotypeswith the desired combination offavorable responses.

If the occurrence or intensity of thelimiting factors varies across the targetregion, this will lead to genotype xenvironment (G×E) interactions inmulti-environment trials. Large G×Einteractions may result in smallerrealized gains from selection if the

Yes

Core breeding program

Selecting andcrossing of parent

genotypes

Production ofprogeny populations

Selection programcomprising:

• Selection directly foreconomic performance

• Selection for traitscorrelated witheconomic performance

Identification ofenvironments for

selection trials

Assess worth oftraits for indirect

selection

Is theregenetic variation

in response to factor incurrent populations?

Improved geneticpopulations

Resolution of factorslimiting performance intargeted environments

Determine candidatetraits for indirect

selection

Conduct introgressionprogram

Is theregenetic variation for

trait in existingpopulations?

No

Yes

No

Figure 1. Key steps in a generalized breeding program (on the left) and the potential role ofphysiological research or understanding.

P.A. JACKSON

13

factors causing the interactions are notadequately sampled at a particular stageof selection. If there is adequatesampling of such factors in selectiontrials, selection should produce goodresults.

A stratified random sampling ofenvironments will usually be conductedacross the target region if limiting factorsare not known or poorly defined. Thestratification may be based on, forexample, geographic location, soil type,management regimes (e.g., irrigationversus rainfed), or other factors thatmight possibly affect the relativeresponse of genotypes. This approachwill be effective to some extent;however, sampling of some factors maybe deficient with such a hit-and-missapproach, and the relative weighting toapply to results from different trialsduring selection will be unknown.

In this situation physiological researchclearly has a role to play by helping toidentify the significant constraints tohigher yields and determine the level ofgenetic variation for response to thoseconstraints in particular populations.Using the approaches outlined in theprevious section, physiological researchon one or a few genotypes may helpidentify economically importantconstraints (determined as previouslydescribed). However, comparingresponses of an adequate sample ofgenotypes representative of those beingevaluated is necessary to determine theextent to which the various factors elicitgenetic variation. This may sometimesbe done in conjunction with selectiontrials already established in the breedingprogram (e.g., Jackson et al., 1994).Improved knowledge of the factorsinvolved in generating variation amongmaterials being tested will nearly alwaysfacilitate more focused and efficientselection strategies.

One way this information may be used inbreeding programs is via the use of

managed environments for selectiontrials. In Queensland, Australia, wheat(Cooper et al., 1995) and barley(Jackson et al., 1994) lines were foundto exhibit large variation in grainnumber and grain yield underfavorable growing conditions(influenced by water and nitrogenavailability) at around anthesis. Theresponsiveness of different lines togood conditions at around anthesisaccounted for a large proportion ofG×E interactions exhibited by breedingpopulations in a sample of productionenvironments. It was thereforesuggested that wheat and barley linesshould be evaluated in one or morehigh input (water and N) environmentsin the early stages of selection(Jackson et al., 1994). This wouldallow effective discrimination amonglines for adaptation across the targetenvironments.

Cooper et al. (1995) tested the value ofusing a small number of high inputenvironments for selection trials inwheat. They showed that mean yieldsof wheat lines across three suchenvironments had a high geneticcorrelation with mean yields across alarger number (16) of randomproduction environments. Theyconcluded that high input managedenvironments could be used at thepreliminary yield evaluation stage tofacilitate efficient and effectivediscrimination among unselected lines.This approach should provide morereliable selection data, at a lower cost,than using a larger number of randomproduction environments for selectiontrials.

Using physiological traits asindirect selection criteriaIdentification of yield limiting factorsmay suggest physiological traits thatbreeders could use as indirect selectioncriteria. Although this topic hasreceived considerable attention in

physiological research and literature,successful application in breedingprograms appears rare. Possible reasonsfor this lack of success and ways toenhance application in breedingprograms are discussed below.

Once identified, physiological traitsaffecting response to important limitingfactors may be used in two ways: 1) asindirect selection criteria in progenypopulations in core breeding programs,and 2) to define objectives forintrogression activities (Figure 1).

Genetic response for yield using anothertrait as an indirect selection criterion maybe predicted from the following formula(Falconer and Mackay, 1996):

CR = i hx . rg . σgy, (1)

where CR is the correlated response foryield, from selection based on characterx; i is the standardized selectiondifferential (related to the proportion ofgenetic population selected); hx is thesquare root of the heritability forcharacter x; rg is the genetic correlationbetween character y and character x; andσgy is the genetic standard deviation forcharacter y.

The genetic correlation is the correlationbetween genetic effects for yield and thetrait used for selection, and may beestimated from analyses of covariance.Many physiological studies reported inthe literature have described onlyphenotypic correlations (i.e., correlationsbased on means), which may be seriouslybiased (either upward or downward) byerror effects or correlated environmentaleffects. Genetic correlations are morerelevant than phenotypic correlations forexamining the value of traits to be usedas selection criteria.

Equation 1 suggests that using a trait asan indirect selection criterion will beeffective only under rare circumstances.First, the heritability (i.e., the ratio of

DIRECTIONS FOR PHYSIOLOGICAL RESEARCH IN BREEDING: ISSUES FROM A BREEDING PERSPECTIVE

14

variation due to genetic variancecompared with error, G×E interaction)of the trait has to be high. Second, thegenetic correlation between the trait andyield (or other character of primaryinterest) also has to be high. A furtherconsideration is the cost of measuringthe trait. When this cost is high(assuming fixed budgets), smallernumbers of genotypes may be screened,reducing the intensity of selection andthe component i in equation 1. Despitethe fundamental importance of theseparameters in determining the value oftraits for use as selection criteria, theyare often ignored in discussions of thesubject in the literature.

For a trait to be relevant to breedingprograms, it is important that itsheritability and genetic correlations beestimated in genetic populationsrepresentative of those being evaluated.This is because such parameters willdiffer among different geneticpopulations (Falconer and Mackay,1996), and misleading or irrelevantinformation could be obtained if atypicalpopulations are used. For example, theuse of highly diverse genotypes wouldproduce estimates of genetic variance,heritability, and, probably, geneticcorrelations that are greater than thoseof more homogeneous geneticpopulations in advanced breeding trials.Similarly, these genetic parameters inhighly selected varieties or lines inadvanced breeding trials would probablybe very different from those in lessselected early generation materials.

It is also important that an adequatesample of the representative geneticpopulation(s) be used in estimatinggenetic parameters. Breeders usuallyconsider a minimum of 30-40 genotypesto be an adequate sample for estimatingvariance components and otherstatistical data. Methods for determiningstandard errors of estimated geneticstatistics have been developed (Falconer

and Mackay, 1996) and are helpful inassessing whether sampling andexperimental design are adequate forprecise estimation.

In practice, there are few cases whereindirect selection alone will be moreeffective than direct selection for yield,particularly if labor-efficient and low-costmethods have been developed forconducting large scale yield trials, as isusually the case. However, sometimes itis possible to identify traits having all thedesired features—high heritability, highgenetic correlation with characters ofeconomic interest, low cost ofmeasurement. Greater gains can be madeby using such traits together with yield asa selection index than by using yieldalone. The use of selection indices hasbeen reviewed elsewhere (Baker, 1986).

Well documented examples of thesuccessful application of indirectselection have been reported by Fischer etal. (1989) and Bolanos and Edmeades(1993a; 1993b). The latter selected maizepopulations using a selection index basedon grain yield across environments wherewater was managed and other traits suchas relative leaf elongation under stress,anthesis to silking interval, and leaf deathscore. Fischer et al. (1989) compared

recurrent selection using a selectionindex to selection for grain yield perse. Gains in grain yield under severemoisture stress conditions were greaterwhen using the selection index thanbased on yield per se (Table 1). Theseauthors suggest that traits other thanyield were more useful as selectioncriteria because they had greaterheritability than yield under severemoisture stress. This may be becausesuch traits were less influenced bycompetition effects in small plots andby soil variability, which is sometimesa problem in trials under severemoisture deficits.

By extension, the identification ofindividual traits for use as selectioncriteria is selection toward an ideotypein which the ideotype predicts what thecharacteristics of a genotype ideal for atarget environmental domain shouldbe. Perhaps the major limitation ofsuch an approach is that it does not, byitself, account for the level of geneticvariation in genetic populations, norfor genetic correlations among traits. Insome cases, there may be little geneticvariation for traits viewed as desirableand, therefore, selection based on suchtraits will be ineffective and wasteful.

Table 1. Performance of maize populations after recurrent selection using different criteria.

Grain yield at soil moisturedeficit level Character†

Population‡ Mild Medium Severe ASI RLE LDS

Original 5240 2780 1520 6.5 75 4.2Yield-mild 6170 2580 1330 6.3 69 4.0Yield-severe 5420 3160 1680 4.7 77 3.8Index 5950 3670 2300 3.7 81 3.1SE 496 399 297 0.6 4 0.3

† ASI: anthesis to silking interval (days); ALE: relative leaf extension (%); LDS: leaf death score.‡ Original: the population before any selection; yield-mild: the population selected based on grain yield under limited

moisture stress; yield-severe: the population selected based on grain yield under severe moisture stress; index: thepopulation selected based on a selection index.

Source: Adapted from Table 3 in Fischer et al. (1989).

P.A. JACKSON

15

Negative genetic correlations may existbetween many physiological traits andother useful characters (e.g., Misken andRasmusson, 1970); this may result in lowor zero genetic correlations witheconomically important characters. Forexample, many yield components arenegatively correlated (Adams, 1967), sothat gains from selecting for onecomponent will inevitably result in adecrease in other important components.Genetic linkage or pleiotropy may causenegative genetic correlations and willeither reduce the rate of progress made inintrogression (in the case of linkage) orreduce the value of the trait beingintrogressed (with pleiotropy).Rasmusson (1991) found pleiotropy andtrait compensation were major factorslimiting progress in an extensive barleybreeding program applying an ideotypeapproach.

Thus selection for traits that could beuseful as selection criteria simply on thebasis of physiological understanding mayresult in small or no gains for charactersof direct interest, such as yield. Further,this approach may not identify traitshaving a positive genetic correlation withyield and in which gains via breedingmay be easiest to achieve. If yield itself(or other characters of economic interest)is used as the key selection criterion,physiological traits influencing yield, andfor which genetic variation exists, willautomatically be changed.

In summary, when considering the use ofphysiological traits as indirect selectioncriteria, expected results should becompared with predicted gains usingyield itself. The search for, andassessment of, traits that may be useful asselection criteria should be based onestimation of their heritability, theirmeasurement cost, and their genetic (notphenotypic) correlation with yield. Usingtraits in association with yield as acombined selection index should also beconsidered.

Using physiologicalunderstanding to defineobjectives of introgressionprogramsThe aim of the main steps in a corebreeding program (Figure 1) is normallythe direct development of new cultivars,and the selection of parents and progenyis usually based on overall estimates ofeconomic performance. However, in amore strategic phase of geneticimprovement, the breeder may selectparent germplasm on the basis of aspecific trait such as disease resistance ortolerance to some abiotic stress. The aimhere is to introgress—i.e., introduce—aspecific trait into locally adaptedbreeding stocks (Simmonds, 1993).Successful introgression of germplasmfrom diverse outside sources has oftenresulted in quantum gains inimprovement. Examples include N.E.Borlaug’s use of dwarf wheat germplasmas parental material to reduce lodging,and the use of Saccharum spontaneum insugarcane to improve ratooning andstress tolerance in noble varieties(Berding and Roach, 1987).

Donor germplasm may be identified fromany source outside the locally adaptedmaterials being selected. For example, itmay include improved materials fromother breeding programs which, thoughnot locally adapted, may have desirablecharacters; materials from germplasmcollections of the same species; ormaterials from related species.Physiological understanding may beuseful or even necessary for choosingdonor material to be used in introgression.Its role here would be to define specifictraits or responses of particular value forwhich little genetic variation exists in corebreeding populations (see chapter bySkovmand et al.).

It should be emphasized thatintrogression is sometimes a difficult,long-term process and often unsuccessful;it therefore requires careful consideration

and planning. The donor germplasm ofthe trait being introduced will nearlyalways be inferior from an overallagronomic point of view. For this reason,several generations of backcrossingtoward locally adapted material andselection between each generation ofcrossing are required to combineadequate expression of the trait in asuitable agronomic background.

In the future, genetic engineeringapproaches will increasingly be used toprovide ‘new’ genes, efficientlyincorporate them into adaptedgermplasm, and control their expression.The aim here is in many ways similar tothat of introgression based on moreconventional breeding approaches.However, the new approaches arepotentially more powerful in that theymake a wider range of genes accessiblefor improving plant traits. This researchwill require an important andcomplementary effort by plantphysiologists and biochemists to defineor suggest specific genetic manipulationsneeded to overcome constraints to betterproductivity or improved quality in targetenvironments.

Conclusions

This chapter outlines three ways in whichknowledge developed from physiologicalresearch may be used to assist plantbreeding programs:

• to improve sampling of environmentsfor selection trials;

• to identify traits that may be used asindirect selection criteria, mostcommonly in an index combined withdirect measurements of economicperformance;

• to assist in determining the objectivesof introgression programs and,increasingly in the future, of geneticengineering approaches.

DIRECTIONS FOR PHYSIOLOGICAL RESEARCH IN BREEDING: ISSUES FROM A BREEDING PERSPECTIVE

16

Physiological research needs to be moreclosely integrated with active breedingprograms for its application to yieldpractical outcomes. If this occurs,physiological research will beconducted on relevant geneticpopulations and, more importantly,outputs and suggestions fromphysiological research can be rapidlyevaluated, compared, and redevelopedin the context of existing breedingapproaches. A plant breedingperspective of the research and potentialopportunities for its application shouldbe determined using simple quantitativegenetics models (see equation 1). Boththe physiologist and the breeder need tocontinually address hard questionsabout how physiological research willimprove existing breeding approaches.This should help maintain a focus onproducing practical breeding outcomes.

References andSuggested ReadingAdams, M.W. 1967. Basis of yield component

compensation in crop plants with specialreference to the field bean, Phaseolusvulgaris. Crop Sci. 7:505-510.

Baker, R.J. 1986. Selection indices in plantbreeding. CRC Press, Boca Raton, FL.

Berding, N., and Roach, B.T. 1987. Germplasm,collection, maintenance and use. In:Sugarcane improvement through breeding.D.J. Heinz (ed.). Elsevier. pp. 143-211.

Bolanos, J., and Edmeades, G.O. 1993a. Eightcycles of selection for drought tolerance inlowland tropical maize. I. Responses ingrain yield, biomass and radiationutilization. Field Crops Res. 31:233-252.

Bolanos, J., and Edmeades, G.O. 1993b. Eightcycles of selection for drought tolerance inlowland tropical maize. II. Responses inreproductive behaviour. Field Crops Res.31:253-268.

Carver, B.F., and Ownby, J.D. 1995. Acid soiltolerance in wheat. Adv. Agron. 54:117-173.

Cooper, M., Woodruffe, D.R., Eisemann, R.L.,Brennan, P.S., and DeLacy, I.H. 1995. Aselection strategy to accommodategenotype-by-environment interaction forgrain yield of wheat: managed environmentsfor selection among genotypes. Theor. Appl.Genet. 90:492-502.

Falconer, D.S., and Mackay, T.F.C. 1996.Introduction to quantitative genetics. FourthEd. Longman.

Fischer, K.S., Edmeades, G.O., and Johnson, E.C.1989. Selection for the improvement ofmaize yield under moisture deficits. FieldCrops Res. 22:227-243.

Fischer, R.A. 1979. Growth and water limitationto dryland wheat yield in Australia: Aphysiological framework. J. Agric. Inst.Agric. Sci. 45:83-94.

Gallagher, J.N., Biscoe, P.V., and Scott, R.K.1975. Barley and its environment. V.Stability of grain weight. J. Appl. Ecol.12:319-336.

Jackson, P.A., Byth, D.E., Fischer, K.S., andJohnston, R.P. 1994. Genotype xenvironment interactions in progeny from abarley cross. II. Variation in grain yield,yield components and dry matter productionamong lines with similar times to anthesis.Field Crops Res. 37:11-23.

Jackson, P.A., Robertson, M., Cooper, M., andHammer, G. 1996. The role of physiologicalunderstanding in plant breeding; from abreeding perspective. Field Crops Res.49:11-37.

Lawn, R.J., and Byth, D.E. 1974. Response ofsoybeans to planting date in S.E.Queensland. II. Vegetative and reproductiveresponses of cultivars. Aust. J. Agric. Res.25:723-737.

Misken, K.E., and Rasmusson, D.C. 1970.Frequency and distribution of stomata inbarley. Crop Sci. 10:575-578.

Muchow, R.C., and Carberry, P.S. 1993.Designing improved plant types forsemiarid tropics: Agronomist’s viewpoints.In: Systems Approaches for AgriculturalDevelopment. F.W.T. Penning de Vries etal. (eds.). Kluwer Academic Publishers,Dordecht, The Netherlands.

Nix, H.A. 1980. Strategies for crop research.Proc. Agron. Soc. NZ. 10:107-110.

Rasmusson, D.C. 1991. A plant breeder’sexperience with ideotype breeding. FieldCrops Res. 26:191-200.

Simmonds, N.W. 1993. Introgression andincorporation strategies for use of cropgenetic resources. Biol. Rev. 68:539-562.

Woodruffe, D.R., and Tonks, J. 1983.Relationship between time of anthesis andgrain yield of wheat genotypes withdiffering developmental patterns. Aust. J.Agric. Res. 34:1-11.

P.A. JACKSON

17

Searching Genetic Resources forPhysiological Traits with Potentialfor Increasing YieldB. Skovmand,1 M.P. Reynolds,1 and I.H. Delacy2

World demand for wheat is growing atapproximately 2% per year (Rosegrant etal., 1995), while genetic gains in yieldpotential of irrigated wheat stand at lessthan 1% (Sayre et al., 1997). Thus globaldemand for wheat is growing at abouttwice the current rate of gain in geneticyield potential, with progress in rainfedenvironments being even lower. Meetingthese demands by continuing to expandagricultural production into remainingnatural ecosystems is environmentallyunacceptable, and the economic costs ofincreasing yields through theintensification of agronomicinfrastructure are high. Hence a cost-effective and environmentally soundmeans of meeting global demand forgrain is through the genetic improvementof wheat.

Increases in wheat yield potential to datehave resulted mostly from manipulationof a few major genes, such as thoseaffecting height reduction (Rht),adaptation to photoperiod (Ppd) andgrowth habit (Vrn). Future gains in yieldpotential, especially under stressedconditions, will almost certainly requireexploitation of the largely untappedsources of genetic diversity housed incollections of wheat landraces and wildrelatives. Though these sources ofgenetic diversity have been exploited toimprove disease resistance in wheat (e.g.Villareal et al., 1995), little use has beenmade of them for physiological

improvement. Nonetheless, many traitsreportedly have potential to enhanceyield, and high expression of these canbe found in germplasm collections. Seedmultiplication nurseries can be used forcharacterizing and evaluating germplasmcollections for non-disease and non-destructive traits (DeLacy et al., 2000).Since seed regeneration activities arecarried out anyway, they can be aneconomic way of collecting data. Recentwork (Hede et al., 1999; DeLacy et al.,2000) has indicated that agronomic traits(including those with low heritability)measured on small, seed-increasehillplots or miniplots can be used forsuch purposes.

Genetic Resources

The genetic resources available to plantphysiologists and breeders are found inseveral Triticeae genepools recognizedby Von Botmer et al. (1992) anddescribed as concentric circles (Figure1). The concept of the genepool wasfirst proposed in 1971 by Harlan anddeWet (Harlan, 1992), who suggested acircular way of demonstrating therelationships among genepools. Theprimary genepool consists of a givenbiological species including, in the caseof a crop species, its cultivated, wild,and weedy forms. Gene transfer withinspecies of the primary genepool is not

Figure 1. Schematic diagram of the concentric circles illustrating genepools of the Triticeae.Source: Adapted from Botmer et al. (1992).

1 CIMMYT, Apartado Postal 6-641, Mexico, D.F., 06600.2 School of Land and Food Science, The University of Queensland, Brisbane, Australia.

C HAPTER 2

RYE

WHEAT

Aegilops

FORAGEGRASSES

18

difficult. Table 1 lists the diploid,tetraploid, and hexaploid species incultivated wheat’s primary genepool,listing their common names andindicating the genomes of each. Thesecondary genepool comprises thecenospecies, i.e., a group of relatedspecies with which gene transfer ispossible but difficult. Most species inwheat’s secondary genepool, along withtheir synonyms and genomicconstitution, are given in Table 2.Wheat’s tertiary genepool is composedof related genera of annual and perennial

grasses from which gene transfer canonly be achieved through the use ofspecial techniques. The genera andspecies in the tertiary genepool,described by Dewey in 1984, are toonumerous to be listed here.

Genetic resources of cultivated plantspecies were categorized by Frankel(1977) and the Food and AgricultureOrganization of the United Nations(FAO) Commission on Plant GeneticResources (FAO, 1983); however, thiscategorization is not followed by all

centers involved in genetic resourceconservation and utilization. Thecategories are:

• modern cultivars in current use• obsolete cultivars, often the elite

cultivars of the past, many of whichare found in the pedigrees of moderncultivars

• landraces• wild relatives of crop species• genetic and cytogenetic stocks• breeding lines

Recently the International Plant GeneticResources Institute (IPGRI) and FAOjointly developed a list of multi-croppassport descriptors to provide codingschemes consistent across crops. Thesedescriptors should be compatible withfuture IPGRI crop descriptors and withthose used by the FAO WorldInformation and Early Warning System(WIEWS) on Plant Genetic Resources(PGR) (Hazekamp et al., 1997). Thedescriptors are: 1) unknown; 2) wild; 3)weedy; 4) traditional cultivar/landrace;5) breeding/research material; 6)advanced cultivar; and 7) other. Becausethey are more generic (so as to fit multi-crop classification), these descriptors arealso less useful in single-cropclassification systems.

The classification system used inCIMMYT’s wheat collection is based onthe categories outlined by Frankel andthe FAO Commission on PGR(Skovmand et al., 1992). Recently,however, a list including 21 categorieswas defined in the GRIP project(Skovmand et al., 2000) to describe thebiological status of materials inCIMMYT’s collection and other geneticresources. When such specific categoriesare applied to collections, utilizationefficiency is enhanced, making it easierfor users to know exactly what they areworking with.

Table 1. Summary of taxa in the primary and secondary genepools of cultivated wheat.

I. Sect. Monococcon Dumont; Ploidy level: diploid; genome type (female x male parent): AA (‘A’)

1. Triticum monococcum L.a. ssp. monococcum; cultivated form; einkorn or small spelt wheatb. ssp. aegilopoides (Link) Thell.; wild form; synonym: T. boeoticum

2. Triticum urartu Tumanian ex Gandilyan; wild form

II. Sect. Dicoccoidea Flaksb.; Ploidy level: tetraploid; genome type (female [B] x male [A] parent): BBAA (‘BA’)

3. Triticum turgidum L.; cultivated and wild formsa. ssp. turgidu; rivet, cone, or pollard wheatb. ssp. carthlicum (Nevski) A. Loeve & D. Loeve; Persian wheat, Persian black wheatc. ssp. dicoccon (Schrank) Thell.; emmer wheatd. ssp. durum (Desf.) Husn.; macaroni wheat, hard wheat, or durum wheate. ssp. paleocolchicum (Menabde) A. Loeve & D. Loevef. ssp. polonicum (L.) Thell.; Polish wheatg. ssp. turanicum (Jakubz.) A. Loeve & D. Loeve; Khorassan wheath. ssp. dicoccoides (Koern. ex Asch. & Graebn.) Thell.; wild emmer wheat

4. Triticum timopheevii (Zhuk.) Zhuk.a. ssp. timopheevii; cultivated and wild formsb. ssp. armeniacum (Jakubz.) van Slageren; wild form

III. Sect. Triticum; Ploidy level: hexaploid; genome type (female [BA] x male [D] parent): BBAADD (‘BAD’)

5. Triticum aestivum L.a. ssp. aestivum; bread wheatb. ssp. compactum (Host) Mackey; club, dwarf, cluster, or hedgehog wheatc. ssp. macha (Dekapr. & Menabde) Mackeyd. ssp. spelta (l.) Thell.; large spelt, spelt or dinkel wheate. ssp. sphaerococcum (Percival) Mackey; Indian dwarf wheat or shot wheat

6. Triticum zhukovskyi Menabde & Ericz.; Ploidy level: hexaploid; genome type (female [GA] x male [A]parent: GGAAAA (‘GAA’)

Source: van Slageren (1994).

B. SKOVMAND, M.P. REYNOLDS, AND I.H. DELACY

19

The concentric circles proposed byHarlan and deWet (Harlan 1992) todescribe the different genepools havebeen very useful, and the concept hasprovided a rational basis for comparingtaxonomies. However, it gives theimpression that separations among thepools are clear-cut, with distinctdivisions between one pool and another,though Harlan (1992) did state that the

lines of demarcation may be fuzzy.Furthermore, the circles do not reflectthe relative difficulty of utilizing thegenepools, nor the cost of utilizinggenetic resources within a genepool orwithin a species.

A schematic diagram of the effort neededto transfer traits from genetic resourcesto farmers’ fields is given in Figure 2.Within the primary genepool, the

utilization cost increases as the geneticdistance increases. Within a speciesthere are also levels of geneticresources (from current high yieldingcultivars to landraces) that maydetermine the cost of using thoseresources.

As one moves away from the primarygenepool, the effort required to utilizegenetic resources in the secondary andtertiary genepools increasesgeometrically. It is difficult to release acommercially acceptable cultivar thatdoes not have previously releasedcultivars in its pedigree (Rajaram, pers.comm.) because crosses with species inthe secondary and tertiary genepoolstend to disunite favorable genecomplexes, which affects performance.Technology extends the genepools andreduces costs, as, for example, embryorescue has done in the recent past andgenetic engineering promises to do inthe future. Also, species in thesecondary genepool, such as Aegilopstauschii, can now be used as readily asspecies in the primary genepoolthrough the production of hexaploidsynthetic wheats using embryo rescuefollowed by chromosome doublingusing colchicine (Mujeeb-Kazi, 1995).

Global wheat geneticresources and theiravailabilityAbout 640,000 accessions of Triticumspp., Aegilops spp., and X Triticosecale(a man-made crop, a cross betweenwheat and rye) can be found incollections around the world (Table 3).The degree of duplication in thesecollections is difficult to ascertainwithout some type of global wheatgenetic resources database. Given thissituation, the level of priority thatshould be placed on collecting morematerials is uncertain, except where

Table 2. Species of Aegilops, their genomic formula and synonyms (when available) whenAegilops and Amblyopyrum are placed within Triticum emend.

Species of Aegilops Genome Species of Triticum

1 Aegilops bicornis (Forssk.) Jaub. & Spach Sb Triticum bicorne Forssk.2 Aegilops biunciales Vis. UM Triticum macrochaetum (Shuttlew. & A. Huet

ex Duval-Jouve) K. Richt.3 Aegilops caudata L. C Triticum dichasians Bowden4 Aegilops columnaris Zhuk. UM Triticum – none5 Aegilops comosa Sm. in Sibth. & Sm. M Triticum comosum (Sm. in Sibth. & Sm.)

K. Richt.6 Aegilops crassa Boiss. DM

DDM Triticum crassum (Boiss.) Aitch. & Hemsl.7 Aegilops cylindrica Host DC Triticum cylindricum (Host) Ces.,

Pass. & Gibelli8 Aegilops geniculata Roth (Ae. ovata) MU Triticum – none9 Aegilops juvenalis (Thell.) Eig DMU Triticum juvenale Thell.10 Aegilops kotschyi Boiss. SU Triticum kotschyi (Boiss.) Bowden11 Aegilops longissima Schweinf. & Muschl. Sl Triticum longissimum (Schweinf. & Muschl.)

Bowden12 Aegilops neglecta Req. ex Bertol. UM

UMN Triticum neclectum (Req. ex Bertol.) GreuterTriticum recta (Zhuk.) Chennav.

13 Aegilops peregrina (Hack. in J. Fraser) SU Triticum peregrinum Hack. in J. FraserMaire & Weiller

14 Aegilops searsii Feldman & Kislev ex Hammer Ss Triticum – none15 Aegilops sharonensis Eig Sl Triticum longissimum (Schweinf. & Muschl.)

Bowden spp. Sharonense (Eig) Chennav.16 Aegilops speltoides Tausch S Triticum speltoides (Tausch) Gren. ex K.Richt.17 Aegilops tauschii Coss. D Triticum aegilops P.Beauv. ex Roem. Ex Schult.18 Aegilops triuncialis L. UC

CU Triticum triunciale (L.) Rasp. (var. triunciale)(T. triunciale spp. Persicum)

19 Aegilops umbellulata Zhuk. U Triticum umbellulatum (Zhuk.) Bowden20 Aegilops uniaristata Vis. N Triticum uniaristatum (Vis.) K.Richt.21 Aegilops vavilovii (Zhuk.) Chennav. DMS Triticum syriacum Bowden22 Aegilops ventricosa Tausch DN Triticum ventricosum (Tausch) Ces.

Pass. & Gibelli

Species of Amblyopyrum1 Amblyopyrum muticum (Boiss.) Eig T Triticum tripsacoides (Jaub. & Spach) Bowden

Source: van Slageren (1994).

SEARCHING GENETIC RESOURCES FOR PHYSIOLOGICAL TRAITS WITH POTENTIAL FOR INCREASING YIELD

20

evaluation data. In the early 1990s, theCIMMYT Wheat Program establishedjust such a strategy for integrating andmanaging all data pertaining togermplasm regardless of where theywere generated (Skovmand et al.,1998). The goal was to facilitate theunambiguous identification of wheatgenetic resources and remove barriersto handling and accessing information.The resulting database, theInternational Wheat InformationSystem (IWIS), seamlessly joins theconservation, utilization, and exchangeof genetic materials. The system is fast,user-friendly, and available on anannually updated CD-ROM (Skovmandet al., 2000a).

IWIS has two major components: theWheat Pedigree Management System,which assigns and maintains uniquewheat identifiers and genealogies, andthe Wheat Data Management System,which manages performanceinformation and data on known genes.Another information tool, the GeneticResource Information Package (GRIP),was developed using IWIS for datawarehousing. One of the functions ofGRIP attempts to collate passportinformation across genebanks toidentify duplications and uniquegenetic resources (Table 4) (Skovmandet al., 2000b).

The Genetic Resources InformationNetwork (GRIN) and the System-WideInformation Network for GeneticResources (SINGER) are other publiclyavailable databases on wheat andTriticeae genetic resources. GRINcontains information about the USDASmall Grains Collection stored inAberdeen, Idaho, and can be reachedon the Internet (http://www.ars-grin.gov/) . SINGER (http://singer.cgiar.org/), developed and madeavailable under the leadership ofSystem Wide Genetic Resources

n

n-1

1

2

a

31 . . . . . . . . . . n-2 n-1 n

n-2 n-1 n31 2 . . . . . . . . . . . . . .

1 2 n-1 n. . .

n

n-1

n-2

4

3

2

1

Geneticdistance

Geneticstock

Breedingline

Advancedcultivar

Obsoletecultivar

Landrace

.

Genepoolspecies†

Gp-1Breadwheat

Clubwheat

DurumwheatSpelt Emmer Einkorn Other

Triticum spp.

GP-1Triticum spp.

GP-1-2X-Triticosecale

Gp-3Related genera ofannual and perennialTriticeae:D,M,T,U,R,C,N,S,PS,N,H,J,J ,W,X,Y

n

n-1

n-2

4o

o

o

3 o

2 o

1o

o

o

E

o

o

Genetic

Genome A , B, D aA , B, D

aA , B, D,Ra aA , A , B, D,G

aA , B, DaA , B, D aA , B aA , B aA

distance

GP-2Secale spp.

GP-2-3Aegilops spp.

D,C,M,N,S,UR

Geneticdistance

Genepoolspecies†

Deg

ree

of d

iffic

ulty

(nom

inal

sca

le) o

f util

izin

g a

gene

tic r

esou

rce

Figure 2. Schematic diagram of the effort required to transfer adaptive traits from genepoolsof wheat to farmers’ fields.† Not in strict phylogenetic order.

Conservation can be either in situ or exsitu, but most wheat genetic resourceshave been conserved ex situ. Only in thelast few years has in-situ conservationbeen seriously considered; the WorldBank recently supported such anundertaking in Turkey. The exception isthe natural habitats in Eastern Galilee,Israel, where the Ammiad wild wheatstudy was undertaken in the 1980s.Shands (1991) and Hawkes (1991)summarized a symposium where thefindings in this in-situ field laboratorywere discussed. Ex-situ conservation ofTriticeae genetic resources is easy andcost effective (Pardey et al., 1998), sincethey are adapted to long-term storageconditions.

The key to accessing wheat geneticresources is the development of adatabase, or interconnected databasesystems, with the capacity to manageand integrate all wheat information,including passport, characterization, and

there is a real threat of genetic erosion tonative species. Accessions in collectionsaround the world may or may not beproperly preserved, and some may noteven be catalogued. It may thus be morecost effective to place such collections insecure storage than to collect morematerials in the field. Most major wheat-producing countries have ex-situcollections, and genetic resources can beobtained from these collections bywriting to the curator.


Table 3. Number of accessions available incollections around the world.

Type of wheat Number of accessions

Hexaploid 266,589Tetraploid 78,726Diploid 11,314Unspecified Triticum 252,530Aegilops spp. 17,748Triticale 23,659Total 640,603

Source: Information collated from IBPGR (1990).

21

Program (SGRP), gives access to allFAO/CGIAR Center accessions,including wheat and other cereals.

In the 1980s there was an increasingtrend towards greater application ofintellectual property protection (IPP),which contrasted with the pervadingattitude during the 1960s and 1970s,where IPP within the context ofinternational plant improvement wasseen as an obstacle to progress. Sincethen, the view that strong IPP can helpmaintain technological leadership hasgained respectability, especially in theUnited States (Siebeck, 1994). Severalinternational initiatives have resulted,such as the strengthening of the UPOVConvention in 1991, which narrowed thebreeder’s privilege to use protectedcultivars as breeding parents. However,according to Siebeck (1994), the mostsignificant initiative was instigated as

part of the Multilateral TradeNegotiating Round in the GeneralAgreement on Tariffs and Trade thatended in 1993. At the insistence ofindustrialized nations, strengthening ofIPP was included as a key negotiatingpoint. Efforts in UPOV and GATT towiden IPP on inventions and breedingtechnologies were paralleled by effortsto regulate international access togenetic resources.

The “International Undertaking on PlantGenetic Resources,” established by FAOin 1983, was an attempt to stop geneticerosion and protect plant geneticresources. The Undertaking initiallysubscribed to the rule of free germplasmexchange and recognized plant geneticresources as the “heritage of mankind.”However, disagreements later arose overthe issue of genetic resourcesownership. The idea of compensationwas introduced in 1989 and modified in1991, when FAO adopted the commonheritage principle but subordinated it to“the sovereignty of states.”

Unlike the FAO Undertaking, whichwas voluntary, the Convention onBiological Diversity (CBD) of 1992 isan internationally ratified treaty amongnations. The CBD officially recognizessovereign control by individual nationsover biological diversity and resourceswithin their territories. The conventionexcludes material collected before 29December 1993, when the CBD tookeffect, but any germplasm collectedafter that date in a country that hassigned the CBD comes under theprovisions of the Convention. A resultof the discussions on the ownership ofgenetic resources was the signing of anagreement between the CGIAR andFAO that places the germplasmcollections held in trust by the CGIARsystem under the auspices of FAO.

Consequently, plant genetic resourcesmay not be freely available to everyonein the future but likely to be madeavailable under some type ofintellectual property rights (IPR)agreement. For example, the accessionsin the CIMMYT collection that comeunder the FAO/CIMMYT in-trustagreement are shared under a MaterialsTransfer Agreement that states theaccessions can be utilized but notprotected by IPR. However, productsderived from research and breedingwith such materials can be protected,since they are deemed to be different andbelong to the scientist or breeder whodeveloped them.

The Search for NewGenetic Variation

A classic method of identifying newgenetic variation is the recognition ofpotentially useful traits by experiencedscientists and research staff in thecourse of routine maintenance activities,special studies, or as an offshoot ofprebreeding and breeding exercises.This should not be underestimated,given that much of the useful novelvariation deployed in cultivated cropstoday was recognized in this way.

Augmented use of seedmultiplication nurseriesSeed multiplication nurseries can beused to characterize and evaluategermplasm collections for non-diseaseand non-destructive traits. Since routineseed regeneration activities have to becarried out anyway, they can be aninexpensive means of collecting data.Recent work has indicated thattraditional agronomic traits (includingthose with low heritability) measured onsmall, seed-increase hillplots can beused for such purposes (Hede et al.,1999; DeLacy et al., 2000). Curators of

Table 4. Biological status classification usedin GRIP II.

GRIP code Status

BL Breeding LineCV CultivarLV LandraceX No dataAL Addition LineBL Apomixis LineBL Breeding PopulationBL CrossBL Genetic PopulationGS Genetic StockML Multi LineMTL Mutation LineNIL Near Isogenic LineRCMS CMS RestorerRF Fertility Restorer, non specific

SL Substitution LineTL Translocation LineCMS Cytoplasmic Male SterileGMS Genetic Male SterileRG Genetic RestorerMS Male Sterile, non specific

Source: Skovmand et al. (2000b).


22

germplasm banks have traditionallyavoided these traits, which are useful forplant improvement programs.Descriptions of germplasm based on“useful” attributes are immediatelyadvantageous to practical plantimprovement programs because theyindicate where useful variation may befound in the collection.

A description based on useful attributesalso allows more directed searchstrategies than those derived fromtraditional characterization attributes orrandom DNA markers with highheritabilities. Provided that randommarkers adequately cover the genome,they give information on the amount ofvariation at and between sites, thusindicating whether adequate collectionhas been done. However, until adequatelinkages to known functions have beenestablished, they, like traditionalcharacterization attributes, provide littleinformation on the type of variationpresent.

When low(er) heritability data are used,means and variances change in differentseasons, years, and places. This haslimited their use for germplasmdescription, but many, if not most,attributes useful for plant improvementprograms are of this type. Much of thedifficulty encountered in integratingsuch information from sets of dataacquired at different times can beavoided by appropriate data analysis.After standardization by the range orstandard deviation within sets, meansand variances for each attribute arethe same.

As an example, DeLacy et al. (2000)reported on an analysis of a seedmultiplication nursery made up of 465accessions of bread wheat landracescollected in 1992 from 24 sites in threestates of Mexico. They were examinedin unreplicated hillplots in ascreenhouse for 15 morphological,

agronomic, and grain quality attributesas part of the routine regenerationprocess conducted by the CIMMYTWheat Genetic Resources Program. Apattern analysis (combined use ofclassification and ordination methods) ofthe data provided a good description ofthe accessions and collection sites(Figure 3). Since economically usefulattributes were used, the analysisprovided relevant information for bothgermplasm curators and potential users,who now have a description of theaccessions from which to chooserelevant breeding material.

The data were analyzed using rangestandardized squared Euclidean distance(rsSED) as the dissimilarity measure.These SEDs are calculated amongattributes that are range standardized(Williams, 1976) and employed toensure each attribute contributes equallyto the analysis. Ordination wasperformed by singular valuedecomposition (Eckart and Young, 1936)of the Gower complement similaritymeasure to the rsSED (Gower, 1967;DeLacy et al., 1996). The relationshipsbetween accessions and attributes fromthe ordination were displayed on a biplot(Gabriel, 1971).

Both the accession and attribute plottingpoints can be interpreted as vectors onthe biplot, but since the accessions wereinvestigated in terms of the attributes,attributes were represented as vectorsand accessions as points. As the datawere centered (i.e., the attribute meanwas subtracted from all values for thatattribute so the grand mean becomeszero), the origin on the biplot representsaverage values for all attributes. Thepercentages on each axis represent theproportion of total variation, measuredby the total sum of squares (TotSS),accounted for by each vector. In this casethe aim is to represent the original 15dimensional space defined by the 15attributes in a low dimensional space.

Since not all the variation is modeled,some distortion of the relationshipsamong attributes and accessions willoccur when depicted on the biplot.

The attribute vectors are drawn in apositive direction, i.e., in the direction ofincreasing value for that attribute. Thelength of each vector is proportional tohow well each attribute was modeled,since each vector should be the same

Vector 2 (13.7%)0.2

0.1

0.0

-0.1

-0.2-0.2 -0.1 0.0 0.1 0.2

Vector 1 (25.4%)

Vector 2 (13.7%)0.2

0.1

0.0

-0.1

-0.2-0.2 -0.1 0.0 0.1 0.2

Vector 1 (25.4%)

Figure 3. Attribute vectors and accessionplotting points for the biplot for vectors 1and 2 from the ordination based on 15morphological, agronomic, and grain qualityattributes of 465 individual spike accessionsof wheat landraces collected from 24 sitescovering four states in Mexico.

Attribute vectors

Accessions and attribute vectors

Hard

SDS

HIGrWt / S

GrNo / SNoSpikes

Pr%Flw Mat


23

length if they were all equally wellmodeled. The angles of the vectors toeach other in the biplot represent thephenotypic correlation between theattributes over all values of theaccessions for each attribute. An angle ofzero indicates a correlation of +1, anangle of 90o a correlation of 0, and anangle of 180o represents a correlation of-1. The length on the attribute vector tothe point where a perpendicular droppedfrom the genotype plotting point to thevector is proportional to the modeled(predicted) value of that genotype forthat attribute.

Grain hardness was well modeled, eightattributes (Flw, Mat, Pr%, SDS, GrWt/S,HI, GrSize, GrY; see Table 5) weremodeled reasonably well, and six(SpikeS, GlumeS, FlagS, Ht, NoS,GrNo/S) were poorly modeled byvectors 1 and 2 (Figure 3). As the twodimensional representation of the 15attribute space accounts for 39% of theoriginal variation, some distortion willoccur. Grain yield, its components (No/S, GrNo/S, GrWt/S, GrSize), and harvestindex are positively correlated andhighly negatively correlated withmaturity (Flw, Mat), protein content ofthe grain, size of the flag leaf, and plantheight as the two groups of attributeshave vectors at close to 1800 to eachother. In contrast, the two qualityattributes (Hard, SDS) and glume andspike size are positively correlated witheach other (parallel vectors) butindependent of the other two groupingsof attributes (vectors at 90o).

Accessions plotted to the right in Figure3 have high yield and high values foryield components, but are early maturingand have low protein percentage in thegrain. Clearly those to the left are late,have higher protein percent, but are lowyielding. Vector 2 separates those withhard grain and high SDS, at the bottom,from soft wheats with low SDS (Figure 3)

at the top. Vector 3 separates those withhigh grain size, grain weight per spike,SDS, and large glume and flag leaf size(bottom) from those with low values forthese attributes and which are hard andhave a high number of spikes (top ofgraph). Hence, accessions occupyingdifferent positions in the biplots havedifferent character combinations that canbe read directly from the plots, alwaysremembering that some distortion mustoccur, as not all variation is representedin the low dimensional space.Accessions in different positions on thebiplot have different attributecombinations in the “description space.”

Augmented use of disease or“special attribute” nurseriesUseful, low-heritability traits can bemeasured in trials that are routinelyplanted by the genetic resources programfor other purposes, as well as in seedmultiplication nurseries with the samevalue-added characteristic. Thus evennurseries planted for evaluating accessionsfor disease resistance can be utilized formeasuring other traits, the exception beingtrials where disease is so severe that plantsare heavily damaged or dead.

Molecular approaches foridentifying useful geneticdiversityGenetic diversity from wheat’s wildrelatives has already been exploitedthrough wide-crossing to improvedisease resistance (e.g., Villareal et al.,1995). Useful characteristics also exist inprimitive or landrace varieties, of whichthere are over 66,000 in the CIMMYTgenebank. It would be extremely time-consuming to evaluate all theselandraces, wide crosses, and wildrelatives for all useful yield traits, suchas those described above, in field trials(Figure 2). Potential exists foridentifying the loci encodingquantitatively-inherited yield traits usingQTL analysis in mapping of delayedbackcross generations (Tanksley andNelson, 1996). When molecular markerslinked to traits of interest are identified,they could be used to screenuncharacterized germplasm collectionsfor the same marker and linked alleles.These lines could then be evaluated incontrolled experiments to observe howwell the molecular marker is linked tophenotypic expression of useful traits.Where there are reasonable associations,


Table 5. Fifteen morphological, agronomic, and grain quality attributes measured on 465individual spike accessions of wheat landraces collected from four states in Mexico.

Name of attribute Abbreviation Description

Flag leaf size FlagS Flag leaf length (cm)Spike size SpikeS Spike length (cm)Glume size GlumeS Glume length (cm)Days to maturity Mat Days from sowingDays to anthesis Flw Days from sowingHeight of plant Ht Height to tip of glume (cm)Number of spikelets No/S Number of spikelets per spikeGrain number per spike GrNo/S Number of grains per spikeGrain weight per spike GrWt/S GramsGrain size GrSize 1000 kernel weight (g)Grain weight per plot GrY GramsHarvest index HI Grain weight as a proportion of total biomassGrain hardness Hard Percent hardness (NIR analysis, calibrated with particle size

index using 0.5 mm sieve in grinder)Grain protein percentage Pr% Percent protein (NIR analysis, calibrated against Kjeldahl N x 5.7)SDS sedimentation SDS Sodium dodecyl sulfate (SDS) sedimentation volume (ml/ lg flour)

24

markers could be used to screenuntapped genetic stocks, enabling newsources of genes with potentially usefulalleles to be exploited in breeding.

How to use theidentified traitsGenetic resources with desirable traitsusually need to be tested and improvedto be of use in wheat improvement(Figure 2). Most often these resourceshave many undesirable characteristics,such as extreme disease susceptibility,

low yield, and highly specificenvironmental adaptation, in addition tothe needed trait.

These resources therefore need to undergoprebreeding before they can be used inimprovement work. Figure 4 demonstratestwo prebreeding schemes, each with adifferent purpose: the open-parent,cyclical crossing program and abackcrossing program aimed at producingisogenic lines. These two program havedifferent purposes and different end

results; moreover, the first isprogressive, while the second isunprogressive in terms of yield potential.

The open-parent, cyclical crossingprogram described by Rasmusson (2001)is utilized when the introgressing a traitknown to be of value. Rasmusson wasstriving to introgress characters fromtwo-row barley into six-row barley andfound that the initial cross yieldedgermplasm with no putative candidatesfor cultivar release, with the best linesyielding about 20% less than theimproved parent. The second cycle ofthe program, where the improved parentwas the best current cultivar, producedprogenies that yielded about 98% of thebest parent’s yield. The third cycle, againusing the best current cultivar as aparent, yielded 112-119% of the checks.Using this scheme, germplasm with thedesired trait is produced that could becompetitive in a cultivar-releaseprogram.

A backcrossing program to generateisogenic lines is applied when theidentified trait has as yet no provenvalue. The recurrent parent is crossedrepeatedly to the genetic resource withthe desired trait. In each backcrossgeneration, selection is done for the tailsof the populations, i.e., lines with thetrait and lines without the trait. Linesthat differ genetically only for the trait inquestion are the end result of thisprogram. Additional trials can beconducted to assess the value of the trait,but bearing in mind that the germplasmproduced will not outperform therecurrent parent.

Future utilization of geneticresourcesAs evidenced by the above, geneticresources have played a significant rolein wheat improvement and will continueto do so, by providing breeders with thegenetic variation they require to effect

Open Parent BackcrossingCyclical Crossing to recurrent

Program† parent

GR x Parent 1 (Best cultivar at the time) GR x Rec Parent

Selection cycle 1 F1 x Rec Parent

Selection cycle n Select plant of BC1F1 x Rec Parent

Yield test selection ofAdvanced Line 1 (AL1)

AL 1 x Parent 2 (Best cultivar at the time) Select plant of BC2F1 x Rec parent

Selection cycle 1 Select plant of BC3F1 x Rec parent

Selection cycle n Select plant of BC4F1 x Rec Parent

Yield test selection ofAdvanced Line 2 (AL)

AL 2 x Parent 3 (Best cultivar at the time) Select plant of BC5F1 x Rec Parent

Selection cycle 1 Selection cycle 1

Selection cycle n Selection cycle n

Advanced line Isogenic lineswith trait (or gene) in differing onlyimproved background in the desired trait

Figure 4. Utilization of genetic resources: prebreeding schemes.† Source: Rasmusson (2001).


25

future improvements. Variation will beneeded 1) to further increase wheat’s yieldpotential; 2) to provide new sources ofdisease and pest resistance and maintainthe yield levels achieved so far; 3) todevelop germplasm adapted to moremarginal environments; and 4) toimprove quality. To date the maincontribution of genetic resources has beenas new sources of disease and pestresistance, thanks to which achieved yieldlevels have been maintained.

There are few examples of geneticresources contributing to the three otherobjectives. One example is the dwarfinggenes, especially Rht1 and Rht2, thatbecame available through the Japanesewheat Norin 10, which in turn inheritedthem from Shiro Daruma, a Japaneselandrace (Kihara, 1983). Persistent effortswere required to transfer these dwarfinggenes into a genotype of value (Borlaug,1988; Krull and Borlaug, 1970), whichillustrates the difficulty of using genesfrom unadapted materials. It also showsthat desirable characteristics other thanapparent ones may result from suchgermplasm, as evidenced by the fact thatwhile incorporating strong straw to avoidlodging, Krull and Borlaug (1970)obtained better fertility and tilleringcapacity. It is now obvious that dwarfinggenes Rht1 and Rht2 have a direct effecton yield over and above the benefitsderived from diminished lodging (Galeand Youssefian, 1986).

A survey conducted by Cox (1991)revealed that most introductions to theUnited States were used to improvedisease and pest resistance (Table 6). Theonly yield-related traits listed in the tableare reduced height, stiff straw, large seed,and yield per se. No instances ofimproving yield in marginalenvironments are listed and only twowhere quality was improved: higherprotein and gluten strength.

In another report (Fischer, 1996), traitsinvolved in improving yield that wereintroduced from genetic resources aredescribed. Erect leaf habit wasintroduced into CIMMYT germplasmfrom Triticum sphaerococcum, and anumber of lines were developedthrough prebreeding. This germplasmwas used in both the bread and durumwheat programs and let to the release ofone bread wheat (Bacanora 88) and twodurum wheat cultivars (Altar 84 andAconchi 89).

Physiological traits are often identifiedas having contributed to improvingyield potential, but usually inretrospective, after the germplasm hasbeen developed. We need to be moreproactive and identify potentially usefultraits and then introduce the trait intothe improvement program.

Traits to raise yield potentialof irrigated wheatTo boost yield in irrigated situations, itis widely believed that genetic improve-ment must come about throughsimultaneously increasing bothphotosynthetic assimilation capacity

and partitioning of assimilates topromote high grain number and growthrate (Richards, 1996). However, anotherway to increase grain number could be toincrease the intrinsic fertility of the spike.

Multi-ovary florets is a trait beingstudied in CIMMYT’s wheat germplasmbank. Spikes with this trait may have upto six kernels per flower (Chen et al.,1998), but individual kernel weights tendto be low. The trait is currently beingintrogressed into high yielding lines withgood agronomic traits. Data for the F1shows that the trait is expressed better insome backgrounds than others. However,average kernel weight of the F1s was inall cases higher than that of the multi-ovary donor and, in many cases, of bothparents. Total grain weight per spike wasgenerally higher than that of the parents(Table 7).

High leaf chlorophyll content has beenidentified in landrace collections: thebest genotypes showed substantiallygreater leaf chlorophyll concentrationthan the check Seri-M82. While the traitdoes not guarantee higher leafphotosynthetic rate in all backgrounds, it

Table 6. Contributions to germplasm improvement of introduced genetic resources.

Yield Marginalpotential Cases Resistance Cases environments Cases Quality Cases

Reduced height 15 Strawbreaker 2 None High protein 2Yield 6 Powdery mildew 9 Gluten strength 1Large seed 1 Stripe rust 4Stiff straw 1 Leaf rust 12

Stem rust 12Septoria leaf blotch 3Bunt 3Soilborne mosaic virus 1Cereal leaf beetle 1Hessian fly 3Snow mold 1Greenbug 1Wheat curl mite 1

Source: Adapted from Cox (1991).


26

has been shown to be associated withincreased leaf photosynthetic rate andhigher yield in improved durum wheatcultivars grown under irrigated conditions(Pfeiffer, pers. comm.). These twofindings suggest that combining higherchlorophyll content with greater spikefertility (for example, due to multi-ovaryflorets), which creates higher demand forphotosynthesis, may help increase yieldpotential under irrigated conditions.

Traits to raise yield understress conditionsWheat yields are reduced by 50-90% oftheir irrigated potential by drought on atleast 60 million ha in the developingworld. At CIMMYT attempts areunderway to improve drought toleranceby introgressing stress adaptive traits intoempirically selected drought tolerantgermplasm. Our current conceptualmodel of a drought resistant cultivarencompasses high expression of thefollowing traits: seed size, coleoptilelength, early ground cover, pre-anthesisbiomass, stem reserves/remobilization,spike photosynthesis, stomatalconductance, osmotic adjustment,accumulation of abscisic acid, heattolerance, leaf anatomical traits (such asglaucousness, pubescence, rolling,

thickness), high tiller survival, and stay-green (Reynolds et al., 1999).CIMMYT’s germplasm collection isbeing screened, as resources allow it, forhigh expression of many of these traits.

High stomatal conductance permits leafcooling through evapotranspiration; this,along with higher leaf chlorophyllcontent and stay-green, is associated withheat tolerance (Reynolds et al., 1994).Recent studies identified high expressionof these traits in bank accessions, andboth traits showed high levels ofheritability under heat stress(Villhelmsen et al., 2001). As a result,these accessions are currently beingcrossed into good heat tolerantbackgrounds.

Pubescence and glaucousness protectplant organs from excess radiation understressful conditions (see Loss andSiddique, 1994). Searches are under wayfor these and a number of other leaftraits, such as leaf rolling, leaf thickness,and upright posture, which may well playsimilar roles under stress.

Osmotic adjustment (Blum et al., 1999)and stored stem fructans (Blum, 1998)have been implicated in stress tolerance.Searches are underway for highexpression of these traits among

germplasm bank accessions, althoughlaboratory protocols are required fortheir identification. High spikephotosynthesis is another trait that couldcontribute to yield under stress butwhich is very time consuming tomeasure. For traits that are difficult tomeasure (and/or that show markedgenotype by environment interaction), itis logical to develop genetic markers,which can be used to confirm theirpresence more unequivocally than bymeasuring phenotypic expression.

Conclusions

The last 30 years have witnessed anunprecedented level of internationalwheat germplasm exchange and thedevelopment of a greater degree ofgenetic relatedness among successfulcultivars all over the world. The conceptof broad adaptation has thus been wellvindicated. However, greater geneticrelatedness is seen by some as increasinggenetic vulnerability to pathogens,although such vulnerability dependsmore on similarities in resistance genes,which may actually be more diverse nowthan before. Various new factors(including the growing strength ofnational breeding programs in thedeveloping world and the advent ofbreeders’ rights) should result inincreased diversity among cultivars andmay lead to the exploitation of hithertooverlooked specific adaptation in wheat.

This would be especially important ifclimate change accelerates. Just asincreasing nitrogen supply andimproving weed control have beenalmost universal factors driving wheatcultivation in the last 50 years, higheratmospheric concentrations of CO2 andglobal warming with resulting warmertemperatures could significantlyinfluence breeding objectives in the next50 years.


Table 7. Expression of the multi-ovary trait and yield components in F1 lines, wheatscreenhouse, Mexico, 1999.

Yield component: Kernel Kernels / Kernel Grain wt /Line / cross no. / spike floret wt (mg) spike (g)

Multi-ovary line 124.0 2.17 37.5 4.65Pastor 69.3 1.00 51.5 3.57Multi-ovary line/Pastor 125.9 1.81 42.1 5.30Pastor/Multi-ovary line 108.5 1.65 45.0 4.88Baviacora M 92 72.7 1.00 57.5 4.18Multi-ovary line/Baviacora M 92 84.6 1.04 62.8 5.31Baviacora M 92/Multi-ovary line 73.5 1.03 60.0 4.41Esmeralda M 86 95.3 1.00 53.0 5.05Multi-ovary line/Esmeralda M 86 91.8 1.13 59.1 5.43Esmeralda M 86/Yanglin 96.3 1.20 53.8 5.18

27

To boost yield in irrigated situations,spike fertility must be improvedsimultaneously with photosyntheticcapacity. CIMMYT’s wheat germplasmbank has identified a source of multi-ovary florets that have up to six kernelsper flower. Other lines from landracecollections have very high chlorophyllconcentrations, which may increasephotosynthetic capacity. Highchlorophyll concentration and highstomatal conductance (which permitsleaf cooling) are associated with heattolerance. Recent studies identified highexpression of these traits in bankaccessions, and both traits were heritableunder heat stress. Searches are underwayfor drought tolerance traits related toremobilization of stem fructans, awnphotosynthesis, osmotic adjustment, andpubescence.

Seed multiplication nurseries can beused for characterizing and evaluatinggermplasm collections for physiologicaltraits. Characterization data can beanalyzed using pattern analysis, whichprovides a good description of theaccessions. The advantage of using theseaugmented seed nurseries is that cohortsof high(er) yielding lines are identifiedthat can be used directly or examined for“new” traits. Genetic diversity fromwheat’s wild relatives has been exploitedthrough wide-crossing to improvedisease resistance. Further potentialexists for identifying quantitative traitsusing QTL analysis in delayed backcrossgenerations. Once markers linked totraits of interest are identified,germplasm collections could be rapidlyscreened for unique alleles at thesemarkers.

Genetic resources are fundamental to theworld’s food security and central toefforts to alleviate poverty. Theycontribute to the development ofsustainable production systems andsupplement the natural resource base.Conserved germplasm is especially rich

in wild crop relatives, traditional farmercultivars, and old varieties, whichtogether represent an immense reserve ofgenetic diversity. Materials conservedboth ex and in situ are a safeguardagainst genetic erosion and a source ofresistance to biotic and abiotic stresses,improved quality, and yield traits forfuture crop improvement. As D.C.Rasmusson recently stated (pers. comm.,2000), “a little genetic diversity goes along way.”

ReferencesBlum, A. 1998. Improving wheat grain filling

under stress by stem reserve mobilization.Euphytica 100:77-83.

Blum, A., J. Zhang, and H.T. Nguyen. 1999.Consistent differences among wheatcultivars in osmotic adjustment and theirrelationship to plant production. Field Cropsresearch 64:287-291.

Borlaug, N.E. 1988. Challenges for global foodand fiber production. Journal of the RoyalSwedish Academy of Agriculture andForestry (Supplement) 21:15-55.

Chen, T.Y., B. Skovmand, S. Rajaram, and M.P.Reynolds. 1998. Novel source of increasedspike fertility in wheat multi-seeded flowers.Agronomy Abstracts p. 161.

Cox, T.S. 1991. The contribution of introducedgermplasm to the development of U.S.wheat cultivars. In: Use of PlantIntroductions in Cultivar Development. Part1. H.L. Shands and L.E. Wiesner (eds.).CSSA Special Publication No. 17. pp. 25-47.

DeLacy, I.H., K.E. Bassford, M. Cooper, J.K.Bull, and C.G. McLaren. 1996. Analysis ofmulti-environment trials – an historicalperspective. In: Plant adaptation and cropimprovement. M. Cooper and G.L. Hammer(eds.). Wallingford, UK: CAB International,IRRI and ICRISAT. pp. 39-124.

DeLacy, I.H., B. Skovmand, and J. Huerta. 2000.Characterization of Mexican landraces usingagronomically useful attributes. Acceptedfor publication in Genetic Resources andCrop Evolution.

Dewey, D.R. 1984. The genomic system ofclassification as a guide to intergenerichybridization with the perennial Triticeae.In: Gene manipulation in plantimprovement. J.P. Gustafson (ed.). NewYork: Plenum Press. pp. 209-279.

Eckart, C., and G. Young. 1936. Theapproximation of one matrix by another oflower rank. Psychometrika 1:211-218.

FAO. 1983. Commission on plant geneticresources. Resolution 8/83 of the 22ndSession of the FAO Conference, Rome.

Fischer, R.A. 1996. Wheat physiology atCIMMYT and raising the yield plateau. In:Increasing yield potential in wheat:Breaking the barrier. M.P. Reynolds, S.Rajaram, and A. McNab (eds.). Mexico,D.F.: CIMMYT. pp. 195-202.

Frankel, O.H. 1977. Natural variation and itsconservation. In: Genetic Diversity ofPlants. A. Muhammed and R.C. von Botstel(eds.). New York: Plenum Press. pp. 21-24.

Gabriel, K.R. 1971. The biplot-graphical displayof matrices with application to principalcomponent analysis. Biometrika58:453-467.

Gale, M.D., and S. Youssefian. 1986. Dwarfinggenes in wheat. In: Progress in PlantBreeding. G.E. Russell (ed.). London, UK:Butterworths.

Gower, J.C. 1966. Some distance properties oflatent root and vector methods used inmultivariate analysis. Biometrika53:325-338.

Harlan, J.R. 1992. Crops and Man. AmericanSociety of Agronomy, Madison, WI, USA.pp. 106-113.

Hawkes, J.G. 1991. International workshop ondynamic in-situ conservation of wildrelatives of major cultivated plants:summary and final discussion andrecommendations. Israel Journal of Botany40:529-536.

Hazekamp, T., J. Serwinski, and A. Alercia. 1997.Multi-crop Passport Descriptors. In: CentralCrop Databases: Tools for Plant GeneticResources Management. Lipman, E.,M.W.M. Jongen, T.J.L. van Hintum, T. Gass,and L. Maggioni (compilers). IPGRI/CGN.pp. 35-39.

Hede, A., B. Skovmand, M.P. Reynolds, J. Crossa,A.L. Vilhelmsen, and O. Stoelen. 1999.Evaluating genetic diversity for heattolerance traits in Mexican wheat landraces.Genetic Resources and Crop Evolution46:37-45.

IBPGR. 1990. Directory of crop germplasmcollections. 3. Cereals: Avena, Hordeum,Millets, Oryza, Secale, Sorghum, Triticum,Zea and Pseudocereals. E. Bettencourt and J.Konopka. International Board for PlantGenetic Resources, Rome.

Kihara, H. 1983. Origin and history of “Daruma,”a parental variety of Norin 10. In:Proceedings of the 6th International WheatGenetics Symposium. S. Sakamoto, (ed.).Plant Germplasm Institute, University ofKyoto. Kyoto, Japan.

Krull, C.F., and N.E. Borlaug. 1970. Theutilization of collections in plant breedingand production. In: Genetic Resources inPlants: Their Exploration and Conservation.O.H. Frankel, and E. Bennett (eds.). Oxford,UK: Blackwell Scientific Publications.


28

Loss, S.P., and K.H.M Siddique. 1994.Morphological and physiological traitsassociated with wheat yield increases inMediterranean environments. Adv. Agron.52:229-276.

Maes, B., Trethowan, R.M., Reynolds, M.P., vanGinkel, M., and Skovmand, B. 2001. Theinfluence of glume pubescence on spikelettemperature of wheat under freezingconditions. Aust. J. Plant Physiol.28:141-148.

Mujeeb-Kazi, A. 1995. Interspecific crosses:hybrid production and utilization. In: A.Mujeeb-Kazi, and G.P. Hettel, eds. 1995.Utilizing wild grass biodiversity in wheatimprovement: 15 years of wide crossresearch at CIMMYT. CIMMYT ResearchReport No. 2. Mexico, D.F.: CIMMYT.

Pardey, P.G., B. Skovmand, S. Taba, M.E. VanDusen, and B.D. Wright. 1998. The cost ofconserving maize and wheat geneticresources ex situ. In: M. Smale (ed.).Farmers, gene banks, and crop breeding:Economic analysis of diversity in rice,wheat, and maize. Dordecht, TheNetherlands: Kluwer Academic Press.pp. 35-55.

Pardey, P.G., B. Koo, B.D. Wright, M.E. VanDusen, B. Skovmand, and S. Taba. 2000.Costing the Conservation of GeneticResources: CIMMYT’s Ex Situ Maize andWheat Collection. Crop Sci. (in press).

Rasmusson, D.C. 2001. Learning about barleybreeding. In: Breeding Barley in the NewMillenium. H.E. Vivar and A. McNab (eds.).Mexico, D.F.: CIMMYT. pp. 1-6.

Reynolds, M.P., M. Balota, M.I.B. Delgado, I.Amani, and R.A. Fischer. 1994.Physiological and morphological traitsassociated with spring wheat yield underhot, irrigated conditions. Aust. J. PlantPhysiol. 21:717-30.

Reynolds, M.P., B. Skovmand, R. Trethowan, andW.H. Pfeiffer. 1999. Evaluating aConceptual Model for Drought Tolerance.In: Using Molecular Markers to ImproveDrought Tolerance. J.M. Ribaut (ed.).Mexico, D.F.: CIMMYT.

Richards, R.A. 1996. Increasing the yieldpotential of wheat: Manipulating sourcesand sinks. In: Increasing Yield Potential inWheat: Breaking the Barriers. M.P.Reynolds, S. Rajaram, and A. McNab (eds.).Mexico, D.F.:CIMMYT. pp. 134-149.

Rosegrant, M.W., M. Agcaoili-Sombilla, and N.D.Perez. 1995. Global Food Projections to2020: Implications for Investment.Washington, D.C.: IFPRI.

Sayre, K.D., S. Rajaram, and R.A. Fischer. 1997.Yield potential progress in short breadwheats in northwest Mexico. Crop Sci.37:36-42.

Shands, H.L. 1991. Complementarity of in-situand ex-situ germplasm conservation fromthe standpoint of the future user. IsraelJournal of Botany 40:521-528.

Siebeck, W.E. 1994. Intellectual Property Rightsand CGIAR Research: Predicament orChallenge. CGIAR Annual Report 1993-1994. pp. 17-20.

Skovmand, B., P.N. Fox, and J.W. White. 1998.Integrating research on genetic resourceswith the international wheat informationsystem. In: Wheat Prospects for globalimprovement. H.-J. Braun, F. Altay, W.E.Kronstad, S.P.S. Beniwal, and A. McNab(eds.). Dordecht, The Netherlands: KluwerAcademic Publishers. pp. 387-391.

Skovmand, B., M.C. Mackay, H. Sanchez, H. vanNiekerk, Zonghu He, M. Flores, R. Herrera,A. Clavel, C.G. Lopez, J.C. Alarcon, G.Grimes, and P.N. Fox. 2000a. GRIP II:Genetic resources package for Triticum andrelated species. In: Tools for the NewMillenium. B. Skovmand, M.C. Mackay, C.Lopez, and A. McNab (eds.). On compactdisk. Mexico, D.F.: CIMMYT.

Skovmand, B., Lopez, C., Sanchez, H., Herrera,R., Vicarte, V., Fox. P.N., Trethowan, R.,Gomez, M.L., Magana, R.I., Gonzalez, S.,van Ginkel, M., Pfeiffer, W., and Mackay,M.C. 2000b. The International WheatInformation System (IWIS). Version 3. In:Tools for the New Millenium. B. Skovmand,M.C. Mackay, C. Lopez, and A. McNab(eds.). On compact disk. Mexico, D.F.:CIMMYT.


Skovmand, B., G. Varughese, and G.P. Hettel.1992. Wheat Genetic Resources atCIMMYT: Their Preservation,Documentation, Enrichment, andDistribution. Mexico, D.F.: CIMMYT.

Tanksley, S.D., and J.C. Nelson. 1996. Advancedback-cross QTL analysis: A method for thesimultaneous discovery and transfer ofvaluable QTLs from unadapted germplasminto elite breeding lines. Theor. Appl. Genet.92:191-203.

Trethowan, R.M., M.P. Reynolds, B. Skovmand,and M. van Ginkel. 1998. The effect ofglume pubescence on floret temperature inwheat. Agronomy Abstracts p. 161.

van Slageren, M.W. 1994. Wild wheats: amonograph of Aegilops L. andAmblyopyrum (Jaub. & Spach) Eig(Poaceae). Wageningen Agric. Univ.Papers 94-97.

Vilhelmsen, A.L., Reynolds, M.P., Skovmand, B.,Mohan, D., Ruwali, K.N., Nagarajan, S., andO. Stoelen. 2001. Genetic diversity andheritability of heat tolerance traits in wheat.Wheat Special Report No. 50. Mexico, D.F.:CIMMYT. (In preparation.)

Villareal, R.L., G. Fuentes Davila, and A. MujeebKazi. 1995. Synthetic hexaploid x Triticumaestivum advanced derivatives resistant toKarnal bunt (Tilletia indica Mitra). Cer. Res.Commun. 23:127-132.

Von Botmer, R., O. Seberg, and N. Jacobsen.1992. Genetic resources in the Triticeae.Hereditas 116:141-150.

Williams, W.T. 1976. Pattern analysis inagricultural science. Amsterdam, TheNetherlands: Elsevier Scientific PublishingCompany.

29

During the past two decades, moleculartools have aided tremendously in theidentification, mapping, and isolation ofgenes in a wide range of crop species.The vast knowledge generated throughthe application of molecular markers hasenabled scientists to analyze the plantgenome and have better insight as tohow genes and pathways controllingimportant biochemical andphysiological parameters are regulated.Three areas of biotechnology have hadsignificant impact: the application ofmolecular markers, tissue culture, andincorporation of genes via planttransformation.

Molecular markers have enabled theidentification of genes or genomicregions associated with the expressionof qualitative and quantitative traits andmade manipulating genomic regionsfeasible through marker assistedselection. Molecular marker applicationshave also helped us understand thephysiological parameters controllingplant responses to biotic and abioticstress or, more generally, those involvedin plant development. This chapterdiscusses different types of molecularmarkers, the basic principles andpractical considerations involved in theirapplication in plant improvement, andsome contributions they have made towheat molecular genetics.

The Genome

Although the expression of genes can bemodified by environmental factors, thenuclear genome of plant cells carries agenetic blueprint in the form ofdeoxyribonucleic acid (DNA) thatcontains information for cell maintenanceand replication. The nuclear genomecontains the largest amount of DNA andthe highest number of genes encoded, butplant cells also contain DNA in theirchloroplasts and mitochondria. Nucleargenomes of crop species are estimated tocontain thousands of genes, some uniqueand others in multiple copies. However,the amount of DNA in the nuclear genomerepresented by transcribed genes is only afraction of total DNA found in thegenome.

Nuclear DNA is packaged and organizedinto chromosomes along with histones andnon-histone proteins. The interactionsbetween DNA and proteins play an

important role in gene expression. WhileDNA encodes genetic information in theform of messenger RNA (mRNA),proteins are involved in the packaging ofDNA and in regulating its availability fortranscription. Transcribed gene productsare transported across the nuclearenvelope to be translated into proteinsusing the cellular apparatus.

Genes are distributed along thechromosomes, and the number ofchromosomes a plant cell contains variesamong crop species. There isconsiderable diversity in genomecomposition and organization of differentorganisms (Table 1). With the aid ofmolecular techniques, it has been possibleto study and understand the organizationof the nuclear genome of several plantspecies. Plant genome analysisencompasses genome mapping, genetagging, quantitative trait (QTL) analysis,and synteny mapping.

Genetic Basis ofPhysiological TraitsJ.-M. Ribaut, H.M. William, M. Khairallah,A.J. Worland,2 and D. Hoisington1

1 CIMMYT Applied Biotechnology Laboratory, Apdo. Postal 6-641, Mexico, D.F., Mexico, 06600.2 John Innes Centre, UK.

Table 1. DNA content per haploid genome in different organisms.

Mega base pairsOrganism 2n Picograms † 106bp / 1C Length (cm)

E. coli (1) 0.0047 4.2 0.14Chloroplast (maize) (c) 0.0002 0.160 0.006Mitochondrion (maize) (m) 0.0007 0.570 0.02Arabidopsis thaliana 10 0.15 150 4.4Oryza sativa 24 0.45 430 13.1Triticum aestivum 42 5.96 5,700 173Zea mays 20 2.6 2,500 75Homo sapiens 46 3.2 3,900 102

† 1 picogram = 1 pg = 0.965 x 109 bp = 29 cm.

C HAPTER 3

30

The DNA MoleculeIn higher organisms, a DNA moleculeconsists of a sequence of nuclear acidslinked by chemical bonds. Eachnucleotide contains a heterocyclic ringcomposed of carbon and nitrogen atoms(the nitrogenous base), a five-carbonsugar in ring form (a pentose), and aphosphate group. There are two kinds ofnitrogenous bases: purines andpyrimidines. Each nucleic acid iscomposed of only four types of bases:two kinds of purines, known as adenine(A) and guanine (G), and two kinds ofpyrimidines, cytosine (C) and thymine (T).

The nitrogenous base is linked to thepentose sugar by glycosidic bonds.When the phosphate group is added tothe pentose sugar, the base-sugar-phosphate complex is called anucleotide. Nucleotides are linkedtogether into a chain by a backboneconsisting of an alternating series ofsugar and phosphate residues with thebases attached to the sugar molecules. Inhigher organisms, DNA consists of twostrands of nucleic acids that are wrappedaround each other in antiparallel form ina double helix. The sides of the twostrands are composed of sugar andphosphate molecules, and the bases areinside the double helices. The twostrands are held together by hydrogenbonding between the purine of onestrand with a pyrimidine of the oppositestrand. Base A always pairs with a T viatwo hydrogen bonds, whereas a Galways pairs with a C via three hydrogenbonds. The composition of bases alongone strand of the DNA chain is exactlycomplementary to its partner strand,which allows both strands to carry thesame genetic information. This isessential for the self replicatingcapability of DNA.

The particular order of the basesarranged along the sugar-phosphatebackbone is called the DNA sequence.This sequence provides precise genetic

instructions for creating a particularorganism with its own unique traits. Thesize of a genome is usually stated as thetotal number of base pairs in the haploidgenome (Table 1).

Genes and ChromosomesThe gene is the basic physical andfunctional unit of heredity. Each gene is anucleic acid sequence that carriesinformation encoded to represent aparticular polypeptide. Polypeptidesprovide the structural components of cellsand tissues as well as enzymes foressential biochemical functions. The plantgenome is estimated to comprise 20,000to 100,000 genes.

Genes vary widely in length, oftenextending over thousands of bases, butonly about 10% of the genome is knownto include protein-coding sequences(exons) of genes. Interspersed withingenes are intron sequences, which haveno coding function. The rest of thegenome is thought to consist of othernoncoding regions (such as controlsequences and intergenic regions), whosefunctions are still obscure. Theconfiguration and methyation level of aDNA molecule play a role in geneexpression, since expressed regions aregenerally characterized by a high level ofmethylation. Some genes have fewcopies; others may be present in multiplecopies per haploid genome. Suchrepeated sequences may be present intandem copies at a chromosomal locus orin different chromosomes dispersedthroughout the genome.

The vast amount of DNA present in eachplant cell is tightly packaged, with thehelp of histone and non-histone proteinsin the nucleus, into microscopicstructures known as chromosomes. Genesare scattered along the chromosomes,which vary in number from species tospecies. During gametic formation thesomatic chromosome number is dividedin half by cell division (meiosis), which

ensures that the zygote (after the maleand female gametes unite) will containthe same number of somaticchromosomes as the parents.

The chromosomal form of the nucleargenome varies significantly during celldivision. During interphase, thechromatin in the chromosomes remainsdiffuse and therefore less visible underthe microscope, it becomes morecondensed and highly visible forcytogenetic manipulation during meiosisand mitosis. Cytological studies ofindividual chromosomes duringmetaphase through chromosomebanding techniques have helpedcharacterize and identify the individualchromosomes of the wheat karyotype.Moreover, classical cytological studiesand current molecular cytogenetictechniques have aided in identifyingchromosomal abnormalities and subtleinterchanges.

The Wheat GenomeThe numerous species of the genusTriticum can be classified into threeploidy groups: diploids (2n=2X=14),tetraploids (2n=4X=28), and hexaploids(2n=6X=42). Of the Triticum species,cultivated T. aestivum, known as breadwheat, is the principal commercial type,whereas T. turgidum (durum wheat) isprincipally used for making pasta.Cultivated bread wheat is anallohexaploid (2n=6X=42), composedof three distinct genomes, A, B and D.Current evidence suggests that itoriginated from natural hybrids of threediploid wild progenitors native to theMiddle East. Triticum urartu Tum. isrecognized as the donor of the Agenome. Although Aegilops speltoideswas considered the donor of the Bgenome, current evidence suggest thatthe real donor is either extinct or anundiscovered species belonging to theSitopsis section of Aegilops (Pathak,1940; Kimber and Athwal, 1972; Milleret al., 1982). Triticum tauschii, also

J.-M. RIBAUT ET AL.

31

known as Aegilops tauschii, is widelyrecognized to be the donor of the Dgenome (Kimber and Feldman, 1987).

Among crops, wheat possesses one of thelargest (about 16 billion bp per haploidgenome) and most complex (hexaploid)genomes, with a high percentage ofrepetitive sequences (90%), which makesit quite challenging to study andmanipulate at the molecular level.However, polyploids have a greaterability to tolerate loss or higher dosagesof chromosomes, referred to asaneuploidy. Because of its hexaploidnature and economic importance as afood source, bread wheat is the mostcytogenetically studied of the cropspecies. The complete range of aneuploidlines (nullisomics, monosomics,trisomics, and tetrasomics; Sears, 1953,1954) and a great diversity ofchromosome deletion stocks (Endo andGill, 1996) have been made available inwheat. These cytogenetic stocks havebeen utilized in numerous studies aimedat locating genes on chromosomes andchromosome arms, as well asestablishing relationships among thechromosomes of hexaploid wheat basedon their origin and function.

DNA Markers

Markers are “characters” whoseinheritance pattern can be followed at themorphological (e.g., flower color),biochemical (e.g., proteins and/orisozymes), or molecular (DNA markers)levels. These characters are calledmarkers because they provide, althoughindirectly, information about the geneticsof other traits of interest in a givenorganism. The main disadvantage ofmorphological markers is that they areeasily influenced by the environment. Incontrast, molecular markers are based onvariations in genomic DNA sequences;since they are neutral, they have nophenotypic effect on the plant. The main

advantages of molecular markers are thatthey can be numerous, are not affected bythe environment, and can be scored atvirtually any stage of plant development.

DNA markers can be based on restrictionfragment length polymorphisms (RFLPs)or on the polymerase chain reaction(PCR) technique.

Restriction Fragment LengthPolymorphismsThe RFLP technique was the first to bewidely used in plant genome analysis.RFLP linkage maps of a number ofspecies including wheat and maize havealready been made. In this technique, aDNA sample taken from a particularplant is treated with restriction enzymes.Restriction enzymes recognize uniquesequences in the double-stranded DNAand cleave both strands to producenumerous DNA fragments of varyinglength. These DNA fragments areseparated, based on their size, on anagarose matrix in gel electrophoresis,denatured to make the DNA single-stranded, and then blotted onto nylon ornitrocellulose membranes using theSouthern transfer technique. The DNA inthe membrane is then hybridized with aprobe isolated from the same, or arelated, plant species, and whosechromosomal location is known. Theprobe, labelled with radioactivity orchemilluminiscent substances, hybridizesto complementary sequences in thefragmented DNA sample. Because ofmolecular differences in the plants beingstudied, the tagged or hybridizedfragments will differ in length, whichallows the samples to be uniquelycharacterized as molecularpolymorphisms. Since the chromosomallocation or “map position” of the probesis known, researchers can trace the lengthpolymorphisms to chromosomal regions.These molecular polymorphisms ormolecular markers can then be treated asany other Mendelian difference betweencontrasting samples.

Although the RFLP technique is time-consuming and somewhat cumbersomecompared to more recent markertechnologies, it is still extensively used ina wide range of crop species.

Markers Based onPolymerase Chain ReactionDescribed in the early 1980s, PCR-basedassays have revolutionized molecularmarker assay systems. PCR-basedtechniques are robust, amenable toautomation, and widely applied in large-scale marker development orimplementation procedures.

PCR-based assays are based on an in vitroprocedure for the enzymatic synthesis ofDNA, in which two oligonucleotideprimers hybridize to opposite strandsflanking the region of interest in the targetDNA (Figure 1). The procedure enablessmall amounts of specific DNA fragments(which may be mixed with large amountsof contaminating DNA) to be amplifiedexponentially. In a typical PCR-basedassay, the “building blocks” required tosynthesize a new strand of DNA aremixed with the template containing thetarget DNA together with primers in atube along with thermostable DNApolymerase. They pass through cycles ofdifferential temperatures involvingtemplate denaturation, primer annealing,and extending the annealed primers byDNA polymerase. The end result is anexponential accumulation of the targetsequence, which can then be resolved onseparation matrices such as agarose oracrylamide and viewed as discrete bandsafter staining.

Several types of PCR-based markers arebeing used in plant genome analysis:

• Random amplified polymorphic DNA(RAPDs)

• Sequence-tagged sites (STSs)• Simple sequence repeat (SSR) or

microsatellite• Amplified fragment length

polymorphism (AFLP)

GENETIC BASIS OF PHYSIOLOGICAL TRAITS

32

Random amplifiedpolymorphic DNAAs its name implies, the randomamplified polymorphic DNA (RAPD)technique (Williams et al., 1990; Welshand McClelland, 1990) is used torandomly amplify certain sequences.The primers used contain randomlysynthesized oligonucleotides and areusually short (about 10 bp). RAPDpolymorphisms are the result of either anucleotide base change that alters theprimer binding site or an insertion ordeletion within the amplified regions.The major advantage of RAPDs is thatthey are suitable for all species because

randomly synthesized primers (whichare widely available) are not speciesspecific. Disadvantages include thedominant nature of RAPD markers(only presence or absence of a bond,which means that the heterozygouscannot be identified), their randomness,and the resulting lack of repeatabilitydue to non-specificity of theamplification products, specially inspecies such as wheat, where thegenome is very large.

Sequence tagged sitesSequence tagged sites (STSs) aremapped loci for which all or part of the

corresponding DNA sequences havebeen determined (Olson et al., 1989;Talbert et al., 1994). This informationcan be used to synthesize PCR primersthat amplify all or part of the originalsequence. Since the primers aredesigned to amplify one specific locusand are longer than those used in RAPDanalysis, STS assays are more robustand therefore more reproducible andreliable than RAPD analysis.Differences in the length of amplifiedsequences from different individuals canserve as genetic markers of the locus.

If no polymorphism is detected uponPCR amplification, the amplifiedfragments can be cut with restrictionenzymes to observe length differencesamong samples, which can then be usedas markers. This technique is sometimesreferred to as cleaved amplifiedpolymorphic sequences (CAPS).

The STS technique holds great promisefor marker-assisted selection schemes,since it can be applied on a large scaleand specific loci can be followedthrough successive plant generations inconventional breeding programs.

Simple sequence repeatsSimple sequence repeats (SSRs), alsoknown as microsatellites, are composedof tandem repeats of two to fivenucleotide DNA core sequences such as(AT)n, (GT)n, (ATT)n, or (GACA)nspread throughout eukaryotic genomes(Tautz and Renz, 1984). The DNAsequences flanking microsatellites aregenerally conserved within individualsof a given species, allowing the designof PCR primers that amplify theintervening SSRs in all genotypes(Weber and May, 1989; Litt and Luty,1989). Variation in the number oftandem repeats results in different PCRproduct lengths (Figure 2). Theserepeats are highly polymorphic, evenamong closely related cultivars, due tomutations causing variation in thenumber of repeating units. The main

Figure 1. Polymerase chain reaction: DNA amplification.

5’ 3’

3’ 5’

Template DNA Denature Step 1: 95°C

5’ 3’

3’ 5’

Anneal oligonucleotide Step 2: 30-55°Cprimers

5’ 3’

3’OH 5’P5’P 3’OH

3’ 5’P

Polymerization(Taq) Step 3: 72°C

5’ 3’

3’ 5’

5’ 3’

3’ 5’

Repeat steps 1-3 for 30 to 45 cycles

N cycles = 2n Molecules

J.-M. RIBAUT ET AL.

33

advantages of SSRs are the co-dominant nature of the observedpolymorphisms (which means thathomozygous A and B, as well asheterozygous AB, can be identified),the robustness of the assay, and thelarge number of polymorphismsobserved. Their main disadvantage isthe significantly high cost involved insequencing genomic libraries in thedevelopment of SSRs.

Amplified fragment lengthpolymorphismsAmplified fragment lengthpolymorphisms (AFLPs) combine thespecificity of RFLP analysis with therobustness of the PCR assay and aredesigned to amplify a subset ofrestriction digested DNA (Vos et al.,1995). Usually two restrictionenzymes, a rare cutter and a frequentcutter, are used in combination todigest genomic DNA (Figure 3). TheDNA fragments thus generated areligated with double-stranded adaptorsequences. The adapter-ligatedfragments are subjected to two roundsof PCR amplifications. In the firstround, primers complementary to theadapter sequences, plus an additionalnucleotide at the 3’ end, are used. Asecond PCR reaction is performed onthe modified fragments with primershaving the same sequence used in pre-amplification, plus one to three

The invention of PCR-based assays hasprovided the basis for a large number ofinnovative methods for recognizingDNA polymorphisms amongindividuals, as described above. Formapping and large-scale screening,SSRs are the most desirable PCR-basedmarkers. Once large numbers of SSRsare available that provide good coveragefor a crop genome, large-scale SSRassays can be reliably performed at anearly plant development stage, because:1) a small amount of tissue is required;2) DNA preparation is faster due to thesmall amount of template DNArequired; and 3) large sample sizes arehandled more efficiently. Moreover,SSRs are reliable, co-dominant,abundant, and uniformly dispersedwithin plant genomes. In July 2000, a

additional nucleotides, to amplify asubset of the pre-amplified DNAproducts. The numerous amplifiedsequences are sorted using highresolution electrophoresis. Differences inthe length of the amplified segments arerelated to differences in the DNAcomposition of two given individuals.The primary advantage of AFLPs is thelarge number of fragments that can becompared per analysis.

Utility of DNA MarkersRFLP markers have been used toconstruct linkage maps for crop species,such as maize, tomato, and rice. ManyRFLP markers with tight linkage togenes controlling economicallyimportant traits in various crop specieshave been identified. Once the sequenceof an RFLP marker of interest is known,a PCR-based marker (STS) can bedeveloped for large-scale screening(Ribaut et al., 1997). RFLPs are reliablemarkers, and the same probe can usuallybe hybridized on different crop genomes,making RFLP markers useful forcomparative mapping studies. However,RFLP analysis requires large quantitiesof quality DNA, and detection of RFLPsby Southern blot hybridization may belaborious and time consuming, whichmay make this assay undesirable forplant breeding projects with highthroughput requirements.

Figure 2. Example of a microsatellite: a dinucleotide repeat showing a polymorphismbetween two different individuals.

Amplified SSR DNA fragments have a 6 bp difference in length, a DNApolymorphism that can be resolved by high resolution electrophoresis

Amplified SSRs

DNA, individual 1

Conservedprimed region 1

DNA, individual 2

Conservedprimed region 2

+ EcoRI Msel

EcoRI adapter Msel adapter primer +1

Preselective primer +1amplification withEcoRI primer +AMsel primer +C

primer +3

primer +3Selectiveamplification withprimers +3

Sequencing gel

EcoRI adapter sequenceMsel adapter sequence

Figure 3. AFLP method.


34

collection of 500 SSRs becameavailable in wheat, but 1000 to 1500SSRs would be needed to develop acomplete linkage map for QTLidentification.

Geneticists today have powerful tools toconduct genomic analysis and traitdissection in crop species. The mostsuitable marker will depend mainly onthe purpose of the investigation and thetype of markers available for a particularcrop (see a comparison of markersystems in Table 2).

GenomicsThe newest area of investigation aimedat understanding the plant genomeencompasses genome-wide approaches.“Functional genomics” can be definedas the development and application ofgenome-wide experimental approachesto assess gene function (Heiter andBoguski, 1997). The ultimate goal ofgenomics would be to characterizeevery gene present in a given genome.Approaches being utilized to achievethis goal are large-scale sequencing ofexpressed sequence tags (ESTs), large-scale functional analysis of plant genes(where thousands of DNA or RNAsequences can be analyzed onmicroscopic slides), and application ofinsertional mutagenesis or reversegenetics. These technologies are highthroughput and require automation.

Innovative tools such as DNA chips andmicroarray have been developed toserve the new approaches. DNA chiptechnology provides efficient access togenetic information using miniaturized,high-density arrays of oligonucleotideprobes. A set of oligonucleotides isdefined, artificially synthesized, andimmobilized on silica wafers or chips toconstruct a high-density array; eachprobe has a predefined position in thearray. Labeled (fluorescence) nucleicacids from the analyzed plant sample arehybridized on the array, and

hybridization intensities are detected bya scanner that reports quantitativeassessment of RNA levels in the samplefor each gene represented in the array(Lemieux et al., 1998). Microarrays aresimilar to the DNA chip, except thatthey use cDNAs (Ex. EST cloneinserts). These innovative approachesare expected to provide insight into howthe plant genome functions and toidentify more genes involved inregulating different pathways inresponse to stress conditions.

Application ofMolecular Markers inPlant Breeding

Conventional plant breeding is based onthe selection of superior individualsamong segregating progenies of sexualmatings. Selection for plantimprovement has largely been carriedout on the whole-plant or phenotype,which is the result of genotypic andenvironmental effects. Althoughconventional plant breeding has madetremendous progress in many crop

species, it is often hampered bydifficulties in selecting foragronomically important traits,especially when they are influenced bythe environment. Moreover, testingprocedures may be difficult, unreliable,or expensive, due to the nature of thetarget traits or the target environment(e.g., abiotic stresses). For thosereasons, selection through molecularmarkers might be an efficientcomplementary breeding tool, especiallywhen selection is done underunfavorable conditions. If individualgenes influencing target traits can beidentified and associated with molecularmarkers, the efficiency of incorporatingthem into new varieties could be greatlyenhanced.

FingerprintingA fingerprint specifically andunambiguously identifies a livingorganism. Identification can be achievedbased on polymorphisms determinedthrough molecular markers. In cropspecies, fingerprinting is a valuable toolfor establishing varietal purity, which isimportant for varietal protection,

Table 2. Characteristics and usefulness of different types of molecular markers in wheatmolecular genetics.

RFLPs RAPDs SSRs STSs AFLPs

Fingerprinting ++ - / + +++ + +++Genetic diversity ++ + ++ + ++Qualitative gene tagging ++ ++ ++ ++ ++QTL mapping ++ - / + ++ ++ ++MAS + - +++ ++ + / ++Comparative mapping +++ - - + -

Types of probe / primers gDNA, cDNA Random 10-mer Specific Specific Specific adaptersoligonucleotides 16-30 -mer 20-25 -mer and selective

oligonucleotides oligonucleotides primersLevel of polymorphism Medium Medium High Medium HighInheritance Codominant Dominant Codominant Codominant Dominant/co-

dominantTechnical difficulty Medium Low Low Medium Medium/HighReliability High Low High High Medium/High

Source: Modified from Rafalski and Tingey (1993).

J.-M. RIBAUT ET AL.

35

Figure 4. Description of how molecular markers can be used in genetic linkage mapping.

currently a concern for the commercialseed industry and public breedingenterprises. Fingerprinting can also beused to estimate the genetic diversity ofa set of cultivars or landraces and toestablish phylogenetic relationships forevolutionary studies. Other applicationsof molecular fingerprinting include:

• genomic characterization andidentification for propriety purposes;

• identification of superior alleles ingenebank accessions (e.g.,landraces);

• identification of duplications withingenebank accessions to ensure thebest use of available resources;

• correlation of genetic diversity withheterotic patterns.

Fingerprinting studies have beenreported for specific wheat germplasmusing sets of DNA markers includingRFLPs, microsatellites, and AFLPs(Barrett and Kidwell, 1998; Bohn et al.,1999; Fahima et al., 1998). The geneticdistance among accessions can beevaluated based on fingerprinting. Thisinformation allows bettercharacterization of genetic relationshipsamong accessions (e.g., establishment ofgene pools) and can be used to identifyparental lines with good alleliccomplementarity.

Genetic Mapping ofTarget TraitsGenetic dissection of a target trait canbe defined as identifying andcharacterizing the genomic segments orgenes involved in its phenotypicexpression. Before genetic manipulationusing molecular markers, genes orquantitative trait loci (QTLs) must beidentified and characterized via a two-step process: 1) construction of asuitable segregating population bycrossing two parental lines contrastingfor the target trait(s), and 2)identification of markers closely linkedto the gene(s) of interest for furtherallelic manipulation (for a summary of

the process, see Figure 4). This providesuseful information, such as:

• the number of genes or QTLssignificantly involved in theexpression of the target trait;

• the effect (additivity, dominance) ofthe identified genomic regions andtheir impact on phenotypic expressionof the trait;

• the stability of gene expressionacross environments (QxE); and

• the presence of pleiotropic effects atsome target genomic regions.

Unfortunately, the evaluation ofepistatic effects remains difficult, duemainly to the reduced number ofgenotypes used for this kind of geneticdissection.


PROBLEMIDENTIFICATION

MOLECULARDIVERSITY

PARENTALSELECTION

SEGREGATINGPOPULATION(S)

PHENOTYPICEVALUATION

GENOTYPICIDENTIFICATION

VERIFICATIONEXPERIMENTS

BREEDING APPLICATIONS

STATISTICALCORRELATIONS

PHENOTYPICDIFFERENCES

• Importance of trait?• Difficult/costly to screen?• Strategic advantage?

• Good contrast between parents (eg, resistant vs susceptible)• Field methodology yields precise measurement of trait(s)• Other traits?

• Backcross• Doubled Haploids• F2s• F3 families• Recombinant Inbred Lines (RILs)

• Measurement of trait(s) in field /greenhouse replicated trials

• ANOVAs (SAS)• MAPMAKER QTL• QTL Cartographer• qGene

• MAPMAKER• JOINMAP• LINKAGE

• Assay of all individuals in segregating population for known differences in isozymes, RFLPs, RAPDs...

• Repetition of experiments on crosses involving different base germplasm

• Marker-Assisted Selection (MAS) - recurrent genome selection - single-segment introgressions - differential QTL selection - reduction/avoidance of linkage drag

• Screening for isozyme differences, RFLPs, RAPDs, SSRs, AFLPs...• Detectable molecular polymorphisms covering the genome (20cM density) or the segments of interest?

MOLECULAR MARKER MAPPING PROJECT OUTLINE

36

important in locating biochemical ormolecular markers to chromosomes.Initial screening of a new markeragainst a set of wheat lines, eachlacking a different chromosome, willdetermine the marker’s chromosomallocation since in the absence of thechromosome carrying the gene, themarker will not be expressed. Althoughnullisomic (2n = 6x = 40 = 20") plantswould be ideal for this analysis, theyare difficult to maintain due to their lowfertility; consequently, compensatingnullisomic tetrasomic lines are utilizedin which the absence of onechromosome is compensated for by twoextra doses of a related homoeologouschromosome (Sears, 1953).

Germplasm for molecular analysisSegregating populations. The mostcommonly used materials for geneticallydissecting and mapping traits aresegregating populations descended fromtwo varieties showing divergence for thetarget trait(s). Effective markeridentification in populations segregatingfor target traits depends upon laboratorydata on the allelic composition ofmolecular markers at genomic locationsdistributed evenly within the genomeand field evaluation of the trait(s). Basedon both types of data, statisticalprocedures are used to find associationsbetween markers and traits. When targettraits are governed or influenced byseveral genetic factors, a genetic linkagemap of the complete genome must bedeveloped and a QTL analysis conductedto associate traits with markers over thecomplete genome. If the target trait isinfluenced by one or a few genes, linescan be classified for the trait and bulksegregant analysis used to associate traitalleles to molecular markers.

To develop a complete linkage map,genetically stable populations areadvanced through several recombinationcycles using self-pollination. In wheat,recombinant inbred lines (RILs) are bestsuited for such analyses, but a doubledhaploid population can be developed toobtain stable, completely homozygouslines for marker analysis and fieldevaluations. In both cases the size of thesegregating population has to becarefully considered, since populationsthat are too small will not allow precisegene characterization, specially whenmapping quantitative traits, and largepopulations will consume resourcesunnecessarily. To genetically dissect apolygenic trait, a RIL population ofabout 200 to 300 families is consideredsuitable, but the number can be reducedif the trait is controlled by major genes.

Genetic stocks. The allohexaploid natureof bread wheat has significantdisadvantages, but also some advantages

in genetic analysis. For example, thepresence of more than one set of genesallows wheat to tolerate the loss ofcomplete chromosomes or chromosomearms (aneuploidy). Wheat’s ability totolerate aneuploidy has resulted innumerous genetic studies aimed atlocating genes to chromosomes. Oncegenes are located on chromosomes, it ispossible to produce detailed linkagemaps for individual chromosomes andassociate genes with markers. Thechromosome constitution of commonlyoccurring wheat aneuploids is shown inFigure 5.

Aneuploids that lack a completechromosome or a chromosome arm havein recent years become extremely

Group 1 Chromosome dosageType of aneuploid

1A 1B 1DEuploid2n=42=21II

Monosomic 1D2n=41=20II+1I

Nullisomic 1D

2n=40=20II

Di-Telocentric 1D2n=42=20II+2t

Trisomic 1D2n=43=20II+1III

Tetrasomic 1D

2n=44=20II+1IV

Nulli 1B Tetra 1D2n=42=19II+1IV

Figure 5. Genetic stocks in wheat.†† Chromosome constitution assumes complete sets of groups 2-7 chromosomes.

J.-M. RIBAUT ET AL.

37

Genetically stable stocks should be usedwhere available to locate complex traitsto individual chromosomes. The mostsuitable stocks are single-chromosomeintervarietal substitution lines developedto introduce individual chromosomesfrom a donor variety into the geneticbackground of a recipient variety.

To develop intervarietal substitutionlines, a series of plants of the recipientvariety is needed in which the dosage ofindividual chromosomes has beenreduced from 2 to 1. Known asmonosomics, plants missing a singlechromosome (2n = 6x = 41 = 20~i x 1~)are the most commonly occurringaneuploids. About 70 monosomic seriesare now available in different varietiesworldwide (Worland, 1988). For simplyinherited characters, monosomic seriescan be used to locate genes using test-cross procedures such as monosomicanalysis (Sears, 1953), reciprocalmonosomic analysis (McKewan andKaltsikes, 1970), or backcross reciprocalmonosomic analysis (Snape and Law,1980). For more complex characters,monosomics are used as a base fordeveloping intervarietal substitutionlines by backcrossing individualchromosomes from a donor variety intothe background of a recipientmonosomic (Figure 6a) (Law andWorland, 1973). Once developed,intervarietal substitution lines are stableand true-breeding, and ideal for geneticanalysis. By screening a complete seriesof 21 chromosome substitution lines anygene or trait can be readily located toindividual chromosomes (Law andWorland, 1996).

Once the chromosomal location of agene or trait has been determined usingintervarietal substitution lines, these canbe used to develop extremely precisegenetic stocks known as single-chromosome recombinant lines (Law,1966; Law and Worland, 1973). Theselines are developed by initiallyproducing an F1 between the critical

substitution line and its recipient variety.In this F1, recombination is restricted tothe single critical chromosome in anotherwise genetically homozygousbackground. Products of recombinationof the critical chromosome are thenfixed by crossing the recombining F1plant onto a plant of the recipient varietymonosomic for the critical chromosome.Monosomic progeny are extracted fromthe backcross and selfed to permitselection of disomic plants carrying ahomozygous recombined chromosome(Figure 6b).

An alternative method of fixingrecombination products is to pollinatethe F1 between the recipient parent andthe substitution line with maize pollento produce haploid progeny that can bedoubled with colchicine. Normallyabout 100 single-chromosomerecombinant lines would be producedfor the critical chromosome. The linescan then be classified for the trait underinvestigation in replicated field orgrowth room experiments. The traitallele can be readily associated withmolecular markers by screening therecombinant lines with markers knownto be polymorphic between the twoparents and located on the criticalchromosome.

Linkage mapConstruction of a linkage map. Duringlinkage map development, polymorphicmolecular markers are used to genotypea segregating population. By statisticallyevaluating segregating marker allelesand linkages among different markeralleles from previous studies, markerscan be placed in “linkage groups.”When marker locations in the genomeare known (e.g., RFLPs or SSRs),linkage groups can be assigned tochromosomes. When the genome of acrop species has adequate coverage withmarkers, the number of linkage groupsobserved should match the number ofhaploid chromosomes in the genome(i.e., the maize linkage map should have

10 linkage groups, that of wheat shouldhave 21, etc.). Although constructing alinkage map is necessary for identifyinggenes controlling quantitative traits, a fulllinkage map is not always required toidentify genes and associate them withmarkers when the target trait is regulatedby major genes.

Principles of linkage map construction.To construct a linkage map, first potentialparental lines are screened usingmolecular markers (one or a combinationof molecular markers mentioned earlier)to identify DNA polymorphisms betweenthe two. Once a suitable number ofmarkers has been identified, they areused to determine the allelic compositionfor all genotypes in the segregatingpopulation. The segregation of a markerin a given population depends on the typeof population. Segregation ratios arebased on Mendel’s first law ofindependent assortment. In an F2population, a dominant marker shouldsegregate 3:1, whereas a codominantmarker that would allow theidentification of heterozygotes shouldsegregate 1:2:1. If a recombinant inbredline (RIL) or a doubled haploid (DH)population is used, segregation ratiosshould be 1:1 irrespective of whether themarker is dominant or co-dominant. AChi-square test can be performed todetermine the nature of the segregation ofa marker.

After a number of markers has beengenotyped across the population, thelinkage among them is determined takinginto account Mendel’s second law ofindependent assortment (Figure 7). Table 3presents the expected segregation for twounlinked loci in different populations. Ifthe two loci are linked, significantdeviations from the expected segregationratios can be observed and confirmedstatistically by performing Chi-squaretests. If the linkage is confirmed, thefrequency of recombination between thetwo loci can be calculated to establish thegenetic distance between the two


38

markers. When a large number ofmarkers has been screened across apopulation, it is not feasible to useconventional statistical parameters suchas Chi-square tests or computingrecombination frequencies to establishlinkage among markers. Furthermore,the presence of one recombination eventbetween two adjacent loci woulddecrease the probability of anotherrecombination event in adjacent loci.Computer programs that take intoaccount all statistical parameters areavailable for use in linkage mapconstruction.

Gene/QTL identificationOnce the genetic map has beenconstructed, the next step is to find out ifmarker segregation within the populationis associated with segregation of thetarget trait(s). Effective mapping studiesaimed identifying molecular markersassociated with target traits depend ontwo types of data: laboratory data onmarker segregation and field data on thesegregation of the trait(s). For example,in a population of RILs segregating fordisease resistance, if a marker segregatesin such a way that when a particularallele of the marker is present in a line,and that line shows disease resistance, astrong association between the markerallele and the trait can be inferred. Inother words, the molecular marker hastagged the gene involved in theexpression of resistance.

Phenotypic evaluation. Regardless ofthe type of data being evaluated, thequality of the phenotypic data is crucialfor the success of gene/QTL analysis,since laboratory data on markersegregation within a population has to becorrelated with field data to identify theQTLs. Therefore, phenotypic evaluation,whether in the field, greenhouse, growth-chamber, or laboratory, must be carefullyplanned and conducted with adequatereplications to reduce the error. Toevaluate a segregating population in the

Figure 6. (a) Simplified scheme showing wheat with only 3 of its 21 pairs of homologouschromosomes. By repeatedly backcrossing the donor variety onto the recipient monosomicand selecting monosomic progeny after each backcross, a line monosomic for chromosome 3is developed in the donor variety. (b) Chromosome 3 of a donor variety is introduced bybackcrossing into the recipient variety. Initially the donor variety is crossed onto achromosome 3 monosomic line in the recipient variety. Monosomic progeny are selectedafter each cross and backcrossed repeatedly onto the recipient monosomic to reconstituteits genetic background.

Intervarietal chromosome substitution

1 2 3 1 2 3

F1

B1

Bn

SELF

X

X

X

b

Development of a monosomic series

1 2 3 1 2 3

F1

B1

Bn

X

X

X

a

J.-M. RIBAUT ET AL.

39

Figure 7. Expected genotypic classes for two codominant independent lociin an F2 population.

Table 3. Expected allelic segregation of two unlinked loci in various populations.

Dominant CodominantPopulation markers markers

BC1F1 1:1:1:1 1:1:1:1

F2 9:3:3:1 1:2:1:2:4:2:1:2:1

RIL 1:1:1:1 1:1:1:1

DH 1:1:1:1 1:1:1:1

field, we strongly recommend the use offield designs that include severalreplications, an alpha (0,1) lattice being themost commonly used to produce phenotypicdata for QTL analysis. The accuracy of theprotocols used in data collection is also veryimportant.

Except for monogenic traits, phenotypicevaluation should not be conducted on asingle-plant basis, but always on severalplants of a family representing a givengenotype, whether it be an F3 familyrepresenting an F2 genotype or severalplants of a RIL family which, by definition,have the same genotype. As an example, toidentify genes involved in the expression ofosmotic potential in a segregatingpopulation for drought tolerance, thefollowing must be determined:

• at what vegetative stage should leaftissue be harvested;

• how to harvest genotypes at the samevegetative stage when they mightsegregate for precocity;

• what time of day to harvest, sincetemperature changes will induce changesin plant-water status;

• what kind of plant tissue is most suitablefor the analysis;

• how to harvest many samples in a shortperiod of time;

• how to avoid changes in water content ofthe tissue sample between harvest andcell sap extraction;

• how to extract cell sap from differenttissue samples in a reproducible fashion;

• how often the osmometer should becalibrated to obtain reproduciblemeasurements;

• how many replicates are needed toensure, for example, increased accuracy.

Depending on the type of physiological test(e.g., hormone quantification), the numberof samples that can be reasonably analyzedmight be limited. In this case, the size of thesegregating population has to be carefullyconsidered. In a small population (e.g.,fewer than 100 F2 plants or than 60 RILs),only genomic regions expressing a largepercentage of phenotypic variance might bereliably identified. The complexity of taking


40

accurate measurements of mostphysiological traits makes identifyingmarkers linked to genes involved inphysiological responses more attractive.

Bulked segregant analysis (BSA).When a trait is regulated by a majorgene, bulked segregant analysis (BSA)might be useful for identifying thelocation of the target genomic region(Michelmore et al., 1991). Bulkedsegregant analysis has been used toidentify DNA sequences linked to atarget region in several crop species(Michelmore et al., 1991; Eastwood etal., 1994). Any segregating populationoriginating from a single cross can beused for BSA. Bulked segregants can bemade for any locus or genomic regiononce the segregating population hasbeen constructed.

Phenotypic distribution within asegregating population should indicatewhether the trait is regulated by one or afew major genes (e.g., 1:3 distribution)or several minor genes (normaldistribution). When the target trait isregulated by major gene(s), the two tailsof the distribution can be safelyidentified through careful phenotypicselection; this material would besuitable for BSA (Figure 8).

The BSA method involves screeningtwo pooled DNA samples fromindividuals with contrasting traits from asegregating population originating froma single cross. Each pool, or bulk,contains individuals selected to haveidentical putative genotypes for aparticular genomic region (target locusor region) but also random genotypes atloci unlinked to the selected region.Therefore, the two bulked DNA samplesdiffer genetically only in the selectedregion and present random allelicsegregation for all other loci. Forexample, if markers are to be identifiedfor disease resistance, equal amounts ofDNA from the 5-10 most resistantindividuals are bulked and taken as a

“resistant” pool. Similarly, DNA fromthe 5-10 most susceptible individualsfrom the same population is bulked andtreated as the “susceptible” pool. Thetwo pools contrasting for the trait arethen analyzed to identify markers thatdistinguish them. Markers that arepolymorphic between the pools wouldmost likely be genetically linked to theloci determining the trait used toconstruct the pools.

Once polymorphic markers for the twopools have been identified, the linkagebetween a marker and the target locus isconfirmed and quantified using thesegregating population from which thebulks were generated. It is oftennecessary to find the marker’s genomic

location, which also establishes thegenomic location of factors controllingthe trait of interest. If RFLPs or SSRsare used in the BSA, their genomiclocation is often known. However, ifmarkers such as RAPDs or AFLPs areused, their genomic locations have to beestablished using several approaches. Ifthese markers segregate in anotherpopulation for which a linkage map hasalready been developed, map locationscan be established using this secondarypopulation. Using single-chromosomeintervarietal substitution lines is anotheralternative.

The last and most tedious alternative isto develop a complete linkage map forthe cross from which the bulks were

Resistant Segregating Susceptible

F2

Parent A Parent B F1 Resistant bulk Susceptible bulk

BSA

Resistant Susceptible

RILs

Figure 8. Bulk segregant analysis.

J.-M. RIBAUT ET AL.

41

generated. If the linkage between themarker of interest and the target gene isconfirmed using the population, themarker may be used for marker-assistedselection. The success of the approachwill depend on 1) the genetic divergencebetween the parents in the target region,2) the accuracy of phenotypicobservations, and 3) the number ofmajor genes involved in the expressionof the target trait.

Identification of QTLs. For polygenictraits, phenotypic distribution within asegregating population is usuallynormal, which implies that several genesare involved in the expression of thetarget trait, each of them expressing aportion of total phenotypic variance.Bulked segregant analysis is notnormally appropriate when target trait(s)are governed by several genes; in thiscase, constructing a complete linkagemap is preferable. If the linkage map isconstructed using DNA extracted fromF2 plants, field evaluation can beconducted on F3 families derived byself-pollinating each individual F2 plant.Once the linkage map is constructed andphenotypic evaluation conducted,phenotypic correlations are commonlyused to associate markers with traits andto genetically dissect complex traits intoMendelian factors. Computer programsare used to assess the correlationbetween phenotypic values of differentgenotypes within the segregatingpopulation and the allelic composition ateach loci used to produce the linkagemap. If this correlation is statisticallysignificant at a given locus, the genomicregion is assumed to be involved in theexpression of the phenotypic trait(Figure 9). The statistical packages usedin this procedure can be as simple as anF test or as complex as compositeinterval mapping.

It is not our purpose in this chapter todescribe in detail the differentmathematical approaches to QTLidentification. However, commonly usedapproaches can be divided into threecategories: simple correlation test,simple interval mapping (SIM, Landerand Botstein, 1989), and compositeinterval mapping (CIM, Zeng, 1994),and will be described briefly.

The simplest test to identify if there isany statistically significant associationbetween the markers and a phenotypicdata is a t –test (2 variables:homozygous parent 1 and homozygousparent 2) or an analysis of variance (Ftest, 3 variables: homozygous parent 1,homozygous parent 2 andheterozygous). The statistical test isconducted on each molecular markerindependently to identify markersassociated with the trait.

In SIM, which employs computerprograms such as mapmaker/QTLsoftware (Lander and Botstein, 1989)using mixture models and maximumlikelihood techniques, a test value canbe attributed to each cM on the linkagemap. Therefore, a QTL peak (i.e., thepoint where the highest level ofstatistical significance is obtained) canbe identified at any point on the map,

not at a specific marker position, as in theprevious approach. The SIM procedurediffers from the previous approach in thatit considers more than one marker at atime. Although SIM is a more “integrated”method than a simple correlations test, itsmajor limitation is that it does not identifyQTLs when they are linked together. Fortwo linked QTL in coupling phase(favorable genetic contribution from thesame parental line at two QTLs), SIM willidentify only one QTL covering a largechromosome segment and overestimate itsimpact on trait expression. When twolinked QTLs are in repulsion (favorablegenetic contribution from differentparental lines at two QTLs), SIM may notidentify any of the QTLs.

Composite interval mapping, the thirdapproach, takes into account thelimitations of SIM (mentioned above). Itconsiders markers as cofactors and hasthree phases:

• Unlinked markers close to QTL peaks(one marker per QTL) are identifiedusing SIM analysis.

• The analysis is conducted again, butusing the identified markers ascofactors to reduce residual variationthroughout the genome, therebyeliminating false positive QTLs andidentifying “new”minor QTLs.

mean 1 mean 2

If

If

no QTL

QTL!

=

≠

RFLPpatternsat locus

A

Figure 9. Illustration of a t-test for QTL detection at one RFLP marker.


42

• At all intervals throughout thegenome, markers flanking a testedinterval are used as cofactors toblock the effects of possible QTLslinked to the interval of interest. Thechosen distance between the testedinterval and a cofactor is defined asa “window” for testing for thepresence of a QTL.

When conducted on a detailed linkagemap, CIM allows more preciseidentification of a QTL in the genomeand better identification of coupledQTLs. In addition, it allows analysis ofindividual field data sets, as well asanalysis of combined phenotypic datafrom different environments (locations,years, or treatments), and therefore alsoevaluates the QTL by environmentinteraction (Q x E). However, accurateevaluation of Q x E interactions, whichrequires top quality molecular andphenotypic data, remains a majorconstraint for validating marker-assisted selection (MAS) experiments.Moreover, even with new approacheslike CIM, there remain clear limitationsin evaluating epistatic effects betweendifferent regions of a genome.

Characterizing a QTL involves findingits precise location on a linkage map,determining the percentage ofphenotypic variance expressed by eachQTL independently or by several QTLstogether, and quantifying the geneticeffect (dominance and additivity) perQTL. Usually, a QTL is defined asmajor when it explains more then 30%of phenotypic variation. Althoughmajor QTLs may be involved in theexpression of disease resistance, grainquality, or tolerance to abiotic stressessuch as aluminum tolerance, they donot usually regulate the expression ofvery complex traits such as yield inwater limited environments.

Progress in wheat moleculargeneticsUse of molecular markers for mappingand gene identification. Progress ingene identification and markerdevelopment has been slow in wheatdue to its hexaploid nature and the largesize of its genome. However, in therecent past, a significant number ofgenes involved in various functionshave been mapped to specific wheatchromosomal regions. Characterizinggenes that control flowering in wheathas benefited from chromosomemanipulations involving aneuploidy aswell as molecular markers.

Using intervarietal chromosomesubstitution lines and single-chromosome recombinant linepopulations, genes controllingvernalization response Vrn1 and Vrn3have been located on the long arms ofchromosomes 5A and 5D, respectively(Law et al., 1976), and VrnB1 onchromosome 5B (Zhuang, 1989).Similar procedures have been utilized toidentify genes controlling photoperiodresponse (Ppd genes) (Worland andLaw, 1986). Plant height, important fordetermining adaptation and yield inwheat, is genetically complex; so farabout 21 genes have been identified tobe associated with this trait (McIntosh etal., 1995). A microsatellite marker hasrecently been developed that is linked toRht8 (Korzun et al., 1998).

Efforts have been made to geneticallydissect complex physiological traitsassociated with drought tolerance suchas accumulation of abscisic acid in riceand to investigate possible relationshipbetween rice and wheat homeologousloci controlling abscisic acidaccumulation (Quarrie et al., 1997).Using single-chromosome recombinantline populations and mapping, Quarrieet al. (1994) located a genetic factorcontrolling drought-induced abscisic

acid production on the long arm ofchromosome 5A in wheat. Moleculargenetic tools have also been used tostudy complex traits such ascarbohydrate metabolism and theassociation between abscisic acidconcentration and stomatalconductance (Prioul et al., 1997).Comparative RFLP mapping incultivated and wild wheat (Triticumdicocoides) has led to the identificationof molecular markers associated withresistance to the herbicidechlorotoluron which is a selectivephenylurea herbicide (Krugman et al.,1997). A list of wheat genes that controlvarious physiological and agronomicparameters that have been identifiedwith the use of molecular markers ispresented in Table 4.

The existence of numerous sets ofwheat near-isogenic lines (NILs)differing in the presence/absence of aresistance allele for various biotic stressfactors (diseases and pests) hasfacilitated the mapping of genes forwhich such lines exist. Large numbersof genes conferring disease or pestresistance have been identified andassociated with molecular markers(reviewed in Hoisington et al., 1999).When the chromosomal location of aparticular gene is known from previousgenetic studies but no NILs areavailable, markers mapped to thatchromosome (Anderson et al., 1992)can still be used to score parental linesfor polymorphisms, construct a single-chromosome map, and determinewhich markers are close to the gene ofinterest. This strategy was followed byDubcovsky et al. (1996) to tag theKna1 locus in wheat, which isresponsible for higher K+/Na+accumulation in leaves, a traitcorrelated with higher salt tolerance.

In wheat, bulked segregant analysis,initially used mostly with RAPDs, cannow be used with any type of marker

J.-M. RIBAUT ET AL.

43

including AFLPs (Goodwin et al., 1998;Hartl et al., 1998), which have theadvantage that a high number of DNAfragments can be amplified with oneprimer combination. Also, with AFLPsthe problem of highly repetitive DNA isovercome by using methylation sensitiveendonucleases such as PstI and SseI.

Many genes that have been tagged withmolecular markers in wheat have beenintrogressed from alien species(Hoisington et al., 1999). In the case oftranslocations from wheat’s wildrelatives known to carry genes foragronomically important traits, markerscan be successfully established due to thehigh level of polymorphisms between thewheat and introgressed genome and thelow level of recombination between thetranslocated segment and thecorresponding wheat chromosomes.

Mapping QTLs in wheat. Utilizing abase map and linkage data from a rangeof other segregating wheat, rye, andbarley populations, a consensus map withmore than 1000 data points has been

developed (Gale et al., 1995). Thisdetailed linkage map has confirmed thatthe order of genetic loci across the A, B,and D genomes has been conserved (Galeet al., 1995).

A RIL mapping population developedutilizing ‘Opata 85’ and a synthetichexaploid from CIMMYT has been usedextensively in mapping and genomerelationship studies (Van Deynze et al.,1995; Nelson et al., 1995a, b, c). Thegenetic map of this population,developed by the International TriticeaeMapping Initiative (ITMI), contains over1000 RFLP loci. Two other publishedmaps are available in wheat (Liu andTsunewaki, 1991; Cadalen et al., 1997).Linkage maps in wheat have confirmedevolutionary chromosomal translocationrearrangements involving chromosomes2B, 4A, 5A, 6B, and 7B, which werebased on cytological evidence, and haveestablished synteny among closelyrelated grass species such as rice, maize,oats, and wheat (Ahn et al., 1993; Devoset al. 1994; Van Deynze et al., 1995;Borner et al., 1998).

The low number of quantitative traitsdissected into their QTLs in wheat is areflection of the focus on simply inheritedtraits and the difficulty of buildingcomprehensive linkage maps. Given thatthe ITMI map is one of the densest and thepopulation from which it was developed issegregating for a number of traits, it hasbeen used to map important traits andseveral major genes. Known genes includevernalization (Vrn1 and Vrn3), red-coleoptile (Rc1), kernel hardness (Ha), andpowdery mildew (Pm1 and Pm2) genes(Nelson et al., 1995a), as well as genesconferring and suppressing leaf rustresistance (Nelson et al., 1997).Quantitative trait loci have been identifiedfor kernel hardness (Sourdille et al., 1996),Karnal bunt (Nelson et al., 1998), and tanspot (Faris et al., 1997).

Research on developing molecular markersfor traits associated with drought tolerancein wheat started recently at CIMMYT. ARIL population is being utilized to identifygenomic regions associated with a range ofphysiological parameters controllingdrought tolerance.


Table 4. Genes identified and mapped with molecular markers for physiological and agronomic traits in wheat.

Traits Genes Species Markers Chromosomes References

Physiological and agronomicPreharvest sprouting QTL Triticum aestivum RFLP Anderson et al., 1993Vernalization Vrn1 RFLP 5AS Galiba et al., 1995; Korzun et al., 1997

Kato et al., 1998Vrn3 RFLP 5DS Nelson et al., 1995a

Photoperiod response Ppd1 T. aestivum RFLP 2DS Worland et al., 1997Ppd2 T. aestivum RFLP 2BS Worland et al., 1997

Dwarfing Rht8 SSR 2DS Korzun et al.,1998Rht12 SSR 5AL Korzun et al., 1997

Cadmium uptake RAPD Penner et al., 1995Aluminum tolerance Alt2 RFLP RFLP 4D 4DL Luo and Dvorak, 1996; Riede and

Anderson, 1996Drought induced ABA RFLP 5A Quarrie et al., 1994Na+/K+ discrimination Kna1 T. aestivum RFLP RFLP 4D 4DL Allen et al., 1995; Dubcovsky et al., 1996

QualityKernel hardness Ha Hn and QTL RFLP RFLP 5D 5DS, 2A, 2D, 5B, 6D Nelson et al., 1995a; Sourdille et al., 1996Grain protein QTL T. turgidum RFLP 4BS, 5AL, 6AS, 6BS, 7BS Blanco et al., 1996LMW glutenins T. turgidum 1B D’ovidio and Porceddu, 1996HMW glutenins Glu -D1 -1 T. aestivum ASA 1DL D’ovidio and Anderson, 1994Flour color RFLP/AFLP 7A Parker et al., 1998

44

Marker-Assisted SelectionBased on information provided bymapping and genetic dissection of agiven trait, the efficiency of usingmarker-assisted selection (MAS) can beevaluated and a suitable experimentdesigned. Molecular markers can be usedfor 1) tracing a favorable allele(including recessive) across generations,and 2) identifying the most suitableindividual among segregating progeny,based on allelic composition across partor the entire genome. For tracingfavorable alleles, it is critical to havemolecular markers very close to thegene of interest.

Markers can sometimes be identifiedwithin genes that have been sequenced.For example, in maize, Opaque-2 mutantgene, which confers high lysine andtryptophan in the kernel, has beensequenced (Schmidt et al., 1990).Microsatellite sequences have beenidentified within the gene, and primers toamplify the microsatellites have beendesigned. This is optimal for tracingfavorable alleles, since the marker islocated within the gene sequence itselfand co-segregates with the target gene.

However, in most cases, especially forpolygenic traits, target genes have notbeen characterized at the molecular level.Therefore, the selected genomic regionsinvolved in a marker-assisted selectionexperiment are chromosome segments orQTLs, identified as described previously.Two polymorphic DNA markers flankingthe QTL region are required to follow thepresence of elite alleles within theselected QTLs. When a QTL represents alarge chromosome segment (more than20cM), there should be a marker withinthe QTL to eliminate genotypespresenting a double recombinationbetween the two flanking markers.

Depending on the nature of the genomicregion (cloned gene, major or minorQTLs) involved in the expression of a

target trait and the number of selectedQTLs or genomic regions that need to bemanipulated, several MAS schemes canbe considered. Optimal strategies, interms of the breeding approaches and thetype of the molecular marker to be usedefficiently in a MAS experiment, shouldalso consider, apart from the technicalconstraints, the objectives of theexperiments and the resources available.A favorable allele can be introgressedusing the marker in target germplasmthrough backcrossing (BC), or can beselected in a segregating populationwhatever the level of recombinationpresent (e.g., F2 or advancedpopulations). In a“marginal” MASapproach, quite important in open-pollinated crops, the most suitableparental lines for developing newmaterials through recurrent selection areidentified.

Marker-assisted selectionstrategiesMAS for parental selection. Molecularmarkers can be used to genotype a set ofgermplasm, and the data used to estimatethe genetic distance among evaluatedmaterials. The degree of heterosisbetween lines can be predicted based ongenetic distance. Moreover, germplasmcan be characterized at specific lociknown to be involved in the expressionof a target trait, such as leaf rust inwheat, provided the germplasm has beenwell characterized phenotypically andmolecular polymorphisms identified in aset of materials that differ in theexpression of leaf rust resistance. Thisallows the identification of linespossessing the most suitable alleliccomposition for leaf rust at different loci.

Fingerprinting of potential parental linescan be very informative for breedersplanning to make new segregatingcrosses. Although the informationprovided by molecular markers may notidentify the best cross, it does helpreduce the number of needed crosses. If

by combining phenotypic and genotypicdata the number of crosses can bereduced by half, this will greatly increasebreeding efficiency.

The backcross-MAS approach. Marker-assisted selection using backcrossing(BC-MAS) has been used extensively—for example, to introgress target alleleswhen dealing with cloned genes or majorQTLs. In a BC scheme, an elite allele attarget loci is transferred from a donor toa recipient line. The use of DNAmarkers, which permit genotyping theprogeny at each cycle, increases thespeed of the selection process (Tanksleyet al., 1989). An excellent example of theintrogression of an elite allele at a targetgene is the introgression of a transgene intoelite maize inbreds (Ragot et al., 1994).

The following parameters must beconsidered in planning and executing aBC-MAS experiment: the number oftarget genomic regions involved in theselection, the size of the populationscreened at each cycle, the number ofgenotypes selected at each cycle, and thelevel of line conversion desired. Theexpected level of conversion is related tothe number and distribution of DNAmarkers at non-selected loci, and to therecombination frequency between thetarget loci and the two flanking markers.All these parameters interactively havean impact on the number of cyclesrequired to do BC-MAS. With the recentdevelopment of reliable PCR-basedmarkers and single-nucleotidepolymorphisms (SNP; Gilles et al.,1999), the capacity for screening largepopulations has been substantiallyimproved (Ribaut et al., 1997).

In view of the non-linear relationshipbetween reduction of the donor genomecontribution at non-selected loci fordifferent population sizes, in a BC-MASexperiment the number of target genes tobe introgressed must first be defined.This facilitates calculating the

J.-M. RIBAUT ET AL.

45

population size to be screened at eachcycle, considering a target effectivepopulation size (defined as the numberof individuals with favorable alleles atthe target genes from which selectionwith markers can be carried out on therest of the genome at non-target loci) of50-100 genotypes. The effectivepopulation size and number ofgenotypes heterozygous at the target lociare essential for determining theselection model. Once the effectivepopulation size is defined, the desiredrecombination frequency betweenflanking markers and the target geneshould be determined, as well as thenumber of genotypes selected at eachcycle, based on the objectives andconstraints of each project. Followingthis strategy, the number of BCsrequired to achieve the introgression canbe easily predicted based on simulations(Frish et al., 1999).

In case of limited resources, or forconcomitant introgression into a largenumber of recipient lines, as is oftennecessary in a breeding program,including an MAS step at an advancedBC cycle should be considered.Independently of the BC-MAS schemeconsidered, it is critical to make someeffort to identify the most convenient setof markers.

New MAS strategiesThe limitation of MAS for polygenictrait improvement. Despite the successof BC-MAS in certain applications, ithas limited utility when several QTLsmust be manipulated concomitently.Quantitative traits are difficult tomanipulate due to their geneticcomplexity, principally the number ofgenes involved in their expression andepistatic interactions between genes (seeRibaut and Hoisington, 1998, for areview). Since several genes areinvolved in the expression of polygenictraits, they generally have smallerindividual effects on plant phenotype.

This implies that several regions, orQTLs, must be manipulated at the sametime to have a significant impact, andthat the phenotypic effect of individualregions is not easily identified. Fieldtrials repeated over multiple years arerequired to accurately characterize QTLeffects and evaluate their stability acrossenvironments. Epistatic interactions caninduce a skewed evaluation of QTLeffects per se, and if all the genomicregions involved in the interactions arenot incorporated in the selection scheme,they can bias the MAS. In conclusion,although QTL identification hasimproved significantly, currently usedMAS strategies have their limitations(especially related to cost effectiveness)and new ones should be considered.

Single large-scale MAS. In thisproposed new approach, the selection ofsuitable parents and development of newlines are overseen by plant breeders, andDNA markers are used at an early stageof recombination to fix alleles at selectedgenomic regions (Ribaut and Betran,1999). The MAS step, conducted onlyonce, is based on the use of reliablePCR-based markers on large segregatingpopulations derived from crossesbetween elite lines. Parental lines mustbe outstanding for the target trait and/orenvironment, and should have goodallelic complementarity.

The new strategy offers two majoradvantages: 1) favorable alleles selectedfor improving a specific trait are derivedfrom two or more sources of eliteparental materials in a complementaryscheme, disregarding the “recipient/donor” line concept, and 2) specificgenomic segments are fixed only atregions previously identified for eachdonor line in the selection scheme, withno selection pressure applied outside thetargeted genomic regions. This ensuresgood allelic variability in the rest of thegenome for future line developmentunder various conditions. This approach

is also relevant for pyramiding favorablealleles at cloned genes or major QTLs innew germplasm.

Pedigree MAS. This approach isespecially relevant for crops such aswheat, where pedigrees of elitegermplasm are known. Fingerprintingelite wheat materials must be conductedin a set of lines actively used in thebreeding program, and in elite materials tobe released in subsequent years.Fingerprinting data may be combinedwith phenotypic data collected duringdifferent selection cycles to identifyalleles favorable for traits of interest. Forexample, if an elite line contains allelesfor yield performance in a targetenvironment, their frequency should behigher than the expected randomfrequency in offspring derived from thiselite parental line. This shift in allelicfrequency reflects phenotypic selection bybreeders and may be identified bycomparing fingerprinting data of bothparents and their offspring. Once thefavorable alleles are identified, DNAmarkers closely linked to the targetgenomic regions can be used to acceleratefixation of favorable alleles in the nextselection step: a new set of elite materials(offspring 1) to derive the next set of elitelines (offspring 2). Such MAS willprobably be most efficient whenconducted on F2 or F3 segregatingpopulations.

References andSuggested Reading

Ahn, S., Anderson, J.E., Sorrells, M.E., and Tanksley,S.D. 1993. Homoeologous relationships of rice,wheat and maize chromosomes. Molec and GenGenet 241:483-490.

Allen, G.J., Jones, R.G.W., and Leigh, R.A. 1995.Sodium transport measured in plasmamembrane vesicles isolated from wheatgenotypes with differing K+/Na+

discrimination traits. Plant Cell andEnvironment 18:105-115.

Anderson, J.A., Ogihara, Y., Sorrells, M.E., andTanksley, S.D. 1992. Development of achromosomal arm map for wheat based on RFLPmarkers. Theor Appl Genet 83:1035-1043.


46

Anderson, J.A., Sorrells, M.E., and Tanksley, S.D.1993. RFLP analysis of genomic regionsassociated with resistance to preharvestsprouting in wheat. Crop Sci 33:453-459.

Barrett, B.A., and K.K. Kidwell. 1998. AFLP-based genetic diversity assessment amongwheat cultivars from the Pacific Northwest.Crop Sci 38:1261-1271.

Bohn, M., Friedrich, H., and Melchinger, H.E.1999. Genetic similarities among winterwheat cultivars determined on the basis ofRFLPs, AFLPs and SSRs and their use forpredicting progeny variance. Crop Sci39:228-237.

Börner, A., Korzun, V., and Worland, A.J. 1998.Comparative genetic mapping of lociaffecting plant height and development incereals. Euphytica 100:245-248.

Blanco, A., Degiovanni, C., Laddomada, B.,Sciancalepore, A., Simeone, R., Devos, K.M.,and Gale, M.D. 1996. Quantitative trait lociinfluencing grain protein content in tetraploidwheats. Plant Breeding 115:310- 316.

Cadalen, T., Boeuf, C., Bernard, S., and Bernard,M. 1997. An intervarietal molecular markermap in Triticum aestivum L. Em. Thell. andcomparison with a map from a wide cross.Theor Appl Genet 94:367-377.

Chao, S., Sharp, P.J., Worland, A.J., Warham, E.J.,Koebner, R.M.D., and Gale, M.D. 1989.RFLP-based genetic maps of wheathomoeologous group 7 chromosomes. TheorAppl Genet 78:495-504.

D’Ovidio, R., and Porceddu, E. 1996. PCR-basedassay for detecting 1B-genes for lowmolecular weight glutenin subunits related togluten quality properties in durum wheat.Plant Breeding 115:413-415.

D’Ovidio, R., and Anderson, O.D. 1994. PCRanalysis to distinguish between alleles of amember of a multigene family correlated withwheat bread-making quality. Theor ApplGenet 88:759-763.

Devos, K.M., Atkinson, M.D., Chinoy, C.N., Liu,C., and Gale, M.D. 1992. RFLP based geneticmap of the homeologous group 3chormosomes of wheat and rye. Theor ApplGenet 83:931-939.

Devos, K.M., Chao, S., Li, Q.Y., Simonetti, M.C.,and Gale, M.D. 1994. Relationships betweenchromosome 9 of maize and wheathomeologous group 7 chromosomes.Genetics 138:1287-1292.

Devos, K.M., Dubcovsky, J., Dvorák, J., Chinoy,C.N., and Gale, M.D. 1995. Structuralevolution of wheat chromosomes 4A, 5A, and7B and its impact on recombination. TheorAppl Genet 91:282-288.

Dubcovsky, J., Santa María, G., Epstein, E., Luo,M.-C., and Dvorák, J. 1996. Mapping of theK+/Na+ discrimination locus Kna1 in wheat.Theor Appl Genet 92:448-454.

Eastwood, R.F., Lagudah, E.S., and Appels, R.1994. A direct search for DNA sequenceslinked to cereal cyst nematode resistancegenes in Triticum tauschii. Genome37:311-319.

Endo, T.R., and Gill, B.S. 1996. The deletion stocksof common wheat. J. Hered 87:295-307.

Fahima, T., Roder, M.S., Grama, A., and Nevo, E.1998. Microsatellite DNA polymorphismdivergence in Triticum dicoccoidesaccessions highly resistant to yellow rust.Theor Appl Genet 96:187-195.

Faris, J.D., Anderson, J.A., Francl, L.J., andJordahl, J.G. 1997. RFLP mapping ofresistance to chlorosis induction byPyrenophora tritici-repentis in wheat. TheorAppl Genet 94:98-103.

Frish, M., Bohn, M., and Melchinger, A.E. 1999.Comparison of selection strategies formarker-assisted backcrossing of a gene. CropSci 39(5) 1295-1301.

Gale, M.D., Atkinson, M.D., Chinoy, C.N.,Harcourt, R.L., Jia, J., Li, Q.Y., and Devos,K.M. 1995. Genetic maps of hexaploidwheat. In: Li, Z.S. and Xin, Z.I. (eds.).Proceeding of the 8th International WheatGenetics Symposium. China AgriculturalScientech Press, Beijing, China. pp. 29-40.

Galiba, G., Quarrie, S.A., Sutka, J., Morgounov,A., and Snape, J.W. 1995. RFLP mapping ofthe vernalization (Vrn1) and frost resistance(Fr1) genes on chromosome 5A of wheat.Theor Appl Genet 90:1174-1179.

Gilles, P.N., Wu, D.J., Foster, C.B., Dillon, P.J., andChanock, S.J. 1999. Single nucleotidepolymorphic discrimination by an electronicdot blot assay on semiconductor microchips.Nature Biotech 17:365-370.

Goodwin, S.B., Hu, X., and Shaner, G. 1998. AnAFLP marker linked to a gene for resistanceto Septoria tritici blotch in wheat.Proceedings of the 9th International WheatGenetics Symposium, Saskatoon,Saskatchewan, Canada, 2-7 August, 1998.Vol. 1 edited by and v.2-4 compiled by A.E.Slinkard 3:108-110.

Hartl, L., Mori, S., and Schweizer, G. 1998.Identification of a diagnostic molecularmarker for the powdery mildew resistancegene Pm4b based on fluorescently labelledAFLPs. Proceedings of the 9th InternationalWheat Genetics Symposium, Saskatoon,Saskatchewan, Canada, 2-7 August, 1998.Vol. 1 edited by and v.2-4 compiled by A.E.Slinkard 3:111-113.

Hieter, P., and Boguski, M. 1997. Functionalgenomics: It’s all how you read it. Science278:601-602.

Hoisington, D., Khairallah, M., Reeves, T., Ribaut,J.-M., Skovmand, B., Taba, S., andWarburton, M. 1999. Plant genetic resources:What can they contribute toward increasingcrop productivity? Proceedings NationalAcademia of Science 96:5937-5943.

Jia, J., Devos, K.M., Chao, S., Miller, T.E., Reader,S.M., and Gale, M.D. 1994. RFLP-basedmaps of the homoeologous group-6chromosomes of wheat and their applicationin the tagging of Pm12, a powdery mildewresistance gene transferred from Aegilopsspeltoides to wheat. Theor Appl Genet92:559-565.

Kato, K., Miura, H., Akiyama, M., Kuroshima,M., and Sawada, S. 1998. RFLP mapping ofthe three major genes, Vrn1, Q and B1 onthe long arm of chromosome 5A of wheat.Euphytica 101:91-95.

Kimber, G., and Athwal, R.S. 1972. Areassessment of the course of evolution ofwheat. Proc Natl Acad Sci (USA)69:912-915.

Kimber, G., and Feldman, M. 1987. Wild wheat.An introduction. Special report 353, Collegeof Agriculture, Univ. Missouri, Columbia,USA.

Korzun, V., Röder, M.S., Ganal, M.W., Worland,A.J., and Law, C.N. 1998. Genetic analysisof the dwarfing gene (Rht8) in wheat. Part I.Molecular mapping of Rht8 on the short armof chromosome 2D of bread wheat (Triticumaestivum L.). Theor Appl Genet96:1104-1109.

Korzun, V., Röder, M., Worland, A.J., and Börner,A. 1997. Intrachromosomal mapping ofgenes for dwarfing (Rht12) andvernalization response (Vrn1) in wheat byusing RFLP and microsatellite markers.Plant Breeding 116:227-232.

Krugman, T., Levy, O., Snape, J.W., Rubin, B.,Korol, A., and Nevo, E. 1997. ComparativeRFLP mapping of the chlorotoluronresistance gene (Su1) in cultivated wheat(Triticum aestivum) and wild wheat(Triticum dicoccoides). Theor Appl Genet94(1)46-51.

Lander, E. S., and D. Botstein. 1989. MappingMendelian factors underlying quantitativetraits using RFLP linkage maps. Genetics121:185-199.

Langridge, P., and Chalmers, K. 1998. Techniquesfor marker development. Proceedings of the9th International Wheat GeneticsSymposium, Saskatoon, Saskatchewan,Canada, 2-7 August, 1998. Vol. 1 edited byand v.2-4 compiled by A.E. Slinkard1:107-117.

Law, C.N. 1966. The location of genetic factorsaffecting a quantitative character in wheat.Genetics 53:487-498.

Law, C.N., and Worland A.J. 1996. Intervarietalchromosome substitution lines in wheat-revisited. Euphytica 89:1-10.

Law, C.N., and A.J. Worland. 1973. Aneuploidy inwheat and its use in genetic analysis. PlantBreed. Inst. Annual Report for 1972.pp. 25-65.

Law, C.N., Worland, A.J., and Giorgi, B. 1976.The genetic control of ear emergence timeby chromosome 5A and 5D of wheat.Heredity 36:49-58.

Lemieux, B., Aharoni, A., and Schene, M. 1998.Overview of DNA chip technology.Molecular Breeding 4:277-289.

Litt, M., and Luty, J.A. 1989. A hypervariablemicrosatellite revealed by in vitroamplification of a dinucleotide repeat withinthe cardiac muscle actin gene. Am J HumGenet 44:397-401.

J.-M. RIBAUT ET AL.

47

Liu, Y., and Tsunewaki, K. 1991. Restrictionfragment length polymorphism (RFLP)analysis in wheat. II. Linkage maps of theRFLP sites in common wheat. Japan J.Genet 66:617-633.

Luo, M.C., and Dvorak, J. 1996. Molecularmapping of an aluminium tolerance locuson chromosome 4D of Chinese Springwheat. Euphytica 91:31-35.

McKewan, J.M., and Kaltsikes, P.J. 1970. Earlygeneration testing as a means of predictingthe value of specific chromosomesubstitutions into common wheat. Can JGenet Cytol 12:711-723.

McIntosh, R.A., Hart, G.E., and Gale, M.D. 1995.Catalogue of gene symbols for wheat. Proc.8th Wheat Genetics Symposium. Beijing,China. Li, Z.S. and Xin, Z.Y. (eds.). pp.1333-1500.

Michelmore, R.W., Paran, I., and Kesseli, R.V.1991. Identification of markers linked todisease-resistance genes by bulkedsegregant analysis: A rapid method to detectmarkers in specific genomic regions byusing segregating populations. Proc NatlAcad Sci 88:9828-9832.

Miller, T.E., Hutchinson, J., and Chapman, V.1982. Investigation of a preferentiallytransmitted Aegilops sharonensischromosome in wheat. Theor Appl Genet61:27-33.

Nelson, J.C., Singh, R.P., Autrique, J.E., andSorrells, M.E. 1997. Mapping genesconferring and suppressing leaf rustresistance in wheat. Crop Sci 37:1928-1935.

Nelson, J.C., Sorrells, M.E., Van Deynze, A.E.,Lu, Y.H., Atkinson, M., Bernard, M., Leroy,P., Faris, J.D., and Anderson, J.A. 1995a.Molecular mapping of wheat: major genesand rearrangements in homoeologousgroups 4, 5, and 7. Genetics 141:721-731.

Nelson, J.C., Van Deynze, A.E., Autrique, E.,Sorrells, M.E., Lu, Y.H., Merlino, M.,Atkinson, M., and Leroy, P. 1995b.Molecular mapping of wheat. homoeologousgroup 2. Genome 38:516-524.

Nelson, J.C., Van Deynze, A.E. Autrique, E.,Sorrells, M.E., Lu,Y.H., Negre, S., Bernard,M., and Leroy, P. 1995c. Molecularmapping of wheat homoelogous group 3.Genome 38:525-533.

Nelson, J.C., Autrique, J.E., Fuentes-Dávila, G.,and Sorrells, M.E. 1998. Chromosomallocation of genes for resistance to Karnalbunt in wheat. Crop Sci 38:231-236.

Olson, M., L. Hood, C. Cantor, and D. Botstein.1989. A common language for physicalmapping of the human genome. Science245:1434-1435.

Parker, G.D., Chalmers, K.J., Rathjen, A.J., andLangridge, P. 1998. Mapping loci associatedwith flour colour in wheat (Triticumaestivum L.) Theor Appl Genet97 (1-2):238-245.

Pathak, G.N. 1940. Studies in the cytology ofcereals. Indian J Genet 39:437-467.

Penner, G.A., Clarke, J., Bezte, L.J., and Leisle, D.1995. Identification of RAPD markers linkedto a gene governing cadmium uptake indurum wheat. Genome 38:543-547.

Prioul, J.L., Quarrie, S., Causse, M., and Vienne, D.1997. Dissecting complex physiological functionsthrough the use of molecular quantitative genetics.J Exp Bot 48(311)1151-1163.

Quarrie, S.A., Laurie, D.A., Zhu, J., Lebreton, C.,Semikhodskii, A., Steed, A., Witsenboer, H.,Calestani, C., and Zhu, J.H. 1997. QTLanalysis to study the association between leafsize and abscisic acid accumulation indroughted rice leaves and comparisons acrosscereals. Plant Mol Biol 35:155-165.

Quarrie, S.A., Gulli, M., Calestani, C., Steed, A.,and Marmiroli, N. 1994. Location of a generegulating drought-induced abscisic acidproduction on the long arm of chromosome5A of wheat. Theor Appl Genet 89:794-800.

Rafalski, J.A., and Tingey, S.V. 1993. Geneticdiagnostics in plant breeding: RAPDs,microsatellites and machines. Trends in Genet9:275-280.

Ragot, M., Biasiolli, M., Delbut, M.F., Dell’orco,A., Malgarini, L., Thevenin, P., Vernoy, J.,Vivant, J., Zimmermann, R., and Gay, G.1994. Marker-assisted backcrossing: apractical example. In: Techniques etUtilisations des Marqueurs. Les Colloques,Vol. 72. Berville, A. and Tersac, M. (eds.).Institut National de la RechercheAgronomique. pp. 45-56.

Ribaut, J.-M., Hu, X., Hoisington, D.A., andGonzález-de-León, D. 1997 Use of STSs andSSRs as rapid and reliable preselection toolsin a marker assisted selection backcrossscheme. Plant Molecular Biology Reporter15:154-162.

Ribaut, J.-M., and Hoisington, D.A. 1998. Marker-assisted selection: new tools and strategies.Trends in Plant Science 3:236-239.

Ribaut, J.-M., and Betrán, F.J. 1999. Single large-scale marker-assisted selection (SLS-MAS).Mol Breeding 5:531-541.

Riede, C.R., and Anderson, J.A. 1996. Linkage ofRFLP markers to an aluminium tolerancegene in wheat. Crop Sci 36:905-909.

Röder, M.S., Korzun, V., Wendehake, K., Plaschke,J., Tixier, M-H., Leroy, P., and Ganal, M.W.1998. A microsatellite map of wheat. Genetics149:1-17.

Schmidt, R., Burr, F.A., Aukerman, M.J., and Burr,B. 1990. Maize regulatory gene opaque-2encodes a protein with a “leucine-zipper”motif that binds to zein DNA. Proc Natl AcadSci 87:46-50.

Sears, E.R. 1953. Nullisomic analysis in commonwheat. Am Nat 87:245-252.

Sears, E.R. 1954. The aneuploids of commonwheat. Missouri Agric. Experimental StationResearch Bulletin 572:1-58.

Snape J.W., and Law, C.N. 1980. The detection ofhomologous chromosome variation in wheatusing backcross reciprocal monosomic lines.Heredity 45:187-200.

Sourdille, P., Perretant, M.R., Charmet, G., Leroy,P., Gautier, M.F., Joudrier, P., Nelson, J.C.,Sorrells, M.E., and Bernard, M. 1996.Linkage between RFLP markers and genesaffecting kernel hardness in wheat. TheorAppl Genet 93:580-586.

Talbert, L.E., Blake, N.K., Chee, P.W., Blake,T.K., and Magyar, G.M. 1994. Evaluation of«sequence-tagged-site» PCR products asmolecular markers in wheat. Theor ApplGenet 87:789-794.

Tanksley, S.D., Young, N.D., Paterson, A.H., andBonierbale, M.W. 1989. RFLP mapping inplant breeding: new tools for an old science.Biotech 7:257-264.

Tautz, D., and Renz, M. 1984. Simple sequencesare ubiquitious repetitive components ifeukaryotic genomes. Nucleic Acids Res12:4127-4137.

Van Deynze, A.E., Dubcovsky, J., Gill, K.S.,Nelson, J.C., Sorrells, M.E., Dvorák, J., Gill,B.S., Lagudah, E.S., McCouch, S.R., andAppels, R. 1995. Molecular-genetic mapsfor group 1 chromosomes of Triticeaespecies and their relation to chromosomes inrice and oat. Genome 38:45-59.

Vos, P., Hogers, R., Bleeker, M., Reijas, M., Van deLee, T., Hornes, M., Frijters, A., Pot, J.,Peleman, J., Kuiper, M., and Zabeau, M. 1995.AFLP: A new technique for DNAfingerprinting. Nucleic Acids Res 23:4407-4414.

Weber, J.L., and May, P.E. 1989. Abundant classof human DNA polymorphisms which canbe typed using the polymerase chainreaction. Am J Hum Genet 44:388-396.

Welsh, J., and M. McClelland. 1990.Fingerprinting genomes using PCR witharbitrary primers. Nucleic Acids Res18:7213-7218.

Williams, J.G.K., Kubelik, A.R., Livak, K.J.,Rafalski, J.A., and Tingey. S.V. 1990. DNApolymorphism amplified by arbitraryprimers are useful as genetic markers.Nucleic Acid Res 18:6531-6535.

Worland, A.J. 1998. Catalogue of monosomicseries. Proc. 7th International Wheat GeneticSymposium. Cambridge. pp. 1399-1403.

Worland, A.J., Börner, A., Korzun, V., Li, W.M.,Petrovíc, S., and Sayers, E.J. 1997. Theinfluence of photoperiod genes on theadaptability of European winter wheats. In:Braun, H.J. et al. (eds.). Wheat: Prospectsfor Global Improvement. pp. 517-526.

Worland, A.J., and Law, C. N. 1986. Geneticanalysis of chromosome 2D of wheat. 1. Thelocation of genes affecting height, daylengthinsensitivity, hybrid dwarfism and yellowrust resistance. Zeitschrift Pflan 96:331-345.

Zeng, Z.B. 1994. Precision mapping ofquantitative trait loci. Genetics 136:1457-1468.

Zhuang, J. 1989. The control of heading date bychromosome 5B of wheat. M. Phil thesis,Cambridge University, UK.


The goal of most breeding trials is toassess the performance of a number ofgenetically diverse breeding materials insuch a way that superior lines can beselected that perform better than localchecks under specific farming conditions.This must be done without bias underconditions that mimic the conditionswhere the material will be recommended.If the breeder is going to select materialfor a specific biotic or abiotic stress, hemust conduct the trial in an area that hasthis stress or condition. This will ensure ahigher rate of success than planting onlyunder favorable conditions that do notrepresent the target environment.

It is also important to manage breedingtrials as carefully as possible to minimizeexperimental error to be able to evaluatedifferences between materialsstatistically. Confounding factors need tobe kept to a minimum so that the breederhas confidence that his selections willstand up to testing under the clientsituation. This chapter will look at someof the factors that need to be consideredwhen planning breeding trials.

Choosing theExperimental Site

Selecting the experimental site is criticalfor success. If a breeder wishes to selectmaterials for salinity tolerance, he mustselect a site that represents the salinitysituation in the areas where the varietywill be released. His chances of successare increased by carefully selecting thesite for the nursery. This selection willtake into consideration the soil, climate,water regime, and biotic and abioticstresses that will be encountered in thetarget environment. Factors that need tobe considered include:

• Soil factors such as texture, pH,conductivity (salinity and alkalinity),and nutrient status. Soil texture willaffect soil physical properties and thepermeability and drainage of the soil.Soil pH and conductivity can have alarge effect on plant growth, andcrops and different cultivars willperform differently under differentvalues for these parameters. Soilnutrient status will affect the yieldpotential of the germplasm. It is alsoimportant to consider soil nutrientstatus when breeders are specificallylooking for tolerance to variousnutrient factors such as phosphorusefficient lines or lines tolerant tomicronutrient deficiencies.

• Climate is an important considerationespecially when looking for abiotic orbiotic stress tolerance. Should thebreeder select a high or low rainfallarea, cool or hot temperatures at thebeginning or end of the growingseason? Is frost an importantconsideration or heat duringgrainfilling?

• Is the crop being recommended for anirrigated or rainfed situation? Iswaterlogging or poor drainage acharacteristic of the recommendationdomain? If selecting for rainfed areas,how do you handle the problem ofsoil moisture at planting? Do you pre-irrigate the plot so germination isgood and then leave the rest of theseason to natural rainfall? Is irrigationwater available and is it of the qualityyou need or similar to that of thetarget environment?

• If the breeder is selecting material forspecific biotic stresses, he shouldselect an area where this stressoccurs. Incidence of certain diseasesand insect pests comes to mind.

• Abiotic stresses such as heat, salinity,and waterlogging also need to beaddressed when selecting a site. Areabiotic stresses (for example, salinityor waterlogging) consistent across theselected field? If not, can they bestatistically handled by experimentaldesign and layout?

1 CIMMYT Natural Resources Group, P.O. Box 5186, Lazimpat, Kathmandu, Nepal.2 CIMMYT Wheat Program, Apdo. Postal 6-641, Mexico, D.F., Mexico, 06600.

Managing ExperimentalBreeding TrialsP.R. Hobbs1 and K.D. Sayre2

C HAPTER 4

4 8

49

Another question is whether theexperimental area should be on stationor in farmers’ fields. Most breedingtrials have traditionally been conductedon station, mainly because it gives thebreeder control over experimentalconditions, e.g., land preparation,fertilization, irrigation, crop protection,and security of the trial. However, thereis often the problem how representativethe station is of the client farmersituation. Unfortunately, many stationsare not chosen for their similarity withfarmers’ fields but rather forconvenience or availability. Experimentstations usually represent one or at mosta few of the soils or environments facedby farmers.

There is therefore a tradeoff betweenexperimental control and howrepresentative on-station conditions areof the target environment. Probably thebest solution is to conduct breedingtrials under both situations. The firstphase selects promising materials underthe control of the station; later thematerials are assessed under actualfarmer situations. The latter shouldpreferably include farmer participationand experimentation. The extra benefitof farmer participation is the feedbackand assessment he can give the breeder.The new emphasis on farmerparticipatory breeding by some donorand development agencies utilizes thisimportant feedback mechanism.

Influence of CropRotations

In many developing countries,especially in subtropical environments,wheat is often grown in double or triplecropping patterns. For example, in Asia,wheat is grown sequentially with rice,cotton, soybeans, and maize in the samecalendar year. In Mexico, where theCIMMYT wheat varieties originate,

wheat usually follows cotton, maize, orsoybeans. It is therefore important thatsome breeding trials be grown on landwith similar cropping patterns, forseveral reasons:

• The previous crop can stronglyinfluence the harvest date andtherefore the planting date for thewheat crop. For example, in Asia,long-maturing “basmati” rice ispreferred in some locations becauseof its high quality and market price,its straw value and the fact that itneeds less fertilizer. The next wheatcrop will inevitably be planted late.Since there is genetic variability forperformance of wheat varieties underlate planting, breeding trials should atsome point in the selection process beplanted late after rice.

• The previous crop can have an effecton soil physical properties. When riceis grown in South Asia, the soil ispuddled (wet cultivation) to lowerpermeability and water use. Thisprofoundly affects the soil structurefor the next upland crop, such aswheat. The poor structure followingrice affects rooting and soilpermeability. Waterlogging iscommon in wheat following rice,with plants turning yellow due tooxygen stress. Thus a breeder shouldevaluate his germplasm under theseconditions to improve the chances ofselecting adaptable varieties for rice-wheat farmers.

• Soil nutrient status is also affected.Some crops like rice are veryexhaustive of nitrogen. For others,like potato, soil nutrient status is oftenbetter than normal due to highfertilizer application rate, whichleaves residual fertility.

• Soil water status can also be affected.Deep-rooted crops such as sugarcaneand cotton can deplete the soil profileof water. If the breeder grows hisnursery after fallow, he will getdifferent results than if he plants afterthe previous crop, unless hecompensates for the low water statusby irrigation.

• The previous crop will have a specificspectrum of pests and diseases. Ifweeds from the previous crop carryover to the wheat crop, measuresmust be taken to control the problem.Residues from the previous crop canalso influence disease and insectincidence. In many countriescombine-harvesting is becomingpopular. This system leaves looseresidue on the soil surface. In somecases, this residue can harbordiseases. In others, there are benefitsto this mulch (especially rainfedareas) in the form of coolertemperatures and better waterinfiltration and conservation. Thus itis important to screen materials underthese conditions.

Preparing the Nursery

Unless germplasm is being selected forreduced or zero-tillage situations, landpreparation is a major first step innursery management. The objective is toprepare the soil to favor wheatgermination. Several factors need to beconsidered:

• Selection of the appropriate plow orharrow for the specific soil to ensurea good tilth, favorable for wheatgermination. This may involve aseries of steps from deep plowing, toharrowing, to compaction of thesurface soil to ensure good seed-soilcontact.

• Fields should be leveled as much aspossible to reduce variability due tosoil moisture, especially if applyingirrigation. If germplasm is beingselected under rainfed conditions andthe fields are sloping, appropriateexperimental blocking designs will beneeded to reduce experimentalvariability and error.

• If the nursery is sown in farmers’fields, the plots should be prepared inthe same way that farmers do, usingthe most representative areas (awayfrom field edges, buildings, trees,etc.).

MANAGING EXPERIMENTAL BREEDING TRIALS

50

• Care should be taken to control anycarryover weeds from the previouscrop, to avoid creating a problem inthe wheat crop.

• It may be necessary to pre-irrigate thefield to ensure adequate soil moisturefor germination of the wheat seed.

If farmers in the target area arebeginning to use or are already usingzero-till planting, strong considerationmust be given to managing breedingnurseries in a similar fashion.

Planting Methods

If available, it is better to usecommercially available seed drillsspecifically designed for plantingbreeding trials (Hege, Wintersteiger,Almaco, etc.). These systems can plantsmall plots with many different varietieswithout having to clean the seed drillafter each plot. They also ensureconsistent planting depth, evendistribution of seed in rows, andoptimize the chances of goodgermination, all if which reduceexperimental error. If machinery is notavailable, hand planting is possible, butgermination will be less regular anderrors may be more common.

The trials should have a sufficientnumber of rows to eliminate bordereffects, i.e., the middle three or fourrows of wheat to be harvested should besurrounded by at least one row of wheatto prevent bias in the experiment (Figure1). Border rows will yield more thaninner rows due to less competition forwater, light, and nutrients. Where there isa lot of variability in plant height in thegermplasm, even more space is neededbetween varieties to preventexperimental bias. This can be achievedby growing more border rows or byleaving a space between the plots.

Statistical Methodsand Considerations

Error associated with yield estimates infield trials is increased by spatialvariation at the experimental site.Factors causing variation may includeuneven application of water andfertilizer, natural differences in soilfertility or water-holding capacity,spatial variation in soil physicalproperties, etc. Such variation can resultin poor precision when estimatingtreatment effects. It is thereforeimportant to reduce, as much as possible,the residual variation not accounted forby the variety effects. Residual variation

can be reduced through the use of anappropriate experimental design (a prioriapproach) and of spatial methods ofanalysis (a posteriori approach), alsocalled nearest neighbor analysis.

Lattice designsMost variety trials use complete orincomplete block designs and areanalyzed by the traditional analysis ofvariance. Block designs attempt an apriori reduction of the experimental errorconsidering spatial heterogeneity amongblocks. If it is probable that environmentalheterogeneity exists within anexperimental complete block (i.e.,replicate), for example in trials with largenumbers of treatments, lattice designs canbe used to adjust the means. The design ofthe experiment involves the restrictedrandomization of entries within sub-blocks so that means can be adjustedaccording to the variation among sub-blocks (Yates, 1938).

Incomplete block designs (such as alphalattices) can regularly achieve better yieldestimates and consume fewer resourcesthan randomized complete blocks. Suchefficient designs must be applied hand inhand with excellent field plot technique toproduce accurate estimates of yield withinlocations. Designs and programs foranalyzing lattice designs are included inthe statistical package MSTATC, forexample. Because they improve precision,lattice designs are highly recommendedfor field studies. The increased precisionresulting from lattice adjusted data maypermit trials to be grown with fewerreplicates than would otherwise benecessary to establish significantdifferences. With today’s analyticalcapabilities, it is rarely useful to considermore than two replicates for yield trials,and we believe unreplicated trials (inwhich only checks are replicated) mayprove increasingly useful.

Figure 1. Layout of trials to avoid experimental bias and border effects.

P.R. HOBBS AND K.D. SAYRE

Border rows and spacing between plots

Larger space between varietiesHarvested differing in height

area

Border rows Border rows

51

Spatial designsGiven that lattices do not consider thepresence of spatial variability withinsub-blocks, researchers have to findhomogeneous sub-blocks in the field,without knowing their most appropriateshape, dimension, and orientation. The aposteriori approach for reducing residualvariation can be applied when fieldvariety trials are laid out in a rectangulararray of rows and columns withreplicates allocated contiguously. Spatialanalysis can be performed to improveprecision of estimates of variety effectsand variety contrasts.

One approach is to adjust a plot forspatial variability by using informationfrom its immediate neighbors. A usefulmeasure for examining soilheterogeneity patterns is spatialautocorrelation of neighboring plotswithin rows or within columns (i.e., thecorrelation between residuals separatedby various distances). If there is nospatial pattern, all correlations will below. If there is pattern in the residuals,neighboring residuals will be moresimilar and so have higher correlation.Gleeson and Cullis (1987) proposed tosequentially fit a class of autoregressive,integrated, moving averages to the ploterrors in one direction (rows orcolumns). This was in the context ofrandomized complete block experiments.They found that differentiating along theblock and then fitting a moving averagecorrelation structure to the residuals inthat direction resulted in big gains in trialefficiency. Cullis and Gleeson (1991)extended the previous model to twodirections (rows and columns) assumingthat, in the field, rows and columns areregularly spaced.

Fertilizer Use

To express the full potential of a variety,sufficient nutrients must be applied tothe experimental area. A soil test isuseful for determining how muchfertilizer to apply. Sufficient nitrogen,phosphorus, and potassium should beapplied (as inorganic or organic sources,depending upon availability) to ensurethese elements are not limiting. Careshould also be taken to apply sufficientquantities of other nutrients, includingmicronutrients if they are known to belimiting. If varieties are planted whennutrients are limiting, they will beunable to express their true geneticpotential, making it difficult to separateout the better lines since they will allyield poorly.

To prevent variation in the field,fertilizer should be applied as uniformlyas possible, preferably by using afertilizer spreader. When a spreader isnot available, it is better to divide thefertilizer into smaller parcels orreplications and apply each dosecarefully to smaller areas. This is easilydone for the basal application offertilizer, but for the topdressapplication, the dose should be split byreplication or plot, and the fertilizercarefully broadcast. If wheat is plantedin a bed and furrow system, topdressfertilizer is best applied by machinery.

In farmers’ fields, the fertilizer levelrecommended for assessment should beused and, in some cases, 50% above therecommendation, so the scientist canevaluate the potential of the variety infarmers’ circumstances. The exceptionwould be where the breeder is screeningmaterial for tolerance to specificnutrients. For example, boron is aknown factor in wheat sterility. If abreeder wants to screen wheat lines fortolerance to low boron, he should havetwo sub-plots, one with applied boronand one without.

Irrigation

In rainfed trials, the experiment shouldbe blocked to ensure uniform moisturedistribution within replications. Whereirrigation is applied, uniform applicationis necessary to avoid experimental error.Providing basins for each replicationensures that each plot will receive thesame amount of water. In bed and furrowirrigation, replicating across theirrigation run ensures uniform waterapplication within a replication. Theadvantage of bed and furrowconfigurations comes from moreuniform water distribution, savings inwater, and savings in land that wouldotherwise be needed to construct thebasins and water distribution system.

Crop ProtectionStrategies

Unless the breeding trial is forevaluating germplasm against bioticstresses, crop protection measures shouldbe used to ensure the varieties expresstheir full yield potential. That may meanapplying herbicides, insecticides, orfungicides, depending on the stressespresent. Care should be taken whenchoosing the herbicide, since some linesmay be susceptible. Labels should becarefully read and, when in doubt, otherherbicides should be used or weedscontrolled manually. Agronomists in theprogram should be screening germplasmagainst various herbicides so that thisinformation is available to breeders.

Since breeders often wish to see theresponse of the lines to disease andinsects, it is rarely necessary to spray forthese factors.


52

Lodging

In the early years of the greenrevolution, lodging resistance was whatenabled plants to break the yield barrier.The old, tall, traditional varieties lodgedwhen fertility improved, and thisrestricted their yield potential. Whennew, shorter varieties with stiffer strawwere introduced, fertilizer levels couldbe increased without lodging, whichresulted in higher yields. However, eventhe new varieties lodged at higherfertilizer levels with late irrigation.Lodging is therefore still a major factorlimiting yield potential and should bereceiving more attention.

Several management methods can beused to reduce lodging and allow the fullyield potential of the germplasm to beexpressed. The following are somesuggested options:

• A major reason for lodging is that thesoil around the crown of the stem iswet and will not support the weight ofthe plant, especially when it is windy.This happens when wheat is plantedon the flat and irrigation is given afterflowering or during grainfilling.There are few options when wheat isplanted on the flat except to applyirrigation in the late afternoon whenwinds tend to subside or to forego thisirrigation. This may cause moisturestress and lead to lower yield,especially due to lower thousand-grain weight.

• Planting wheat on beds, especiallywith lower seed rates and two rowsper 70-cm bed, can result in lesslodging. In this system, laterirrigation does not saturate the soilaround the base of the plant as muchas on the flat and so the plant is bettersupported by the soil. Plants are alsoexposed to more sunlight, much likeborder rows in solid stands. Thismakes the plants stronger and moreable to resist lodging. In fact, planting

wheat on beds will enable the plantsto better express their yield potentialthan when planted on the flat becauseof better lodging resistance whenmore nitrogen is applied orgrainfilling irrigation is given.

• Fertilizer timing can also be used toreduce lodging. If all the nitrogen isapplied basally, the plants will beluxurious and competition for lightwill be high. This results in weakerplants more prone to lodging. Ifnitrogen can be reduced at plantingand delayed until just around the firstnode stage (DC31, Zadoks’ scale), itwill be utilized more for grain thanfor excess foliage, and lodging will beless. Experiments have also shownthat delaying nitrogen applicationuntil this later stage does not sacrificeyield potential but instead, because ofless lodging, can give higher yields.Grain protein content will also behigher.

• Seed rate can be manipulated to helpreduce lodging. If the plant stand istoo thick, independent plants will becompeting for light with adjoiningplants and be weaker in the stem andmore prone to lodging. By reducingthe seed rate, lodging can be reduced.

If lodging is a factor in an experiment, itis important to take a good measureof it. Several factors are needed:

• Estimate of the area lodged in thesampled plot. The angle of the stemin relation to the vertical is alsoimportant. This datum can be used asa covariate in the analysis.

• The date when lodging occurred isimportant, though it is better to notethe growth stage at which ithappened. The timing and cause oflodging can usually be determined byreference to the occurrence of a stormor an irrigation event. Data should becollected as soon as lodging occurssince the angle and extent of lodgingmay change with time.

If breeders wish to measure theimportance of lodging, they mayconsider growing one set of materialswith support (e.g., using netting) andcomparing with an unsupported check.Growth hormone sprays that reduceinternode length may also reducelodging and help measure lodging losses.

Harvesting andSampling

Harvesting the crop and taking reliablesamples are critical for breeding trials. Ifthis is not done properly, all efforts togrow a good crop may be wasted. Takingsamples as accurately as possible isessential for reducing experimental error.This will allow smaller differencesbetween varietal means to be separatedstatistically.

In breeding trials, a sample area—ratherthan the whole plot—is generally usedfor estimating yield. As mentionedearlier, to reduce bias in the experiment,the border rows are removed and onlythe inner rows used for yield estimation.It is best to remove the outer rows andhalf a meter from either end of the plot.The remaining area can be cut and usedfor measuring yield and yieldcomponents. Plants should be cut closeto the ground so that accurate harvestindex and straw yield can also bedetermined.

Calculating Yield andYield Components

Yield components can be calculated ormeasured individually. Three mainharvesting methods are suggested for usewhen calculating yield components(Table 1). After harvest, yieldcomponents can be calculated accordingto formulas presented in Table 2.


53

Harvesting steps are similar across thethree methodologies (Table 1); theprocedure is described below underMethod 1.

Method 1: Calculating yieldcomponents by harvestingtotal biomass• Cut all aboveground biomass in a pre-

determined area (A). Avoid bordereffects by sampling away from edgesof plot.

• Sub-sample a set number (e.g., 50 or100) of spike-bearing culms (i.e.,spike, leaves, and stem) randomly

from the plot sample and measurefresh weight (sub-sample freshweight, or SSfw).

• Measure fresh weight of remaining plotsample (plot fresh weight, or Pfw).

• Dry the sub-sample of culms at 70ºC(facilitate by putting culms in aclosed bag to avoid loss of grain, etc.)and weigh (sub-sample dry weight, orSSdw).

• Thresh plot sample for fresh grainweight (plot grain weight, or P-GW).

• Count 200 representative kernels andweigh fresh and after drying (200-grain fresh and dry weights, or200fw/dw).

Method 2: Calculating yieldcomponents by harvesting arandom sub-sample of culmsYield components can be determineddirectly by taking random spike-bearingculms from the crop at physiologicalmaturity.

• Take about 50 culms from rows (orarea) to be harvested by grabbing anumber of random sub-samples andcutting them off at ground level. Allharvested rows should be representedin the sample.

• All culms in the sub-sample are driedat 70oC.

• After measuring dry weight of thesub-sample (SS-dw), thresh tomeasure grain dry weight (sub-sample grain weight, or SS-GW).

This method has the advantage that handharvesting is quick (<5 minutes/plot),and samples can be readily stored forprocessing when time is available. Notethat with this method, measurement ofharvest index is statistically independentof measurement of grain yield, whereasthat of biomass is not independentof yield.

Table 1. Samples to measure when using three alternative harvesting methods forestimating yield, biomass, and yield components from experimental yield plots.

Samples to be measured†

Plot S-sample S-sample S-sample Plot 200 200biomass culms culms grain grain kernels kernels

Method fw (g) fw (g) dw (g) dw (g) fw (g) fw (g) dw (g)

1. Harvest total biomass X X X X X X2. Harvest sub-sample of culms X X X X X X3. Reduced threshing X X X X X X

† S-sample= sub-sample; fw=fresh weight; dw=dry weight.

Table 2. Formulas for calculating yield components using three different harvesting methods.†

Harvesting method

Yield component Method 1 Method 2 Method 3

Harvest index (HI) P-GW*(200dw/200fw)/P-fw*(SSdw/SSfw) SS-GW/SSdw SS-GW/SSdwYield (g m-2) [(P-GW*200dw/200fw)+(SSdw*HI)]/A [(P-GW*200dw/200fw)+SS-GW]/A Biomass*HIBiomass (g m-2) [(Pfw+SSfw)*(SSdw/SSfw)]/A Yield/HI [(Pfw+SSfw)*(SSdw/SSfw)]/A1000-GW (TGW) (g) 200dw*5 200dw*5 200dw*5Grains m-2 Yield/TGW*1000 Yield/TGW*1000 Yield/TGW*1000Culm dw (g) SSdw/#culms sampled SSdw/#culms sampled SSdw/#culms sampledSpikes m-2 Biomass/culm-dw Biomass/culm-dw Biomass/culm-dwGrains/spike Grains m-2/spike m-2 Grains m-2/spike m-2 Grains m-2/spike m-2

† A=plot area harvested (m-2); SS=sub-sample; fw=fresh weight; dw=dry weight; P=plot; GW=grain weight.Note: Formulas assume that grain dried at 70oC is at 0% moisture. Grain yield at x% moisture = Yield * [100/(100 – x)] (g/m2).


54

Method 3: Calculating yieldcomponents with reducedgrain-threshing requirementMeasuring grain yield and yieldcomponents with this method requiresthreshing only a sub-sample. The procedureis useful when larger-scale threshingmachinery is lacking or when working withspecies that are difficult to thresh, such asTriticum dicoccum or synthetic hexaploidwheats. Measure the samples indicated forMethod 3 in Table 1, and follow therelevant sampling procedures describedunder Methods 1 and 2.

Sampling Biomass inLodged Crops

It is difficult to cut a given area in alodged crop, especially if it has beensown by broadcasting. The process isfacilitated by folding back spikes andstems to establish a starting referenceline before inserting a quadrat. Greatcare must then be taken to collect onlythose plants whose crowns fall withinthe randomly located quadrat. If one sideof a quadrat is separated from the otherthree sides, it is easier to insert thequadrat into the crop. In this case, thefree fourth side is used to ensure theframe is properly square by placing itbetween the two free ends of the 3-sided frame.

Measuring IndividualYield Components inthe Field

Plant populationA count of plant population should bemade after the maximum number ofplants has emerged and before tilleringoccurs (usually 10-14 days afteradequate moisture becomes available forgermination).

If plants are sown in rows, then 0.5 mlength from each sampling row or fromat least six such rows should be countedper plot. If broadcast, then samples of atleast 0.5-1.0 m should be taken fromeach plot. The number of samplesrequired will depend on the degree ofvariation within the plots, but at leasttwo per plot should be recorded.

The mean plant density may disguiseimportant variability in plant distribution(i.e., gaps that will cause yieldreduction). This should be noted andmeasured by estimating the percent ofthe plot with missing plants.

Plant population can be used to assessthe germination, vigor, and emergence ofsown seed, and/or the extent ofcompensation under conditionsadvantageous to tillering. It is alsoneeded if early growth per unit area isgoing to be monitored by successivemeasurements of growth per plant. Plantpopulation typically varies between 50and 300 plants/m2. The number ofplants/m2 has a broad optimum and willvary with variety, climate, andmanagement. However, under goodrainfed conditions, 100 plants/m2 couldbe considered the minimum formaximum yield, unless the crop isgrowing on residual moisture, in whichcase optimum density may be lower.

Spikes / m2

The number of productive spikes can bemeasured nondestructively by countingin a given area or length of row, orcalculated from sampling asdemonstrated above.3 Spikes per m2 canbe measured most easily beforephysiological maturity, which alsoreduces yield loss due to shatteringcaused by movement in the plots. Inbroadcast planting, direct measurementcan be difficult, especially if crops lodge.

Spikes/m2 can be used to assess the finalnumber of productive spikes/ m2 andcan be combined with plant populationcount to assess the extent of tillering.Tillering typically ranges from 1 to 10per plant. Spikes/m2 is determined byevents occurring from sowing toflowering and is dependent on variety,management, and environment.

Spikelets / spikeSample a minimum of 10 spikes per plotat random (aim for a total of 30-40spikes/treatment); select the culms fromthe base and count the number ofspikelets. Take the average based onsample size. Most commonly, count thefully developed or grain-bearingspikelets (or at least those large enoughto have at least one grain). Potentialspikelet number is obtained by countingall the nodes on the rachis; it can exceedthe developed spikelet number becauseof aborted spikelets at the base or tip ofthe spikes. Alternatively, under excellentenvironmental conditions, all potentialspikelets may develop into grain-bearingspikelets.

The potential number of spikelets perspike is determined by the time ofterminal spikelet formation (in wheatand triticale; barley does not form aterminal spikelet) around first nodeappearance. Subsequently, primordialspikelets at the base (and, later, at thetip) of the spike may abort because ofstress. Normally, 10-25 spikelets formon each spike.

Grains / spikeletSample as for spikelets per spike. Countthe spikelets, thresh, count grains, andcalculate; or less accurately simplycalculate from calculated grains perspike and measured spikelet number.When sampling large numbers ofsamples or plots, time may be saved byrandomly counting one side of the spikeand multiplying by two.

3 If measured directly, the procedure and number of sub-samples are as for plant population.


55

Grains per spikelet is the result of boththe number of competent florets/spikeletand kernels/competent floret (or grainset). Values for competent florets perspikelet typically vary from 1.5 to 5.0and for kernels per competent floretfrom 0.6 to 0.99.

Grain setGrain set refers to the percentage ofcompetent or entire florets (florets withfully formed, plump green/yellowanthers at flowering) that actuallyproduce grain (the opposite of percentsterility) and should reflect conditionsaround anthesis (e.g., pollen fertility,early grain survival), in contrast to grainsper spikelet, which can be influenced byearlier conditions as well. However, atmaturity, it is difficult to know whichflorets were competent. It is suggestedthat the basal two florets of the 6-10central spikelets are always competent,and therefore a grain set index can beobtained by observing the percentage ofsuch florets with grains. Sample asspecified for spikelets or count 10 suchspikelets in five random spikes per plot;the total of missed florets is the %sterility (= 100 – % grain set).

Alternatively, one can use matchedspikes (i.e., spikes showing equal sizeand development). One spike is sampledat anthesis and the other at maturity,counting (destructively) competentflorets at anthesis in one spike (i.e.,florets showing normal antherdevelopment; non-competent florets willshow whitened, flattened anthers with nofertile pollen, whose stamens neverelongate) and grains at maturity in theother spike. At least 20 matched spikesper treatment are needed for reasonableaccuracy; selection of matched spikesand counting at anthesis are timeconsuming.

Grain set is an indicator of stress eventsoccurring around anthesis (e.g., drought,temperature extremes, boron deficiency, orgenetic sterility, which can interact withthe environment). It is a more precise and,hence, more useful measure than grains perspikelet or grains per spike.

Thousand-grain weight atmaturity and duringgrainfillingTo measure thousand grain weight(TGW), count out two random samples of100 entire grains (i.e., those possessing anembryo). Dry the grains at 70oC (48 hoursshould be sufficient) and weigh. This willusually give sufficient accuracy. Ifweights differ by more than 10%, a thirdsample of 100 should be taken or recheckthe counts.

To study grain growth during grainfilling,maximum accuracy is achieved byselecting groups of a sufficient number ofspikes, matched for anthesis date and size,so that one can be randomly sampledfrom each group on each sampling date.Four to eight such groups (four to eightspikes each date) per treatment should besufficient for accurately calculating graingrowth rate by linear or curvilinearregression. The study can be based on allgrains on the spike or in a given positionon the spike (e.g., basal florets of centralspikelets).

A reduction in TGW may be due toclimatic or biological (e.g., pathogen)stress during grainfilling. Kernel weight(calculated as TGW/1000) is usually 20-50 mg. However, lower grain weight maynot indicate stress during grainfilling, dueto the plasticity of yield components. Forexample, if the plant population is high(leading to a high number of kernels/m2),TGW may decrease without seriouslyaffecting yield. TGW tends to becharacteristic of a variety, and there are

large differences between varieties evenunder good conditions. Within a variety,kernel weight usually shows a negativelinear relationship to mean grainfillingtemperature. Potassium deficiency canalso result in low thousand grain weight.

Grain or kernel number (perm2)Kernel number per m2 (KNO) is usuallycalculated by dividing grain yield (GY;in g/m2) by kernel weight (KW; in mg):4

KNO GY * (1000/KW)

Kernel number can also beindependently measured by directlydetermining spike number (SNO/m2),and kernels per spike (KPS) from at least20 randomly sampled spikes per plot(aim for 60-100 spikes):

KNO=SNO * KPS

Kernel number per m2 acts as a summaryof all events up to and a little beyondanthesis. For example, the combinedeffects of management and climate onplants/m2, spikes/plant, spikelets/spike,and grains per spikelet are all combinedin this single term. Competent floretnumber (the precursor of kernel number)is also well correlated with spike(inflorescence only) dry weight atanthesis, the relationship being on theorder of 100 florets/1.0 g spike (10 mg/floret), although the range acrossvarieties for grain number is 70-140kernels/g spike dry weight at anthesis.

Under many conditions, yield is afunction of KNO, which is particularlydependent on crop growth rate duringrapid spike growth (emergence of thesecond-to-last leaf—or about 1 monthbefore anthesis for spring wheat—untiljust after anthesis).

4 Using this calculation, KNO is statistically linked to GY and may give rise to spurious correlations between GY and GNO if GY is not determinedaccurately.


56

Yield Estimation andMeasurement

Visual estimatesYield can be estimated by visualassessment (this requires experience anda knowledge of the variety and area) orbased on yield components from mid-grainfilling onward.

Yield estimates from yieldcomponentsTo calculate yield from yieldcomponents, first estimate the numberof spikes/m2 from in situ counts(outlined above). Next, randomlysample and then count the number ofgrains/spike. Then, assume a TGWbased on the variety and conditionsexpected during the rest of thegrainfilling period (typical TGW underreasonable grainfilling conditions andtemperatures: 30-40). Calculate yieldwith the following equations:

Yield (g/m2) = spikes/m2 * grains/spike*(TGW/1000)

Yield (kg/ha) = yield (g/m2) * 10

Variability of the field or treatment anddesired accuracy will determine thenumber of spike counts made. Forgreatest accuracy many small samplingsare best and feasible when dealing withnon-destructive sampling. For example,for a 1-ha drill-planted field, take spikenumber counts in 20 random but welldispersed 2-row x 50-cm quadrats andcombine these counts with kernelnumber counts in 50 random spikes; areasonable estimate of kernels/m2

should result (but be careful to selectsampling sites and spikes at random).Counting kernels per spike (one side x2) while walking between quadrat sitessaves time and should take less than 30minutes. Be sure row spacing isaccurately known and/or measurespacing to confirm; replicate the

measurements for accurate assessment.For example:

• Average row spacing: 15 cm• Average spike count (2 row by 50

cm): 40 spikes• Spike sample area: 2 rows * 15-cm

row spacing * 50 cm = 0.15 m2

• 40 spikes/0.15 m2 = 266.7 spikes/m2

• Average kernel count/spike = 21.5• Kernels/m2 = 266.7 * 21.5 = S734• Assume TGW = 40 (based on

experience)• Yield = 5734 * 40/1000 = 229.4 g/

m2 = 2294 kg/ha

Yield estimates from samplesYield can be estimated as outlined foryield components. Alternatively, borderscan be discarded and yield estimatedfrom the remaining plot that is harvested.In some instances, it is not possible(especially in on-farm trials) to harvestthe entire plot or to thresh the entire plotsample. In these cases, follow theoptions outlined above or sample 5 x 1m2/field or 2 samples of 1 m2/plot ifthere are 3 replicates.

Yield moisture contentThe grain trade and farmers usuallyexpress yield at a given moisture content(e.g., 10% in Australia, 12 or 14% inEurope on a fresh weight basis).Therefore, conversion factors arerequired to adjust moisture. Moisturecontent is calculated as the weight ofmoisture relative to fresh weight:

[Moisture/(moisture + dry weight)]

The following formulas outline variousmoisture calculations. Yield and grainmoisture calculations:

Field weight of harvested grain = FW (kg)Harvested area = A (m2)Fresh weight of sub-sample = WSOven-dry weight of sub-sample = DS

Grain moisture conversions:

Grain moisture content (M%)M% = [(WS – DS) * 100] /WSYield (unknown moisture, GYm)GYm (t/ha) = (FW * 10)/AYield (0% moisture, GY(0%))GY(0%) = [GYm * (100 - M)]/100Yield (X% moisture, GY(X%))GY(X%) = [Y(0%) * 100]/(100 – X)

Throughout the previous discussion, allweights of plant parts (including grain)refer to constant weight after drying at70oC. However, the AmericanAssociation of Cereal Chemistry(AACC) defines 0% moisture as thatachieved by drying ground grain at103oC. Therefore, other conversionfactors are required in addition to theabove to obtain a true 0% moisturereading. To convert the weight of graindried for 20 hours at 70oC to moisturecontent as defined by AACC, divide theweight by 1.025 (because grain dried at70oC has approximately 2.5% moisture).The factor drops to 1.012 as drying timeat 70oC increases to 48 hours. Thismeans that, to express grain at 10%moisture, the oven-dry weight (70oC for24 hours) needs to be multiplied by1.084 (i.e., 1.00/1.025).

Assessing CropResidue

Direct measurementCollect and bulk straw found within fiverandom samples (or two per replicate) ofat least 1.0 m2. Oven dry straw (70oC)and weigh. Convert g/m2 to t/ha bydividing by 100.

Due to the generally large spatialvariation in ground cover, a visualestimate (by an experienced researcher)is often sufficiently accurate using themethod described below.


57

Estimating percent residuecover using the line-transectmethod5

The line-transect method is one of theeasiest methods to use for determiningthe percent residue cover on the soilsurface. Accurate measurement isnecessary to determine if enough coveris present to comply with theconservation plan.

The following is a step-by-stepprocedure for using the line-transectmethod to measure the percentage ofresidue cover.

Step 1. Use a 100- or 50-m measuringtape for measuring residue cover.Other measuring lengths or evenknotted ropes can be used if theappropriate multiplication factor isused to calculate the percentage.

Step 2. Select an area that isrepresentative of the whole field/plot.Avoid end rows or small areas thathave been adversely affected byflooding, drought, weed or insectinfestations, or other factors that mayhave substantially reduced yields.

The most accurate reading of residuecover is obtained by taking an averagefrom at least three differentrepresentative locations in the field.

Step 3. Anchor one end of tape andstretch it diagonally across the rowsso that it crosses several passes by thefarming implements. Readings aremore accurate when the tape isstretched diagonally across the rowsthan when taken in a window ofresidue left by the combine or wherethe amount of residue is smaller.

Step 4. Measure residue cover bycounting the number of “m” marksthat are directly over a piece ofresidue, as shown below. For anaccurate reading,

• do not move the tape while counting;• look at the same side of the tape at

each m mark;• look straight down at the tape and the

m mark, and• count only those m marks that have

residue directly under them.

To effectively reduce erosion, a piece ofresidue must be large enough to dissipatethe energy of a raindrop during an intensestorm. To be counted, the residue must belarger in diameter than this dot•A convenient way to determine if a pieceof residue is large enough to count is touse a 1/4 cm diameter brazing rod orwooden dowel. Touch the edge of therod to the m mark. If the residue extendscompletely beyond all edges of the rodend, count it. If the rod covers the

residue completely or if part of the rodend extends beyond the edge of theresidue at any point, don’t count itbecause a raindrop falling on this pointwould strike some unprotected soil.

On a 100-m tape, the number of m marksthat are directly over residue will be adirect measurement of the percentage ofresidue cover for the field. On a 50-mtape the number of m marks must bemultiplied by 2. If other measurementlengths are used, the appropriatemultiplication factor must be used,especially for plot-level measurement.

Residue amount can be used to assesssoil cover and thus potential evaporationreduction but, more importantly, erosioncontrol. Assuming uniform distribution,approximately 4 t of wheat straw lyinghorizontally are required to give 100%ground cover. Straw is also a potentialsource of disease infection forsubsequent crops and may immobilize Nduring the decomposition process.

5 This step-by-step description of the line-transect method was taken from a fact sheet by David P. Shelton, and Elbert C. Dickey, Extension AgriculturalEngineers, University of Nebraska; Robert Kanable, Conservation Agronomist, Soil Conservation Service; Stewart W. Melvin, Extension AgriculturalEngineer, Iowa State University; and Charles A. Burr, Extension Agricultural Specialist, University of Missouri. The fact sheet was produced throughthe cooperative efforts of representatives of Cooperative Extension, Midwest Plant Service, NACD’s Conservation Technology Information Center,U.S. Environmental Protection Agency, and the USDA Soil Conservation Service.


Does not countas a point of

residue

Counts as apoint ofresidue

Does not count as apoint of residue

Counts as a pointof residue

A 1/4 cm rod can be used to determine which piece of residue to count.

When using the top of the tape, this m mark...

1'

2'

3'

4'

58

Ancillary dataIt is always useful, especially forinterpretating the results, to collect othersite data. The following are someminimum data sets for consideration.

Climate data. The performance of anycrop is very dependent on climate.Usually, there is a weather station closeby that can provide this valuablesupporting information. The followingdata would be useful:

• maximum, minimum, and averagedaily temperature

• precipitation• radiation data or sunshine hours• relative humidity (max-min, if

possible)

Soil data. Soil data such as the followingare useful for interpreting results:

• soil texture• soil pH• conductivity (a measure of salinity)• infiltration (a physical measure of

porosity)• soil moisture holding capacity

(permanent wilting point and fieldcapacity)

• soil carbon level (a measure oforganic matter)

• available P and K levels• micronutrients status• penetrometer measures to assess if

plow pan exists• water table depth• irrigation water applied, numbers,

and, if possible, amounts

ReferencesAnderson, W.P. 1983. Weed Science: Principles.

Second Edition. West Publishing Co.St. Paul.

Bell, M., and R.A. Fischer. 1994. Guide to Plantand Crop Sampling: Measurements andObservations for Agronomic andPhysiological Research in Small GrainCereals. Wheat Special Report No. 32.Mexico, D.F.: CIMMYT.

Brady, N.C. 1974. The Nature and Properties ofSoils. Eighth Edition. Macmillan PublishingCo., Inc. NY.

Cooke, G.W. 1982. Fertilizing for MaximumYield. Third Edition. Macmillan PublishingCo., Inc. NY.

Cullis, B.R., and A.C. Gleeson. 1991. Spatialanalysis of field experiments-an extension totwo dimensions. Biometrics 47:1449-1460.

Dregne, H.E., and W.O. Willis (eds.). 1983.Dryland Agriculture. Agronomy Series No.23. American Society of Agronomy, Inc.Madison, WI.

Counts Does not count Does not countraindrop strikes residue raindrop strikes both soil and residue raindrop strikes both soil and residue

Only those “m” marks having a piece of residue directly under them should be counted.


Gleeson, A.C., and B.R. Cullis. 1987. Residualmaximum likelihood (REML) estimation ofa neighbour model for field experiments.Biometrics 43:277-288.

Leonard, W.H., and J.H. Martin. 1963. CerealCrops. The Macmillan Company, NY.

Martin, J.H., W.H. Leonard, and D.L. Stamp.1976. Principles of Field Crop Production.Third Edition. Macmillar Publishing Co.,Inc. NY.

Mortvedt, J.J., P.M. Giordano, and W.L. Lindsay(eds.). 1972. Micronutrients in Agriculture.Soil Science Society of America, Inc.Madison, WI.

Phillips, R.E., and S.H. Phillips (eds.). 1984. No-Tillage Agriculture. Van Nostrand ReinholdCo. NY.

Russell, E.W., and S.E.J. Russell (eds.). 1973. SoilConditions and Plant Growth. Tenth Edition.Longman, London.

Sanchez, P.A. 1976. Properties and Managementof Soil in the Tropics. A Wiley-IntersciencePublication. John Wiley and Sons, NY.

Sayre, K.D. 1998. Ensuring the Use ofSustainable Crop Management Strategies bySmall Wheat Farmers in the 21st Century.Wheat Special Report No. 48. Mexico, D.F.:CIMMYT.

Sayre, K.D., and O.H. Moreno-Ramos. 1997.Applications of Raised-Bed PlantingSystems to Wheat. Wheat Special ReportNo. 31. Mexico, D.F.: CIMMYT.

Stewart, B.A., and D.R. Nielsen (eds.). 1990.Irrigation of Agricultural Crops. AgronomySeries No. 30. American Society ofAgronomy, Inc. Madison, WI.

Tisdale, S.L., W.L. Nelson, and J.D. Beaton (eds.).1985. Soil Fertility and Fertilizers. FourthEdition. Macmillan PublishingCompany, NY.

Walsh, L.M., and J.D. Beaton (eds.). 1973. SoilTesting and Plant Analysis. Revised Edition.Soil Science Society of America, Inc.Madison, WI.

Yates, F. 1938. The comparative advantages ofsystematic and randomized arrangements inthe design of agricultural and biologicalexperiments Biometrika 30:444-46.

59

This review presents practical guidelinesfor breeders of wheat (and other smallgrain cereals) interested in adopting aphysiological approach to cropimprovement. Some of the mostpromising tools for fast, reliablecharacterization of yield-determiningtraits will be discussed from anecophysiological perspective. We willfocus on the practical aspects andlimitations of using relevant screeningtools or selection criteria. The potentialcontributions of physiological researchto plant breeding, as well as its inherentlimitations, have been extensivelyreviewed from a breeding perspective(for example, Jackson et al., 1996). Atheoretical framework for identifyingyield determinants that are obviouscandidates for evaluation has also beenestablished (see below), although it hasnot been used in practical breeding.

The use of physiological traits asscreening tools in breeding is still largelyexperimental—for different reasons.Sometimes the traits are very indirectlyrelated to yield (Araus, 1996; Richards,1996) or there is little ecophysiologicalunderstanding of the crop, especiallywhen breeding for yield under stress.Nevertheless, breeding for crop escapehas been very successful, andphenological changes have been themost important indirect factor inincreasing wheat yields underMediterranean conditions (Slafer et al.,

1993; Loss and Siddique, 1994).However, in breeding for cropresistance, the evaluated traits andscreening tools are often related totolerance, not avoidance (see definitionsin Larcher, 1995).

An indirect (i.e., physiology-based)breeding strategy could fail to produceyield gains and might even lead to adecrease in yield. Improved planttolerance, though it protects the crop,can limit yield potential. The targetenvironment where selection is carriedout must be defined a priori, and thepossibility of a negative breeding effortshould not be ignored. In fact, plants thatshow the most tolerant responseduring screening may also be the mostsensitive, in terms of yield loss, forexample, because they are unable todelay the effect of stress at the cellularlevel.

The most promising methods allow forquick screening of “integrative”physiological traits (Araus, 1996), socalled because they integratephysiological processes either in time(i.e., during the plant cycle) or at theorganization level (e.g., whole plant,canopy). Other quick screening methodsfor evaluating, for example, thephotosynthetic performance of plantsunder stress conditions have beenproposed—among them, chlorophyllfluorescence measurements on single

leaves. However, under field conditionsfluorescence values may only reflectdifferences in phenology acrossgenotypes. Nevertheless, remote sensingdetection of fluorescence spectra at thecanopy level could become a promisingapproach for breeding purposes(Lichtenthaler, 1996).

IdentifyingPhysiological Traits forUse As SelectionCriteria

One approach to search for traits thatcould be used in breeding programs is toidentify the physiological processesdetermining productivity. A crop’s yieldpotential (or harvestable part, Y) over agiven period of time can be divided intothree major processes (Hay and Walker,1989). First, the interception of incidentsolar radiation by the canopy; second,the conversion of intercepted radiantenergy to potential chemical energy (i.e.,plant dry matter); and third, thepartitioning of dry matter between theharvested parts and the rest of the plant.Whereas the first component depends onthe canopy’s total photosynthetic area,the second relies on the crop’s overallphotosynthetic efficiency (i.e., total drymatter produced per unit of interceptedradiant energy); the third is harvestindex. Total biomass, which is the result

Recent Tools for theScreening of PhysiologicalTraits Determining YieldJ.L. Araus,1 J. Casadesus,2 and J. Bort1

1 Unitat de Fisiologia Vegetal, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain.2 Servei de Camps Experimentals, Universitat de Barcelona, Barcelona, Spain.

C HAPTER 5

60

of the first two components, can bephysiologically defined as the result ofcanopy photosynthesis over time.

Other approaches may be followed,depending on agroecological conditions.For example, under water-limiting (e.g.,Mediterranean) conditions, the mostwidely used framework, proposed byPassioura (1977), allows the study ofindices that maximize yield per rainfallunit. Thus economic yield depends onthe total water transpired by the crop,water use efficiency, and harvest index.Although these three components are nottruly independent, Passioura’s is a usefulframework for searching for criticaltraits to improve yield under drought.

Yield can be divided into severalintegrative components or traits. Yielditself is the most integrative trait,because it is influenced by all factors(known and unknown) that determineproductivity. However, there are manyknown limitations in a purely empiricalbreeding approach based only on yield.Therefore, any breeding strategy basedon a physiological (i.e., analytical)approach should use screening tools orcriteria to evaluate the integrativephysiological parameters that determineharvestable yield with a singlemeasurement. Although harvest indexhas been the most successful trait whenmodified to improve yield, the other twocomponents of the above equations,which are responsible for total cropbiomass, remain (basically) unchanged.In the following pages we will focus ontools used to evaluate physiological traitsdetermining total biomass. We willdiscuss two different kinds of tools forscreening integrative physiological traits:carbon isotope discrimination (Δ) andindices based on canopy spectral-reflectance.

The Δ not only evaluates genotypicdifferences in water use efficiency, butcan also be affected by the total amountof water transpired by the crop (the firstcomponent of Passioura’s equation) orby photosynthetic activity (the secondcomponent of the yield potentialequation). Keeping in mind developingcountry breeders, we have includedunder the generic title “surrogates” otherscreening criteria, such as ashaccumulation in different plant parts orcriteria related to leaf structure. Thoughthese features are not Δ surrogates in astrict sense, they are always related toyield and are both quicker and cheaperthan Δ determinations; furthermore, nolarge facilities or highly qualifiedtechnical support are necessary touse them.

Canopy spectral reflectance is one of themost promising remote sensingtechniques (see also Araus, 1996).Although at present the equipment isvery expensive, in a few years its costshould drop dramatically. Anotherremote sensing technique, canopytemperature, provides integratedinformation on the crop’s stomatalconductance at the canopy level (seechapter by Reynolds) (Reynolds et al.,1994) and has the advantage of beinglow cost. However, its usefulness islimited in severely stressedenvironments, and genotypic differencesin phenology and canopy architecturecan further limit its validity.

Carbon IsotopeDiscrimination

For C3 plants, discrimination (Δ) of theheavier (13C) stable carbon isotope overthe lighter, more abundant (12C) form(99%) during photosynthetic CO2fixation is an integrated measure ofinternal plant physiological properties

and external environmental conditionsthat influence photosynthetic gasexchange (Farquhar et al., 1989). In C3cereals such as wheat, Δ is positivelyrelated to CO2 levels in intercellular airspaces (Diagram 1) (Farquhar et al.,1982; Farquhar and Richards, 1984;Ehdaie et al., 1991) and, given a constantleaf-to-air vapor pressure difference, alsonegatively related to water use efficiency(WUE, measured either as netphotosynthesis/transpiration, also calledtranspiration efficiency, or as plantbiomass produced/water transpired)(Farquhar and Richards, 1984; Hubickand Farquhar, 1989). Plants with highWUE would be less able to discriminateagainst 13C, and thus would accumulatemore of the heavy carbon isotope in theirtissues than less efficient water users.

When measured in plant dry matter, Dprovides an (integrated) indication ofWUE throughout plant growth (Farquharet al., 1982, 1989). D has been proposedas a possible screening tool foridentifying variations in WUE in wheat(Farquhar and Richards, 1984; Ehdaie etal., 1991; Condon and Richards, 1993)and barley (Hubick and Farquhar, 1989).In fact, the permanent relationshipsbetween WUE and D during treatmentand the high broad-sense heritability ofD in wheat, together with otherconsiderations, indicate that D may beuseful for modifying the WUE and yieldof water-limited wheat crops (Condon etal., 1987; Condon and Richards,1992, 1993).

The relationship between Δand water-use efficiencyFollowing on the model of Farquhar etal. (1982), Δ in C3 plants may be definedin its simplest form as:

Δ = a + (b - a)(pi/pa),

where a is the carbon-13 discriminationcaused by diffusion in air (4.4 0/00), b isthat caused by carboxylation by the

J.L. ARAUS, J. CASADESUS, AND J. BORT

61

RuBP carboxylase enzyme (27 0/00), and(pi/pa) is the intercellular to atmosphericCO2 partial pressure ratio. Conversely, pi/pa may be defined as a function of Δ:

pi/pa = (Δ - 4.4)/2.6.

If we assume that the temperature of theleaf (or other photosynthetic organ) isclose to air temperature, and if daytimerelative humidity is also known, WUE(the net assimilation to transpiration ratio)may be defined as a function of pi/pa:

WUE = pa(1 - pi/pa)/V(1 - RH)1.6,

where V is the saturated partial watervapor pressure at a given temperature,RH is relative humidity, and 1.6 is the

psychometric constant (Farquhar et al.,1982). Thus, WUE can be estimated fromΔ values using the following equation:

WUE = pa[1 - (Δ - 4.4)/22.6]/V(1 - RH)1.6.

An agronomic estimation of WUE(considered as the ratio of dry matteraccumulated to total water transpired) canbe also be derived from Δ using differentequations (Hubick et al., 1986; Hubickand Farquhar, 1989; Craufurd et al., 1991;see also Araus et al., 1993).

Methodological considerationsCarbon isotope analysis is performed bymass spectrometry. Although theequipment needed for such testing isexpensive and often beyond the capabilityof many laboratories and research

stations, there are commercial firms thatdo the analyses reliably and at areasonable price. Before sending thematerial to be analyzed, it must beground very finely.

13C/12C ratio values are expressed ascarbon isotope composition (δ13C)values, where

δ13C (0 /00) = [(R sample/R standard)-1] x 1000,

R being the 13C/12C ratio. The standardfor comparison is a secondary standardcalibrated against PeeDee belemnite(PDB) carbonate. Test precision isusually lower than 0.10 0/00. The value ofthe discrimination (Δ) against 13C iscalculated from δa and δp, where a refersto air and p to plant (Farquhar et al.,1989):

Δ = (δa - δp Δ / (1 + δp )

On the PDB scale, free atmospheric CO2has a current deviation, δa, ofapproximately -8.0 0/00 (Farquhar et al.,1989).

Implications for plantbreedingWhat type of sample to take and when?Considerable genotypic variations for Δhave been found in bread wheat (Condonand Richards, 1992), barley (Romagosaand Araus, 1991; Acevedo, 1993), anddurum wheat (Araus et al., 1993a), butenvironmental factors may cause evenlarger changes in the value of Δ, whichcould compromise the effective use of Δin breeding programs. After studyingwheat Condon and Richards (1992)concluded that assessing genotypicvariation in Δ would be most effectiveunder well-watered conditions. In thisregard, Richards and Condon (1993)pointed out that under adequateconditions, Δ is highly heritable andexhibits substantial genetic variation andfew genotype x environmentinteractions.

Diagram 1. Carbon isotope discrimination under irrigated and dry conditions.

RECENT TOOLS FOR THE SCREENING OF PHYSIOLOGICAL TRAITS DETERMINING YIELD

62

4000

3000

2000

100021 22 23 24

4000

3000

2000

100017 18 19 20

4000

3000

2000

100013 14 15 16

Carbon isotope discrimination

As an alternative in rainfedenvironments, Condon and Richards(1992) proposed sampling for Δ at earlycrop stages, when terminal stress islacking. However, the informationavailable on rainfed environments oftendoes not support these expectations.Though correlation between Δ and yieldis usually weak or non-existent whendry material from seedlings is analyzed(Bort et al., 1998), it increases whenupper plant parts are used in Δ analyses(Figure 1). The best genetic correlationsbetween Δ and yield, together with thehigh broad sense heritability of Δ, havealso been reported for the upper parts ofdurum wheat (Araus et al., 1998b).

The correlation between yield and Δincreases with plant age, perhaps due tothe effects of progressive stress(particularly after anthesis) on yield. Infact, Δ usually decreases from the oldestto the youngest plant parts, even underwell-watered conditions (Hubick andFarquhar, 1989; Acevedo, 1993; see alsoFigure 1). Such a decrease may beattributed to stomatal closure inresponse to declining soil water and/orincreasing vapor-pressure deficit duringthe last period of crop growth (Condonand Richards, 1992; Condon et al.,1992; Araus et al., 1993b). Thus, maturekernels could be the most adequate plantpart to sample. Under Mediterraneanconditions, for example, the Δ of kernelsrather than the Δ of lower plant partsmay provide more information on whichgenotype is less affected by stress duringgrain filling.

Higher or lower carbon isotopediscrimination? In water-limitedenvironments, genotypes with low Δshould have greater biomass and hencepotential for higher yields, assumingthat all genotypes use the same amountof water for transpiration (Richards,1996). In fact, selecting for high WUE(Passioura, 1977) or low Δ (Craufurd et

al., 1991) has been proposed as animportant alternative when definingplant breeding strategies under limitedwater conditions. However, Δ valuesoften correlate positively with grainyield and/or total biomass in wheat(Condon et al., 1987; Kirda et al., 1992;Araus et al., 1993c, 1997b; Morgan etal., 1993; Sayre et al., 1995) and barley

(Romagosa and Araus, 1991; Richards,1996) under well-irrigated or rainfedconditions (Figure 1).

From an agronomic point of view, apositive relationship between Δ and yieldmay exist if plants are not using allavailable soil water. Assuming the samephenology, a genotype with high Δ will

Seedlingsy = -59.7x + 5995.6,r = 0.14, n.s.

Figure 1. Relationships between grain yield and carbon isotope discrimination (Δ) measuredin dry matter of seedlings (a), in the penultimate leaf sampled around heading (b) and inmature kernels (c) for a set of 144 durum wheat genotypes grown under rainfed conditionsin the Tel Hadya, northwestern Syria. For more details, see Araus et al. (1997b).

a

Grain

yiel d

(kg h

a –1)

b

c


63

be able to sustain a high level oftranspiration. Therefore, Δ can beconsidered an indicator of WUE, butalso depends on the water transpired bythe crop, which is in fact the firstparameter of Passioura’s identity. Thepositive association between Δ and yieldalso suggests that variations in stomatalconductance rather than in intrinsicphotosynthetic capacity are predominantin determining Δ (Romagosa and Araus,1991; Condon et al., 1992). Higher Δ isrelated to a higher level of CO2 in thecellular air spaces due to greaterstomatal conductance (Farquhar andRichards, 1984), which leads to higherphotosynthetic rates and, hence, higheryield even in the absence of stress. Inthis situation, WUE decreases (and Δincreases) because stomatal limitationreduces transpiration more thanphotosynthesis, even when yield may bepositively affected by low stomatallimitation on photosynthetic rate.

Relatively high transpiration levels mayhave implications for water-limitedenvironments. For example, when watersupply can be provided to the crop underdrought stress, (e.g., by deep soilmoisture extraction), the high-yielding(i.e., high-transpiring) genotype willhave the most advantage (Blum, 1993,1996). In fact, relatively low canopytemperatures resulting from highstomatal conductance and transpirationare typical of the more drought-resistantgenotypes (Garrity and O’Toole, 1995;see also Blum, 1996). In addition, whengrown at above-optimal temperatures,such as the maximum daily temperaturestypical during grain filling, the positivecorrelation observed between stomatalconductance and yield may also berelated to heat avoidance (Reynolds etal., 1994).

Alternatively, mechanisms that preventwater loss, such as inherently lowerstomatal conductance, may limit yield

potential because of the intercellularlevels of CO2, thus decreasingphotosynthesis. These genotypes willconsistently show low Δ values (Morganet al., 1993). In fact, stomata that closeonly in response to severe water stressmay be more useful in terms of yieldthan stomata that permanently show lowconductance values (Jones, 1987).Moreover, selection for low Δ (i.e., highWUE) may favor low-producinggenotypes under drought conditions (i.e.,drought-susceptible genotypes).Therefore, low Δ may not be a goodselection criterion for improving yieldsin dry environments. Plant productionunder drought conditions depends notonly on WUE but largely on thegenotype’s capacity to sustaintranspiration (Blum, 1993).

Blum (1996) pointed out that when soilmoisture is very limited, the high-yielding genotype may be at adisadvantage because of its highstomatal conductance. In fact, it has beenreported for wheat and barley that this(crossover) happens if yield is reduced tobelow 2-3 t ha-1 (Ceccarelli and Grando,1991; Blum, 1993). Other reports ondurum wheat and barley do not supportthe existence of such a crossover inenvironments with a mean grain yield of1.5 t ha-1 or lower (Romagosa and Araus,1991; Araus et al., 1998b). In theseenvironments it may still be worthselecting for yield potential, particularlyif deep extractable soil moisture isavailable to provide a yield above that ofthe crossover of genotype yields(Richards, 1996). Summarizing, theabove results support the hypothesis thatunder Mediterranean conditions high-yielding genotypes, which sustaingreater stomatal conductance andtranspiration losses during grain filling,can provide higher yields in a wide rangeof environments with different levels ofdrought stress.

Optimal yield conditions. Carbonisotope discrimination has also beenproposed as a useful trait to select foryield potential (Araus et al., 1993c;Sayre et al., 1995; Araus, 1996). Thepositive correlation between Δ and yieldwould exist in the same context asabove. A positive relationship between Δand growth has also been reported forseedlings grown under adequate waterconditions (Febrero et al., 1992; Lopez-Castañeda et al., 1995), even when underfield conditions cold stress (usual at thisearly stage) may obscure such arelationship (Bort et al., 1998).Nevertheless, increased early growth andleaf area development may be inherentlylinked to decreased WUE (Turner, 1993)and thus to higher Δ. Blum (1996)pointed out that the data accumulated oncarbon isotope discrimination and yieldappears to support a consistent positiverelationship between crop yield andphotosynthetic capacity for variousgenetic materials of wheat and othercrops (Hall et al., 1994). If selection forhigh photosynthetic capacity or highercrop productivity brings about anincrease in stomatal conductance, then aconcomitant increase in Δ (or a decreasein crop WUE) is expected (Blum, 1996).

Role of phenology in genotypicaldifferences in Δ. In the absence of stress,Δ in wheat is independent ofphenological differences (Sayre et al.,1995). However, under non-optimalconditions the role of phenology in therelationship between Δ and yield must beconsidered. Thus, as pointed out before,in Mediterranean environmentsphenology is the most important factorthat accounts for increased wheat yields,as it affects assimilate partitioning, thepattern of water use, and other traits (seereferences in Slafer et al., 1993; Lossand Siddique, 1994). In addition, someof the genotypical differences in Δ, aswell as their positive association with


64

yield, can be due to phenology. Thusearly flowering lines are more likely tohave high Δ than later-flowering linesdue to the lower transpirative demand,which maintains higher stomatalconductances in the former (Ehdaie etal., 1991; Acevedo, 1993). Summarizing,under Mediterranean conditions early-flowering in wheat and other cereals isrelated to higher yields, which is inaccordance with higher Δ in the earliergenotypes. Alternatively, there isgenotypical variability in Δ that cannotbe explained by phenology (Condon andRichards, 1993; Richards and Condon,1993; Araus et al., 1998b) and is just dueto differences in accumulatedtranspiration.

Δ SurrogatesGiven the cost and technical skillsinvolved in carbon isotope analysis,different “surrogates” of Δ have beeninvestigated, such as mineralaccumulation in vegetative plant parts(Figures 2 and 3) and leaf structure.Instead of being Δ substitutes orsurrogates, these selection criteriaprobably allow the evaluation of traits,other than WUE, that determine yield.Regarding the first criterion, if passivetransport driven by transpiration is (atleast partially) the mechanism of mineralaccumulation in vegetative parts, thenmineral content will also be an indicatorof the first parameter of Passioura’sidentity, i.e. total water transpired. Thesecond trait corresponds to structuralcriteria that indicate the amount ofphotosynthetic tissue per unit leaf areaand is therefore related to photosyntheticcapacity. In addition, the mechanismsunderlying the physiological associationbetween Δ and either mineralaccumulation (Walker and Lance, 1991;Masle et al., 1992) or the amount ofphotosynthetic tissue (Araus et al.,1997a, b) are not fully understood.However, the empirical relationships of

these alternative criteria with Δ and yieldmay justify their use on a routine basis.Such tools might be used during theearly phases of a breeding program,which usually involve large populations.If the facilities or the funds are available,later selections could be based on moreprecise and accurate, yet expensive, Δanalysis (Mayland et al., 1993). Thesetwo alternative criteria are discussed inthe following paragraphs. Interestingly,these surrogates can be used in C4 crops(such as corn or sorghum), where Δ isnot as useful for evaluating WUE andyield itself as in C3 plants (Farquhar etal., 1989; Henderson et al., 1998).

Mineral content invegetative partsPotassium, silicon, total mineral, or ashcontent accumulated in vegetativetissues have been proposed as surrogatesof Δ in cereals, forage crops, andsoybean (Walker and Lance, 1991;Masle et al., 1992; Mayland et al., 1993;Mian et al., 1996). Masle et al. (1992)reported for all the herbaceous C3species they assayed a positive linearrelationship between total mineralcontent of vegetative tissues and theinverse of either WUE or Δ. Therefore,the amount of minerals accumulated byplants in the glasshouse or in the fieldcould be a potentially useful indicator ofΔ and WUE (Walker and Lance, 1991;Masle et al., 1992; Mayland et al., 1993).

In theory, total mineral and ash contentseem to be better surrogates than thecontent of any single mineral, such assilicon or potassium (Masle et al., 1992;Mayland et al., 1993). Therefore,estimating plant mineral content,especially ash content, which requiresonly a muffle furnace, might be anattractive alternative to Δ for preliminaryscreening of large, genetically diversepopulations (Masle et al., 1992; Araus etal., 1998b).

Methodological considerations regardingash content. Samples must be properlyoven-dried and ground. Approximately 1 to1.5 g of dry matter is placed in a pre-weighed porcelain crucible (emptycrucible), the crucible with the sample isweighed (filled crucible), and the sample isburnt in a furnace at 450BC for 12 h. Thenthe crucible with mineral residue (burntcrucible) is weighed again. Ash content isexpressed (on a concentration basis) as apercentage of sample dry weight asfollows:

Ash(burnt crucible weight -

content =empty crucible weight)

(%)

______________

x100

(filled crucible weight -empty crucible weight)

Implications for plant breeding: Choice ofenvironment and type of sample. Thepositive correlation between ash contentand Δ may indicate that plants able tomaintain higher stomatal conductance andtranspiration will accumulate more ash in atranspirative organ, provided entry andaccumulation of minerals in the plant takeplace (at least partially) through thetranspirative stream.

Which are the best growing environmentsfor using this surrogate? Mineralaccumulation seems to be better related toΔ and even to yield under well-wateredconditions (Masle et al., 1992; Mayland etal., 1993), although these traits can beuseful under drought conditions (Figure 2a;Araus et al., 1998b). Measurements takenon plants grown in the greenhouse can givecontradictory results (Walker and Lance,1991). Another question is the kind ofsample to use. Later developed leaves (flagor penultimate leaves) are best. Ashaccumulated in the flag leaf may bepositively related to the Δ of kernels (Arauset al., 1998b). Leaves must be mature to letminerals accumulate, but not senescentbecause minerals can remobilize to otherplant parts. Thus, ash content measured atmaturity in straw did not correlate witheither Δ of mature grains or yield. (Voltaset al., 1998).


65

Mineral content in maturekernels as a criterion tocomplement ΔA negative relationship between ashcontent (on a dry mass basis) in maturekernels and yield has been reported forbarley and durum wheat, under bothoptimum and non-optimum (i.e., rainfed)conditions (Febrero et al., 1994; Araus etal., 1998b; Voltas et al., 1998). This maybe explained by the fact that ash contenton a kernel mass basis may be an indirectindicator of total reproductive sink perculm attained at maturity (Araus et al.,1998b). In fact, total kernel mass perspike is the product of two yield

components developed last during thecrop cycle: number of kernels per spikeand kernel mass. Thus, kernel ash hasbeen proposed as a criterioncomplementary to kernel Δ in assessinggenotype differences in cereal yieldunder Mediterranean conditions (Febreroet al., 1994; Voltas et al., 1998). Intheory (Diagram 2), the pattern of ashaccumulation in kernels is different fromthat in vegetative tissues because, unlikemineral accumulation in vegetativetissues, grain filling does not take placevia the xylem (driven by transpiration)(Slafer et al., 1993). Such differences inmineral accumulation could explain thecomplementarity of ash content and Δ in

kernels as integrative traits predictinggrain yield (Febrero et al., 1994; Voltaset al., 1998).

Summarizing, kernel ash combined witheither kernel Δ or leaf ash can bepartially complementary (i.e. independent)parameters when assessing differences ingrain yield (Febrero et al., 1994; Araus etal., 1998b; Voltas et al., 1998). Selectingfor low ash content in kernels, combinedwith either high Δ in kernels or,alternatively, high ash content in the flagleaf, deserves further attention in wheatbreeding (Figure 3). This approach couldbe particularly interesting if it werecoupled with a new analytical technique


18.5

17

15.5

14.4

12.5

15

10

5

0

2.2

1.8

1.4

1.0

0.61500 3000 4500 6000 7500 9000

Grain yield (kg ha-1)

Figure 2. Relationship between carbon isotope discrimination (Δ) inmature kernels and ash content based on dry mass (a) of the flag leafblades around three weeks after anthesis, and (b) in the same maturekernels. Plants were cultivated in three trials differing in water status:Breda, Tel Hadya rainfed, and Tel Hadya with supplementary irrigation.

20

15

10

5

013 14 15 16 17 18

2.4

2.0

1.6

1.2

0.813 14 15 16 17 18

Δ Kernels (0/00)

Ash k

erne

ls (%

dry m

ass)

Ash f

lag le

af (%

dry m

ass)

r2 = 0.17 y = -3.67 + 0.62xr2 = 0.21 y = -10.12 + 1.19xr2 = 0.27 y = -10.42 + 1.44x

r2 = 0.09 y = 2.28 - 0.078xr2 = 0.12 y = 3.20 - 0.103xr2 = 0.09 y = 3.42 - 0.105x

r2 = 0.28 y = 1.31 - 0.00063xr2 = 0.25 y = 1.42 - 0.00044xr2 = 0.26 y = 0.56 - 0.00018x

r2 = 0.86

r2 = 0.89

a

b

Ash k

erne

ls (%

)As

h fl ag

l eaf

(%)

r2 = 0.12 y = 4.04 + 0.00062xr2 = 0.09 y = 6.35 + 0.00069xr2 = 0.28 y = 10.35 + 0.00052x

r2 = 0.28 y = 1.44 - 0.000167xr2 = 0.19 y = 1.91 - 0.000111xr2 = 0.22 y = 2.03 - 0.000057x

a

b

c

Figure 3. Relationship between grain yield and (a) carbonisotope discrimination (Δ) in mature kernels, (b) ash content(based on dry of the flag leaf around three weeks afteranthesis, and (c) ash content of the same mature kernels.

Δ K

e rne

l s (0 / 0

0)

66

such as near infrared reflectancespectroscopy (NIRS), which wouldallow a fast, reliable estimation of ashcontent and Δ in intact kernels (seeAraus, 1996).

Leaf structural criteriaChanges in Δ can derive from changes inthe balance between leaf stomatalconductance and photosyntheticcapacity. In wheat, genotypic variationsin Δ seem to derive from differences inboth stomatal conductance andphotosynthetic capacity, eachcontributing about the same (Condon etal., 1990; Condon and Richards, 1993;Morgan et al., 1993). If the intrinsicphotosynthetic capacity of leaves is

increased, Δ could decrease and WUEcould be improved, withoutcompromising yield potential (see above).Therefore, a negative relationshipbetween Δ and yield could be expectedeven in the absence of stress.

Genotypic differences in photosyntheticcapacity may depend on the amount ofphotosynthetic tissue per unit leaf area.Thus, single structural parameters suchas dry mass per unit leaf area (LDM, thereciprocal of specific leaf area, alsotermed specific leaf dry weight, orSLDW) or total nitrogen or chlorophyllcontent per unit leaf area may be goodindicators of the strength ofphotosynthetic tissue (see references in

Araus et al., 1989; Nageswara Rao andWright, 1994). For example, totalchlorophyll content per leaf area maybe evaluated in a fast, single and non-destructive way using a portablechlorophyll meter like the SPAD-502(Soil-Plant Analysis DevelopmentSection, Minolta Camera Co., LTd.,Japan). Usually the leaf parameter thatcorrelates negatively best with Δ isLDM, followed by SPAD, whichindicates that genotypes with thickerand/or more compact leaves have lowerΔ. The results suggest that LDM andSPAD measurements can be used assingle, rapid indicators of Δ in barley(Araus et al., 1997a) and durum wheat(Araus et al., 1997b) under optimalgrowing conditions (see also Lopez-Castañeda et al., 1995).

However, some of these correlationsmay exist under drought conditions andcould be useful for breeding, but maybe spurious in nature. In fact, growingconditions have a strong direct effectnot only on Δ, but also on leaf structure,which in turn could lead to spuriousrelationships (Araus et al., 1997b). Thecorrelations between Δ and leafstructure, rather than being sustained bya physiological relationship betweenthe amount of photosynthetic tissue andΔ, may sometimes be indirectassociations caused by a parallel effectof water status and phenology on leafstructure, grain Δ, and yield (Araus etal., 1997a, b). Summarizing, LDMshould be used only in the absence ofdrought to determine segregatingpopulation differences in leaf Δ basedon internal photosynthetic capacity. It isworth selecting for higher kernel Δ andgrain yield based on higher LDM inrainfed trials, although there probably isno direct physiological basis behindsuch relationships (Araus et al., 1997b).

C(H20)

C(H20)Retranslocation

H20

Evapotranspirationof water

K+

K+

C a++

C a++

C a++

N a+

N a+

H20

H20

H20C I-

Accumulation of C(H20)dilutes mineral

concentration in grains

Mineralaccumulation

in tissue

H20 + Mineralstransported by xylem to

leaves

Water & mineralsin soil

Diagram 2. Accumulation of ash in plants.


67

Spectral ReflectanceMethods

The pattern of light reflection on leavesat different wavelengths through thephotosynthetically active radiation(PAR, 400-700 nm) and near infraredradiation (NIR, 700-1200 nm) regions ofthe electromagnetic spectrum is verydifferent from that of soil and othermaterials (Diagram 3). Leaf pigmentsabsorb light strongly in the PAR regionbut not in the NIR, thus reducing thereflection of PAR but not of NIR. Such apattern of pigment absorptiondetermines the characteristic reflectancesignature of leaves (Figure 4). Similarly,the light spectrum reflected by a canopy(either natural or agricultural) differsfrom that reflected by the bare soil andvaries in a way that can be related to theoverall area of leaves and otherphotosynthetic organs in the canopy, aswell as to their pigment composition andother physiological factors (Figure 5).Therefore, the measurement of spectrareflected by vegetation canopiesprovides information that can be used toestimate a large scope of parameters.Some of them are related to the greenbiomass of the canopy, itsphotosynthetic size (i.e., total area ofleaves and other photosynthetic organs),the amount of PAR absorbed by thecanopy, and its photosynthetic potential.Other parameters are more related to thecanopy’s physiological status at the timeof measurement and can be used toassess the extent of some nutrientdeficiencies and environmental stresses.The physiological parameters that canbe estimated by spectral reflectancetechniques include chlorophyll andcarotenoid concentrations,photosynthetic radiation use efficiency(PRUE), and water content.

RADIATIONRADIATION

Refle

ctan c

e

Diagram 3. Spectral reflectance from crop surfaces.

1.0

0.8

0.6

0.4

0.2

0.0400 500 600 700 800 900 1000

Wavelength (nm)

Figure 4. Reflectance signature of two wheat leaves differing in nitrogen status. Note thehigher reflectance in the photosynthetically active radiation region of the nitrogen deficientleaf due to lower chlorophyll content in the leaf area.

N-deficient

Control


68

Spectral reflectance indicesSpectral reflectance indices areformulations based on simple operationsbetween reflectances at givenwavelengths, such as ratios, differences,etc., which are widely used toquantitatively relate changes inreflectance spectra to changes inphysiological variables. These indiceshave the advantage of summing up in afew numbers the large amount ofinformation contained in a reflectancespectrum with narrow wavebandresolution.

Originally used in remote sensing byaircraft and satellites, reflectancesmeasured at the ground level are veryuseful for assessing agrophysiologicaltraits. These traits can be evaluatedsimultaneously in each sample, at a rateof up to one thousand samples per day,which is much more tedious and timeconsuming with other methods. Thismakes spectroradiometric indices idealfor screening for yield potential or forresistance to different stresses.

Sample applicationsPerhaps the most widespread applicationof reflectance indices is for assessingparameters related to canopy greenness.These parameters are related to thecanopy’s photosynthetic size and includegreen biomass, leaf area index (LAI)(total one-side leaf area of the croprelative to soil area), green leaf areaindex (GLAI) (similar to LAI, butincludes only functional green leaves),and green area index (GAI) (similar toGLAI, but includes other photosyntheticorgans such as green stems). The amountof green area in a canopy determinesPAR absorption by photosyntheticorgans, which in turn determines thecanopy’s potential production. Thefraction of the incident PAR that isabsorbed by the canopy (fPAR) can beestimated from LAI-related parametersor directly from reflectancemeasurements. Cumulative PARabsorption, which is one of theparameters determining total biomassand thus final yield (see the beginning of

this chapter), can be assessed bymeasuring reflectance periodicallyduring the growth cycle.

Some physiological parameters can alsobe quantified by spectral indices. Leafpigments can be detected and quantifiedbased on reflectance spectra and can beused as indicators of severalphysiological processes. Thus, thecanopy’s nutritional state can beevaluated through pigmentconcentration, as chlorophyll (Chl)concentration in leaves is (usually)closely correlated to its nitrogen content.Indices that are good indicators of Chlare (usually) also good indicators of N-content. In addition, plants with low Nusually have a high carotenoid (Car) toChl ratio, which can also be assessed byreflectance indices (Figure 5).

Pigment remote sensing can also beused for assessing the crop’sphenological stage (Figure 5) and theoccurrence of several stress factors(Blackburn, 1998; Peñuelas, 1998). Forexample, the Car to Chl ratio can beassociated with senescing processes thatresult from the plant’s natural ontogenypattern or are triggered by differentstresses. Also, phenological stages canbe associated with different Car/Chlvalues. Several indices related tochanges in pigment composition havebeen developed and can be used for theremote detection of nutrientdeficiencies, environmental stresses,pest attacks, etc. In such contexts, byperiodically assessing leaf area, leaf areaduration (LAD) can also be used as anindicator of resistance to certainenvironmental stresses.

The photosynthetic capacity of a canopycan be estimated by using vegetationindices that correlate to thephotosynthetic size of the canopy orindices related to the amount ofchlorophyll. However, actualphotosynthesis may not match

0.6

0.5

0.4

0.3

0.2

0.1

0.0400 500 600 700 800 900

Wavelength (nm)

Figure 5. Changes in the pattern of canopy reflectance in a durum wheat plot.Measurements were taken every three days (a, b, c), during the last week of grainfilling,coinciding with fast crop senescence. Note the decrease, during senescence, in the amplitudeof the change in reflectance in the red-NIR (around 700 nm) edge. Note also the increasewithin the PAR region of the reflectance in the red compared to the blue band due to arelatively faster decrease during senescence in chlorophyll compared to carotenoids. Soilreflectance is included for comparison.

Refle

ct anc

e

a

b

c

soil


69

photosynthetic capacity due to thevariability of photosynthetic useefficiency of the absorbed radiation,especially when plants are exposed tounfavorable conditions. Thephotochemical reflectance index (PRI)was developed to detect pigmentchanges in the xantophyll cycleassociated with changes in PRUE(Filella et al., 1996). PRI has beenshown to track the changes in PRUEinduced by factors such as nutritionalstatus and midday reduction, acrossdifferent species and functional types.

Another potential application ofreflectance indices is remote detection ofrelative water content (RWC) of plants.Different levels of water stress can bedetected indirectly through their effectson vegetation indices related to leaf area,pigment concentration, or photochemicalefficiency. In addition, specific indiceshave been developed for the directassessment of RWC.

Measurement techniquesInstruments. The instruments requiredfor measuring reflectance spectra are: 1)a field spectroradiometer that analyzesthe spectrum of sampled radiation, 2)foreoptics that capture the radiationreflected by a given target, and 3)reference panels, supports, and levels forrepeated sampling of incident radiationand radiation reflected by the canopy.

Modern narrow-band spectro-radiometers measure the irradiance atdifferent wavelengths with a bandwidthof about 2 nm through the PAR and NIRregions of the spectrum. Most spectralindices use specific wavebands in the400-900 nm range; only a few use longerwavelengths, such as the water index,which uses 970 nm (Peñuelas et al.,1993). The use of spectroradiometerswith narrow band resolution allows thecalculation of several parametersobtained from the first and second

derivative of the reflectance spectraagainst wavelength, which can be usedto complement the reflectance indices.

Radiation reflected by the canopy in thePAR and NIR regions is sampled by aforeoptic that limits the field of view to agiven solid angle, usually between 10and 25º. Sampled radiation is conveyedto the spectrum analyser through a fiberoptic cable. Light reflected by thecanopy is measured with the foreopticheld 1-2 m above the canopy on a fixedor hand-held support, such as a boom(Picture 1), and with the help of therequired levels or protractors to ensurethat all measurements are taken at thesame angle between the foreoptics andthe sampled surface.

In order to cross-reference the intensityof reflected radiation at each wavelengthto the intensity of incident radiation atthe same wavelength, all sampled

spectra must be converted toreflectance units, i.e., the ratio betweenthe absolute spectrum reflected by thecanopy and the absolute spectrumincident on the canopy. Regularmeasurements of the spectra incidenton the canopy are then made. Incidentspectra are measured by aiming theforeoptic at a white reference panel inthe same orientation to the sun and tothe foreoptic as the canopy. Referencepanels are commercially availableunder the name Spectralon (Labsphere,PO Box 70, North Sutton, NH 03260)or they can be made of barium sulphate(Jackson et al., 1992).

Factors affecting the estimation ofcanopy parameters byspectroradiometrical methods. Inaddition to canopy variables estimatedusing spectroradiometrical methods,other factors related to the canopy orexternal to it will affect the measured

Picture 1. How to place the foreoptic while measuring radiation reflected by awheat canopy.


70

reflectance spectra. Variation in canopystructure (such as changes in leaferectness or appearance of reproductiveorgans), as well as in the angles betweensun, sensor, and target surface, willaffect the amount of shadow and/or soilbackground appearing in the field ofview; this can cause non-desiredvariation in the measured spectra.

There are no standard methods to copewith the variability introduced byinterference; most researchers usingspectroradiometry adapt the details oftheir experimental protocols to theparticular traits and objectives of theirexperiments. It is important to fix themeasuring conditions used to obtain thespectra. Viewing angle and viewingheight, row orientation, and time of dayhave to be determined when designingan experiment. Disturbance by factorsbeyond the researcher’s control has to beconsidered when interpreting the results.Not all indices are equally affected bythese factors, and indices also differ intheir sensitivity to the parameter beingmeasured. Some indices may be moreappropriate than others, depending onthe aims of the study, canopycharacteristics, and measurementconditions.

To minimize the variability induced bysun position, it is preferable to take allmeasurements at about noon.Nevertheless, the angle of the sun ismost important in canopies with lowLAI (Kollenkark et al., 1982; Ranson etal., 1985). As for the viewing angle,nadir (sensor looking verticallydownward) is perhaps the mostcommonly used set-up. This is becauseit has a lower interaction with sunposition and row orientation, and delaysthe time at which spectra becomesaturated by LAI. On the other hand,nadir viewing is more affected by thesoil background. When an off-nadir

viewing angle is used, variability due tochanges in solar elevation or sensorelevation is minimized if the anglebetween the sensor azimuth and the sunazimuth is 0-90º (Wardley, 1984).

In a row canopy with low soil cover, theamount of shadow within the canopyvaries during the day, depending on theangle between sun azimuth and roworientation. Such angular changes canproduce variation in the measuredreflectance as great as 100% in red andlower in NIR wavelengths (Kollenkarket al., 1982). Peak variability occurswhen the sun is shining down the rows(when sun azimuth equals roworientation), lightening the soil surfaceand thus giving a higher reflectancereading. For that reason, if reflectance ismeasured at about noon, rows orientedeast to west are more appropriate thanrows oriented south to north, especiallyif soil cover is poor.

Ratio indices are usually less sensitive tochanges in viewing geometry and tend tocancel the effects for angular changes(Wardley, 1984). However, they can alsobe altered because at some wavelengths(such as in red) reflectance is (usually)more intensely altered than at otherwavelengths (such as in NIR). Lightincident on shaded leaves is poor in thewavelengths that have been absorbed byupper leaves, and their reflected spectrais even poorer. For that reason, thehigher the number of shaded leaves thatappears in the field of view, the largerthe differences in the canopy’sreflectance spectra between regionswhere radiation is absorbed byphotosynthetic pigments and regionswhere it is not. If due to external factorssuch as viewing angle, sun angle, orwind, the number of shaded leaves in thefield of view increases, this will lead toan increase in indices related to greenbiomass.

The relationship between indices andestimated canopy parameters has beenreported to be disturbed by phenologicalchanges that affect crop structure, such asthose associated with anthesis in maize(Andrieu and Baret, 1993) or heademergence in wheat (Shibayama et al.,1986). Leaf erectness can also affectcanopy reflectance. Model calculationsand test results show that radiationreflected perpendicularly from plantcanopies is considerably greater fromplanophile canopies than fromerectophile canopies (Jackson and Pinter,1986). The vertical elements of anerectophile canopy trap reflectedradiation within the canopy, while in amore planophile canopy, more radiationis reflected vertically. These structuraleffects can alter indices used forestimating the same canopy parameter ina different way. For example, Jacksonand Pinter (1986) observed that althoughindices SR and PVI (see later in thischapter) are both used for estimatingGLAI, SR was higher in erectophilecanopies of wheat, while PVI was higherin planophile canopies. Opticaldifferences in the surface of plant organs,such as different glaucousness (Febreroet al., 1998), can also have some effecton the canopy’s reflectance spectra.

Clouds increase the proportion ofindirect radiation (i.e., diffuse) to totalradiation incident on the canopy; thisimproves the penetration of light into thecanopy. As a result, a greater proportionof incoming radiation is absorbed byphotosynthetic pigments; this increasesthe vegetation indices and leads to anoverestimation of green biomass. Windduring the measurements canmomentarily alter canopy structure anddisturb the relationship between thereflectance spectra and the canopyparameters to be estimated from thespectra (Lord et al., 1985).


71

Nearby objects, including instrumentsand the people operating them, can alterthe measured spectra by reflectingradiation on the target surface. For thatreason, they should be kept as far aspossible from the field of view; theinstruments should be painted a darkcolor, and people should wear darkclothes (Kimes et al., 1983).

Taking measurements. Systematicmeasurements of incident radiation mustbe made before and during themeasurement of reflected radiation toaccount for possible variation in theincident spectra caused by atmosphericconditions or sun position.

Reference panels should be Lambertiansurfaces, that is, they reflect the incidentlight equally in all directions and for allwavelengths. However, they are notperfect and the intensity of the reflectionchanges in an important way when panelorientation changes. For that reason, caremust be taken to make all incidentmeasurements keeping the panel at thesame angle with the foreoptics and withthe sun. Changes in the distance fromthe panel to the foreoptics are lessimportant. This distance is set to ensurethat the entire field of view is covered bythe panel. Then the reflectance of thecanopy samples can be measuredmaking sure that the field of view of theinstrument is covered with weed-freecanopy and a uniform background, andwith plant material homogeneous instructure (Bellairs et al., 1996).

Use of CanopyReflectance Indices

Assessing the photosyntheticsize of canopies usingvegetation indicesVegetation indices (VI) estimateparameters related to the photosyntheticsize of a canopy based on thereflectances in the red and near infrared

regions. Green biomass, LAI, GAI,GLAI, fPAR, etc., can be estimatedthrough their positive correlation (eitherlinear or logarithmic) with vegetationindices (Wiegand and Richardson,1990a, b; Baret and Guyot, 1991; Priceand Bausch, 1995). Measuringvegetation indices periodically duringthe crop growing cycle allows theestimation of LAD (which can be usedas an indicator of environmental stresstolerance) and the total PAR absorbed bythe canopy, which is one of the mostimportant factors for predicting yield(Wiegand and Richardson, 1990).

Vegetation indices take advantage of thegreat differences in reflectance at redand NIR caused by vegetation. The mostwidely used VI are the simple ratio (SR)and the normalized difference vegetationindex (NDVI), which are defined as:

SR = RNIR / RRed, with a range of 0 to ∞,

where RNIR is the reflectance at NIR andRRed is the reflectance at red.

NDVI = (RNIR - RRed) / (RNIR + RRed), with a rangeof -1 to 1.

SR and NDVI were originally used withthe wide wavebands of formerradiometers (for example, 550-670 nmfor red and 710-980 nm for near infraredin AVHRR radiometers in satellites ofNOAA series). With the high spectralresolution of today’s radiometers,wavebands can be much narrower. Hallet al. (1990) used a waveband centeredat 770 nm for NIR and another at 660nm for red, while Peñuelas et al. (1997b)used 900 nm and 680 nm for NIR andred, respectively.

Some authors have reportedimprovements in NDVI performanceafter changing the wavebands used inthe index. Carter (1998) describes animproved correlation with leafphotosynthetic capacity when using a

modified NDVI where R701 (+/-2nm)and R520 (+/-2nm) were used for NIRand red, respectively.

Variations of these indices have beenproposed to compensate for the effect ofsoil background. Thus the soil adjustedvegetation index (SAVI) was defined byHuete (1988) as:

SAVI = [ (RNIR - RRed) / RNIR + RRed + L) ] (1 + L),

where the parameter L was adjusted tominimize noise caused by soil for a largerange of soil covers. For most cropconditions L=0.5, while for very low soilcovers L=1 would be more appropriate,and L=0.25 would be appropriate forvery high covers (Huete, 1988).

Other indices include parametersobtained from the soil’s reflectancespectrum. One of them is thetransformed soil adjusted vegetationindex (TSAVI) which was defined byBaret and Guyot (1991) as:

TSAVI = a(RNIR - aRRed - b) / [RRed +

a(RNIR - b) + 0.08(1+a2)],

where a is the slope and b is the interceptof the linear equation

RNIRsoil = a*RRed soil + b.

An important drawback in estimatingLAI by VI is the saturation of the VIwith LAI. Saturation of NDVI starts atabout LAI=1, and beyond LAI=2 itbecomes insensitive to further increasesin LAI (Gamon et al., 1995).Perpendicular vegetation index (PVI)partly overcomes the saturation probleminherent to NDVI (Richardson andWiegand, 1977):

PVI ={(RRed.soio-RRed.vegetation)2 +

(RNIR.vegetation-R NIR.soil) 2}1/2

Although PVI is more sensitive thanNDVI to changes in the viewing


72

geometry, PVI does not become asclearly saturated as NDVI with changesin GLAI (Shibayama et al., 1986).

Examples of assessing LAI-relatedparameters by VI can be found in theliterature (Baret and Guyot, 1991; Fieldet al., 1994; Price and Bausch, 1995).Ground level measurement of VI hasbeen used successfully as a tool forassessing early biomass and vigor ofdifferent wheat genotypes (Elliott andRegan, 1993; Bellairs et al., 1996).Under experimental conditions of awheat breeding program, Bellairs et al.(1996) reported young wheat canopieswhere LAI was less than 1.5, acoefficient determination (r2) of 0.90-0.95 between biomass and NDVI. As forassessing the intensity of different plantstresses, Peñuelas et al. (1997b) showedthat NDVI was a useful tool formeasuring agronomic responses ofbarley to salinity.

A practical use of vegetation indices isfor making yield predictions. Yield canbe predicted from successive VImeasurements taken during the growingseason, based on the followingassumptions (Wiegand et al., 1991): 1)plant stands integrate the growingconditions experienced and express netassimilation achieved through thecanopy, 2) stresses severe enough toaffect economic yield will be detectablethrough their effects on cropdevelopment and the persistence ofphotosynthetically active tissue in thecanopy, 3) high economic yields cannotbe achieved unless plant canopies fullyutilize available solar radiation as theplants enter the reproductive stage, and4) vegetation indices calculated fromremote observations in appropriatewavelengths effectively measure thephotosynthetic size of the canopy.Wiegand and Richardson (1990b)reported an r2 of 0.5 for predicting wheatgrain yield from PVI measured on four

dates during vegetative growth.Similarly, Rudorff and Batista (1990)reported an r2 of 0.66 between wheatyield and integrated VI from booting tocompletely senesced plants. If mostuncertainty in yield prediction by VI issite-dependent, then calibrations of yieldvs. VI across good and poor growingconditions within production areas candescribe the results of past and futuregrowing seasons acceptably (Wiegand etal., 1991).

Remote sensing of pigmentsEstimating chlorophyll concentration.Several indices have been developed forestimating Chl concentration usingcanopy reflectance methods. Thesimplest indices are just reflectance at675 and 550 nm. Reflectance at 675 nm(R675) is very sensitive to changes inChl content. However, the relationshipbecomes saturated at relatively low Chlvalues (around 10 µg cm-2) and is a goodindicator of chlorophyll content only atvery low concentrations. Absorption byChl at 550 nm is lower than at 675 nm;therefore, the reflectance at thiswavelength (R550) is less sensitive tochanges in Chl content but is notsaturated at such low concentrations,thus covering a range of higher Chlvalues (Thomas and Gausman, 1977;Jacquemoud and Baret, 1990;Lichtenthaler et al., 1996).

Both R675 and R550 are non-normalizedindices that can be affected by externalfactors (Curran, 1983). Other indices usemore than one wavelength. Analyzingwavelengths that were more sensitive tochanges in Chla, Chlb, and Cars insoybean leaves grown at different Nlevels, Chapelle et al. (1992) developedthe ratio analysis of reflectance spectra(RARS) indices, RARSa, RARSb andRARSc, which optimized the estimation ofChla, Chlb, and Cars, respectively, insoybean leaves.

RARSa = R675 / R700 showed adetermination coefficient of 0.93, withChla ranging from 0.4 to 27 µg cm-2;RARSb = R675 / (R650 * R700) showed anr2 of 0.82, with Chlb ranging from 1 to 7µg cm-2; and RARSc = R760 / R500showed an r2 of 0.94, with Cars rangingfrom 1.5 to 6µg cm-2 (Chapelle et al.,1992). Blackburn (1998) reported thatusing R680 and R800, instead of R675and R700, in RARSa significantlyimproved the prediction of Chla in arange of leaves from different specieswith different degrees of senescence.Other reflectance indices that can be usedfor estimating pigment concentration aresummarized in Table 1.

Leaf chlorophyll content can also beassessed through its relationship withparameters derived from the position ofthe red edge. The red edge position(REP) is the wavelength in the 680-780nm range where the change inreflectance when increasing thewavelength from red to NIR reaches itsmaximum. The REP shifts to slightlylonger wavelengths as Chla valuesincrease (Curran et al., 1990; Filella etal., 1995). By obtaining the first andsecond derivatives of the spectra in thisarea, several parameters that are goodindicators of Chl content can becalculated. Among these parameters arethe wavelength of the red edge (λre), themaximum amplitude in the firstderivative of the reflectance spectra(dRre), and the sum of amplitudesbetween 680 and 780 nm in the firstderivative of the reflectance spectra(∑dR680-780). These REP-relatedparameters are suitable indicators ofchlorophyll content in a wider andhigher range of concentration than R675and R550, with the additional advantagethat they are less affected by externalfactors such as the geometry, incidentintensity, and soil background (Filellaand Peñuelas, 1995).


73

In addition to the wide variety of indicesrelated to absolute Chl concentration, thenormalized phaeophytinization index(NPQI) can be used to detect chlorophylldegradation.

NPQI = (R415 - R435) / (R415 + R435)

(Peñuelas et al., 1995c)

NPQI was introduced as an indicator ofpest attacks on apple trees (Peñuelas etal., 1995c). In some cases it also seems toindicate different phenological statesin wheat (Casadesús and Araus,unpublished data).

One practical approach for estimatingChl concentration using reflectanceindices is to test the performance of morethan one index and choose the one mostappropriate for the experiment. Anotherapproach is to pool the informationcontained in a number of indices. In thissense, Filella et al. (1995) were able toassign different reflectance spectra todifferent N-status classes using adiscriminant analysis based on R430,

R550, R680, λre, dRre, and NDPI(defined later in this chapter). Non-destructive portable chlorophyll metersbased on absorbance measurementsthrough the leaf provide fast and easydeterminations of chlorophyll contentand are commercially available at arelatively low price. For example, theSPAD-502 mentioned above calculatesthe ratio of absorbances at 650 nm λ(chlorophyll absorbance peak) and at940 nm (non-chlorophyll absorbance)(Monje and Bugbee, 1992). Estimates ofchlorophyll using canopy spectralreflectance methods are in generalclosely related to the amount ofchlorophyll per soil area calculated fromthe reading of portable chlorophyllmeters multiplied by the LAI (Filella etal., 1995). Chlorophyll assessment usingcanopy reflectance methods has theadvantage that it directly integrates thechlorophyll content of all the leaves inthe canopy. It also offers additionalinformation such as canopy size andcontent of pigments other thanchlorophyll.

Carotenoid to chlorophyll ratios.Estimating the Car: Chl ratio byreflectance indices can be useful forassessing the extent of some plantstresses, given that increases in Carsconcentration relative to Chl are oftenobserved when plants are subjected tostress (Young and Britton, 1990).

Both Chl and Car absorb in the blue, butonly Chl absorbs in the red. Indices thatare combinations of the reflectance inthese two regions are correlated to theCar : Chl ratio. The simplest indices arepigment simple ratio (PSR) andnormalized difference pigment index(NDPI), which are formulated in ananalog way to SR and NDVI and definedto estimate the ratio of total pigments toChla (Peñuelas et al., 1993):

PSR = R430 / R680 , NDPI = (R680 - R430) /

(R680 + R430)

Both PSR and NDPI are affected bydisrupting effects introduced by leafsurface and structure. A new index wasdeveloped to avoid such problems: thestructural independent pigment index(SIPI), which was defined by Peñuelas etal. (1995a) as:

SIPI = (R800 - R435) / (R415 + R435)

SIPI uses wavelengths showing the bestsemi-empirical estimation of the Car :Chla ratio, and its formulation minimizesthe disrupting effects of leaf surface andmesophyll structure (Peñuelas et al.,1995a). R800 is used as a referencewhere neither Cars nor Chl absorb andare only affected by the structure. Thebest fit between the Cars : Chla ratio andSIPI for a variety of plants with Chlaranging from 0.06 to 54 µg cm-2 andCars from 1 to 16 µg cm-2 wasexponential, with an r2 of 0.98 and theform, Cars : Chla = 4.44 - 6.77exp-0.48

SIPI (Peñuelas et al., 1995a).


Table 1. Reflectance indices for estimating pigment concentration.

Pigment Definition Reference

Chl R675 Jacquemoud and Baret, 1990

R550 Jacquemoud and Baret, 1990

R750/550 Lichtenthaler et al., 1996Gitelson and Merzlyak, 1997

R750/700 Lichtenthaler et al., 1996Gitelson and Merzlyak, 1997

NDVI green = (RNIR – R540–570)/RNIR +R540 –570) Gitelson and Merzlyak, 1997

λre, dRre and ∑dR680-780 Filella et al., 1995

Chla RARSa = R675/R700 Chapelle et al., 1992RARSa* = R680/R800 Blackburn, 1998PSSRa = R800/R675 Blackburn, 1998

Chlb RARSb = R675/(R650 * R700) Chapelle et al., 1992PSSRb = R800/R650 Blackburn, 1998

Cars RARSc = R760/ R500 Chapelle et al., 1992

Cars/Chla SIPI = (R800–R435)/(R415+R435) Peñuelas et al., 1992

74

Indices related to the Cars : Chl ratiochange during the crop growing cycle.They are low during vegetative growthand start to increase before thebeginning of senescence (Filella et al.,1995). They can be used in assessingthe nutritional state of a crop (Filella etal., 1995), shown by high values of theindices when N is low, and fordetecting pest attacks (Peñuelas et al.,1995c).

Assessing radiation useefficiency by PRICanopy photosynthesis can be roughlyestimated based on the estimation ofthe canopy’s photosynthetic size or Chlconcentration. However, theseparameters are associated withpotential canopy photosynthesis, whichdoes not always correspond to actualphotosynthesis, especially for plantsgrowing in stressful environments.While VI are correlated with PARabsorption by the canopy (a slowlyvarying trait, in a range of days toweeks), the photochemical reflectanceindex (PRI) is correlated withphotosynthetic radiation use efficiency(PRUE) of absorbed PAR, a rapidlyvarying process, in a range of hours.

Part of the PAR absorbed by Chlcannot be used for photosynthesis andis lost mainly through heat dissipation,which is linked to the xanthophyll-de-epoxidation cycle (Demmig-Adamsand Adams, 1996). PRI reflectschanges in reflectance of around 531nm, which have been associated withpigment changes in the xanthophyllscycle (Gamon et al., 1992).

PRI was originally defined asphysiological reflectance index(Gamon et al., 1992) but later thedefinition was slightly modified (itssign was changed) and the name of the

index was revised as photochemicalreflectance index (Peñuelas et al.,1995b). Here PRI refers to the seconddefinition.

PRI = (R531 - R570) / (R531 + R570)

(Peñuelas et al., 1995b)

PRI is correlated with the de-epoxidation stage of the xanthophyllscycle, with zeaxanthin, and withradiation-use efficiency (Filella et al.,1996). Higher PRI values indicategreater efficiency.

PRI has been shown to track changes inphotosynthetic radiation use efficiencyinduced by different factors such asnutritional state and midday reduction,across different species and functionaltypes (Gamon et al., 1997). However, itdoes not properly track changes inPRUE if there are structural changes inthe canopy associated with stress, suchas leaf wilting (Gamon et al., 1992).Also, this index is valid only for fullyilluminated canopies and does notperform properly across wide ranges ofillumination from shade to sun (Gamonet al., 1997).

Directly assessing plantwater statusSome bands of radiation absorption bywater exist in the 1300-2500 nm region,but due to its high absorptance in thisregion, reflectance becomes saturated(i.e., it does not respond to furtherincreases in RWC) even in a canopywith low water content. In the 950-970nm region, there is some weakabsorption of radiation by water that isnot saturated for a moderately drycanopy. The reflectance at 970 has beenused in the definition of the waterindex (WI).

WI = R900 / R970 (Peñuelas et al., 1993, 1997)

In WI, reflectance at 970 nm is takenas a wavelength sensitive to watercontent, while reflectance at 900 nm istaken as a reference which is similarlyaffected by canopy and leaf structuresbut with null absorption by water.

WI has been used to track changes inRWC, leaf water potential, stomatalconductance, and foliage minus airtemperature differences when plantwater stress is well developed(RWC<0.85) (Peñuelas et al., 1993).Peñuelas et al. (1997a) reported acorrelation coefficient of around 0.55between WI and RWC for a range ofspecies measured at different times ofthe year in their natural Mediterraneanenvironment. However, WI appears tobe quite insensitive until the dryingprocess is well advanced. For thatreason, WI can be useful for assessingwild fire risk but has less utility inirrigation scheduling. As for stressdetection, Peñuelas et al. (1997b)showed that WI was a good indicatorof water status in response to salinity.

NDVI is also affected by the dryingprocess and by structural and colorchanges in the plants. The ratio of WIand NDVI has a better correlation withRWC increases, especially in speciesthat undergo important changes in NDVIthroughout the year (Peñuelas et al.,1997a).

Acknowledgments

This work was supported in part byCICYT (Spain), grants AGF95-1008-C05-03 and AGF96-1137-C02-01. Weare grateful to Dr. M.M. Nachit for hisgenerosity in providing unpublishedresults on durum wheat.


75

ReferencesAcevedo, E. 1993. Potential of carbon isotope

discrimination as a selection criterion inbarley breeding. In: Stable Isotopes and PlantCarbon-Water Relations. J.R. Ehleringer, A.E.Hall, and G.D. Farquhar (eds.). AcademicPress, New York. pp. 399-417.

Andrieu, B., and Baret, F. 1993. Indirect methodsof estimating crop structure from opticalmeasurements. In: Crop Structure and LightMicroclimate. C. Varlet-Grancher, R.Bonhomme, and H. Sinoquet (eds.). INRAEditions, Paris. pp. 285-322.

Araus, J.L. 1996. Integrative physiological criteriaassociated with yield potential. In:Increasing Yield Potential in Wheat:Breaking the Barriers. M.P. Reynolds, S.Rajaram, and A. McNab (eds.). Mexico,D.F.: CIMMYT. pp. 150-167.

Araus, J.L., Tapia, L. and Alegre, L. 1989. Theeffect of changing sowing date on leafstructure and gas exchange characteristics ofwheat flag leaves grown underMediterranean conditions. J. Exp. Botany40:639-646.

Araus, J.L., Febrero, A., Bort, J., Santiveri, P., andRomagosa, I. 1993a. Carbon isotopediscrimination, water use efficiency andyield in cereals: Some case studies. In:Tolérance a la Sécheresse des Céréales enZone Méditerranéenne. Diversité Génétiqueet Amélioration Variétale. Ph. Monneveuxand M. Ben Salem (eds.). Les Colloques n164, Ed. INRA, Paris. pp. 47-60.

Araus, J.L., Brown, H.R., Febrero, A., Bort, J.,and Serret, M.D. 1993b. Ear photosynthesis,carbon isotope discrimination and thecontribution of respiratory CO2 todifferences in grain mass in durum wheat.Plant, Cell and Environment 16:383-392.

Araus, J.L., Reynolds, M.P., and Acevedo, E.1993c. Leaf posture, grain yield, growth,leaf structure and carbon isotopediscrimination in wheat. Crop Sci.33:1273-1279.

Araus, J.L., Bort, J., Ceccarelli, S., and Grando, S.1997a. Relationship between leaf structureand carbon isotope discrimination in fieldgrown barley. Plant Physiology andBiochemistry 35:533-541.

Araus, J.L., Amaro, T., Zuhair, Y., and Nachit,M.M. 1997b. Effect of leaf structure andwater status on carbon isotopediscrimination in field-grown durum wheat.Plant Cell and Environment 20:1484-1494.

Araus, J.L., Amaro, T., Voltas, J., Nakkoul, H.,and Nachit, M.M. 1998a. Chlorophyllfluorescence as a selection criterion for grainyield in durum wheat under Mediterraneanconditions. Field Crops Res. 55:209-223.

Araus, J.L., Amaro, T., Casadesús, J., Asbati, A.,and Nachit, M.M. 1998b. Relationshipsbetween ash content, carbon isotopediscrimination and yield in durum wheat.Austr. J. Plant Phys. 25:835-842.

Baret, F., and Guyot, G. 1991. Potentials andlimits of vegetation indices for LAI andAPAR estimation. Remote Sensing ofEnvironment 35:161-173.

Bellairs, S.M., Turner, N.C., Hick, P.T., and Smith,R.C.G. 1996. Plant and soil influences onestimating biomass of wheat in plant-breedingplots using field spectral radiometers Aust. J.Agric. Res. 47: 1017-1034.

Blackburn, G.A. 1998. Spectral indexes forestimating photosynthetic pigmentconcentrations - A test using senescent treeleaves. International Journal of RemoteSensing 19:657-675.

Blum, A. 1993. Selection for sustainedproduction in water-deficit environments.In: International Crop Science. I. D.R.Buxton, R. Shibles, R.A. Forsberg, B.L.Blad, K.H. Asay, G.M. Paulsen, and R.F.Wilson (eds.). Crop Science Society ofAmerica, Madison. pp. 343-347.

Blum, A. 1996. Yield potential and droughtresistance: are they mutually exclusive? In:Increasing Yield Potential in Wheat:Breaking the Barriers. M.P. Reynolds, S.Rajaram, and A. McNab (eds.). Mexico,D.F.: CIMMYT. pp. 76-89.

Bort, J., Araus, J.L., Hazzam, H., Grando, S., andCeccarelli, S. 1998. Relationships betweenearly vigour, grain yield, leaf structure andstable isotope composition in field grownbarley. Plant Physiology and Biochemistry36 (12):889-897.

Carter, G.A. 1998. Reflectance wavebands andindexes for remote estimation ofphotosynthesis and stomatal conductance inpine canopies Remote Sensing ofEnvironment 63:61-72.

Ceccarelli, S., and Grando, S. 1991. Selectionenvironmentand environmental sensitivityin barley. Euphytica 57:157-167.

Chappelle, E.W., Kim, M.S., and McMurtrey, J.E.1992. Ratio analysis of reflectance spectra(RARS): an algorithm for the remoteestimation of the concentrations of Chl a, band carotenoids in soybean leaves. RemoteSensing of Environment 39:239-247.

Condon, A.G., Richards, R.A., and Farquhar,G.D. 1987. Carbon isotope discrimination ispositively correlated with grain yield anddry matter production in field-grown wheat.Crop Sci. 27:996-1001.

Condon, A.G., Farquhar, G.D., and Richards,R.A. 1990. Genotypic variation in carbonisotope discrimination and transpirationefficiency in wheat. Leaf gas exchange andwhole plant studies. Aust. J. Plant Phys.17:9-22.

Condon, A.G., and Richards, R.A. 1992. Broadsense heritability and genotype xenvironment interaction for carbon isotopediscrimination in field-grown wheat. Aust. J.Agric. Res. 43:921-934.

Condon, A.G., Richards, R.A., and Farquhar, G.D.1992. The effect of variation in soil wateravailability, vapour pressure deficit andnitrogen nutrition on carbon isotopediscrimination in wheat. Aust. J. Agric. Res.43:935-947.

Condon, A.G., and Richards, R.A. 1993.Exploiting genetic variation in transpirationefficiency in wheat: an agronomic view. In:Stable Isotopes and Plant Carbon-WaterRelations J.R. Ehleringer, A.E. Hall, andG.D. Farquhar (eds.). Academic Press, NewYork. pp. 435-450.

Craufurd, P.Q., Austin, R.B., Acevedo, E., andHall, M.A. 1991. Carbon isotopediscrimination and grain yield in barley.Field Crops Res. 27:301-313.

Curran, P.J. 1983. Multispectral remote sensingfor the estimation of green leaf area index.Philosophical Transactions of the RoyalSociety of London Series A: PhysicalSciences and Engineering 309:257-270.

Curran, P.J., Dungan, J.L., and Gholz, H.L. 1990.Exploring the relatioonship betweenreflectance red edge and chlorophyll contentin slash pine. Tree Physiology 7:33-48.

Demmig-Adams, B., and Adams, W.W. 1996.Xanthophyll cycle and light stress in nature:uniform response to excess direct sunlightamong higher plant species. Planta198:460-470.

Ehdaie, B., Hall, A.E., Farquhar, G.D., Nguyen,H.T., and Waines, J.G. 1991. Water-useefficiency and carbon isotope discriminationin wheat. Crop Sci. 31:1282-1288.

Elliott, G.A., and Regan, K.L. 1993. Use ofreflectance measurements to estimate earlycereal biomass production on sandplainsoils. Aust. J. Exp. Agric. 33:179-183.

Farquhar, G.D., O’Leary, M.H., and Berry, J.A.1982. On the relationship between carbonisotope discrimination and the intercellularcarbon dioxide concentration in leaves.Aust. J. Plant Phys. 9:121-137.

Farquhar, G.D., and Richards, R.A. 1984. Isotopiccomposition of plant carbon correlates withwater-use-efficiency of wheat genotypes.Aust. J. Plant Phys. 11:539-552.

Farquhar, G.D., Ehleringer, J.R., and Hubick, K.T.1989. Carbon isotope discrimination andphotosynthesis. Annual Review of PlantPhysiology and Plant Molecular Biology40:503-537.

Febrero, A., Blum, A., Romagosa, I., and Araus,J.L. 1992. Relationships between carbonisotope discrimination in field grown barleyand some physiological traits of juvenileplants in growth chambers. In: AbstractsSupplement of the First International CropScience Congress. Ames, IA. p. 26.


76

Febrero, A., Bort, T., Voltas, J., and Araus, J.L.1994. Grain yield, carbon isotopediscrimination and mineral content in maturekernels of barley, under irrigated and rain-fedconditions. Agronomie 2:127-132.

Febrero, A., Fernández, S, Molina-Cano, J.L., andAraus, J.L. 1998. Yield, carbon isotopediscrimination, canopy reflectance andcuticular conductance of barley isolines ofdiffering glaucosness. J. Exp. Botany49:1575-1581.

Field, C.B., Gamon, J.A., and Peñuelas, J. 1994.Remote sensing of terrestrial photosynthesis.In Ecophysiology of Photosynthesis. E.D.Schulze and MM Caldwell (eds.). Springer-Verlag. Berlin Heidelberg. pp. 511-528.

Filella, I., and Peñuelas, J. 1994. The red edgeposition and shape as indicators of plantchlorophyll content, biomass and hydricstatus. International Journal of RemoteSensing 15:1459-1470.

Filella, I., Serrano, L., Serra, J., and Peñuelas, J.1995. Evaluating wheat nitrogen status withcanopy reflectance indices and discriminantanalysis. Crop Sci. 35:1400-1405.

Filella, I., Amaro, T., Araus, J.L., and Peñuelas, J.1996. Relationship between photosyntheticradiation-use efficiency of barley canopiesand the photochemical reflectance index(PRI). Physiologia Plantarum 96:211-216.

Gamon, J.A., Peñuelas, J., and Field, C.B. 1992. Anarrow-waveband spectral index that tracksdiurnal changes in photosynthetic efficiency.Remote Sensing of Environment 41:35-44.

Gamon, J.A., Field, C.B., Goulden, M.L., Griffin,K.L., Hartley, A.E., Joel, G., Peñuelas, J., andValentini, R. 1995. Relationships betweenNdvi, canopy structure, and photosynthesis in3 Californian vegetation types. EcologicalApplications 5:28-41.

Gamon, J.A., Serrano, L., and Surfus, J. 1997. Thephotochemical reflectance index: an opticalindicator of photosynthetic radiation-useefficiency across species, functional typesand nutrient levels. Oecologia 112:492-501.

Garrity, D.P., and O’Toole, J.C. 1995. Selection forreproductive stage drought avoidance in rice,using infrared thermometry. Agron. J. 87:773-779.

Gitelson, A.A., and Merzlyak, M.N. 1997. Remoteestimation of chlorophyll content in higher-plant leaves. International Journal of RemoteSensing 18:2691-2697.

Hall, F.G., Huemmrich, K.F., and Goward, S.N.1990. Use of narrow-band spectra to stimatethe fraction of absorved photosyntheticallyactive radiation. Remote Sensing ofEnvironment 34:273-288.

Hall, A.E., Richards, R.A., Condon, A.G., Wright,G.C., and Farquhar, G.D. 1994. Carbonisotope discrimination and plant breeding.Plant Breeding Rev. 12:81-113.

Hay, R.K.M., and Walker, A.J. (eds). 1989. Anintroduction to the physiology of crop yield.Longman Scientific and Technical, Harlow,England.

Henderson, S., von Caemmerer, S., Farquhar,G.D., Wade, L., and Hammer, G. 1998.Correlation between carbon isotopediscrimination and transpiration efficiency inlines of the C4 species Sorghum bicolor inthe glasshouse and in the field. Aust. J. PlantPhys. 25:111-123.

Hubick, K., Farquhar, G.D., and Shorter, R. 1986.Correlation between water use efficiency andcarbon isotope discrimination in diversepeanut (Arachis) germplasm. Aust. J. PlantPhys. 13:803-816.

Hubick, K.T., and Farquhar, G.D. 1989. Carbonisotope discrimination and the ratio of carbongains to water lost in barley cultivars. Plant,Cell and Environment 12:795-804.

Huete, A.R. 1988. A soil adjusted vegetation index(SAVI). Remote Sensing of Environment25:295-309.

Jackson, R.D., and Pinter, P.J. Jr. 1986. Spectralresponse of architecturally different wheatcanopies. Remote Sensing of Environment20:43-56.

Jackson, R.D., Clarke, T.R., and Moran, M.S.1992. Bidirectional calibration results from11 Spectralon and 16 BaSO4 referencereflectance panels Remote Sensing ofEnvironment 49:231-239.

Jackson P., Robertson, M., Cooper, M., and Hammer,G. 1996. The role of physiological understandingin plant breeding; from a breeding perspective.Field Crops Res. 49:11-39.

Jacquemoud, S., and Baret, F. 1990. PROSPECT: amodel of leaf optical properties. RemoteSensing of Environment 34:75-91.

Jones, H.G. 1987. Breeding for stomatal characters.In: Stomatal Function. E. Zeiger, G.D.Farquhar, and I.R. Cowan (eds.). StanfordUniversity Press, Stanford. pp. 431-443.

Kimes, D.S., Kirchner, J.A., and Newcombe, W.W.1983. Spectral radiance errors in remotesensing ground studies due to nearly objects.Applied Optics 22:8-10.

Kirda, C., Mohamed, A.R.A.G., Kumarasinghe,K.S., Montenegro, A., and Zapata F. 1992.Carbon isotope discrimination at vegetativestage as an indicator of yield and water useefficiency of spring wheat (Triticum turgidumL. var. durum). Plant and Soil 7:217-223.

Kollenkark, J.C., Vanderbilt, V.C., Daughtry,C.S.T., and Bauer, M.E. 1982. Influence ofsolar illumination angle on soybean canopyreflectance. Applied Optics 21:1179-1184.

Larcher, W. (ed.). 1995. Physiological PlantEcology. Third edition. Springer, Berlin.

Lichtenthaler, H.K. (ed). 1996. Vegetation Stress.Based on the International Symposium onVegetation Stress. Gustav Fischer, Stuttgart.

Lichtenthaler, H.K., Gitelson, A., and Lang, M.1996. Non-destructive determination ofchlorophyll content of leaves of a green andan aurea mutant of tobacco by reflectancemeasurements. J. Plant Phys. 148:483-493.

López-Castañeda, C., Richards, R.A., andFarquhar, G.D. 1995. Variation in early vigorbetween wheat and barley. Crop Sci. 35:472-479.

Lord, D., Desjardins, R.L., and Dube, P.A. 1985.Influence of wind on crop canopyreflectance measurements. Remote Sensingof Environment 18:113-123.

Loss, S.P., and Siddique, K.H.M. 1994.Morphological and physiological traitsassociated with wheat yield increases inMediterranean environments. Advances inAgronomy 52:229-276.

Masle, J., Farquhar, G.D., and Wong, S.C. 1992.Transpiration ratio and plant mineral contentare related among genotypes of a range ofspecies. Aust. J. Plant Phys. 19:709-721.

Mayland, H.F., Johnson, D.A., Asay, K.H., andRead, J.J. 1993. Ash, carbon isotopediscrimination and silicon as estimators oftranspiration efficiency in crestedwheatgrass. Aust. J. Plant Phys. 20:361-369.

Mian, M.A.R., Bailey, M.A., Ashley, D.A., Wells,R., Carter, Jr., T.E, Parrott, W.A., andBoerma, H.R. 1996. Molecular markersassociated with water use efficiency and leafash in soybean. Crop Sci. 36:1252-1257.

Monje, OA., and Bugbee, B. 1992. Inherentlimitations of nondestructive chlorophyllmeters: a comparison of two types of meters.HortScience 27:69-71.

Morgan, J.A., LeCain, D.R., McCaig, T.N., andQuick, J.S. 1993. Gas exchange, carbonisotope discrimination and productivity inwinter wheat. Crop Sci. 33:178-186.

Nageswara Rao, R.C., and Wright, G.C. 1994.Stability of the relationship between specificleaf area and carbon isotope discriminationacross environments in peanut. Crop Sci.34:98-103.

Passioura, J.B. 1977. Grain yield, harvest indexand water use of wheat. J. Aust. Inst. Agric.Sci. 43:117-121.

Peñuelas, J. 1998. Visible and near-infrarereflectance techniques for diagnosing plantphysiological status. Trends in Plant Science3:151-156.

Peñuelas, J., Gamon, J.A., Griffinand, K.L., andField, C.B. 1993a. Assessing type, biomass,pigment composition and photosyntheticefficiency of aquatic vegetation fromspectral reflectance. Remote Sensing ofEnvironment 46:110-118.

Peñuelas, J., Filella, I., Biel, C., Serrano, L., andSavé, R. 1993b. The reflectance at the 950-970 nm region as an indicator of plant waterstatus. International Journal of RemoteSensing 14:1887-1905.


77

Peñuelas, J., Filella, I., and Baret, F. 1995a.Semiempirical indices to assess carotenoids/chlorophyll a ratio from leaf spectralreflectance. Photosynthetica 31:221-230.

Peñuelas, J., Filella, I., and Gamon, J.A. 1995b.Assessment of photosynthetic radiation-useefficiency with spectral reflectance. NewPhytologist 131:291-296.

Peñuelas, J., Filella, I., Lloret, P., Muñoz, F., andVilajeliu, M. 1995c. Reflectance assessmentof plant mite attack on apple trees.International Journal of Remote Sensing16:2727-2733.

Peñuelas, J., Piñol, J., Ogaya, R., and Filella, I.1997a. Estimation of plant waterconcentration by the reflectance water indexWi (R900/R970). International Journal ofRemote Sensing 18:2869-2875.

Peñuelas, J., Isla, R., Filella, I., and Araus, J.L.1997b. Visible and near-infrared reflectanceassessment of salinity effects on barley.Crop Sci. 37:198-202.

Price, J.C., and Bausch, W.C. 1995. Leaf-areaindex estimation from visible and near-infrared reflectance data. Remote Sensing ofEnvironment 52:55-65.

Ranson, K.J., Daughtry, C.S.T., Biehl, L.L., andBauer, M.E. 1985. Sun-view angle effects onreflectance factors of corn canopies. RemoteSensing of Environment 18:147-161.

Reynolds, M.P., Balota, M., Delgado, M.I.B.,Amani, I., and Fischer, R.A. 1994.Physiological and morphological traitsassociated with spring wheat yield underhot, irrigated conditions. Aust. J. Plant Phys.21:717-730.

Richards, R.A. 1996. Defining selection criteria toimprove yield under drought. Plant GrowthRegulation 20:157-166.

Richards, R.A., and Condon, A.G. 1993.Challenges ahead in using carbon isotopediscrimination in plant breeding programs.In: Stable Isotopes and Plant Carbon-WaterRelations J.R. Ehleringer, A.E. Hall, andG.D. Farquhar (eds.). Academic Press, NewYork. pp. 451-462.

Richardson, A.J., and Wiegand, C.L. 1977.Distinguishing vegetation from soilbackground information. PhotogrammetricEngineering and Remote Sensing43:1541-1552.

Romagosa, I., and Araus, J.L. 1991.Genotype-environment interaction for grainyield and 13C discrimination in barley.Barley Genetics VI:563-567.

Rudorff, B.F.T., and Batista, G.T. 1990. Spectralresponse of wheat and its relationship toAgronomic variables in the tropical region.Remote Sensing of Environment 31:53-63.

Sayre, K.D., Acevedo, E., and Austin, R.B. 1995.Carbon isotope discrimination and grainyield for three bread wheat germplasmgroups grown at different levels of waterstress. Field Crops Res. 41:45-54.

Shibayama, M., Wiegand, C.L., and Richardson,A.J. 1986. Diurnal patterns of bidirectionalvegetation indices for wheat canopies.International Journal of Remote Sensing7:233-246.

Slafer, G.A., Satorre, E.H., and Andrade, F.H.1993. Increases in grain yield in bread wheatfrom breeding and associated physiologicalchanges. In: Genetic Improvements of FieldCrops: Current Status and Development. G.Slafer (ed.). Marcel Dekker, New York.pp. 1-68.

Thomas, J.R., and Gausman, H.W. 1977. Leafreflectance vs. leaf chlorophyllconcentration for eight crops. Agron. J.69:799-802.

Turner, N.C. 1993. Water use efficiency of cropplants: potential for improvement. InInternational Crop Science. I. D.R. Buxton,R. Shibles, R.A. Forsberg, B.L. Blad, K.H.Asay, G.M. Paulsen, and R.F. Wilson (eds.).Crop Science Society of America, Madison.pp. 75-82.

Voltas, J., Romagosa, I., Muñoz, P., and Araus,J.L. 1998. Mineral accumulation, carbonisotope discrimination and indirect selectionfor grain yield in two-rowed barley grownunder semiarid conditions. Eur. J. Agron.9:145-153.

Walker, C.D., and Lance, R.C.M. 1991. Siliconaccumulation and 13C composition asindices of water-use efficiency in barleycultivars. Aust. J. Plant Phys. 18:427-434.

Wardley, N.W. 1984. Vegetation index variabilityas a function of viewing geometry.International Journal of Remote Sensing5:861-870.

Wiegand, C.L., and Richardson, A.J. 1990a. Useof spectral vegetation indices to infer leafarea, evapotranspiration and yield: I.Rationale. Agron. J. 82:623-629.

Wiegand, C.L., and Richardson, A.J. 1990b. Useof spectral vegetation indices to infer leafarea, evapotranspiration and yield: II.Results. Agron. J. 82:630-636.

Wiegand, C.L., Richardson, A.J., Escobar, D.E.,and Gerbermann, A.H. 1991. Vegetationindices in crop assessments. Remote Sensingof Environment 35:105-119.

Young, A., and Britton, G. 1990. Carotenoids andstress. In: Stress Responses in Plants:Adaptation and Acclimation Mechanisms.R.G. Alscher and J.R. Cumming (eds.).Wiley-Liss, New York. pp. 87-112.


78

Economic Issues in Assessing theRole of Physiology in WheatBreeding ProgramsJ.P. Brennan1 and M.L. Morris2

Wheat breeding can be thought of as aneconomic activity, in the sense that itinvolves processes of physicaltransformation that are characterized bystreams of costs and benefits. Decisionstaken about the organization andoperation of a wheat breeding program(including technical decisions such asthe choice of parental materials, crossingtechniques, selection methods, andevaluation procedures) are likely to haveeconomic implications. To the extent thatchanges in the organization or operationof a wheat breeding program affect thestreams of costs and benefits, theeconomic outcomes that can be expectedfrom the program will increase ordecrease.

Plant breeders are viewed by many,especially by those in the foodproduction and processing industries, asa resource that can be used to enhancethe overall performance of theagricultural sector (Brennan, 1997).Precisely because they have thiscapacity, plant breeders often facedemands on their services that far exceedwhat they can realistically expect todeliver. In a world of limited resources,plant breeders therefore need some basisfor deciding which among the manydemands being placed on them shouldhave priority. Although economic factorsare often taken into account (explicitlyor implicitly) when research priorities

are established, the informal and ad hocmanner in which this is done frequentlyleads to decisions that are far fromoptimal in an economic sense. Basiceconomic analysis, grounded in thecareful assessment of benefits and costs,can provide the foundation for makingthose decisions in a more informed anddefensible manner.

Assessing PotentialChanges to a WheatBreeding Program

Under what circumstances might it beadvisable to incorporate physiology intoa wheat breeding program? In assessingthe organization and management of anexisting breeding program and decidingwhether or not changes may be neededto meet a particular objective, it willoften be useful to review the followingpreliminary questions beforeundertaking formal economic analysis.

Is the problem bestaddressed throughbreeding?Before any changes are made to abreeding program, it is important todetermine whether the results beingsought could be obtained more quicklyand/or cheaply by some other means.For example, if the research objective is

to increase protein content in wheat,experience suggests that it will often bebetter to concentrate on breeding forhigher yield, while leaving the challengeof raising protein content to agronomicmanagement. This is because eventhough cultivars differ in their proteincontent, prospects for increasing proteincontent through breeding are limited.Research has shown that protein contentis mainly influenced by environment(and by genotype x environmentinteraction), so the genotype effect isgenerally very small (Bingham, 1979).Furthermore, given the known negativerelationship between yield and proteincontent (O’Brien and Ronalds, 1984),any increases in protein obtained throughselecting higher-protein cultivars arelikely to result in lower yields.

What level of breeding inputis appropriate?Once it has been decided that theresearch objective is best addressedthrough breeding, it is necessary todetermine what level of breeding input isappropriate. An appropriate level ofbreeding input is one that can be justifiedin terms of the size of the expectedbenefits. Generally these will be relatedto the size of the target region: As thesize of the target region increases, so willthe level of breeding effort that isjustified. Brennan (1992)

1 Agricultural Research Institute, New South Wales, Australia.2 CIMMYT Economics Program, Apdo. Postal 6-641, Mexico, D.F., Mexico, 06600.

C HAPTER 6

79

and Maredia (1993) have shown that forsmall target regions, it often will beappropriate only to select from amongbreeding lines that have been importedfrom elsewhere. But once the targetregion increases beyond a certain size, itwill be appropriate to establish a full-fledged local crossing program. Theprecise threshold values needed to justifyexpansion of an existing breedingprogram will depend, among otherthings, on the characteristics of the targetenvironment (area, average yield levels,use of improved varieties, etc.) and theyield gains that can be expected byincreasing the breeding input.

What breeding strategy islikely to be most efficient?Once it has been determined that theproblem is best handled throughbreeding and that the size of the targetregion justifies a full-fledged crossingprogram, a breeding strategy must bedecided. Wheat breeding can be pursuedin many different ways. The initialchoice of source materials is of coursecritical; if the source materials selectedfor improvement contain a highproportion of favorable alleles for theproblem being addressed, the chances ofsuccess are greatly enhanced. Afterchoosing the source materials, thebreeder must determine how much effortwill be put into crossing, as compared toselection and evaluation of the resultinglines. In addition to a wide range of so-called conventional breeding methods,modern breeders also have the option ofincorporating biotechnology techniques,such as the use of genetic markers, tissueculture, etc. In deciding what types ofselection methods will be optimal, it isimportant that the decision be driven bywhat is best for the program and itsoutcomes, not simply by the availabilityof new tools or techniques (Brennan,1997). Innovative technologies thatappear to be tremendously appealing inthe short run often turn out to be far less

attractive over the longer term. Forexample, Brennan and O’Brien (1991)found that the incorporation of early-generation, small-scale quality testinginto an Australian commercial wheatbreeding program, while initiallyattractive, led to a lower economic returnfor the breeding program (Box A).

Box A: Incorporating Quality Selection into aWheat Breeding Program

A recurring question facing any wheat breeding program concerns when to begin selecting for qualitycharacteristics—for example, protein content or gluten level. Opinions differ as to what the appropriatetime is for starting quality testing. Some breeders feel that selecting for quality should be left until late inthe improvement process, after significant progress has been achieved in raising yield potential. Othersfeel that selecting for quality in addition to yield should commence early in the breeding process, so thatlow-quality materials are screened out early on.

Brennan and O’Brien (1991) used an economic framework to evaluate the efficiency of two alternativeapproaches to the problem. Their study focused on two Australian wheat breeding programs, one whichperformed early generation quality testing, and another which did not (see table). Both programs used thesame amount of labor, and the number of crosses and lines sown in the F2 generation was identical. Thetwo programs differed mainly in the stage at which quality testing was initiated, which caused differentsets of lines to move through each program.

Small-scale tests for quality carried out early in the breeding process (F2 stage) were found to be lessexpensive than tests carried out in later generations (F6 stage). This led to the prediction that earlygeneration testing would prove more cost effective. But when the economic returns of the program doingearly generation quality testing were compared to those of the program in which quality testing wasintroduced at a later stage, they were found to be lower. The costs associated with the program thatincluded early generation testing were slightlyhigher, but the benefits were markedly lower overthe longer term. The program without earlygeneration quality testing was able to concentrateexclusively on yield potential, enabling it toachieve much more rapid rates of yield gains.Admittedly, it also had a lower rate of qualityincrease, since less selection pressure was placedon quality, but when economic values wereassigned to yield levels and quality factors, theadditional yield gains more than compensated forthe lower levels of quality. The study thus showedthat in the absence of a substantial premium forquality, wheat producers and consumers will bedisadvantaged if breeding programs opt to selectfor quality in the early generations at theexpense of yield improvement.

Economic returns to early versuslate selection for grain quality in wheat.

Program ProgramA† B‡

Expected increase overcurrent varieties (%):- yield 4.6 2.3- quality 0.2 1.1

Total costs‡ (US$ 000) 353 369Total benefits‡ (US$ 000) 3,710 2,557Benefit-cost ratio 10.5 6.9

† Quality-testing introduced in F6 for Program A, and F2for Program B.

‡ Discounted to year of crossing at 5% per annum.Source: Derived from Brennan and O’Brien (1991).

On what basis will varietiesbe released for use byfarmers?The procedures used to evaluate newvarieties prior to their release also meritconsideration, because evaluationprocedures can have important economicimplications for breeding programs.

ECONOMIC ISSUES IN ASSESSING THE ROLE OF PHYSIOLOGY IN WHEAT BREEDING PROGRAMS

80 J.P. BRENNAN AND M.L. MORRIS

Before a new variety is released, breedersmust decide how widely it should betested and over what period of time, howwell it need perform relative to othercultivars that are already in use, and whatlevel of genetic diversity is desirablewithin and among released varieties.3Subjecting experimental varieties torigorous testing prior to their releaseincreases the likelihood that they will becommercially successful, but extensivetesting can also significantly raise overalldevelopment costs.

If after reviewing these preliminaryquestions it still seems worthwhile toproceed, it may be appropriate toundertake more rigorous economicanalyses. Although space limitationsprevent us from presenting detailed step-by-step instructions, the next twosections provide a broad overview of keyeconomic concepts needed for formallyevaluating the desirability ofincorporating physiology into a wheatbreeding program. They provide a briefdescription of the basic procedures thatwould have to be followed.

Key Economic ConceptsRelating to InvestmentAnalysis

Basis for economicassessmentThe decision of whether or not toincorporate physiology into a wheatbreeding program can be approached likeany other investment decision. In thisrespect, the key issue concerns theeconomic returns that will be generatedas a result of the proposed organizationalchange. The basic economic question thatneeds to be addressed is really quitesimple: What level of investment is

justified by the expected benefits, takinginto account alternative investmentopportunities?

Before any formal economic analysis isundertaken, two important concepts needto be understood: opportunity cost andtime value of money.

Opportunity cost. An opportunity cost isthe benefit foregone by using a scarceresource for one purpose instead of itsnext best alternative use (Gittinger1982). Opportunity costs play animportant role in investment analysis,because most investments involvechoices between mutually exclusivealternatives. Since the resourcesavailable to a breeding program areusually limited, whenever additionalemphasis is put on one breedingobjective, less emphasis must necessarilybe put on other objectives. To return tothe example cited earlier, if the decisionis taken to target higher protein content,this will probably slow the expected rateof progress in breeding for higher yield.Thus, the opportunity cost of breedingfor enhanced protein content will be theprogress that would have been achieved(but had to be given up) in breeding forhigher yield.

Of course, the tradeoffs may not alwaysbe so evident. In plant breeding,targeting one objective does notnecessarily mean that progress towardother objectives will be suppressed, atleast not directly. The rationale foradding a physiology component in factmay be to achieve the same outcomewith greater efficiency. Thus, data fromphysiological measurements may beused to complement yield trial data; ifthe additional information improves thebreeder’s ability to predict cultivarperformance in target environments, theneed for extensive yield trials may be

reduced or even eliminated (Reynolds etal., 1997). But even in cases such as this,the concept of opportunity cost remainsvalid, because the resources invested intaking the additional physiologicalmeasurements presumably could havebeen used in some other way to generateother types of benefits.

Time value of money. Economicanalysis must take into account oneimportant facet of value that stems fromhuman behavior, namely, the time valueof money. The time value of moneyrefers to the fact that people place ahigher worth on values realized earlier ascompared to values realized later(Gittinger, 1982). Asked to choosebetween receiving $100 today andreceiving $100 one year from today,most people would choose today. Ineconomic analysis, the time value ofmoney is taken into account throughdiscounting, whereby costs and benefitsexpected to occur in the future areassigned lower values.

Discounting is important in any type ofinvestment analysis, which by definitiondeals with streams of costs and benefitsthrough time. It is particularly importantin the analysis of agricultural researchinvestments, given 1) the unequaldistribution through time of expectedcosts and benefits, and 2) the uncertaintyabout future outcomes of agriculturalresearch. In plant breeding, there areusually long lags—often 10 years ormore—between the initial crossing andselection activities (which imply costs)and the eventual adoption of improvedvarieties by farmers (which generatesbenefits). Under those circumstances, theexpected benefits may be discounted bydecision makers, to the extent that theinvestment may no longer seemattractive (Box B).

3 The important issue of how many varieties should be released may also have to be decided by breeders, although usually this matter is left to some sortof government-appointed varietal certification and release committee.

81

Discounting can dramatically alter theattractiveness of any investmentopportunity. For example, spending $1now to bring about a $2 return after 10years at first glance might seem a soundinvestment. But with a 12% discount rate(often used to approximate theopportunity cost of capital in developingcountries), $2 received 10 years in thefuture is worth only $2/(1.1210) = $0.64at today’s prices. Seen in this light, theinvestment does not appear to beattractive.

Significantly, the effects of discountingare very sensitive to the distributionthrough time of the expected costs andbenefits. If the same $1 investment isexpected to generate the same $2 returnafter only five years rather than ten, thepresent value of the benefits (i.e., thevalue at today’s prices) is $2/(1.125) =$1.13. Depending on whether or notmore attractive investment opportunitiesare available, this return might seemquite attractive.

Approaches to economicanalysisHow can economic analysis help aresearch administrator assess thedesirability of establishing a physiologycomponent to the breeding program?Economists would characterize theincorporation of physiology into anexisting breeding program as a marginalchange, since it does not involvereorganization of the entire breedingprogram, but only an incremental change,or a change “at the margin.”

Two alternative approaches can be usedto analyze marginal changes: 1)comparing only changes in costs andbenefits expected to result from theaddition of the physiology component(partial budget analysis), or 2) comparingthe costs and benefits of the entirebreeding program with the physiologycomponent to the costs and benefits of

Box B: Stream of Costs and BenefitsAssociated with a Breeding Program

To calculate the economic returns to a plant breeding program, it is necessary to estimate the costs andbenefits associated with it. Figure 1 illustrates the stream of costs and benefits typically associated witha plant breeding program. During an initial period, net benefits remain negative because research costsare being incurred (for example, in the crossing, selection and evaluation of experimental materials)without any benefits being realized (Morris et al., 1992). Eventually the research produces an improvedvariety, which after undergoing a certification process is approved for release. Following a lagnecessary for the production and distribution of seed, the variety is taken up by farmers, with the rateof adoption typically following an S-shaped (or logistic) curve. Providing the variety leads to improvedyields in farmers’ fields, the original research investment (made years earlier in most cases) now beginsto generate benefits in the form of increased production. The stream of net benefits consequently turnspositive, increasing as the area planted to the new variety expands, reaching a maximum at peakadoption, and then declining as the variety is gradually replaced by another, newer variety.

While the relative sizes of costs and benefits are obviously important in evaluating a researchinvestment, their distribution through time is also important. Benefits realized far in the future areconsidered less valuable than benefits realized in the short term. To accommodate the time value ofmoney, research costs and benefits are discounted. Figure 2 illustrates how discounting depresses thevalue of net benefits realized near the end of the period of analysis relative to those realized near thebeginning. Because of the long time lags involved in research such as plant breeding, discounting is animportant concept used in analyzing returns to investments in research.

Net benefits ($)400

300

200

100

0

-1000 5 10 15 20 25 30 35 40 45 50

Years

Figure 2. Effects of discounting the stream ofcosts and benefits associated with a plantbreeding program.Source: Figure 13 in Morris et al. (1992).

Research costsResearch benefits

(Discontinued at 8% per year)

Certificationbegins

Researchbegins

Year ofrelease Peak adoption

Net benefits ($)600

500

400

300

200

100

0

-1000 5 10 15 20 25 30 35 40 45 50

Years

Figure 1. Undiscounted stream of costs andbenefits associated with a plant breedingprogram.Source: Figure 12 in Morris et al. (1992).

Research costsResearch benefits Peak adoption

Certificationbegins

Researchbegins

Year ofrelease


82

the entire breeding program without thephysiology component (whole budgetanalysis). Which of these twoapproaches is preferable will depend onthe quantity and quality of data that canbe collected, the time available for theanalysis, and perhaps the level ofeconomic expertise on hand. The resultsobtained through the two approacheswill generally be similar, although theymay differ somewhat depending on thelevel at which the analysis is undertaken.

Whether the approach being used isbased on partial budget analysis orwhole budget analysis, the key toassessing the economic desirability ofthe proposed change is to correctlyidentify the costs and benefits that can beexpected to vary—known as themarginal costs and marginal benefits.

Certain types of cost changes can readilybe identified and estimated in advance.For example, if incorporating physiologyinvolves the hiring of a physiologist,then it can safely be predicted that oneadditional cost will be the physiologist’ssalary and associated costs. Similarly, ifthe incorporation of physiology requiresthe construction of a new laboratoryfacility with specialized equipment andmaterials, that cost, too, can easily beforeseen and quantified.

But other types of cost changes will bemuch more difficult to predict. Decisionmaking in plant breeding programsusually proceeds in a sequential manner,with decisions taken at early stages ofselection often leading to unpredictableoutcomes that in turn affect decisionstaken in subsequent stages. For thisreason, determining future cost streamsis not always easy. Because of the“snowballing” effect, the cumulativechanges in the cost structure of abreeding program over the long run mayappear quite different from what couldrealistically have been anticipated at thetime a management decision wasoriginally taken. In studying two wheat

breeding programs in Australia, Brennanand O’Brien (1991) found that eventhough the introduction of earlygeneration quality testing reducedoverall breeding costs in the firstgeneration of plants, over the longerterm total costs were higher because thesequential cost effect increased costs insubsequent generations (see Box A). Theoutcomes of research investments beinginherently more unpredictable than theoutcomes of many other types ofinvestments, careful thought needs to beput into assessing the extent to whichfuture costs are likely to be affected byany marginal changes.

Just as marginal costs are often difficultto predict, so, too, are marginal benefits.With investments that affect theorganization and management of a plantbreeding program, in some cases it maybe relatively easy to identify andestimate expected marginal benefits.That will occur especially when themarginal benefits relate to changes in thevalue of final outputs of the program(e.g., the acquisition of a new source ofgermplasm that contributes directly tothe development of higher yieldingvarieties). In other cases, it is extremelydifficult to identify and quantifyexpected marginal benefits. That isparticularly so when they will resultfrom changes in current researchprocedures that are likely to affect theway future research is carried out (e.g.,modifying early generation evaluationprocedures in ways that are likely tohave consequences for the numbers andquality of materials available in latergenerations).

When the research investment beingconsidered is the addition of aphysiology component to the breedingprogram, identification andquantification of expected marginal costsand benefits may be difficult because somany aspects of the existing program arelikely to be affected. Under thesecircumstances, partial budget analysis

based on expected changes in marginalcosts and marginal benefits will often beinadequate, and it may be desirable tocarry out a more complete economicanalysis based on the program’s totalcosts and returns. Brennan (1989a)describes a method for developingdetailed estimates of costs and returnsfor an entire plant breeding program;these detailed estimates can be used toevaluate the economic attractiveness ofsignificant (i.e., non-marginal) changesto the organization of the program. Thetotal-budget approach described byBrennan is based on a comparison ofexpected costs and returns without aphysiological component and expectedcosts and returns with the physiologicalcomponent.

Evaluating theDesirability of UsingPhysiology inBreeding Programs

Estimating costs and benefitsof the current programWhenever a detailed economic analysisis to be carried out to determine thedesirability of incorporating physiologyinto an existing breeding program, as astarting point it will often be useful todevelop broad estimates of the totalcosts and returns of the currentprogram.

A broad estimate of the currentprogram’s costs can be made in a “top-down” fashion based on aggregatebudget information about the program’stotal operating costs, total annual capitalcosts, total annual salary costs, and totaloverhead costs. Alternatively, thecurrent program’s costs can be broadlyestimated in a “bottom-up” fashionbased on disaggregated cost datarelating to each individual activity ofthe program (see Brennan and Khan,1991). Either way, it is important to

J.P. BRENNAN AND M.L. MORRIS

83

include all relevant costs likely to beaffected by the proposed change,including costs associated with crossing,evaluation, selection, disease screening,quality evaluation, and regional trials, aswell as costs relating to variety releaseand registration activities.

The extent to which overhead andadministrative costs (such as the salariesand support costs of head officepersonnel, finance officers, and humanresources staff; library costs; informationtechnology costs) need to be taken intoaccount depends on whether or not thesecosts are likely to be affected by theproposed changes. If administrativeoverheads are likely to remain unaffected,it may be convenient to overlook them.

The benefits generated by a wheatbreeding program can be measured atdifferent levels. For researchers workingin a breeding program, the benefitsinclude not only improved varieties perse, but also scientific benefits such asnovel research techniques, specializedlaboratory equipment, and originalknowledge. For the organization thatsponsors the breeding program, especiallyif it is a profit-oriented private company,the benefits will often be measured interms of the additional income earnedthrough the sale of improved varieties(whether directly through commercialseed sales or indirectly through royaltiesor licenses). For society as a whole, themost important benefits that flow from awheat breeding program tend to be theproductivity gains achieved by farmerswhen they grow improved varietiesproduced by the program (measured asincome increases or as cost reductions).4For simplicity, the emphasis in thischapter is on the latter type of benefit(farm-level productivity gains), althoughour analysis also applies to other typesof benefits.

The benefits (or returns) generated bythe current program can be broadlyestimated based on the outputs from theprogram. In most cases, these willconsist of improved varieties. In order toestimate the economic benefitsassociated with the adoption of improvedvarieties, usually it is necessary toanswer the following questions:

• To what target regions are thevarieties adapted?

• What advantage do the varietiesconfer (e.g., higher yield, improvedquality)?

• What will be the average price ofeach incremental ton of grainproduced, or the average priceincrease attributable to improvedquality?

• What will be the rate and extent ofadoption of the varieties followingtheir release?

Once these questions have beenanswered (to the extent that they can be),it should be possible to estimate thebenefits likely to be generated by theproposed change, if only in broad terms.For example, in his study of a publicwheat breeding program in New SouthWales, Australia, Brennan (1989a,b)made the following estimates:

• the program serves a target regionthat includes about 1.0 million haplanted to wheat each year, withaverage yields of 1.7 t/ha;

• each new variety produced by theprogram generates an average yieldincrease of 2.25% and an averageimprovement in the quality index of1.09%;

• each 1% yield increase is worth$1.11/t, while each 1% improvementin quality is worth $0.81/t;

• on average, adoption of each varietypeaks at about 16% of the target areain the seventh year following release;and

• each new variety has a productive lifeof 20 years.

Based on these broad estimates, Brennanwas able to calculate the economicbenefits generated by the breedingprogram, which total approximately$920,000 per year at peak adoption.

Estimating marginal costsand benefits of projectedchangesThe next task is to estimate the changesto both costs and benefits that wouldflow from incorporating physiology intothe program. To a certain extent, thesewill necessarily be speculative, and theywill in any case depend on the roleenvisioned for physiology within thelarger breeding program.

Estimating future cost changes isfrequently complicated by short-runversus long-run issues. In some cases,the physiology component can beexpected to lead directly to cost savings,for example when the introduction ofearly generation screening methods islikely to reduce the number of lines thatwill have to be evaluated in latergenerations. Similarly, it has been shownthat the introduction of certain tissueculture techniques can dramaticallyreduce the costs of multiplyingexperimental materials (Brennan, 1989b)(Box C). In other cases, however, thephysiology component can be expectedto lead to cost increases, at least in theshort run, for example when thephysiological tests undertaken can beexpected to add measurably to theexpense and/or time involved inscreening.

4 Depending on the degree to which farmers sell their products, and depending on the nature of the markets in which they sell their products, theseproductivity gains may be transmitted in part or in whole to consumers.


84

Particularly in these cases, it is importantto determine whether the additional costsincurred in the short run are likely to leadto even greater cost savings over thelonger term. Although additionalexpenditure in the short run is oftenjustified on the grounds that the long-runpayoffs will be large, this is not alwaysthe case. In their comparative analysis oftwo Australian wheat breeding programs,Brennan and O’Brien (1991) found thatthe slowdown in yield gains that resulted

Box C: Using Doubled Haploid Tissue Culture in aWheat Breeding Program

Of the many forms of tissue culture available for use by wheat breeders, one of the most valuable isdoubled haploid culture, which involves in vitro development of fixed lines from parental material. Thetechnique is attractive because development of each generation of progeny can be initiated before theparents have achieved physiological maturity, thus accelerating the breeding process. Furthermore, it iscarried out in the laboratory, rather than in the field, thus reducing the need for costly grow-out trials. Byusing doubled haploid tissue culturing techniques, wheat breeders can reduce the number of years neededto produce lines for advanced testing, while at the same time saving considerably on field productioncosts.

Brennan (1989b) examined the potential returns to a conventional wheat breeding program of adoptingdoubled haploid tissue culture techniques (see table). The anticipated impacts of adopting the techniqueswere modeled by assuming that: 1) the production of generations F1 to F5 would be compressed into justtwo years, as opposed to the usual five years, and 2) field production costs would be slightly reduced. Inaddition, it was implicitly assumed that linesproduced using doubled haploid culture wouldbe identical to the lines emanating from aconventional breeding program.

Brennan estimated the costs and benefits of thetwo alternative scenarios (conventional breedingwithout tissue culture and conventional breedingwith tissue culture). His analysis showed that theuse of tissue culture to accelerate the productionof advanced breeding lines could be expected togenerate handsome economic returns. Followingthe adoption of tissue culture, the net presentvalue increased by more than $600,000 perline, and the benefit-cost ratio rose from 6.9 to9.0. Thus, by slightly reducing production costsand significantly accelerating the production ofadvanced generation materials, doubled haploidtissue culture was shown to significantlyincrease the expected profitability of thebreeding program.

Economic returns to the use of tissue culturein wheat breeding.

BreedingConventional program

breeding with tissueprogram† culture‡

Discounted costs 550 489 ($A 000)‡

Discounted benefits 3,816 4,418 ($A 000)‡

Net present value 3,266 3,929 ($A 000)Benefit-cost ratio 6.9 9.0

† All values in 1986 Australian dollars.‡ Discounted at 5% per annum.Source: Brennan (1989b).

from incorporating early generationquality testing was not sufficientlycompensated by the resulting gains ingrain quality resulting from that higherselection pressure.

Estimating future benefits is complicatedby the difficulty of precisely anticipatingresearch outputs. All research is to someextent speculative, so the outcomes ofany research investment can never beknown with certainty. Nevertheless, it is

usually possible to make educatedguesses about the future values of keyparameters that will determine the size(and distribution through time) ofeconomic benefits. As stated earlier, theimpact of incorporating physiology intoan existing wheat breeding programpotentially will be reflected in: 1) higheryielding varieties, 2) higher qualityvarieties, 3) better adapted (andtherefore more widely grown) varieties,and/or 4) earlier release of newvarieties. To the extent that it is possibleto relate the incorporation of physiologyto expected changes in the values ofthese key parameters, it will be possibleto arrive at a rough estimate of expectedbenefits.

In estimating the benefits from a changeto the program, it is important toremember that resources are limited, soany gains made in selecting for oneobjective must come at the expense ofgains in selecting for another objectiveor objectives. These trade-offs must notbe overlooked when benefits are beingestimated.

Analyzing anticipated futureflows of costs and benefitsOnce the size of marginal costs andmarginal benefits have been estimated,it is necessary to project theirdistribution through time. The simplestway to do this is by making year-by-year projections of marginal costs andmarginal benefits. Given the relativelylong period that can elapse between thetime a research investment is initiatedand the time tangible benefits are firstrealized in farmers’ fields (known as the“research lag”), it is usually desirable toproject marginal costs over a period ofat least 10 years. The exact duration ofthe expected research lag will depend onthe type of research that is beingcontemplated. In wheat breeding, someactivities can be expected to have arelatively short research lag—forexample, 3-4 years to introduce a grain


85

color characteristic that is controlled bya single gene. Other activities can beexpected to have a much longer lag of10 years or more—for example,incorporating drought tolerance, whichis controlled by complex interactionsamong several different genes.

Once the size and distribution throughtime of research costs have beenestimated, equivalent estimates must bemade about expected flows of futurebenefits. In the case of a wheat breedingprogram, these will generally depend onthe diffusion pattern of the new varietiesproduced by the program. Varietaldiffusion patterns can vary widely,depending on the characteristics of thenew variety or varieties, the degree towhich farmers recognize and value thenew characteristics, the effectiveness ofthe seed production and distributionsystem, and other factors. Someimproved varieties are adopted veryrapidly by a large proportion of farmers,resulting in a short, steep diffusion curvethat reaches a ceiling level approaching100% of the target region. Otherimproved varieties are adopted muchmore slowly and only by a relativelysmall proportion of farmers, resulting in along, flat diffusion curve that tops out at aceiling level well below 100% of thetarget area. Based on the assumptionsmade about the technology diffusionpattern (known as the “adoption lag”), thesize and distribution through time ofmarginal benefits can be estimated.

Given the two types of lag involved inwheat breeding (research lag andadoption lag), it is advisable to consideran extended period when evaluating thedesirability of incorporating physiologyinto an existing breeding program. As ageneral rule of thumb, marginal costs andbenefits should be projected out over a30-year period.

Next, the projected flows of costs andbenefits must be discounted. Discountingis necessary to take into account the time

value of money, i.e., the fact that costs andbenefits realized in the future are valuedless than costs and benefits realized in thepresent. To take into account the timevalue of money, discount factors areapplied to future costs and benefits toconvert them to their present value.

Discounting is carried out using thefollowing formula:

Dn = Un / ((1+r)n-1),

where:

D = discounted value of cost(benefit) in year n

U = undiscounted value of cost(benefit) in year n

r = discount raten = year (where n = 1 is the

present year, n = 2 is nextyear, etc.).

Alternatively, discount factors may beapplied to future costs and benefits.Standard discount factors are readilyavailable in most handbooks on projectanalysis and are readily generated bymost financial spreadsheet programs.

Calculating measures ofproject worthAfter projected costs and benefits havebeen discounted, they can be summed toobtain total discounted costs (TDC) andtotal discounted benefits (TDB). TheTDC and TDB can be used to calculatetwo simple measures for use in assessingthe attractiveness of any potentialinvestment:

• net present value (NPV = TDB -TDC), and

• benefit-cost ratio (B/C ratio = TDB /TDC).

If the objective of the economic analysisis simply to determine whether or not theincorporation of physiology will beprofitable, then it may be appropriate toproceed with the investment if it can beestablished that it will add to the overall

economic returns generated by thebreeding program. This will be the case ifthe NPV is positive (NPV > 0).

The main advantage of using NPV as adecision criterion is that it is easy tocompute. One big disadvantage of theNPV measure, however, is that it fails totake into account the size of the proposedinvestment; without further investigation,there is no way to tell whether a givenNPV was generated by a largeinvestment or by a small one. This limitsthe usefulness of the NPV as a tool fordeciding between alternative investmentopportunities, because simply choosingthe alternative with the highest NPV maynot always be desirable. When asked tochoose between investing $100 in oneproject expected to generate a NPV of$200 and investing $200 in anotherproject expected to generate a NPV of$210, most people would prefer the firstproject–even though the NPV is lower.

If the objective of the economic analysisis to select among two or morealternative investment opportunities, thenit will be preferable to use the B/C ratioas a decision criterion. Since the B/Cratio expresses (discounted) benefits perunit cost, it provides a measure of projectworth that is not affected by the size ofthe project. Choosing the project with thehighest B/C ratio, regardless of the sizeof the absolute size of the project, willensure the highest possible returns to theinvestment.

Factoring in non-economicconsiderationsMeasures of project worth such as theNPV and the B/C ratio provide usefulinformation that can help in decidingwhether or not to proceed with aproposed investment, but they should notbe the sole basis for the decision. Mostpotential investments are characterizedby costs and benefits whose value cannoteasily be assessed, meaning they cannotbe incorporated into the economic


86

“bottom line.” For this reason, beforetaking a decision based on conventionalmeasures of project worth such as theNPV and the B/C ratio, it is important toassess the extent to which non-economicconsiderations should be allowed toinfluence the final decision. Only afterthese non-economic considerations havebeen carefully considered can a balancedjudgment be made concerning howto proceed.

Conclusions

With funds for agricultural researchbecoming increasingly scarce in mostcountries, research administrators facemounting pressure to ensure that availableresources are used efficiently. Althoughthere can be no question that a properlyorganized and managed physiologycomponent has the potential to add valueto wheat breeding activities, this does notnecessarily mean that every wheatbreeding program should include one.

This chapter has reviewed some basicconcepts from investment analysis thatcan be used to assess the desirability ofinvesting in a physiology component ofa breeding program. We have described aseries of steps that may be useful inhelping to formalize decisions that alltoo often are still left to the “gut feeling”of scientists and research administrators:

• establish that the problem isappropriately addressed throughbreeding;

• estimate in rough terms the costs andbenefits of the current breedingprogram;

• identify activities that are likely tochange with the incorporation ofphysiology;

• estimate the economic consequencesin terms of changes in costs andbenefits;

• calculate economic measures ofproject worth (NPV and B/C ratio);and

• factor in any non-economicconsiderations.

Economic analysis is not infallible, sofollowing these steps will not necessarilyensure that the “correct” decision will betaken. And as we have pointed out, theoutcomes of research are by natureuncertain, so some of the parameters usedin the economic analysis will necessarilybe tenuous. But one big advantage ofinvoking an economic framework ofanalysis is that it forces decision makersto think somewhat more systematicallyabout the many factors that are likely toinfluence the outcome of investmentdecisions; this in turn increases thelikelihood of achieving a favorableoutcome.

If decisions about the role of physiologyin wheat breeding are taken based partlyon economic considerations, thenchanges made to the organization andmanagement of existing breedingprograms are likely to lead to genuineimprovements in efficiency.Improvements in efficiency in turnenhance the flow of new varietiesemanating from the breeding programs,leading to increases in farm-levelproductivity that will eventually benefitboth producers and consumers.

References

Bingham, J. 1979. Wheat breeding objectives andprospects. Agricultural Progress 54:1-17.

Brennan, J.P. 1989a. An analytical model of a wheatbreeding program. Agricultural Systems31(4):349-66.

Brennan, J.P. 1989b. An analysis of the economicpotential of some innovations in a wheatbreeding programme. Aust. J. Agric. Economics33(1):48-55.

Brennan, J.P. 1992. Economic Criteria forEstablishing Plant Breeding Programs.CIMMYT Economics Working Paper 92-01.Mexico, D.F.: CIMMYT.

Brennan, J.P. 1997. Economic aspects of qualityissues in wheat variety improvement. In: Steele,J.L. and Chung, O.K. (eds.). Proceedings,International Wheat Quality Conference.pp. 363-76.

Brennan, J.P. (forthcoming). Efficiency in wheatimprovement research: A case study of wheatimprovement research in Australia. In: Maredia,M. and Byerlee, D. (eds.). Efficiency ofInvestment in National and International WheatImprovement Research. CIMMYT ResearchReport. Mexico, D.F.: CIMMYT.

Brennan, J.P., and Khan, M.A. 1989. Costs ofOperating a Wheat Breeding Program. Ruraland Resource Economics Report No. 5.Division of Rural and Resource Economics.N.S.W. Agriculture and Fisheries, Sydney.

Brennan, J.P., and O’Brien, L. 1991. An economicinvestigation of early-generation quality testingin a wheat breeding program. Plant Breeding106(2):132-40.

Gittinger, J.P. 1982. Economic Analysis ofAgricultural Projects. Second Edition. JohnsHopkins, Baltimore, Maryland.

Maredia, M.K. 1993. The economics of internationalagricultural research spillovers: Implications forthe efficient design of wheat improvementresearch programs. Unpublished PhDdissertation, Michigan State University, EastLansing, MI.

Morris, M.L., Clancy, C., and Lopez-Pereira, M.A.1992. Maize research investment and impacts indeveloping countries. Part I of 1991-92CIMMYT World Maize Facts and Trends: MaizeResearch Investment and Impacts in DevelopingCountries. Mexico, D.F.: CIMMYT.

O’Brien, L., and Ronalds, J.A. 1984. Yield andquality interrelationships amongst randomF3 lines and their implications for wheatbreeding. Austr. J. Agric. Res. 35(6):443-51.

Reynolds, M.P., S. Nagarajan, M.A. Razzaque, andO.A.A. Ageeb (eds.). 1997. Using CanopyTemperature Depression to Select for YieldPotential of Wheat in Heat StressedEnvironments. Wheat Special Report No. 42.Mexico, D.F.: CIMMYT.


87

BREEDING FOR ADAPTATION

TO ENVIRONMENTAL

FACTORS

88

Traits to Improve Yield inDry EnvironmentsR.A. Richards, A.G. Condon, and G.J. Rebetzke1

Genetic increases in wheat yields in dryareas have not been as great as in morefavorable environments or whereirrigation is available. A likely reasonfor this is that dry environments arecharacterized by unpredictable andhighly variable seasonal rainfall and,hence, highly variable yields. Thisresults in slow genetic advances inbreeding programs because geneticvariation in yield is masked by largegenotype x year and/or genotype xlocation interactions.

It is interesting, but not surprising, thatgenetic increases in yield potential madeby selection in predictable irrigatedenvironments have resulted in broadlyadapted wheats that are often wellsuited to both favorable and lowyielding, rainfed environments. Thisarises because genetic variation in traitsthat contribute to high yield in allenvironments, such as a high harvestindex, is greater in predictableenvironments and, therefore, morelikely to be selected under favorableconditions. There is no reason whygenetic advance in favorableenvironments should not continue tocontribute to yield in less favorableenvironments, provided germplasm iswidely evaluated in rainfedenvironments. However, a host ofspecific adaptations that may beuniquely important in rainfedenvironments can also be targeted toachieve higher regional yields.

Breeding for specific physiological traitsthat are expected to impart a yieldadvantage in dry environments has beennotoriously difficult and unsuccessful.Among the reasons for the lack ofsuccess is that considerable researcheffort has been directed towards traitsthat are unlikely to improve productivity,and traits that have a low heritability orare difficult to measure. There may alsobe negative correlations between droughtadaptive traits and with yield. Forexample, earlier flowering may result inreduced biomass accumulation or mayincrease the risk of frost damage.

Traits selected may also be inappropriatefor the target region. For example, unlessthe trait conditioning survival at theseedling stage also conditions responseto drought at later stages, breeding forsurvival during drought at the seedlingstage may be totally inappropriate ifdrought only occurs around flowering orgrainfilling. Another reason any attemptto breed for yield under drought is likelyto fail is that in many dry environmentslow rainfall is not always the primaryfactor responsible for low yields; otherfactors, such as soil mineral nutrition orsoilborne diseases, may be theoverriding constraints. Since it is oftendifficult to identify those other, lessobvious factors, this may also limitgenetic progress in dry environments.

Only two traits have had a major impacton improving yields in rainfedenvironments: flowering time and plant

height. Genetic manipulation offlowering time has been important toadjust the duration of vegetative growth,reproductive growth, and grain growth inrelation to water supply, frost, andevaporative demand. A reduction in plantheight has been universally important inincreasing the proportion of grain tobiomass (harvest index), providedbiomass growth is not compromised.

The understanding of physiological andmorphological traits that limit yieldunder drought has improved in recentyears, and this has opened up newopportunities. Before discussing these,and how they may be used in a breedingprogram, it is important, first, toestablish how widespread drought maybe and in which environments yield maybe limited by insufficient water. It is alsoimportant to establish, in anyenvironment, whether water is theprimary factor limiting yields or whetherother factors may override the waterlimitation.

The Extent of Droughtand Its Nature

Globally, CIMMYT recognizes 12distinct mega-environments (MEs)where wheat is produced. Only three areirrigated (Rajaram et al., 1995). Severalof the rainfed environments have highrainfall (>500 mm) but may experienceintermittent drought or terminal droughtin years when rainfall is below average.

1 CSIRO Plant Industry, Canberra, Australia.

C HAPTER 7

89

Three MEs experience frequent droughtsand are considered marginal with respectto food production.

There are subclassifications within eachof the MEs. For example, ME4, one ofthe largest rainfed environments,comprising about 33 million hectares, isfurther subdivided into environmentswith late drought (e.g., Mediterraneanclimates), early drought (e.g., Argentina)and regions where wheat is grown onresidual moisture (e.g., parts of Indiafollowing monsoon rains).

Since drought can also be a problem infavorable environments, physiologicalmeans of minimizing drought stress orimproving water use efficiency may alsoinfluence yield in high yielding, rainfedenvironments. Indeed, the best farmerswill often have the best yields as well asexperience the most severe droughtsbecause their practices lead to thegreatest use of available soil water.Reducing the impact of drought may alsobe important in irrigated environments ifit results in less water being used toachieve high yield (i.e., high water useefficiency).

Is Drought the PrimaryDeterminant of Yield ina Dry Environment?

Low rainfall is usually perceived to bethe most important factor resulting in lowyields in dry environments. However,this may not always be true. Otherfactors, such as disease, soil nutritionalproblems, or even waterlogging at certaintimes, may limit yields and shouldprobably be overcome as far as possiblebefore applying physiologicalunderstanding to improve yields underdrought. Because foliar diseases areeasily observed, they are targeted inbreeding programs. However, other moreinsidious, not readily identifiableproblems may also result in low yields

and mistakenly be attributed to drought.There may be nutritional problems,perhaps mineral toxicities due to soil pH;nutrients may be chronically low or eventoo high; there may be soilbornepathogens such as nematodes, root orcrown fungal diseases such as take-all,or rhizoctonia. All of these factors resultin either poor root growth or diseasedroots, and reduce water uptake, inducingsymptoms of drought (Picture 1).

There are a number of ways to determinewhether soil-based factors limit yields(Table 1). First, soil tests can beconducted to assess whether a range ofdisorders such as pH or micro- andmacro-nutrient deficiencies or toxicitiesmay limit yields. Second, roots can beexamined around mid-tillering todetermine the presence of take-all,rhizoctonia, or some nematode species(e.g., cereal cyst nematode). Third, tillerdevelopment can be monitored. Tilleringin temperate cereals follows a verypredictable pattern in favorableconditions and therefore can be used todetect poor plant health. For example,primary tillers appear in the axils ofleaves 1, 2, 3 etc., and secondary tillersappear in the axils of primary tillerleaves. Missing tillers under favorable

light and macro-nutrient conditionsprobably indicate some disorder.

Finally, a very effective way to identifylimiting factors is to grow a range ofprobe genotypes, or other cereal species,that are known to vary in their toleranceor resistance to soil mineral disorders orsoilborne pathogens (Cooper and Fox,1996). Relatively greater growth or yieldof one or more of the probe genotypesmay make it possible to diagnose aparticular problem.

Another valuable way to determinewhether factors other than drought aremore important in limiting yields is to

Table 1. Methods to assess whether bioticor abiotic factors other than drought arelimiting yields.†

Tests for soil pH or micro- and macro-nutrientdeficiencies or toxicities

Planting and subsequent growth of probe genotypesor species

Tiller development

Examination of roots

Soil moisture availability at harvest

Calculation of water-use efficiency

† Methods are listed in the order in which they should beconducted during the crop cycle.

Picture 1. Droughted sections of fields caused by the root disease take-all andwaterlogging during the vegetative period. Limiting the effects of these factors, principallythrough management, will increase yields in dry environments.

TRAITS TO IMPROVE YIELD IN DRY ENVIRONMENTS

90

calculate the water use efficiency of thecrop and/or determine whether crops areleaving water behind in the soil afterharvest. Soil moisture measurementsdown to at least 1 m will establish thelatter. The simplest calculation of wateruse efficiency is the quotient of grainyield to cumulative rainfall from amonth before sowing to physiologicalmaturity. A more accurate value isobtained by using the sum of seasonalrainfall and the difference in soilmoisture content at sowing and atharvest from at least the top 1 m of soil.This value will vary depending on thearidity of the environment, the amount ofwater stored in the soil prior to sowing,and the seasonal distribution of rainfall.Maximum values are around 20 kg ha-1

mm-1 for cooler environments and maybe as low as 10 kg ha-1 mm-1 in hotter,drier environments. However, comparingmoisture content in different soil typesand among farmers within a regionshould identify likely anomalies.

If factors other than drought are found tobe the primary determinants of yield,both breeding and management may beimportant to overcome them. Breedingmay be used to increase tolerance orresistance to diseases and to increasetolerance to nutrient disorders, some ofwhich are discussed in this volume.Rotations may be the most effective wayto reduce root diseases in the absence ofgenetic resistance.

Breeding for Yield inWater-StressedEnvironments

Yields are likely to continue to increase,although possibly at a slower rate than inthe past few decades. Increases in wheatyields in rainfed environments have beenachieved during most of this centurythrough the use of conventional breedingmethods. Direct selection for yield in

target environments will remain thecornerstone of wheat improvementbecause of the great integrating capacityof final grain weight for genotypicadaptation and the efficiency with whichtrials can now be conducted. Theeffectiveness of conventional breedingmethods will depend on non-geneticfactors such as how precise experimentsare, on land and land preparation, sowingand harvesting equipment, trialmaintenance, as well as statisticalprocedures to more precisely discriminatebetween entries in less than ideal trials orfield conditions. Their effectiveness willalso depend on trials being conducted inrepresentative regions using standardfarming practices and adequate plot sizesto estimate yields.

It is expected that physiologicalapproaches to breeding will become moreimportant. These will develop from animproved understanding of the factorsregulating wheat production in theprevailing farming systems, and of wheatphysiology and ways to manipulate it inrelation to the climate where wheat isgrown. This knowledge will make iteasier to identify the major limitingfactors and ways to overcome them bymore precise targeting of physiologicaltraits to reduce the impact of drought andthereby increase yields.

A physiological approach may increasethe rate of yield improvement in anumber of ways. First, by identifyingtraits for which there is inadequategenetic variation in breeders’populations. This, in turn, will facilitatethe identification of new parental lines toincrease variability for key traits. Second,large seasonal variation in yield and G×Emay make direct selection for yieldineffective, so that more specifictargeting of physiological characters thatlimit yield and have high heritability maybe more effective than direct selection foryield. Third, greater yield stability may

be important to minimize yield losses inthe driest years and could be achievedthrough selection for physiological traits.Fourth, selection for physiological traits,particularly in early generations or out ofseason, may be more cost effective thandirect yield selection. Yield trials areexpensive to conduct, and if thepopulation can be culled in earliergenerations using critical physiologicalcriteria, this will allow either more highyielding, adapted entries to be tested orgreater replication to increase selectionprecision. Fifth, selection may beconducted out of season, which wouldallow several generations to becompleted each year.

If the physiological trait has reasonablyhigh heritability and is not too difficultto select, backcrossing can be veryeffective for incorporating the trait intoan already well adapted cultivar withgood grain quality and diseaseresistance. This will ensure rapidprogress in breeding and should result ina high frequency of progeny with highyield, good quality, and appropriatedisease resistance.

Trait identificationThe most appropriate way to identifytraits that may limit wheat yields in dryenvironments is to use the frameworkproposed by Passioura (1977). Thisframework is based on grain yield, noton drought protection or survival underdrought, which were popular in the pastbut largely unsuccessful. Passiouraproposed that when water is limiting,grain yield is a function of: 1) theamount of water used by the crop, 2)how efficiently the crop uses water forbiomass growth (i.e., water-useefficiency, or above-ground biomass/water use), and 3) the harvest index, i.e.,the proportion of grain yield to above-ground biomass. Since each of thesecomponents is likely to be largelydependent on the others, an

R.A. RICHARDS, A.G. CONDON, AND G.J. REBETZKE

91

improvement in any one of them shouldresult in an increase in yield. Thisidentity is shown below.

The local environment must beconsidered in conjunction with thisidentity, since traits identified as yieldlimiting may only be so in specificenvironments. Indeed, because ofseasonal variability in rainfall in dryenvironments, a particular trait may noteven be important every season in agiven region. Exceptions to this are traitsthat are universally important in water-stressed environments—for example,good crop emergence and establishment,and high water use efficiency.Subsequent tables listing traits willindicate whether they are universal orspecific to a particular type ofenvironment (e.g., having early or end-of-season drought).

Appropriate flowering time is the singlemost important factor to maximize yieldand adaptation in dry environments.Further genetic modification offlowering time in different regions islikely, since management practices areconstantly changing and providing newcrop opportunities. For example, newmachinery and herbicides may facilitateearlier sowing, or the best adaptedcultivars may require differentsensitivities to photoperiod orvernalization than currently growncultivars.

In the rest of this paper we shall tabulatethe traits that may be important toincrease yield in water-stressedenvironments. A comprehensive list oftraits is provided for consideration. Thisis because the choice of traits for use in abreeding program will depend onnumerous factors such as the nature ofdrought, trait expression in currentcultivars, available genetic variation, and

ease of genetic manipulation. We start offwith traits that can be selected in earlygenerations, and then list and discuss insome detail traits that may be importantto increase either water use, water useefficiency, or harvest index.

Selection in F2 populationsTable 2 lists physiological traits that, ingeneral, have high heritability and can bevisually selected in F2 populations. Allare likely to be important in most rainfedenvironments. For selection in the F2,populations should be grown underfavorable moisture conditions tomaximize genetic variation inmorphological traits and to allow thedevelopment of diseases so that selectionfor resistance can also be made.

Water Use

A deep root system is synonymous withdrought resistance and with more wateruptake from the soil. This trait is, ofcourse, difficult to measure. However, itmay be that the root system of currentcultivars is adequate and so furtherimprovement may not be necessary.

However, information on whethercurrent cultivars extract all available soilwater is required to establish this.

If, on the contrary, the root system ofcurrent cultivars does needimprovement, the simplest way toincrease rooting depth and rootdistribution is to increase the duration ofthe vegetative period. This may beachieved by sowing earlier or plantinglater-flowering genotypes. Increasedearly vigor may result in both fastergrowth of deep roots and moreadventitious roots in the top soil. Thelatter may be important to mop up waterand nutrients before evaporative lossesdry the top soil. Appropriate waysto select for greater vigor will bediscussed later.

Table 3 lists plant characteristics thatmay either increase crop water use orroot growth, or indicate genotypes thathave a deep root system. Phenology andearly vigor are listed as traits thatincrease water use, since they may resultin deeper roots. Tiller inhibition is alsoincluded in that category, based onevidence that assimilates normally usedfor the growth of additional tillers could

Grain=

CropX

Water-useX

Harvestyield water use efficiency index

Table 2. Morphological traits for visual selection in an F2 population in regions wheredrought limits yield.

Universalor environment-

Trait Heritability Expected GxE specific trait

Selected at floweringFlowering time high low specificSmall flag leaves intermediate high universalGlaucousness high low universalAwns high low universalDisease resistance disease dependent low universal

Selected near physiological maturityPlant height high low universalFloret fertility low intermediate universalMaintenance of green leaf area low high specificLarge seed size high intermediate universal


92

be used for root growth if a gene fortiller inhibition is present. Osmoticadjustment may also increase rootgrowth and the ability to extract moresoil water. However, selection for thistrait is not easy at the present time.

Traits that indicate deeper rooting maybe used as selection criteria (Table 3).Low canopy temperature or highstomatal conductance (see later), both ofwhich can be measured very simply, mayindicate more favorable soil moistureconditions and, hence, a deeper rootsystem. These characteristics could bevaluable in selection, but measuringthem requires extremely uniform soils toeliminate any subsoil spatial variation.

“Stay-green” leaves may also be anindicator of favorable soil moistureconditions and, therefore, deeper roots.This trait is desirable when, after verydry conditions, there is high probabilityof further rainfall. Stay-green capabilitymeans there would be morephotosynthetic tissue for furtherassimilation and additional soil waterextraction. The degree of rolling in theflag leaf, which occurs in unbent leavesof plants growing in dry soils, may alsoindicate plant water status and, hence, adeep root system. Leaf rolling may be anadaptive feature to avoid leaf senescence

and maintain leaf area, which allowsfurther soil water extraction in the eventof late rains. However, care must betaken not to to confound these traits withanthesis date.

Water use efficiencyThe term water use efficiency (WUE) isgenerally used to express the ratio oftotal dry matter to evapotranspiration.An increase in transpiration efficiency(TE, dry matter/transpiration) and/or areduction in soil evaporation willincrease WUE. Both of thesecomponents may readily be improvedthrough breeding.

Reducing water evaporationfrom the soilWheat is often grown in environmentswhere rainfall between sowing and stemelongation is frequent. This is typical ofMediterranean environments around theworld. In these environments, avoidingwater evaporation from the soil isessential to ensure sufficient moisturethroughout the crop season. Any increasein early seedling vigor should reduceevaporative losses from the soil surface(Picture 2). The potential for gains islarge because 1) as much as half of thegrowing-season rainfall may be lost

through evaporation from the soil, and 2)widely-grown semidwarf wheats haveinherently low vigor compared withtaller wheats (Richards, 1992). Greatervigor is not likely to be as important inother environments, although if it resultsin more growth when vapor pressuredeficit is low, this will also result inhigher TE. If crop duration is short,greater vigor is likely to increase finalbiomass and yield. Furthermore, greatercrop vigor may be an effective way toreduce weed growth and, hence,herbicide use in most environments.

Traits that may contribute to increaseseedling vigor are listed in Table 4. Thefirst requirement for maximizing vigor isto establish a high plant population asquickly as possible. This has becomeparticularly important with thewidespread adoption of semidwarfcultivars, as they have a shortercoleoptile and slower emergence thanstandard-height wheats. Short coleoptilesresult in poor emergence, which leads topoor crop establishment. This is true,particularly when seeds of semidwarfwheats are sown deeply to seek moistureor are sown into stubble. Better

Table 3. Plant traits that may increase soil water use or root growth, or indicate deep rooting.

Universal orHeritability Expected GxE environment-specific trait

Traits that increase soil water useDeeper roots low high specificPhenology high low specificSeedling vigor high low specificTiller inhibition high low specificOsmotic adjustment low high specific

Traits that indicate deeper rootsCanopy temperature intermediate high -Stomatal conductance intermediate high -Stay-green intermediate high -Leaf rolling intermediate high -

Picture 2. Greater seedling vigor, such asthat achieved by the crop on the left, willreduce loss of soil water by evaporationfrom the soil surface and limit weed growth.


93

maximizes yield, they also limit thelength of coleoptiles, which often causespoor establishment (Picture 3). However,wheat plant height can be reduced by theuse of either GA-insensitive or GA-sensitive genes. There are GA-sensitivegenes that reduce plant height to that ofplants that possess GA-insensitive Rhtgenes, yet produce coleoptiles with up to100% greater length than coleoptilesfound in GA-insensitive geneticbackgrounds (Rebetzke et al., 1999).Furthermore, plant height and coleoptilelength seem unrelated in GA-sensitivebackgrounds, which facilitatessimultaneous selection for both traits in abreeding population. Our studies showthese GA-sensitive dwarfing genesconfer the same desirable partitioningcharacteristics as in current semidwarfwheats, but may also produce greaterbiomass when soil conditions at sowingare unfavorable (Rebetzke and Richards,2000).

Genetic studies conducted on differentwheat populations have shown that GA-sensitive dwarfing genes have high

Table 4. Traits that may improve plant establishment and early canopy development in wheat.

Universal orTraits Heritability Expected GxE environment-specific trait

Highest priorityLong coleoptiles high low universalBroad seedling leaves high low specific

Émbryo size high low specificSpecific leaf area intermediate high specific

Large coleoptile tiller intermediate high specific

Lower priorityLarge grains high low universalFast emergence low low specificFast leaf expansion rate intermediate low specificLow temperature tolerance intermediate low specificCrown depth intermediate intermediate specificCrown to shoot partitioning intermediate low universalLeaf area ratio intermediate low specific

emergence is achieved by sowing wheatswith long coleoptiles. Increased coleoptilelength can be achieved by selectionwithin semidwarf germplasm, but greaterprogress can be made using parents thatare sensitive to gibberellic acid (GA),although short stature also needs to beselected (Rebetzke and Richards, 1999).Such wheats may have a coleoptile up totwice the length of coleoptiles found incurrent semidwarfs. Wheats with longcoleoptiles also tend to have larger earlyleaves and more rapid rates of emergence,which together contribute to faster leafarea development.

Wheats with large grains also have ahigher rate of emergence and larger, morevigorous plants than small-grain wheats.However, if sowing rate is based onweight per unit area, there may be littleadvantage in sowing large-grain wheats.Of traits that improve early vigor(Table 4), we have found that the breadthof the seedling leaves and the frequencyand size of coleoptile tillers are likely tobe the most effective. These two traits,together with long coleoptiles, should beselected first. Selection protocols forthese are described later.

Other traits that are likely to improveseedling vigor, but which we consider oflower priority, are also shown in Table 4.Although genetic variation exists for eachof these traits, they are less likely toinfluence vigor, are more difficult toselect, or initial findings suggest thatgenetic variation is small. Nevertheless,expression of these lower priority traitsmay in some cases be improved byselection for the high priority traits. Forexample, to improve plant establishmentby increasing coleoptile length, werecommend substituting the GA-insensitive dwarfing genes Rht1 (Rht-Blb)and Rht2 (Rht-Dlb) with GA-sensitivemajor or minor genes. Evidence suggeststhat this will also increase seedlingemergence and leaf expansion rates.Furthermore, selection for broad seedlingleaves, which integrates both embryo sizeand specific leaf area, should alsoincrease leaf area ratio.

Improving establishmentThe height of semidwarf wheat cultivarsis principally due to the GA insensitiveRht1 and Rht2 alleles. Although thesealleles result in a plant height that

Picture 3. Semidwarf GA-sensitive wheatswith long coleoptiles establish better underadverse conditions than GA-insensitive semid-warf wheats with the Rht1 and Rht2 alleles.


94

heritability, making it easy to select forthem. Coleoptile length is also highlyheritable, with long coleoptile selectionsmaintaining their longer coleoptileranking across a wide range of soiltemperatures.

We typically select at 19°C, as largenumbers of families can be sown andassessed in just under two weeks. Seedsof uniform size are sown at a uniformdepth (10 mm) in a deep, wooden traycontaining a fertile potting mix wateredto field capacity. Trays are covered withan opaque plastic sheet to exclude lightand are then placed at 19°C. After200°Cd (approximately 10 days,assuming a base temperature of 0°C), theplastic sheet is removed and longercoleoptile progeny identified eithervisually or by measuring with a ruler(Picture 4). Selected families can then betransplanted into a glasshouse or into thefield, provided good care is given to thetransplanted seedlings. Plant height canbe determined based on primary tillers atmaturity, and plants of the desired heightretained. High heritability for plantheight and coleoptile length indicatesthat screening for these traits cancommence as early as the F2 generation.

Seedling vigorAlthough wheats with high seedlingvigor offer considerable promise, there islittle genetic variation for thischaracteristic among currently grownsemidwarf wheat cultivars.Characterization of seedling vigordifferences between wheat and barley(López-Castañeda et al., 1996) hasshown that barley achieves almostdouble the leaf area of wheat primarilybecause of its earlier emergence (around1 day), larger embryo, and greaterspecific leaf area (leaf area-to-leaf massratio). Barley often produces largecoleoptile tillers that emerge before theprimary tillers and contribute toincreased vigor. Exhaustive screening ofinternational wheat collections hasrevealed two populations containingsuitable genetic variation for leaf areameasured early in the season (Richardsand Lukacs, 2001). As in barley, thegreater early vigor of these wheats arisesfrom their larger embryo and specificleaf area. Pyramiding these twoindependent characteristics has producedprogeny with even greater vigor than theoriginal wheat parents (Richards, 1996).Large coleoptile tillers are also frequentin these high vigor progeny.

Our experiments show that heritabilityfor early leaf area is small. However, leafbreadth averaged across the first twoleaves is highly heritable and has astrong genetic correlation with leaf area(Rebetzke and Richards, 1999a). Thecombination of higher heritability andstrong genetic correlation suggests thatselection for early leaf area developmentusing leaf breadth as a measure of leafarea is as good as selecting for leaf areaitself. Measurement of leaf width hasother advantages: it is rapid and non-destructive, and can be done onseedlings. The larger heritability of leafwidth indicates that it is less sensitive toG×E interaction than leaf area.Nonetheless, genetic variance for leafwidth and, to a greater extent, for plantleaf area and biomass is greatest in thecooler months when wheat is typicallysown in farmers’ fields. Familiesshowing greater seedling vigor in our exsitu tray assessments also show greatervigor in the field. This is also true forcoleoptile tiller assessments, althoughthe appearance of the coleoptile tillerseems to be somewhat dependent on soilfertility and soil strength.

The procedure for assessing lines orfamilies for early vigor is a simple one.It is essential to start with good qualityseed of each line. Because seed size canhave a big effect on early vigor, it isnecessary to discard seeds that are eithertoo small or too large, or to weighindividual seeds for later covarianceadjustment. We measure seed size eithervisually based on seed length × breadthassessments or by weighing seed on abalance. Seed may also be passedthrough different sized screens.

Seeds are planted at uniform depth in adeep tray containing a fertile potting mixand then watered. Time of emergence isscored by measuring seedling height(from the soil surface to the tip of thefirst leaf) when approximately 90% ofseedlings have emerged. Leaf width ofPicture 4. Wooden tray containing seedlings grown in the dark to screen for coleoptile length.


95

the first two leaves is measured with aruler upon full expansion of the secondleaf (usually at 2.4 leaves). Leaf lengthand leaf number are also measured at thistime. Coleoptile tillers are simplyassessed by recording their presence orabsence; if present, the length of the tilleris recorded. More detailed assessment ofcoleoptile tillers might include the timeof tiller emergence from the soil. Thebest plants or families are thentransplanted and used for hybridizationor seed increase. Picture 5 showsseedlings growing in wooden trays at thetime of selection.

Field observations indicate that wheatscontaining GA-insensitive genes forreduced height produce smaller leaf areasand plant dry weights early in the season(Richards, 1992). Preliminary data alsoindicate that the potential of largerembryos and greater SLA to increaseseedling vigor may be somewhatrestricted when expressed in these GA-insensitive Rht genetic backgrounds. Webelieve that the expression of seedlingvigor using larger embryos and greaterSLA would be enhanced in geneticbackgrounds containing GA-sensitivedwarfing genes. Furthermore, longercoleoptiles and greater seedling vigorwould provide a “package” that is bettersuited to production in less favorableenvironments.

Greater transpirationefficiencyTranspiration efficiency (the ratio of drymatter to transpiration), the othercomponent of water use efficiency, isalso amenable to improvement. There arenumerous ways to increase TE in wheat;the more important ones are given inTable 5. Perhaps the simplest way toimprove TE is to ensure that the period ofmaximum biomass increase occursduring the coolest period. This capitalizeson the fact that less water is required forgrowth when it is cool. To achieve thismay require a change in planting time so

that full canopy closure is achieved by theonset of the coolest period. It may alsorequire wheats with a different phenology.Since this is such a simple way toimprove TE, opportunities to altermanagement practices for earlier sowing,including dry seeding, should be exploredwhere practical. Selection for increasedvigor outlined earlier should also result ina higher leaf area index and, hence, morelight interception and growth when it iscool, resulting in higher TE.

Other ways to improve TE are toincrease surface reflectance, therebylowering surface temperatures ofphotosynthetic tissue. Selection forglaucousness and, possibly, pubescenceare two ways to achieve this. Also,smaller photosynthetic surfaces aremore effective at dissipating heat thanlarger surfaces, if it is dry and hot.Selection for awns that maintainphotosynthetic activity and for a smallerect upper canopy of leaves may be an

Table 5. Traits that can be selected to improve transpiration efficiency in wheat.

Universal orTraits Heritability Expected GxE environment-specific trait

Phenology high low specificSeedling vigor high low specificCarbon isotope discrimination high low universal

Ash content ? highNIR ? ?Stomatal conductance intermediate highSPAD intermediate intermediateSLA intermediate intermediateCanopy temperature intermediate intermediate

Glaucousness high low universalPubescence high low universalResidual transpiration ? high universalLeaf size and habit ? high universal

Picture 5. Measuring leaf breadth of seedlings grown in trays to screen for early vigor.


96

effective way of achieving this. Sometranspiration occurs at night throughincompletely closed stomata and throughthe cuticle. This is unlikely to be large inmost environments, but it could result inup to 0.5 mm of water per day being lostfrom a crop at night if the vapor pressuredeficit is high. Cuticular loss during theday may also be important. Substantialgenetic variation for cuticulartranspiration has been found in wheat,and a relatively simple procedure toselect for it has been established (Clarkeand McCaig, 1982).

Carbon isotopediscriminationCarbon isotope discrimination (Δ) is ameasure of the ratio of the stable isotopesof carbon (13C/12C) in plant dry matterrelative to the value of the 13C/12C ratioin the air that plants use inphotosynthesis. About 1% of the CO2 inthe atmosphere contains 13C. Because13CO2 is a larger molecule than 12CO2,plant species such as wheat and barleywith the C3 photosynthetic pathwaydiscriminate against 13CO2 duringphotosynthesis. As a result there isrelatively less 13C in plant dry matterthan in the atmosphere. The extent ofdiscrimination against 13C varies amonggenotypes.

The processes that influence the extent of13C discrimination are also important indetermining the TE of leaf gas exchange.Thus Δ provides a relative measure ofleaf-level TE among genotypes orbreeding lines. Discrimination is less(i.e., the value of Δ is low) when TE ishigh. In terms of leaf gas exchange, TE isa ratio which describes how much CO2 isassimilated in photosynthesis per unitwater lost in transpiration. The rate ofCO2 uptake into the leaf is determined by1) the “sucking power” of the leaf forCO2, i.e., the amount of photosyntheticmachinery per unit leaf area, and 2) howeasily CO2 can move into the leaf, which

is determined by the stomatalconductance. Since CO2 and water areexchanged through the same stomatalpores, stomatal conductance is alsoimportant in determining the rate oftranspiration. The other majordeterminant of transpiration rate is theevaporative demand on the leaf (i.e., the“sucking power” of the air for water),which is most precisely measured as thevapor pressure gradient between the leafand the air.

We have shown that the Δ of plantmaterial is closely related to TEintegrated over the life of the plantmaterial sampled (Farquhar andRichards, 1984; Condon et al., 1992).We are now using this technique to breedwheats with higher TE. Lines developedso far using backcrossing have higheryields than the recurrent parents inwater-stressed environments.

There are several properties of Δ thatmake it appealing as a potential breedingtool (Hall et al., 1994). Mostimportantly, Δ is a lot easier and faster tomeasure than TE itself. Measuring Δ ofplant dry matter sampled from a

collection of genotypes provides anestimate of variation in leaf-level TEintegrated over the time that the drymatter was laid down. We haveestablished that the heritability of Δ ishigh and that Δ can be measured onfreshly sampled dry matter or onsamples that have been storedindefinitely.

Our experience indicates that assessinglines or families for Δ is best done usingplants grown in the field. In essence,leaves of similar age are sampled fromfield plants at full tillering before theonset of drought, when vapor pressuredeficit is low. Sampling plant materiallater is less reliable because of possibledifferences in phenology, soil wateravailability, and translocation ofassimilates formed earlier. We havefound heritability of single-plant Δ to below; hence, it is advisable to takesamples from several plants.

To do this we sow F3 (or latergeneration) families in replicated shortrows (Picture 6). Plant material issampled at full tillering, before stemelongation, by cutting along the row at

Picture 6. Short rows of F3 families from which leaf material will be harvested toscreen for carbon isotope discrimination.


97

Picture 7. Viscous-flow porometer used to measure leaf stomatal conductance. It takesbetween 10 and 20 seconds to do single measurements on healthy wheat plants.

about 5 cm above the soil surface. Theplant tops (mainly leaf material withminimal sheath and stem) are placed inpaper bags for oven drying at about70oC. Because they are cut above theapex, the plants are able to recover, andplant height, flowering time, and diseasereactions can be assessed. If Δ analysesare completed before flowering, it is alsopossible to identify lines that could beused in further crossing (e.g., in abackcrossing program). Once thesampled material is oven dried, it can bestored indefinitely. Prior to Δ analysisthe material should be ground as finelyas possible. It is useful to re-dry thesamples before grinding to removeresidual moisture. Thorough mixing ofthe ground sample is also advisable tominimize subsampling errors, since onlyabout 5 mg of the sample is used forcarbon isotope analysis.

For temperate cereals, including wheat,the utility of Δ in breeding is likely tovary depending on the extent to whichyield is limited by water supply and alsoat what stage of the crop cycle waterstress occurs. In wheat, Δ is a“conservative” trait associated withsomewhat slower water use and possiblyslower growth rate. Consequently, inenvironments where water is largelynon-limiting, low Δ (high TE) has beenassociated with relatively low yieldpotential in wheat (Condon et al., 1987).The implication is that selection for highΔ in these environments may proveuseful in identifying lines with highyield potential (see chapter by Fischer).Selection for high Δ may also beeffective for yield improvement inrainfed environments where crop growthto anthesis is sustained by currentrainfall or where drought is relieved wellbefore anthesis. In these environmentshigh Δ should be associated with moredry matter at maturity.

In environments where crop growth isheavily dependent on moisture stored in

the soil profile from rain that fallsoutside the main crop growth phase,selection for low Δ (rather than high Δ)is likely to be effective for yieldimprovement. In these environments,where transpiration makes up a highproportion of total crop water use, highTE promotes conservation of soil water,which sustains yield determiningprocesses in the period leading up to andafter anthesis. In stored-moistureenvironments, profligate high Δgenotypes are more likely to exhaust thesoil water supply before this criticalphase.

The fact that the direction of selectionbased on Δ may vary depending on thetarget environment may be seen as adisadvantage of Δ. Another disadvantageis that its measurement requiresspecialized equipment (a ratio massspectrometer) and therefore is relativelyexpensive (US$5-15 per sample atcommercial analytical laboratories). Forthis reason several “surrogates” for Δhave been proposed that may be suitablefor use at various stages of a breedingprogram. These surrogates have beenshown either to be correlated with Δ

itself or to provide estimates of theimportant plant processes that determinegenotypic variation for Δ.

Surrogates that may help cull thepopulations that are related to Δ, forreasons not yet clear, are ash content ofdry matter (Masle et al., 1992) and near-infrared reflectance of bulk tissue (Clarket al., 1995). Furthermore, measuringtraits related to the determinants of Δ(viz., stomatal conductance andphotosynthetic capacity) may also beeffective for culling populations duringbreeding. The measurement of leafstomatal conductance, particularly with anew, portable and fast viscous-flowporometer (Picture 7; Rawson, 1996), oran infrared thermometer to measurecanopy temperature (see chapter byReynolds), which in itself is a functionof stomatal conductance, may beadvantageous. Possible surrogates forphotosynthetic capacity are leafchlorophyll content measured using, forexample, the Minolta ‘SPAD Meter’(Araus et al., 1997), or specific leaf area(the leaf area per unit leaf dry weight),which often reflects the amount ofphotosynthetic machinery per unit leafarea (Wright et al., 1988).


98

Drought-dependentharvest indexOnly when the HI of a given genotype inthe absence of drought in the targetenvironment is already high, does thegenetic improvement of the droughtdependent HI become important.Drought dependent HI is a function ofpost-anthesis water use. If post-anthesiswater use as a proportion of total wateruse is large, HI will be large. If soilwater is finite, conserving soil waterbefore flowering so that it can be usedfor grainfilling should increase HI.Achieving high grain yield will thendepend on the balance between growthbefore and after anthesis. Getting thisbalance right is difficult. For example,too little growth before anthesis willlimit total dry matter yield but maximizeHI, whereas too much growth beforeanthesis will ensure that total dry matteryield is maximized, but could result in alow HI.

Water use is a function of evaporativedemand and leaf area. There is little thatcan be done to alter evaporative demand,although crop phenology may be alteredto change the timing of crop growth.However, there are a number of traitsthat may be manipulated to geneticallyreduce leaf area development. Thesemay regulate water use and thereforeeffectively increase the drought-dependent HI. Table 6 lists traits thatachieve this.

Phenology is a major determinant ofdrought independent HI, as it candetermine the amount of pre-anthesisand post-anthesis water use. Forexample, if flowering occurs a few daysearlier, this may mean an extra 5-10 mmof soil water for post-anthesis use(drought escape) and, hence, a higher HIand, perhaps, greater yield. However, ifearlier flowering also results in less pre-anthesis growth, yield may not be greaterdespite the higher HI. Seasonal variationin yields is generally large in waterstressed environments, and sowing

Table 6. Traits that improve the harvest index of wheat.

Universal orTrait Heritability Expected GxE environment-specific trait

Drought-independentPhenology high low specificHeight and peduncle length high low universalTiller inhibition high low specificAssimilate retranslocation intermediate high universal

Drought-dependentPhenology high low specificTiller inhibition high low specificXylem vessel diameter intermediate low specificLeaf conductance intermediate intermediate specificStay-green/ leaf rolling intermediate intermediate specificAssimilate retranslocation intermediate high universal

Harvest Index

Measuring harvest index (HI) is simplerthan measuring water use and water useefficiency. Above-ground biomass andgrain yield of a “grab” sample atmaturity together provide a simplemeasure of HI in a breeding program.Furthermore, when comparinggenotypes, the measurement of HI isgenerally more robust than either of itstwo components (yield and above-ground biomass) and genotype rankingfor HI is relatively stable, providedplants are grown under favorableconditions. However, like water use andwater use efficiency, the geneticmanipulation of HI in variable rainfedenvironments is not simple, given thattwo separate factors determine HI incrops that experience drought and,hence, two factors can be geneticallymanipulated to maximize HI to achieve ahigh grain yield. The first determinant ofHI is independent of drought, i.e., HI inthe absence of drought in a givenenvironment. The second determinant ofHI is drought dependent, i.e., it dependslargely on water availability duringgrainfilling.

Drought-independentharvest indexTraits that result in a high HI underoptimal conditions are likely tocontribute to high yield in allenvironments, provided there is nosacrifice in biomass. This is theadvantage semidwarf wheat varietieshave over standard height varieties andis the reason semidwarfs have been sosuccessful in both favorable and lessfavorable environments. A high droughtindependent HI in a given environmentis a prerequisite to high yield underdrought, as it determines the geneticpotential in that environment. In brief,the drought independent HI is a functionof differential partitioning of dry matterto reproductive and non-reproductiveorgans. Thus, incorporating genes thatcontribute to height reduction and toearly flowering is a simple and effectiveway to increase HI as they result in lessgrowth of vegetative organs. Traits thatincrease the drought independent HI arelisted in Table 6. Factors contributing toa greater HI under favorable conditionsare discussed elsewhere (in this volumeand in Richards, 1996).


99

earlier flowering cultivars may notalways be advantageous. After decadesof breeding and yield testing, it is likelythat anthesis time of successful varietiesin a given region is close to optimum.Nevertheless, earlier flowering willresult in higher water use efficiency inenvironments where temperaturesincrease after anthesis. Combiningearlier flowering with greater vigor orfrost resistance may also assist inimproving HI and yield.

A reduction in tillering may contribute toa higher HI both in the presence andabsence of drought because of the lowernumber of sterile tillers. Reducedtillering may also contribute to a higherHI under drought because a smaller leafarea before anthesis increases thelikelihood of less transpiration and morewater being available for grainfilling.Narrower xylem vessels in the seminalroots would have a similar effect.Seminal roots are responsible for uptakeof water in the deeper soil layers, andreducing the diameter of xylem vesselsin the roots should increase the hydraulicresistance. This in turn slows water usebefore anthesis if it is dry, making moresoil water available for grainfilling. Animportant feature of reducing xylemvessel diameter is that it is likely to beadvantageous under dry conditions butneutral in favorable conditions, as thenodal roots, which are in the topsoil, willsupply the crop with its waterrequirement. Other ways to reduce pre-anthesis transpiration may be to selectfor smaller uppermost leaves, includingthe flag leaf, or for lower stomatalconductance, and/or lower nighttime leafconductance.

There is substantial genetic variationamong wheats for tillering, xylem vesseldiameter, leaf dimensions, and stomatalor cuticular water loss. In the case oftillering, there is a major gene onchromosome 1AS that inhibits tillering(Richards, 1988). The penetrance of this

gene varies with climate and geneticbackground, and so it providessubstantial scope for the regulation oftillering and, therefore, leaf area.

Measuring xylem vessel diameter is notdifficult: a cross-section of the seminalroots is made using a tightly sprung clipto hold the tissue. It is then stained withtoluidine blue. The diameter of thelargest vessel can be measured quicklyunder a microscope (Richards andPassioura, 1981). This is best done onwheat seedlings with two to three leaves.Selected plants can later be grown outfor hybridization or seed increase.Genotypic variation and ways to selectfor stomatal conductance and/ornighttime leaf conductance werediscussed previously.

Manipulating pre- and post-anthesiswater use is one way to increase thedrought dependent HI, but there areothers. Surplus assimilates are stored instems around the time of anthesis. Theseare in the form of water solublecarbohydrates in wheat and can amountto 25% of the total above-ground dryweight at anthesis. Assimilates aretranslocated to the developing kernelsduring grainfilling and, if it is dry, mayform up to 100% of the final grain yield.This redistribution of dry matter can bevery important to increase HI.

There appears to be substantial geneticvariation in wheat for the storage andremobilization of assimilates. Effectiveselection techniques have not yet beendeveloped, although novel proceduresusing senescence agents have been tried(Blum et al., 1983). An effective way toidentify the best parents is to determinethe weight loss in stems betweenanthesis and maturity in genotypesgrown in bordered field plots. This mustbe determined on a ground area basisand should provide an estimate ofassimilate remobilization. Morphologicaltraits could also be effective. Forexample, the tiller inhibition gene results

in thicker stems. Also, there isvariation for the size and anatomy ofthe internode cavity, which may beimportant for assimilate storage.

Some dry environments have a highprobability of rainfall during thegrainfilling period. For these, it may bepossible to extend the duration ofgrainfilling, thereby achieving a higherHI. An extended grainfilling periodwould allow time for furthertranslocation of assimilates to thegrain. Genetic means exist to delayleaf senescence and extend theduration of grainfilling. Selection forleaf rolling is effective in sheddingradiant energy and likely to result incooler leaf temperatures and lesstranspiration, while leaf senescencetolerance or “stay-green” capabilitymay be selected visually.

Concluding Remarks

Making genetic gains in yield underdrought is not an easy task. Even themost assured methods (i.e., empiricalbreeding where plot yield is the unit ofselection) are difficult and slowbecause of the unpredictability ofdrought and the large seasonalvariation. Using a physiologicalapproach (where the underlyingphysiological limitations to yield underdrought are known) to more preciselytarget the yield limiting factors alsohas a substantial risk in a breedingprogram. No single trait is likely to beuniversally important, and the fact thatthere are few examples of success isevidence of the difficulties inherent inthis approach. However, if successful,the resulting benefits would probablybe large.

Identifying yield limiting traits andapplying them effectively in a breedingprogram are major challenges becauseof the different types of drought andseasonal variation in the severity of


100

drought. Also, both predictable andunpredictable pleiotropic effects ofdifferent traits are likely, which adds afurther degree of complexity. Anotherchallenge lies in trait validation, giventhat a character may be important in oneyear but not in the next. Thus, there is nocertainty that targeting so-calledimportant traits will be effective fromone season to the next. For this reason itis important that a physiologicalapproach complement empirical breedingprograms.

A physiological approach has substantialpotential to improve yields and yieldstability over and above that attainablethrough empirical breeding alone. First,it may identify key traits that currentlylimit yield in dry environments andtherefore identify outstanding parentalgermplasm with extreme expression ofthe trait that would not normally befound in a breeding program. Second, ithas the potential to more effectively culllarge populations so that only the mostelite lines are yield tested. Often this canbe done out of season, making it possibleto advance more than one generation peryear. If the trait is correlated with yield,it may be more effective to select for thetrait itself rather than for yield in earlygenerations, owing to large G×Einteractions for yield. Thus, traitselection has the potential to makebreeding more efficient. Some traits maybe difficult to measure. However, thisoften motivates a breeding program todevise fast and effective ways to selectfor the trait.

In the end success may also be difficultto measure. This will not be the casewhere backcrossing the trait into anadapted background results in bettercultivars. But it will not always be clearif a physiological approach has beensuccessful when it is used in a pedigreebreeding program. A physiologicalapproach will stimulate more detailedthinking and a deeper understanding of

crop growth in relation to the prevailingenvironment and, hence, a deeperappreciation of the underlying factorsinfluencing yield. It will also bring abouta broadening of the germplasm base, thecreation of novel germplasm, and moreefficient ways of manipulating andevaluating populations. As with anempirical breeding program, it requires along term investment.

References

Araus, J.L., Bort, J., Ceccarelli, S., and Grando, S.1997. Relationship between leaf structureand carbon isotope discrimination in fieldgrown barley. Plant Physiol. Biochem.35:553-541.

Blum, A., Poiarkova, H., Golan, G., and Mayer, J.1983. Chemical desiccation of wheat plantsas a simulator of post-anthesis stress. I:Effects on translocation and kernel growth.Field Crops Res. 6:51-58.

Clark, D.H., Johnson, D.A., Kephart, K.D., andJackson, J.A. 1995. Near infraredreflectance spectroscopy estimation of 13Cdiscrimination in forages. J. RangeManagement 48:132-136.

Clarke, J.M., and McCaig, T.N. 1982. Excised-leaf water retention capability as anindicator of drought resistance of Triticumgenotypes. Can. J. Plant Sci. 62:571-578.

Condon, A.G., Richards, R.A., and Farquhar, G.D.1987. Carbon isotope discrimination ispositively correlated with grain yield anddry matter production in field-grown wheat.Crop Sci. 27:996-1001.

Condon, A.G., Richards, R.A., and Farquhar, G.D.1992. The effect of variation in soil wateravailability, vapour pressure deficit andnitrogen nutrition on carbon isotopediscrimination in wheat. Aust. J. Agric. Res.43:935-47.

Cooper, M., and Fox, P.N. 1996. Environmentalcharacterisation based on probe andreference genotypes. In: Plant Adaptationand Crop Improvement. M. Cooper and G.L.Hammer (eds.). CAB International. pp.529-547.

Farquhar, G.D., and Richards, R.A. 1984. Isotopiccomposition of plant carbon correlates withwater-use efficiency of wheat genotypes.Aust. J. Plant Physiol. 11:539-552.

Hall, A.E., Richards, R.A., Condon, A.G., Wright,G.C., and Farquhar, G.D. 1994. Carbonisotope discrimination and plant breeding.In: Plant Breeding Reviews. Janick, J. (ed.).John Wiley & Sons. pp. 81-113.

López-Castañeda, C., Richards, R.A., Farquhar,G.D., and Williamson, R.E. 1996. Seed andseedling characteristics contributing tovariation in seedling vigour amongtemperate cereals. Crop Sci. 36:1257-1266.

Masle, J., Farquhar, G.D., and Wong, S.C. 1992.Transpiration ratio and plant mineral contentare related among genotypes of a range ofspecies. Aust. J. Plant Physiol. 19:709-721.

Passioura, J.B. 1977. Grain yield harvest indexand water use of wheat. J. Aust. Inst. Agric.Sci. 43:117-120.

Rajaram, S., van Ginkel, M., and Fischer, R.A.1995. CIMMYT’s wheat breeding mega-environments (ME). Proceedings of the 8th

International Wheat Genetics Symposium,1993. Beijing, China.

Rawson, H.M. 1996. An inexpensive pocket-sizedinstrument for rapid ranking of wheatgenotypes for leaf resistance. In:Proceedings of the 8th Assembly of theWheat Breeding Society of Australia.Richards, R.A. et al. (eds.). Canberra. pp.127-128.

Rebetzke, G.J., and Richards, R.A. 1999. Geneticimprovement of early vigour in wheat. Aust.J. Agric. Res. 50:291-301.

Rebetzke, G.J., and Richards, R.A. 2000.Gibberellic acid sensitive dwarfing genesreduce plant height to increase kernelnumber and grain yield of wheat. Aust. J.Agric. Res. 51:235-245.

Rebetzke, G.J., Richards, R.A., Fischer, V.M., andMickelson, B.J. 1999. Breeding longcoleoptile, reduced height wheats. Euphytica106:158-168.

Richards, R.A. 1988. A tiller inhibitor gene inwheat and its effect on plant growth. Aust. J.Agric. Res. 39:749-757.

Richards, R.A. 1991. Crop improvement fortemperate Australia: Future opportunities.Field Crops Res. 26:141-169.

Richards, R.A. 1992. The effect of dwarfing genesin spring wheat in dry environments. II.Growth, water use and water use efficiency.Aust. J. Agric. Res. 43:529-539.

Richards, R.A. 1996. Increasing the yieldpotential in wheat: manipulating sources andsinks. In: Increasing yield potential inwheat: Breaking the barriers. Reynolds,M.P., Rajaram, S., and McNab, A. (eds.).Mexico, D.F.: CIMMYT. pp. 134-149.

Richards, R.A., and Passioura, J.B. 1981. Seminalroot morphology and water use of wheat. I.Environmental Effects. Crop Sci.21:249-252.

Richards, R.A., and Lukacs, Z. 2001. Seedlingvigour in wheat - sources of variation forgenetic and agronomic improvement. Aust.J. Agric. Res. 52: (in press).

Wright, G.C., Hubick, K.T., and Farquhar, G.D.1988. Discrimination in carbon isotopes ofleaves correlates with water-use efficiencyof field-grown peanut cultivars. Aust. J.Agric. Res. 15:815-825.


101

Salinity ToleranceK.N. Singh and R. Chatrath1

Natural soil salinity predates humancivilization. When early man, lookingfor better sources of livelihood, movedto arid lands along the riverbanks, heresorted to irrigated agriculture. With thepractice of irrigation began salinity, thefirst man-made environmental problem.The earliest written account of salt landsdates back to 2400 BC and was recordedin the Tigris-Euphrates alluvial plains ofIraq (Russel et al., 1965). The first timesalt lands were associated with irrigationwas in northeastern Sumer, in thevicinity of modern Telloh. Salinity isthought to have been partiallyresponsible for the breakdown of theancient Sumarian civilization (Jacobsonand Adams, 1988).

Systematically recorded research on salt-affected soils is only a century old. In theIndian Subcontinent, the British spreadirrigation and constructed a largenetwork of irrigation canals. Thisinitiated the process of secondarysalinization, and several regions startedreporting salt-related problems. Salinityalso emerged in the Deccan Plateau, withthe commissioning of the Nira IrrigationProject in Maharashtra. Many otherareas of India became waterlogged andsaline during the post-independenceperiod due to the rapid commissioning ofseveral large and medium-size irrigationprojects.

Distribution ofSaline Soils

Salt-affected lands occur in practicallyall climatic regions, from the humidtropics to the polar regions. Saline soilscan be found at different altitudes, frombelow sea level (e.g., around the DeadSea) to mountains rising above 5000meters, such as the Tibetan Plateau orthe Rocky Mountains. The occurrence ofsaline soils is not limited to desertconditions; the problem has beenreported in the tropical belts of Africaand Latin America, and even in the polarregions, particularly Antarctica.

Areas affected by soil salinity are notwell defined, since detailed maps areavailable for only a few. Consequently,global estimates vary widely (Flowers etal., 1986). Of nearly 160 million hectaresof cultivated land under irrigationworldwide, about one-third is alreadyaffected by salt (Figure 1), which makessalinity a major constraint to foodproduction. It is the single largest soiltoxicity problem in tropical Asia(Greenland, 1984). In the wheat growingareas of India, the combination of salt-affected soils and poor qualitygroundwater (Figure 2) severely limitsproductivity.

1 Crop Improvement Division, Central Soil Salinity Research Institute, Karnal, 132 001 (Haryana), India.

Figure 1. Global distribution of salt-affected soils.

C HAPTER 8

Mapa en proceso....

102 K.N. SINGH AND R. CHATRATH

Effect of Salinity /Alkalinity on Plants

Crop species show a spectrum ofresponses to salt, although all have theirgrowth and, eventually, their yieldreduced by salt. Salt effects are thecombined result of the complexinteraction among differentmorphological, physiological, andbiochemical processes.

Morphological effectsMorphological symptoms are indicationsof the injurious effects of salt stress. Theextent of inhibitory or adverse effectscan be known only by making criticalcomparisons with plants growing undercomparable conditions in normal soils.

Salinity may directly or indirectly inhibitcell division and enlargement in theplant’s growing point. Reduced shootgrowth caused by salinity originates ingrowing tissues, not in maturephotosynthetic tissues (Munns et al.,1982). As a result, leaves and stems ofthe affected plants appear stunted.Chloride induces elongation of thepalisade cells, which leads to leavesbecoming succulent. Salt stress hastensphenological development induces earlyflowering in wheat (Maas and Poss,1989; Francois et al., 1986; Bernal andBingham, 1973; Rawson, 1988). It also

Table 1. Saline soil classification system usedby the United States Salinity Laboratory.

EC † < 4 EC > 4mmhos mmhos

cm-1 cm-1

ESP ‡ <15% Non-saline, Saline soil non-sodic soil §

ESP>15% Sodic soil Saline sodic soil† EC = electrical conductivity.‡ ESP = exchangeable sodium percentage.§ The pH of saline soils is generally less than 8.5; of

saline-sodic soils, about 8.5; and of sodic soils, morethan 8.5.

Source: USDA (1954).

Classifying Salt-Affected Soils

High salinity levels can damage soilstructure. The action of Na+ ions, whenthey occupy the cation exchange complexof clay particles, makes the soil morecompact, thereby hampering soil aeration.As a result, plants in saline soils not onlysuffer from high Na levels, but are alsoaffected by some degree of hypoxia. It ispossible to leach out salt deposited in theroot zone through extensive irrigation.However, in heavy textured soils leachingeither through irrigation or rainfall isminimal due to the soil’s poor infiltrationcharacteristics. In alkaline soils suchcrusts are formed when soils dry out andharden; as a result, tender seedlings areunable to emerge. This usually happens ifimmediately after sowing there is a periodof rain followed by a dry spell.

The diagnostic parameters for classifyingsaline soils are electrical conductivity(EC) of the soil solution, which detectsosmotic problems, and exchangeablesodium percentage, indicative of aphysical dispersion problem. Salt-affectedsoils are classified as shown in Table 1.

Figure 2. Distribution of salt-affected soils and poor quality ground water in wheat-growingstates of India.

Gujarat Maharashtra Rajastham Madhya Pradesh

Haryana Punjab Bihar Uttar Pradesh

Some salt is introduced into the soil withevery round of irrigation. Part of the saltis leached below the root zone, but partremains in it. The gradual buildup of saltin previously salt-free topsoil is referredto as secondary salinity. The productivelife of the land is limited as a result ofsecondary salinity. When saltconcentration in the soil forcesproduction below the economicthreshold, cultivation soon becomesimpossible.

The expanded use of saline water forirrigation, together with poormanagement practices, aggravates theproblem (Framji, 1976). The immensepotential of salt-affected soils for muchneeded production of food, fiber, fuel,and forage crops is now more relevantthan ever; production demands areincreasing due to the growingpopulation, and there is scant possibilityof bringing new land under cultivation.This points up the urgent need toincrease productivity of salt-affectedsoils and help innumerable low-incomesmall farmers to improve their lot.

Salt-affected soils (Mha) Poor quality ground water (%)

103

reduces dry matter content, increasesroot : shoot ratio, and diminishes leafsize in wheat. As a result, grain yield isreduced. This is attributed to the reducednumbers of seeds, spikelets, and tillers,as well as low grain weight.

Under sodic conditions growth isdrastically affected due mainly to grossnutritional deficiencies or imbalancesand to a root system severely restrictedby poor soil physical conditions and highalkalinity. Restricted growth is typicallyaccompanied by delayed flowering insensitive varieties.

Physiological effectsSalt stress affects many aspects of plantmetabolism and, as a result, growth isreduced. Excess salt in the soil solutionmay adversely affect plant growth eitherthrough osmotic inhibition (Bernsteinand Hayward, 1958) of water uptake byroots or by specific ion effects. Specificion effects may cause direct toxicity or,alternatively, the insolubility orcompetitive absorption of ions mayaffect the plant’s nutritional balance.These effects may be associated withenzyme activity, hormonal imbalance, ormorphological modifications. It shouldbe noted that the relative role of osmoticand specific ion phenomena inexplaining the observed effects isdisputed.

Even at low salinity levels, external saltconcentration is much greater than thatof nutrient ions, so that a considerableconcentration of ions may reach thexylem. Being the actively transpiringparts of the plant, the leaves accumulatesalt, which leads to their premature death(Munns and Termat, 1986).Photosynthesis is reduced because it isaffected by leaf expansion rate, leaf area,and leaf duration, as well as byphotosynthesis and respiration per unitleaf area. Growth may be indirectlyaffected given that salts decrease theamount of photosynthates, water, and

other growth factors reaching thegrowing region. This decrease may bedue to stomatal closure or the directeffect of salt on the photosyntheticapparatus. Transport of photosynthates inthe phloem may also be inhibited.Although photosynthesis is reduced inwheat, this is thought to be caused moreby inhibition of oxygen release ratherthan by stomatal closure (Passera andAlbuzio, 1997).

Mineral uptake by roots is affected as aresult of imbalance in the availability ofdifferent ions. In wheat the cause ofreduced growth was attributed more tothe reduced rate of transport of essentialnutrients to the shoot (Termaat andMunns, 1986). Although salinity mayupset cation nutrition, imbalances tend tobe restricted when mixed salts arepresent.

Proportions of Ca in the medium that areadequate under non-saline conditionsbecome inadequate under salineconditions. In sodic soils, increases inexchangeable sodium are accompaniedby decreases in exchangeable Ca andMg, leading to Ca and/or Mgdeficiencies: When Ca and Mgconcentrations in the soil solution fallbelow critical levels, K uptake alsodecreases, affecting nutritional balance.

Soil alkalinity severely affects zincsolubility and drastically reduces itsavailability to plants. Iron availabilitymay also be adversely affected. Buildupof organic matter may accentuate zincdeficiency in sodic soils through theformation of a chemical complex that tiesup zinc. Therefore, increased nutrientlevels through fertilizer application mayappear to increase salt tolerance underconditions of salinity-induced nutritionaldeficiency (Bernstein et al., 1974).

Growth inhibition by salt also occurs dueto the diversion of energy from growth tomaintenance (Nieman and Maas, 1978).The latter may include the regulation of

ion concentration in various organs and,within the cell, the synthesis of organicsolutes for osmoregulation or protectionof macromolecules, and for maintenanceof membrane integrity. Osmoregulationensures that adequate turgor ismaintained in the cell. Organiccompounds that accumulate in thecytoplasm may function as osmotica andin protecting the conformation ofmacromolecules in the changing ionicenvironment (Borowitzak, 1981; WynJones and Pollard, 1983).

Since salt damage has a broadphysiological spectrum affecting manymetabolic processes, it is difficult toassess the contribution of individualprocesses to plant death or to the finaldamage done to the plant. One approachfor evaluating the contribution ofindividual processes to salt damage hasbeen to compare crop varieties that showdifferential responses to salt stress,varieties being genetically closer to eachother than are different species. Instudying two wheat cultivars, oneregarded as sensitive and the other asresistant to salinity, it is concluded thatosmotic stress is not the major factordiscriminating between the two lines;rather, susceptibility to specific ions maybe what causes the difference. Short-season varieties have been found to havehigher Na and Cl contents than long-season varieties. This has been suggestedby Bernal et al. (1974) to be the cause ofthe poor performance of short-seasonvarieties.

Biochemical effectsGiven that the physiological approachfor identifying traits that confer stressresistance so far has not been verysuccessful (i.e. there is no singlephysiological trait that is stronglyassociated with salt tolerance) (Yeo etal., 1990), studying the pattern of proteinsynthesis under salt stress may help toidentify a protein(s) associated with

SALINITY TOLERANCE

104

stress. Under saline conditions there is achange in the pattern of gene expression,and both qualitative and quantitativechanges in protein synthesis.

Although it is generally agreed that saltstress brings about quantitative changesin protein synthesis, there is somecontroversy as to whether salinityactivates specialized genes that areinvolved in salt tolerance. In comparinga salt tolerant amphiploid of the breadwheat Chinese Spring and Thinophyrumelongatum, Gulick and Dvorak (1987)did not find that novel mRNAs werepresent in the tolerant amphiploid andabsent in Chinese Spring following saltstress. Therefore, it appears thatvariation for salt tolerance, at leastbetween closely related species orvarieties, may be attributed to allelicvariation at the gene level, which givesrise to quantitative changes in the levelsof expression. Therefore, salt tolerancedoes not appear to be conferred byunique gene(s).

Salinity also changes the levels of planthormones, such as abscisic acid (Moorbyand Besford, 1983) and cytokinin. It hasbeen suggested that salt affects cellularand nuclear volume, inducesendopolyploidy, and induces nucleic acidand protein synthesis (Leopold andWilling, 1984). Several steps involved inprotein synthesis are very sensitive tochanges in the ionic environment andmay result in impairment of proteinmetabolism (Wyn Jones et al., 1979).

Mechanisms of SaltTolerance

A salt tolerance breeding program cannotbe successful in the absence of data onthe physiological mechanisms by whichplants cope with salinity stress. A preciseunderstanding of the mechanism ofexpressed tolerance can help to resolvethe genetic base of the trait to

complement knowledge on theinheritance and dominance pattern of salttolerance.

Mechanisms examined to investigate salttolerance include ion transport andlocalization, osmoregulation and ion-induced metabolic shifts, and effects onphotosynthesis and transpiration. Themechanisms that operate to imparttolerance vary with the level of salt stress.

Control of salt uptakeUnder conditions of low salinity,tolerance in a plant can be due tominimized salt uptake (i.e., exclusion).This is achieved through the followingmechanisms:

• Selectivity of the carriers and ionchannels that are responsible,respectively, for the active and passivetransport of ions across the roots andinto the xylem. This has beenreviewed by Rains (1972), Flowers etal. (1977), and Wyn Jones (1980).

• Better water use efficiency, as thisminimizes the potential salt load perunit of new growth, finally reducingthe salt concentration in the tissue(Flowers et al., 1988).

• Vigorous growth and/or continualreplacement of lost leaves results indilution of salt concentration in theplant (Yeo and Flowers, 1984).

Reducing damage underexcessive ion uptakeDue to the effect of transpiration andxylem ionic concentration, the saltconcentration in leaves rises so fast that itcauses damage. Plants use the followingmechanisms to accommodate the saltload without reducing leaf photosyntheticactivity:

• Localization of saline ions in oldleaves so as to protect the relativelynew and actively transpiring leaves(Yeo and Flowers, 1982).

• Compartmentalization of ions withinthe leaf to avoid water deficit in theleaf (Oertli, 1988).

• Removal of excessive salt from theleaf through specialized structures(Fahn, 1979).

Osmotic adjustmentsUnder high salinity levels, osmoticadjustment to the external waterpotential is required. The plant has toaccumulate or synthesize compoundsthat are osmotically active, for example,through:

• Better nutrient acquisition and highion selectivity.

• Synthesis of organic solutes, such assugars and organic acids, proline,glycinebetaine, sorbitol, etc., to adjustthe osmotic potential of the cytoplasmand vacuole.

• Compartmentalization of salt ions inthe vacuoles (Jeschke, 1984).

Strategies for Breedingfor Salt Tolerance

Breeding for salt tolerance is aformidable task, and slow progress maybe due to a combination of many factors:

• Incomplete knowledge of the effectsof salinity on plants.

• Inadequate means of detecting andmeasuring salinity.

• Ineffective selection methods.• Poor understanding of the interactions

of salinity as it affects the plant.• The vague or nonspecific effects

(other than on plant growth) ofmoderate salt stress.

• Interactions of the ionic and osmoticproperties of salt in plants.

• Changes in salt tolerance with plantdevelopment.

Some prerequisites for improving salttolerance are:

• Availability of suitable geneticvariability in the cultivated species ortheir wild relatives.

• Method of screening large numbers ofgenotypes for salt tolerance.

• A suitable breeding methodology.

K.N. SINGH AND R. CHATRATH

105

Genetic diversity forsalt toleranceGenetic variation for salt tolerance, asdefined by parameters such as survivaland yield, has been reported in manycrop species, including wheat. Variationfor salt tolerance has been found inworld collections of bread wheat(Quershi et al., 1980; Kingsbury andEpstein, 1984; Sayed, 1985), andpotential exists for small improvementsusing conventional breeding methods(Rana, 1986). Singh and Chatrath (1992)screened and found variation for salttolerance in tissue-culture-derived wheatlines developed at CIMMYT. Thechromosomal locations of genescontrolling tolerance to several mineralstresses have been identified, and genescontrolling tolerance are located on allseven homoeologous chromosomegroups of the Triticeae. However, group4 and group 5 chromosomes arepredominant for most stresses(Forster, 1994).

Screening and selecting forsalt toleranceThe physiological effects of salinity arenot fully known, measuring salttolerance is difficult, and little is knownabout genes involved in salt tolerance.Since salinity imposes an environmentalrestraint on plant growth, quantitativeparameters of growth and yieldreductions can be measured based on theprinciples of biometrics and quantitativegenetics. Reductions in yield and growthare the only measure of salinity stress.

These difficulties raise the problem ofhow to measure the development of salttolerance in plants. However, they can beovercome. Thus, breeding for salinitytolerance will provide plant materials toaid plant physiologists understand themechanism of salt tolerance.

Field and microplot studiesThough a variety of methods have beenused for screening for salt tolerance, thereis a need to develop a simple, convenient,and rapid screening technique. A largenumber of genotypes can easily bescreened at the germination stage in asmall space after just four weeks ofgrowth. However, the tolerance of agenotype varies with the stage of growth.So before utilizing the genotypes in abreeding program or recommending themfor sowing in saline soils, they should bere-tested in the field.

It has been reported that genotypesevaluated for salt tolerance in onemedium may show a differential responseto the same salinity levels in other media.Final selection should therefore becarried out under field conditions similarto those for which the genotype will berecommended. Moreover, environmentalconditions in the screening field shouldalso correspond to the growing field;salinity was observed to producedifferential effects under differentenvironmental conditions such astemperature and humidity (Salim, 1989).

Selecting salt tolerant plants in salinefields and microplots (Picture 1) is basic,but there are many problems associatedwith this type of screening. Soil salinityvaries substantially with time, location,and soil depth. At high saltconcentrations, certain ions may havespecific toxic effects on plants, and thephysical structure of soils may bechanged by salt-soil chemicalinteractions. Saline irrigation watershould be composed of a realisticmixture of salts, not just NaCl. Saltimbalance is the most commondeficiency in screening studies andvarietal assessments.

Sodium absorption ratios (SAR) must beconsidered in soils that have substantialclay content. The effects of salt on plantsin soils with high SAR and permeabilityproblems are substantially different fromthose on plants in non-sodic saline soils.

Another problem that may complicateselection is that salt tolerant plants mayactually have higher root zone salinity,since their greater growth and water useconcentrate the salt through exclusionprocesses. Plants that happen to be

Picture 1. Screening for salt tolerant wheat genotypes in microplots under controlled saltstress conditions.

SALINITY TOLERANCE

106

located in slightly less saline areas maythus appear more tolerant, but in realityare not. Screening for salt tolerance canbe improved if selections are based onplant response to soil salinity and watercontent measurements.

More refined control of water applicationand use of saline waters is important.Drip and sprinkler systems are easilymodified and can be applied according tobreeding objectives because waterapplication can be closely controlled andmonitored. High frequency irrigationdecreases the variability in soil moisturecontent and soil water salinity. Use of theTriple Line Source Sprinkler, developedat Servicio de Investigación Agraria,Zaragoza, Spain, was however notsuccessful due to foliar uptake of salt.Their studies later indicated that dripirrigation system was found to be morepractical and successful.

Greenhouse and laboratorymethodsGreenhouse and laboratory techniquesare usually used for screening atgermination or early vegetative stages ofgrowth. The seedling stage is generallythe most sensitive phase of plantdevelopment, and almost all work on salttolerance in different crop speciesreported previously (Kingsbury andEpstein, 1984, Norlyn and Epstein, 1984,Allen et al., 1985, Sayed, 1985, Singhand Rana, 1989, Cramer et al., 1991) hasincluded plant assessment at this stage.

Pot culture and solution culture arecommon greenhouse techniques.Screening under controlled conditions inglasshouse or laboratory can overcomethe problem of environmentalinteractions. Such techniques are popularbecause large numbers of genotypes canbe screened in smaller spaces and lesstime. However, these systems may not beuseful in selecting for traits associatedwith root-soil interactions. A wide variety

of criteria and media are used forscreening, including simple devices suchas germination dishes or more advancedtechniques such as tissue culture andrecombinant DNA.

Germination under high osmotic stresshas been advocated as a method forscreening for salt tolerance (Fryxell,1954). However, it is now clearly knownthat there is no correlation betweentolerance at germination and at latergrowth stages. Also, screening duringgermination could eliminate valuablegenetic sources of tolerance. Thus it isbest to select for salt tolerance duringgermination and emergenceindependently from seedling or latergrowth stages. This is because differentgenes may be involved in variousmechanisms of salt tolerance as the plantdevelops.

The emergence of seedlings from sandcultures has been reported as a usefultechnique for determining emergencepotential under salt stress (Sexton andGerard, 1982). Salinized agar gel,nutrient film technique, and otherscreening techniques have used rootgrowth as the criteria. But root growth isnot a reliable indicator of salt tolerancebecause root growth at low salinities isfar less sensitive to salinity thanvegetative growth. Resistance to leafchlorosis as a selection criteria is basedon the fact that absence of chlorosisgenerally indicates varieties that arebetter salt excluders (Ream and Furr,1976; Shannon, 1978).

Biochemical techniquesScreening based on morphological traitsis easier than using physiologicalmarkers, but salt glands and hairs are theonly morphological markers that havebeen associated with increased salttolerance The use of biochemicalmarkers, such as proline, glycinebetaine,sugar accumulation, and Na : K ratio, hasnot shown a consistent trend and hence is

unreliable. This is probably because salttolerance is the combined result of manyplant features, both morphological andphysiological.

Procedures for Breedingunder Salt Stress

The introduction of crops to new areas hasplayed a major role in varietalimprovement. The majority of traditionalvarieties grown on the problem soils ofSouth and Southeast Asia appear to haveoriginated in India and Thailand. Theimportance of plant introductions hasrecently been recognized, and plantmaterials have been exchanged at bothnational and international levels.International organizations have availablefor different crops germplasm withtolerance to salinity, alkalinity, and acidity.

It is important to screen and evaluate thefull spectrum of genetic variabilityavailable for tolerance to salt stress.Landraces of self-pollinated crops containa heterogeneous mixture of highlyhomogeneous lines. It is thereforenecessary to select within the landrace todevelop a pure breeding line. KharchiaLocal, a highly salt tolerant landrace, isprevalent in the Kharchi area of Rajasthanin India.

When variability is limited or unavailable,it can be generated by making crossesbetween salt tolerant donors and highyielding varieties. Breeding methods usedare the pedigree method, the bulk method,backcross breeding, and recurrentselection. The pedigree method was usedto transfer the salt tolerance of Kharchia65 by hybridizing it with a popular highyielding wheat variety WL 71 I which ledto the development of India’s firstsystematically bred salt tolerant varietyKRL I-4 (Picture 2) at Central SoilSalinity Research Institute, Karnal. Thismethod has also been used successfullyin rice.


107

When variability is restricted, it may alsobe increased by inducing polyploidy. Ithas been observed that allopolyploidcrop species are more tolerant to bothalkaline and saline soil conditions thantheir diploid counterparts. It was inferredthat favorable genome interactions andthe availability of a wide range ofgenetic variation arising fromallopolyploidy enable amphiploids toattain a level of tolerance that diploidparents are incapable of attaining (Ranaet al., 1980).

Soil stress varies widely over both timeand space; as a result, population sizeneeds to be increased. In such cases thepedigree method is less efficient than the“modified bulk method” where onlydesirable genotypes are selected andbulked. Segregating materials are bulkevaluated and selected under stressconditions for several successivegenerations. When the desired level oftolerance is achieved, plant selectionsare made and the pedigree method isused for further selection. Final testing isdone in the target environment.

Backcrossing is also used to transfer oneor two major genes from the donorparent. In wheat phosphorus useefficiency characteristics of two varietieswere transferred to high yielding wheatvarieties. Plants in segregatinggenerations have been selected underlow soil pH and low soil phosphorus.Efforts were also made to transfer genesfor P use efficiency from rye to wheatusing the backcross method(Rosa, 1988).

In crop plants, adaptability andproductivity are usually negativelycorrelated. Recurrent selection can beemployed to break undesirable linkagesand to increase the frequency offavorable combinations, particularlywhen the trait has additive geneticeffects. Mutation breeding can beresorted to if tolerance to salt stress isinadequate; however, mutagenesis hasrarely been applied for developingtolerance to problem soils.

Improving BreedingEfficiency

Wide hybridization in wheatThinopyrum bessarabicum is a perennialspecies within the graminaceous tribeTriticeae and is more salt resistant thanthe annual Triticum aestivum (breadwheat). The amphidiploid produced(Forster and Miller, 1985) by hybridizingthe bread wheat Chinese Spring and Th.bessarabicum was found to be moreresistant in terms of survival and abilityto produce grain at moderate salinity(250 mol ma), than Chinese Spring oreven Kharchia. The greater resistance ofthe amphidiploid was attributed to itsinheritance of more efficient exclusionof Na+ and Cl- from younger leaves andreproductive tissue (Gorham et al.,1986). This technique has yet to be fullyutilized by breeders, given that theamphidiploid expresses much of theThinopyrum genome, which has manyother undesirable qualities. Transferringonly the relevant part of the genomeshould be considered.

It is also known that bread wheat(AABBDD) expresses more K+/Na+selectivity than tetraploid (AABB)wheats, as K+/Na+ was found to beassociated with chromosome 4 of the Dgenome (Gorham et al., 1987). Althoughthere appears to be little allelic variationfor this character in hexaploid wheat,variation may exist in Aegilopssquarrosa, donor of the D genome, andother relatives of wheat carrying thisgenome. Studies using wheat tetrasomiclines (2x=44) and wheat/Agropyronjunceum disomic lines (2x=44) haveshown that chromosomes 2A, 2B, and2D of wheat and 2J of A. junceum carrygenes that confer salt susceptibility.However, chromosome 2J also appearsto carry genes for salt tolerance (Forsteret al., 1988). These findings suggest theexistence of genes with major effectsthat might be exploited to increase salttolerance.Picture 2. Salt tolerant Indian wheat variety KRL 1-4.

SALINITY TOLERANCE

108

Tissue cultureIt may be possible to select cells andprotoplasts that show salt resistance, andregenerate salt-resistant plants fromthem. The main advantages of thisapproach are the large numbers ofindividual cells that can be handled, andthe ease of screening. There is alsopotential for increasing the frequency ofgenetic variation through somaclonalvariation.

Since selection based on tissue culture isat the cellular level, it is important toknow whether salt tolerance observed atthe cellular level is reflected in thewhole plant. It is also a matter of greatinterest to determine whether the salttolerance induced in cells taken fromsalt sensitive species will be transmittedto regenerated whole plants. Nabors etal. (1980) regenerated plants from salttolerant tobacco callus tissue that wasfound to be tolerant; however, they didnot obtain unequivocal evidence forgenetic control.

Quantitative trait lociapproachThe existence of genetic variation fortolerance to salt stress in wheat and itsrelatives is clear, but its complexinheritance hinders both its geneticdissection and the incorporation of therelevant genes into commercialvarieties. However, the recent andcontinuing development of marker-based genetic maps (e.g., RFLPs andsubsequent PCR-based molecularmarkers) of major crop species offers away to overcome these difficulties, asthe maps, in principle, allow thepartitioning of quantitative variation intoeffects associated with definedchromosomal locations; this is known asquantitative trait loci (QTL) analysis.Foolad and Jones (1993) made a fewattempts to apply this methodology tothe salt tolerance response in tomatoes,and Lebreton et al. (1995), to thedrought tolerance response in maize.

Wheat x maize doubledhaploidsDeveloping new salt tolerant wheats viathe doubled haploid system using wheat xmaize crosses saves time and is moreefficient than using conventionalbreeding. Wheat x maize sexualhybridization is, to date, the most efficienttechnique for wheat polyhaploidproduction (Laurie and Bennett, 1988).Since maize is insensitive to the Kr allelesof wheat, embryo recovery frequenciesacross different wheat genotypes are high.Haploid wheat embryos result from theelimination of maize chromosomes in thezygote or during the first three celldivisions. The germinated embryos arelater transferred to pots and treated withcolchicine for chromosome doubling at thefour-tiller stage.

Doubled haploids (KTDH series)developed at John Innes Centre, Norwich,UK, involving Kharchia and TW 161, asodium excluding line, have beenevaluated in natural salt-affected fieldsand under controlled microplot conditionsat CSSRI, Karnal. The best performers,KTDH 54, KTDH 6, and KTDH 7, are talland late maturing under Indian conditions.Other doubled haploid lines developed bycrossing salt tolerant Kharchia, KRL 14,and KRL 3-4 lines with high yieldingIndian wheat varieties (Picture 3) are beingmultiplied for field testing.

Conclusions

New, biological, and genetic techniquesshould be vigorously applied to researchefforts aimed at developing salt tolerantcrop varieties. Genetic and physiologicalapproaches have received a fraction of theattention devoted to research anddevelopment on, for example, landreclamation, drainage, and irrigation withgood quality water. The use of tolerantvarieties is likely to make positivecontributions to solving salinity problems.

A discreet choice of species with geneticpotential for salt tolerance has to bemade, and the available germplasm ofeach species should be collected andsystematically screened for salt tolerancebased on physiological factors. Rapidand reliable screening techniques shouldbe developed considering the selectionprocedure and the specific type ofsalinity stress.

When there is little or no diversity forsalt tolerance in the genepool, exotic andwild genotypes should be explored.Other alternatives that may increasegenetic diversity are somatichybridization, mutation breeding, and,eventually, genetic engineering.

Picture 3. Haploid seedlings being developedusing wheat x maize intergeneric crosses.


109

Inheritance studies of new lines andecotypes isolated for tolerance should beconducted. The inheritance anddominance pattern needs to be identifiedbased on a precise understanding of themechanisms of the expressed tolerance.Studies on the heritability of salttolerance derived from wild relativesbackcrossed to the cultivated species, andthe development of isolines for particularphysiological traits contributing to salttolerance are also important.

Newer techniques, such as selection at thecellular level, somaclonal variation,protoplasmic fusion, and mutationbreeding, may contribute to thedevelopment of salt tolerant varieties.

Selection based on the Na and Kpercentage in plants grown on sodic soilsmay prove more useful than selectionbased on quantitative characters.

Although wheat yields under salt stressconditions cannot be expected to equalthose in a highly productive environment,it seems quite feasible to develop wheatvarieties that will withstand moderateincreases in soil salinity and alkalinitywith little yield reduction. Basic researchinto the mechanisms of salt tolerance andsensitivity based on the physiological,metabolic, biochemical, structural, andultra-structural features that enable salttolerant plants to exist and even thriveunder stress conditions must continue.

Exploitable genotypic differences in plantgrowth and yield under saline andalkaline soils conditions are now wellestablished: Breeding efforts aimed atcombining superior salt tolerance withacceptable yield levels are thus on firmerground than research directed atdeveloping tolerance to other abioticstresses. Progress in developing salttolerant varieties of wheat is just thebeginning of a process that we hope willend in a significant contribution tohuman welfare.

References andSuggested ReadingAllen, S.G., Dobrenz, A.K., Schonhorst, M.H.,

and Stoner, J.E. 1985. Heritability of NaCltolerance in germinating alfalfa seed. Agron.J. 77:99-101.

Bernal, T.C., and Bingham 1973. Salt tolerance ofMexican wheat. I. Effect of No.3 and NaClon mineral nutrition, growth and grainproduction of four wheats. Division S4-Soilfertility and Plant Nutrition. Soil Sci. Soc.Amer. Proc. 37:711-715.

Bernal, C.T., Bingham, F.T., and Oertli, J. 1974.Salt tolerance of Mexican wheat. II. Relationto variable sodium chloride and length ofgrowing season. Proc. Soil Sci. Soc. Am.38:777-780.

Bernstein, L., Francois, L.E., and Clark R.A.1974. Interactive effects of salinity andfertility on yields of grains and vegetables.Agron. J. 66:412.

Bernstein, L., and Hayward, H.E. 1958.Physiology of salt tolerance Annu. Rev.Plant Physiol. 9:25.

Borowitzak, L.J. 1981. Solute accumulation andregulation of cell water activity. In: ThePhysiology and Biochemistry of DroughtResistance in Plants. L.G. Paleg and D.Aspinall (eds.). Academic Press, Sidney. pp97-130.

Cramer, G.R., Epstein, E., and Lauchi, A. 1991.Effect of sodium, potassium and calcium onsalt stressed barley. II. Elemental analysis.Physiol. Plant 81:197-202.

Fahn, A. 1979. Secretory Tissues in Plants.London: Academic Press.

Flowers, T.J., Troke, P.F., and Yeo, A.R. 1977. Themechanism of salt tolerance in halophytes.Ann. Rev. Plant Physiol. 28:89.

Flowers, T.J., Hajibagheri, M.A., and Clipson,N.C.W. 1986. Halophytes. Quarterly Reviewof Biology 61:313-337.

Flowers, T.J., Salama, F.M., and Yeo, A.R. 1988.Water use efficiency in rice (Oryza sativaL.) in relation to resistance to salinity. Plant,Cell and Environment 11:453-459.

Foolad, M.R., and Jones, R.A. 1993. Mappingsalt-tolerance genes in tomato (Lycopersiconesculentum) using trait-based markeranalysis. Theor. Appl. Genet. 87:184-192.

Forster, B.P., and Miller, T.E. 1985. 5b deficienthybrid between Triticum aestivum andAgropyron junceum. Cereal Res. Comm.13:93-95.

Forster, B.P., Gorham, J., and Taeb, M. 1988. Theuse of genetic stocks in understanding andimproving the salt tolerance in wheat. In:Cereals Breeding Related to IntegratedCereal Production. M.L. Joma and L.A.J.Slootmaker (eds.). Wageningen: PUDOC.

Forster, B.P. 1994. Cytogenetic manipulations inthe Triticeae. In: Monographs on Theoreticaland Applied Genetics. Vol. 21. A.R. Yeo andT.J. Flowers (eds.). Springer-Verlag, BerlinHeidelberg.

Framji, K.K. (ed.). 1976. Irrigation and Salinity: AWorld Wide Survey. InternationalCommission on Irrigation and Drainage.New Delhi.

Francois, L.E., Maas, E.V., Donovan, T.J., andYoungs, V.L. 1986. Effect of salinity on grainyield and quality, vegetative growth andgermination of semidwarf durum wheat.Agron. J. 78:1053-1058.

Fryxell, P.A. 1954. A procedure of possible valuein plant breeding. Agron. J. 46:433.

Gorham, J., Forster, B.P., Budrewicz E., WynJones, R.G., Miller, T.E., and Law, C.N.1986. Al tolerance in the Triticeae: soluteaccumulation and distribution in anamphidiploid derived from Triticum aestivumcv. Chinese Spring and Thinopyrumbessarabicum J. Exp. Bot. 37:143549.

Gorham, J., Hardy, C., Wyn-Jones, R.G., Joppa,L., and Law, C.N. 1987. Chromosomallocation of a K/Na discrimination characterin the D genome of wheat. Theor. Appl.Genet. 74:484-488.

Greenland, D.J. 1984. Exploited plants: rice.Biologist 31:291-325.

Gulick, P., and Dvorak, J. 1987. Genetic inductionand repression by salt treatment in roots ofthe salinity sensitive Chinese Spring Wheatand the salinity tolerant Chinese Spring xElytrigia elongata amphiploid. Proc. Natl.Acad. Sci. USA 84:99-104.

Jacobson, T., and Adams, R.M. 1968. Salt and saltin ancient Mesopotamian agriculture.Science 128:1251-1258.

Jeschke, W.D. 1984. K+-Na+ exchange at cellularmembranes, intracellular compartmentationof cations and salt tolerance. In: SalinityTolerance in Plants: Strategies for CropImprovement. Staples, R.C. and Toenniessen,G.H. (eds.). Wiley & Sons, New York.pp 37-66.

Kingsbury, R.W., and Epstein, E. 1984. Selectionfor salt-resistant spring wheat. Crop Sci.24:310-315.

Laurie, D.A., and Bennett, M.D. 1988. Theproduction of haploid wheat plants fromwheat x maize crosses. Theor. Appl. Genet.76:393-97.

Lebreton, C., Laziejancie, V., Steed, A., Pekic, S.,and Quarrie, S.A. 1995. Identification ofQTL for drought responses in maize andtheir testing casual relationships betweentraits. J. Exp. Bot. 46:853-865.

Leopold, A.C., and Willing, R.P. 1984. Evidencefor toxicity effects of salt on membranes. In:Salinity Tolerance in Plants: Strategies forCrop Improvement. Staples, R.C. andToenniessen, G.H. (eds.). Wiley & Sons,New York. pp. 77-91.

SALINITY TOLERANCE

110

Maas, E.V., and Poss, J.A. 1989. Salt sensitivityof wheat at various growth stages. Irrig. Sci.10:2940.

Moorby, J., and Besford, R.T. 1983. Mineralnutrition and growth. In: Encyclopedia ofPlant Physiology. New Series. Vol. 15B.Inorganic Plant Nutrition. Lauchli, A. andBieleskio, R.L. (eds.). Springer-Verlag,Berlin. pp. 481-527.

Munns, R., Greenway, H., Delane, R., and Gibbs,R. 1982. Ion concentration and carbohydratestatus of the elongating leaf tissue ofHordeum vulgare growing at high externalNaCl. II. Causes of the growth reduction. J.Exp. Bot. 33:574-583.

Munns, R., and Termat, A. 1986. Whole plantresponses to salinity. Aust. J. Plant Physiol.13:143-160.

Nabors, M.W., Gibbs, S.E., Bernstein, C.S., andMeis, M.M. 1980. NaCl-tolerant tobaccoplants from cultured cells. Zeitschrift furPflanzenphysiologie 97:13-17.

Nieman, R.H., and Maas, E.V. 1978. The energychange of salt stressed plants. Sixth Intern.Biophysics Cong. Abs. p. 121.

Norlyn, J.D., and Epstein, E. 1984. Variability insalt tolerance of four triticale lines atgermination and emergence. Crop Sci.24:1090-1092.

Oertli, J.J. 1968. Extracellular salt accumulation, apossible mechanism of salt injury in plants.Agrochimica 12:461-469.

Passera, C., and Albuzio, A. 1997. Effect ofsalinity on photosynthesis andphotorespiration of two wheat species(Triticum durum cv. PEPE 2122 andTriticum aestivum cv. Marzotto) Can. J.Bot.56:121-126.

Quershi, R.H., Ahmed, R., llyas, M., and Aslam,Z. 1980. Screening wheat (Triticumaestivum L.) for salt tolerance. Pak. J. Agric.Sci. 17:19-25.

Rana, R.S. 1986. Evaluation and utilization oftraditionally grown cereal cultivars of saltaffected areas of India. Indian J. Genet.46(Suppl I):121-135.

Rains, D.W. 1972. Salt transport by plants inrelation to salinity. Ann. Rev. Plant. Physiol.233-267.

Rana, R.S., Singh, K.N., and Ahuja, P.S. 1980.Chromosomal variation and plant toleranceto sodic and saline soils. In: lnt. Symp SaltAffected Soils. Central Soil SalinityResearch Institute, Kamal, India. pp.487-493.

Ream, G.L., and Furr, J.R. 1976. Salt tolerance ofsome citrus species: Relatives and hybridstested as rootstocks. J. Am. Sot. Hort. Sci.101:265.

Rosa, O. 1988. Current special breeding projectsat the National Research Centre for Wheat(CNPT). EMBRAPA, Brazil. In: Wheatbreeding for acid soils. Review of Brazilian/ClMMYT Collaboration 1976-1986. Kohli,M.M. and Rajaram, S. (eds.). Mexico, D.F.:CIMMYT. pp. 6-12.

Rawson, H.M. 1986. Gas exchange and growth inwheat and barley grown in salt. Aust. J.Plant Physiol. 13:475-489.

Russel, J.C., Kadry, L., and Hanna, A.B. 1965.Sodic soils in Iraq. Agrokomia ES Talajtan.Tom 14(Suppl.):91-97.

Salim, M. 1989a. Effects of salinity and relativehumidity on growth and ionic relations ofplants. New Phyton. 113:13-20.

Sayed, J. 1985. Diversity of salt tolerance in agermplasm collection of wheat (Triticumaestivum). Theor. Appl. Genet. 69:651-657.

Shannon, M.C. 1978. Testing salt tolerancevariability among tall wheatgrass lines.Agron. J. 70:719-722.

Sexton, P.D., and Gerard, C.J. 1982. Emergenceforce of cotton seedlings as influenced bysalinity. Agron. J. 74:699.

Singh, K.N., and Chatrath, R. 1992. Geneticvariability in grain yield and its componentcharacters and their associations under saltstress conditions in tissue culture lines ofbread wheat (Triticum aestivum L. emThell.) Wheat Information Service 75:46-53.

Singh, K.N., and Rana, R.S. 1989. Seedlingemergence rating: A criterion fordifferentiating varietal responses to saltstress in cereals. Agric. Sci. Digest 9:71-73.

Termaat, A., and Munns, R. 1986. Use ofconcentrated micronutrient solution toseparate osmotic from NaCl specific effectson plant growth. Aust. J. Plant Physiol.13:509-522.

USDA. 1954. Diagnosis and Improvement ofSaline and Alkali Soils. United StatesSalinity Laboratory Staff. AgricultureHandbook No. 60, United States Departmentof Agriculture. 160 pp.

Wyn Jones, R.G., Brady, C.J., and Speirs, J. 1979.Ionic and osmotic relations in plant cell. In:Recent Advances in the Biochemistry ofCereals. D.L. Laidamn and R.G. Wyn Jones(eds.). Academic Press, London. pp. 63-103.

Wyn Jones, R.G. 1980. Salt tolerance. In:Physiological Processes Limiting PlantProductivity. C.B. Johnson (ed.).Butterworths, London. p. 271.

Wyn Jones, R.G. and Pollard, A. 1983. Proteins,enzymes and inorganic ions. In:Encyclopedia of Plant Physiology. NewSeries. Vol 15B. Inorganic Plant Nutrition.Lauchli, A. and Bieleskio, R.L. Springer-Verlag, Berlin. pp. 528-562.

Yeo, A.R., and Flowers, T.J. 1982. Accumulationand localization of sodium ions within theshoots of rice varieties differing in salinityresistance. Plant Physiol. 56:343-348.

Yeo, A.R., and Flowers, T.J. 1984. Mechanisms ofsalinity resistance in rice and their role asphysiological criteria in plant breeding. In:Salinity Tolerance in Plants: Strategies forCrop Improvement. Staples, R.C. and G.H.Toenniessen (eds.). Wiley & Sons, NewYork. pp. 151-170.

Yeo, A.R., Yea, M.E., Flowers, S.A., and Flowers,T.J. 1990. Screening of rice (Oryza sativaL.) genotypes for physiological characterscontributing to salinity resistance, and theirrelationship to overall performance. Theor.Appl. Genet. 79:377-384.


111

Cold ToleranceN.N. Sãulescu1 and H.-J. Braun2

Wheat is grown across a wide range ofenvironments and is considered to havethe broadest adaptation of all cereal cropspecies (Briggle and Curtis, 1987). Thisbroad adaptation is due, to a large extent,to wheat’s cold tolerance, i.e. the abilityto withstand temperatures much lowerthan 1-4ºC, considered the minimumtemperature for growth (Figure 1).

In a general sense, cold tolerance inwheat should refer to performance attemperatures lower than the optimum for

growth (about 20ºC), and there aredefinitely differences in the growth rateof cultivars at low temperatures and,consequently, in their adaptation to coolclimate. However, the term “coldtolerance” is most frequently used todescribe a plant’s response to freezingtemperatures, which have more dramaticeffects on the crop. Most often, freezingtemperatures affect autumn-sown wheatduring winter. Freezing tolerance refersto the broader term of “winter hardiness,”an attribute of autumn-sown cereals thatis responsible for differences in “wintersurvival” or “overwintering.”

Winter survival is defined by Blum(1988) as “the final integrated plantresponse to a multitude of stressesinvolved during and after freezing stress,including both external-physical andbiotic stresses.” Even if plants are notwinter-killed, they can be affected byfreezing temperatures that may damagethe leaf, causing reduction in leaf area,delayed growth, and plant debilitation.Considerable variation for winterhardiness exists among cultivars, whichjustifies dedicating extensive efforts tothis breeding objective (Pictures 1 and 2).

Less often, freezing temperatures canoccur during late frosts in spring, causingleaf or spike injury. Unhardened leavescan tolerate -4 to -8ºC (Gusta and Chen,1987), but the reproductive tissue of thedeveloping ear is considerably less

1 Research Institute for Cereals and Industrial Crops, Fundulea, 8264 Calarasi, Romania. Email: [email protected] Winter Wheat Breeding, CIMMYT/Turkey.

Figure 1. General range of favorable andunfavorable (stress) temperatures forwheat.

resistant to freezing and may be injuredat -1.8ºC (Single and Marcellos, 1974).Differences in what is generally called“frost tolerance” are less pronounced,although waxy or hairy lemma, palea,and awns are thought to delay formationof ice in the tissue. Due to the limitedgenetic variation for frost tolerance,breeding efforts have been directedmostly to escaping frost by selecting forlater flowering.

Picture 2. Differential winter damage in plots.

Heat killRespiration accelerates;plants may lose weight

Crowns of most winter-hardy varieties killed

Upper limit for photosynthesis

High temperature seed dormancy

Ideal for growth and development

Cold-hardening induced

Spikelets and flowers freezeRoots freezeLeaves freeze

C HAPTER 9

Picture 1. Differential winter damage inhead rows.

112 N.N. SÃULESCU AND H.-J. BRAUN

It should be noted that low, non-freezingtemperatures (below 10ºC) at the criticalstage of meiosis can also have dramaticeffects on wheat by causing male-sterilityand, consequently, low yields. Geneticdifferences in the response to this stressare known to exist (Qian et al., 1986,Saulescu et al., 1997) but, because of itsrelatively rare occurrence, little effort isdirected towards breeding for toleranceother than selecting for an appropriateflowering time that allows plants toescape the stress.

Stress Factors Involvedin Winterkill

The reasons for winterkill in wheat, aswell as the extent of the damage, varygreatly from region to region and fromyear to year. The main factors causingwinterkill (alone or in combination) arerelated to low temperature per se (such asextreme air or soil temperatures, belowthe critical temperature of a particularwheat cultivar):

• inadequate hardening, due to lateemergence in autumn or a suddendrop in temperature;

• long periods of cold-induceddesiccation (Gusta et al., 1997a);

• prolonged periods of low sub-zerotemperatures; in particular, mid-wintertemperatures below -15ºC result in therapid loss of winter hardiness (Gustaet al., 1997b);

• alternate freezing and thawing, whichcauses increased injury from icecrystal growth with each freeze(Olien, 1969).

Another factor responsible for winterkillis ice encasement, a major cause of plantdeath in areas of high rainfall andfluctuating temperatures during winter

(Andrews et al., 1974). Ice has highthermal conductivity and can aggravatethe effect of low temperatures. It alsohas low gas permeability and may, inextreme cases, smother or suffocateplants by depriving them of oxygen(Poltarev et al., 1992).

Finally, low temperatures or snow cancause indirect damage through:

• frost heaving due to the formation ofice in the soil. The ice pushes theplants upward, breaking and exposingthe roots;

• snow mold, caused by fungi in areaswith long-lasting snow cover. Themost damaging fungus affectingwinter survival is pink snow mold(Microdochium nivale (Fries) Samueland Hallet), previously known asFusarium nivale (Fr) Ces. (Hömmö,1994). Although Microdochiumnivale cannot survive freezing, it istolerant to low temperatures andseverely damages plants in the 0-5ºCtemperature range. Other, lessimportant fungi causing snow moldare Typhula spp., the pathogen forspeckled snow mold or typhulablight, and Sclerotinia borealis,which causes sclerotinia snow mold.

The relative importance of stress factorscausing winterkill can vary greatlyamong regions. In the Ukraine, ananalysis of data from the last 100 yearsshowed that winterkill was caused bylow temperatures in 35% of cases, byalternate freezing and thawing in 26% ofcases, and by ice encasement in 22% ofyears when significant winter damageoccurred (Poltarev et al., 1992).Wisniewski et al. (1997) stated that thecritical factors that affect winter survivalin Poland are low temperature, freeze-induced desiccation, and infection bypathogenic fungi.

Gusta et al. (1997a) reported that themain factors responsible for winterkill inthe Great Plains of North America arelong periods of cold-induced desiccation,poor acclimation conditions in autumn,and unpredictable timing and duration ofextremely cold temperatures, whereas theprimary cause of winterkill in westernCanada is freeze-induced desiccation. Foreastern North America, Olien (1967)found that winterkill is most likely tooccur during low temperature stressfollowing a midwinter thaw, when thecrown tissues have high moisture content.Correct evaluation of the frequency withwhich these or other factors can affectwinter survival in the target area isessential for making a better choice ofparents and testing procedures in abreeding program; this can also improveresource allocation efficiency.

Wheat plants can cope with each of theabove mentioned winter stress factorsthrough different genetic andphysiological mechanisms. For example,a plant’s freezing tolerance and snowmold resistance are based on differentgenetic mechanisms (Hömmö, 1994).However, the basic process behind mostevents leading to winterkill is freezing, orformation of ice in plant tissues (Figure 2).Freezing damage is in general not aconsequence of low temperature per se,but rather the result of cellulardehydration brought about by extra-cellular ice crystallization. Cellularmembranes have been recognized as theprimary sites of freezing injury (Hinchaand Schmitt, 1994).

Freezing tolerance is defined as theability of plants to survive ice formationin extracellular tissues without significantdamage to membranes or other cellcomponents.3 It is the result ofphysiological, chemical, and physical

3 For the sake of clarification, it should be noted that intracellular ice formation is always lethal. The chemical potential in the intracellular solutionmust be equal to the chemical potential of the external solution or the ice. This equilibrium is attained through removal of intracellular water. To avoidcellular dehydration under freezing stress, the osmotic potential of intracelluar solutions is increased as the osmotic potential of extracellular solutiondecreases. For a detailed discussion of the physiological processes during freezing stress, see Blum (1988).

113

reactions, and of changes in plant cellstructure that take place at appropriatedevelopmental stages, under suitableenvironmental conditions. This process iscalled hardening or acclimation.

Acclimation proceeds in two stages,depending on the sequential action ofchilling (>0ºC) and freezing (-3 to -5ºC)temperatures. A decrease of waterpotential in tissues, due to decreasedosmotic potential (because of sugaraccumulation in vacuoles), is the mostimportant feature in the first stage ofplant acclimation. It is correlated with asignificant increase in the abscisic acid(ABA) level and results in modificationof protein synthesis. There are greatdifferences among cereals regarding theabove 0oC temperature at whichacclimation is initiated. Winter rye startsat much warmer temperatures, whichmakes its acclimation period longer thanthat of winter wheat. Spring wheat andspring barley do not initiate acclimationat temperatures above 2oC (Gusta et al.,1997a). Reversible modifications ofmembrane properties, which result infurther decrease of water potential inparenchymatic tissue, seem to play amain role in the frost-dependent stage ofacclimation (Kacperska, 1994).

Low temperatures act as the primaryinducing factor, but stresses other thanlow temperature (water stress, wind,etc.) can also induce a certain level offreezing tolerance. Low temperatureitself or secondary factors (ABA,sucrose fatty acids, and water status),produced in response to the primarysignal, can result in conformationalchanges in either membrane and/orproteins and/or an ABA hormonereceptor, and this, in turn, can result inthe regulation of genes involved in coldacclimation (Gusta et al., 1997a).Several genes have been identified invarious plants as low-temperature-inducible (lti) or cold regulated (COR).These genes may have a cryoprotectiveeffect on cellular membranes(Thomashow, 1993).

The fact that frost-resistant cultivarsharden faster and deharden more slowlythan frost-susceptible genotypes suggeststhat acclimation may have differentthreshold induction temperatures incultivars differing in cold tolerance. Inbarley, a higher degree of frost resistanceis associated with a higher thresholdinduction temperature for theaccumulation of COR proteins (Rizza etal., 1994).

Freezing tolerance is not a staticcondition, for it changes with time,temperature, soil and plant moisture,nutrition, and physiological age andstatus. It depends largely on the coldacclimation or hardening processes.Indeed, differences in freezing toleranceof unhardened plants of differentcultivars are negligible, whileconsiderable differences can be detectedafter full hardening. The hardeningprocess can be stopped, reversed, andrestarted. Generally, under naturalconditions, the dynamics of freezingtolerance are characterized by threestages (Prasil et al. 1994):

• a hardening period, in autumn, whencold tolerance is acquired;

• a period of tolerance maintenance,when the critical or lethal temperaturevaries, depending on temperaturefluctuations in winter;

• a dehardening period, generally at theend of winter, when plants lose theircold tolerance.

Each stage is influenced not only bygenotypical (vernalization requirementand photoperiodic response) ordevelopmental (age of plants) factors,but also by environmental ones. Forexample, if water content is altered byflooding or desiccation, the coldhardiness of winter cereals changesdramatically (Metcalfe et al., 1970).Similarly, prolonged periods of low sub-zero temperatures decrease the freezingtolerance of winter wheat seedlingssignificantly. The most cold tolerant,fully hardened winter wheats can tolerate-15ºC for only around six days, andsurvive at -18ºC for only 24 h and at-23ºC for only 12 h (Gusta et al., 1982).After being exposed to low temperaturesfor a longer period, the seedlings’acquired freezing tolerance is greatlyreduced, and they are killed at muchhigher temperatures than in early winter.

Cultivars with similar freezing tolerancein early winter vary greatly in their

Figure 2. Diagram of the processes involved in winterkill and winter hardiness.

GENOTYPE-Vrn, Ppd -disease resistance genes-Fr genes-COR genes

HARDENING

DEHARDENING

Low temperatures chilling (>0oC) freezing (-3 to -5oC)

Water stress

Thawing - highmoisture

Prolonged periods oflow temperature

FREEZING TOLERANCE

ROOT SYSTEM MOLD RESISTANCE

FROST HEAVING

DESICCATION

FREEZING TEMPERATURES

SNOW MOLD

WINTERKILL

ICE ENCASEMENT SMOTHERING

COLD TOLERANCE

114

ability to cope with long periods of sub-freezing temperatures. The duration andintensity of sub-zero temperature istherefore a main factor determining theloss of freezing tolerance and consequentwinterkill in winter wheat. (Gusta et al.,1997a). Freezing tolerance is also knownto be influenced by plant nutrition(Freyman and Kaldy, 1979), herbicides(Freyman and Hamman, 1979), viralinfection (Paliwal and Andrews, 1979),and seed-borne diseases such as commonbunt (Tilletia foetida and T. caries)(Veisz, 1997). This is why, to detectgenotypic differences in freezing tolerance,it is important to keep all environmentalfactors as uniform as possible.

Traits Associated withFreezing Tolerance inWheat

Freezing tolerance is the result ofcomplex physiological mechanismsinvolving many cell and plant traits.Numerous studies have shown that thegenetic control of cold tolerance iscomplex and can be regarded aspolygenic. As many as 15 out of 21chromosomes in wheat have been foundto influence tolerance to lowtemperatures (Stushnoff et al., 1984).Nevertheless, major genes such as Fr1,closely linked but separable from theVrn1 gene on chromosome 5A, and Fr2,linked but easily separated from Vrn3 onchromosome 5D, were shown to have alarge effect on low temperature response(Snape et al., 1997).

The association between freezingtolerance and vernalization requirementscan therefore be partially explained bythe linkage between major genescontrolling freezing tolerance and twogenes that control growth habit. Inaddition, the vernalization genes havebeen identified as key developmentalfactors responsible for the duration ofexpression of low-temperature-induced

structural genes (Fowler et al., 1996b).Recent data show that the regulatoryinfluence exerted by the vernalizationgenes over low-temperature-inducedstructural gene expression occurs at thetranscriptional level (Fowler et al.,1996a).

Winter wheats initiate hardening athigher temperatures than spring wheats,and the latter harden only to a verylimited extent. Similarly, in spring,completely vernalized winter wheatsonly reharden to the level of springcereals. Obviously, a high level ofhardening can only be achieved in“dormant,” not rapidly developing,plants; therefore, a strong associationexists between the degree ofvernalization and the degree of freezingtolerance that can be achieved in wintercereal seedlings (Roberts, 1990a). On theother hand, several northern Europeanwheat cultivars with very longvernalization requirements only havemoderate freezing tolerance (Gusta et al.,1997a). Braun (1997) found a highlysignificant correlation (r=0.67-0.77)between growth habit and freezingtolerance, which suggests that only 45-60% of the variation for cold hardinesscan be attributed to the differences invernalization requirements.

Freezing tolerance was found to beassociated with prostrate growth type. Agene controlling prostrate growth wasfound to be closely linked with Fr1 andVrn1 on chromosome 5A (Roberts,1990b). Genetic linkage is probably notthe only explanation for the observedassociation, since a prostrate plant is alsoless exposed to low temperatures anddesiccation, and better protectedby snow.

It is worth mentioning that prostrategrowth type, found in dormant juvenileplants in autumn, can also occur incultivars with low vernalizationrequirements but high photoperiodresponse. In wheat such cultivars are

usually only moderately winter hardy,but in barley some of the most coldtolerant cultivars are known to be day-length sensitive, with a low vernalizationrequirement.

Both high vernalization requirement andday-length response, as well as othermechanisms that cause “winterdormancy” and delay development to thegenerative stage (differentiation of themeristem), are generally associated withlateness. In an experiment with winteroat, populations selected for higherlevels of winter hardiness were also latermaturing and often taller than desired(Marshall, 1976). A similar associationhas also been observed in wheat. Thisassociation can create difficulties inbreeding early and short winter hardycultivars. However, no effect of heightalone on winter hardiness was found in acomparison of winter wheat heightisolines derived from the cultivar Yogo(Allen et al., 1986). This suggests thatthe association between winter hardinessand plant height may depend on theheight genes involved and on geneticbackground.

Cell length, as measured in stomatalguard cells, was also found to beassociated with cold hardiness, as wasleaf length and height of hardened plants(Limin and Fowler, 1994; Roberts,1990b). Although the association is notvery strong, these traits have theadvantage of being easily determined onindividual plants.

Hardening induces significant changes inmany of the plant’s biochemical andphysiological characters. Phenotypicdifferences in these characters are oftenassociated with freezing tolerance. Forexample, tissue water content (Liminand Fowler, 1994), accumulation ofsimple sugars or polysaccharides (Olienet al., 1986), free proline accumulationin leaves and shoots (Dörffling et al.,1990), and accumulation of specificcold-regulated proteins (Houde et al.,

N.N. SÃULESCU AND H.-J. BRAUN

115

4 Cereals differ greatly in their ability to survive low temperatures. The most cold tolerant rye cultivars are killed at around –34oC, wheat cultivars ataround –23oC, and barley at around –18oC.

1992) in hardened plants were found tobe correlated with freezing tolerance.

Genetic linkage is the most likelyexplanation for the association betweencertain gliadin blocks, as identified byelectrophoresis, and freezing tolerance(Sasek et al., 1984). Such associationscan be useful only in specific crosses,involving specific parents. Despite thelarge number of correlations with othertraits, none is high enough to justifyreplacing direct freezing tests.

Breeding Approaches

Handling a complex trait such as winterhardiness in a breeding program is adifficult task, due to the large number ofgenes involved and the numerousinteractions with the environment. But themain difficulty in breeding cold tolerantwheat is that high freezing tolerance isgenerally associated with lower yieldsand later maturity.

Many traits that are associated withfreezing tolerance, such as delayed springgrowth or small cells, can have negativeeffects on yield, especially in rainfedenvironments where rapid growth in earlyspring and earliness are important toavoid late drought and high temperatures.Besides, every additional breedingobjective will slow down genetic progressfor all other traits of interest. Therefore,the breeding objective should not be tomaximize winter hardiness, but todevelop cultivars with the minimumwinter hardiness necessary for a giventarget area. As Fowler et al. (1981)pointed out, in general the mostsuccessful winter wheat cultivars haveonly marginally greater winter hardinessthan the minimum required for the area inwhich they are grown.

Definition of the minimum hardinessrequired for a given region is not asimple job. It should be based onassessment of the winterkill risk, basedboth on weather data and informationabout cultivar performance in the area. Acareful analysis of historical weatherdata, including minimum temperaturesand time of their occurrence, is useful,but not sufficient. The same lowtemperature can have rather differenteffects on wheat plants, depending on priortemperature regimes and other factors,which determine the level of hardeningachieved. Some of the crop models, suchas CERES, have a sub-routine thatsimulates hardening and winterkill, andthese could be used for a more correctestimation of the winterkill risk.

A simpler, but very useful approach, is touse winterkill data of cultivars withvarying levels of winter hardiness thathave been cultivated in the target area fora long time and for which long-termrecords for overwintering are available.In general, it is not difficult to identify acultivar grown for many years in aregion with only occasional and not veryserious winterkill. A frequency of 1 outof 10 years with some winterkill cangenerally be considered acceptable. Butthe accepted risk can be higher or lower,depending on economical, social, andother factors. If such a cultivar isidentified, it is the best definition of theminimum level of hardiness required forthe area and should be used as a check inall cold hardiness tests. Additionalchecks should be used to provide a rangeof hardiness.

When wheat cultivars with higher levelsof freezing tolerance are selected tolower the risk of winter damage, itshould be remembered that the resultingyield penalty in years without freezing

stress may be far greater than theadvantage of better winter survival inyears with severe winters.

Obviously, the breeding strategy fordeveloping winter hardiness will dependon the ratio between the hardiness levelin the gene pool used and the minimumnecessary for the target area. If mostparents used in crosses have a winterhardiness equal to or higher than theaccepted minimum, maintaining thislevel is a relatively easy task that can beaccomplished through the application ofmild selection pressure against the rare,less hardy segregants. On the other hand,if a large number of parents are notwinter hardy enough, as is the case inprograms that use spring x wintercrosses, higher selection pressure isadvisable, beginning with the earlygenerations, to increase chances ofrecovering an acceptable level ofhardiness. Early generation selectionagainst spring growth habit, as suggestedby Braun (1997), can be very efficient.

Breeding for winter hardiness is muchmore difficult in areas marginal forwinter wheat cropping where theminimum required hardiness is at orabove the maximum available coldhardiness potential. As Grafius (1981)pointed out, there “has been a notableabsence of improvement in themaximum cold hardiness potential ofcereals in this century”, and this“inability of plant breeders to increasemaximum cold tolerance levels suggeststhat all of the available cold tolerancegenes had been previously concentratedin hardy land races within winter cerealspecies.”4 Recovering this maximumlevel of hardiness in higher yieldinggenotypes is only possible by applyingvery high selection pressure in largesegregating populations.

COLD TOLERANCE

116

Transgressive segregation for freezingtolerance has only been recorded incrosses among medium or less hardyparents, but not among the most hardyparents. There are hopes thatinterspecific hybridization can bring innew genes from species with higherfreezing tolerance (such as rye), but todate no such transfer has been successfulfor common wheat.

Durum wheat generally has much lowerwinter hardiness than bread wheat, sobreeding for freezing tolerance is moredifficult. However, for areas wherewinter durums are superior to springdurums, breeding for winter hardinesshas to be a high priority. The best winterhardiness is found in cultivars derivedfrom interspecific crosses with breadwheat, and such crosses, as well astransgressive segregation in intraspecificcrosses, will probably allow furtherprogress in this respect.

Methods andTechniques

Field testingWhenever winter conditions differentiategenotypes for winter survival, evaluationof winter hardiness in the field isdesirable. Field evaluation allows large-scale, inexpensive characterization ofbreeding materials against the full rangeof factors affecting winter survival,whereas controlled freeze tests measureonly low temperature tolerance. For thisreason, most breeding programs,regardless of available resources, favorfield testing to measure winter survival(Fowler et al., 1993), despite thedisadvantages described below.

Levitt (1972) defined a “test winter,” or“differential winter,” as a winter severeenough to kill the most tender plants and

damage those of intermediate hardinessto various degrees. Unfortunately, from aplant breeder’s point of view, winterswith “good” differentiation amonggenotypes for their winter hardiness areinfrequent, even in areas that require ahigh level of cold tolerance.

It should be stressed that winterkill isoften not only the result of lowtemperature stress, but also of theinteraction of a range of factors, whichmost likely will not all occur in a givenyear or location. Therefore,multilocational testing can give betterinformation on winter hardiness,especially if locations are selected toprovide higher probability of winterkill.

Years with milder winters than a “testwinter” may sometimes exert someselection pressure for winter hardiness,based on leaf damage and color(Picture 3). Although the correlation withactual winter hardiness is not very high,scores based on leaf damage are helpfulfor discarding the less cold tolerant lines,provided notes are taken when symptomsare most visible (a very short period, only2-3 days) in spring, before active growthbegins.

An additional problem of field testingfor winter hardiness is the greatvariability within a field due to non-uniform snow cover, soil preparation,planting depth, soil and plant moisture,etc. To cope with this problem, thefollowing are highly recommended:

• Plant one or two check cultivars ofknown winter hardiness every fewrows.

• Take notes on replicated nurseries,preferably in small plots. Marshall etal. (1981) consider short (0.5-1.5 mlong), replicated, single-row plots asthe most efficient and reliable methodof selecting under field conditions.

• Make every effort to improveuniformity in the field (especially soilpreparation). It has even beensuggested that the top soil becompletely replaced with ahomogeneous mixture, carefullyleveled over a coarse base to provideuniform drainage.

• Use special data-handling proceduresthat allow controlling and reducingenvironmental errors.

Fowler and Gusta (1979) developed afield survival index (FSI) based on therelative winter hardiness of winter wheatcultivars tested in more than 60 trialsover a five-year period. The FSI uses:

Picture 3. Leaf damage and discoloring after a mild winter.


117

Table 1. Approaches used for testing freezing tolerance in wheat.

Hardening Exposure to freezing Assessment

Direct - Natural in the field - Field (regular or - Plants in boxes - Plant survival- In growth chambers special locations) - Plants from field, - Leaf damage- Combined - Field enhanced transplanted (in boxes, - Root regrowth

(ridges, boxes above rootrainers, moist sand) - Cell membraneground, snow removal) - Crowns damage

- Freezing cabinets (in polyethylene (electrolyte leakage)- Immersed in a bags, tubes, sand) - Tissue viability

refrigerated solution - Seedlings - Tissue electricalconductivity

- Fluorescence- Enzyme activity

Indirect - Natural in the field No - Tissue water content- In growth chambers - Free proline- Combined - COR proteins

- Tissue electricalresistance of:

- Seedlings- Crowns

No No - Molecular markers

• only data from plots with partialwinterkill;

• differences in percent winterkillamong entries in a block, rather thanactual percent winterkill in each plot;

• a moving average.

Although calculations and effortsinvolved in determining the FSI mayseem tedious, the index provides a veryrobust measure for comparing the winterhardiness of cultivars. Other approachessuch as a “nearest neighbor analysis”may also be useful.

Enhancing winter stressin the fieldThe probability of differentialwinterkilling in a natural winterenvironment can be increased by usingsimple procedures:

• planting wheat on ridges, 20-30 cmhigh, from which snow is usuallyblown out, leaving plants moreexposed to low temperatures anddesiccation (Nam et al., 1982);

• planting wheat in wooden or cementboxes placed above the ground,preferably in an open field, to allowlower temperatures to be reached atthe crown level, producing higherwinterkilling than in field plantedwheat;

• leaving plots without snow cover bytemporarily covering them duringsnowfalls, or by gently removingnewly fallen snow from the plots.

As in normal field testing, use of severalchecks of known winter hardiness,repeated every few plots or rows, ishighly recommended.

Artificial testingThe irregular occurrence of naturalconditions that satisfactorily differentiategenotypes has led many plant breeders todevelop artificial techniques forassessing the freezing resistance of theirmaterials. Already in 1956, Dexterconcluded that the results of such tests

were generally well correlated with fieldassessments of winter hardiness, andrecent methodological refinements haveimproved the correlation. On the otherhand, Fowler et al. (1981, 1993)concluded that, although controlledenvironments should allow more rigidcontrol of freezing conditions,comparative studies suggested that fieldtrials usually provide more repeatableresults and have lower experimentalerrors. The decision on which method toapply should largely depend on howfrequently field testing results in “good”differentiation and on the equipmentavailable for screening under artificialconditions. Wherever possible, bothmethods should be applied.

Most methods used in wheat breedingprograms are direct, i.e., they are basedon exposing plants or seedlings tocontrolled freezing in artificial climatefacilities, such as freezing cabinets,growth chambers, etc. However, there areindirect methods in which plants are notexposed to freezing; instead, their

freezing tolerance is estimated based onbiochemical changes induced byhardening or on the presence of molecularmarkers associated with genes involvedin controlling winter hardiness (Table 1).

Direct freezing testsMany methods have been suggested andare used in winter wheat breedingprograms around the world for artificialtesting of freezing tolerance. They differin the way plants are sown and preparedfor testing, in the hardening and freezingprocedures, and in the way freezingdamage is assessed.

In many cases, sowing is done in boxes,flats, or pots, which makes it easy tohandle, regardless of weather conditions,but is relatively laborious. The otherchoice is to pick up plants from breedingplots in the field. This approach is moreeconomical, but picking plants from thefield is dependent on weather conditionsand can be hampered by snow cover orfrozen soil.

COLD TOLERANCE

118

There are several options for exposingwheat plants to low temperatures.Often boxes or pots in which wheat hasbeen planted are placed directly infreezing chambers. This has theadvantage of not disturbing the plantsbefore stress exposure, but requireslarger freezing cabinets and a longerexposure period, due to thermal inertiaof the large amount of soil.

Most important for plant survival, thecrown meristem must be able toproduce new roots and tillers, so somemethods expose just the crowns tofreezing temperatures. Crowns areprepared by trimming the upper part ofthe plant to 2-3 cm above the crownand the roots to 0.5-1 cm (Figure 3).Crowns are then put into plastic bags,vials, tubes, moist sand, etc. (Fowler etal. 1981, Gusta et al. 1978). To avoidthe work involved in trimming, othermethods expose plants that have beentransplanted from the field into smallboxes, trays, or some type ofsupporting device (e.g., roottrainers)with a small amount of moist sand orsoil (Poltarev, 1990; O’Connor et al.,1993; Ryabchun et al., 1995). Usingyoung seedlings has the advantage ofreducing test duration and the amountof soil needed, but, as differentialsurvival of seedlings is more difficultto obtain, evaluation is usually basedon leaf damage (Figure 4). Larsson(1986) found a very good correlationbetween seedling leaf damage and fieldwinter hardiness.

Most methods use freezing cabinetswith controlled air temperature.However, to achieve better temperaturecontrol, Jenkins and Roffey (1974)used a refrigerated bath with ethyleneglycol, in which pots with plants wereimmersed. Most authors recommend a

Picture 4. Hardening of plants sown in wooden boxes, using a vegetation house.

Hardening is most easily done undernatural conditions, either by placing theboxes or pots outdoors or by picking uphardened plants from the field (Picture 4).The main disadvantage of this approachis the lack of control over the hardeninglevel. Results from such freezing testsare not reproducible, and the testtemperature must be adjusted for eachtest, according to the hardening level, toproperly differentiate genotypes.

Hardening in growth chambers undercontrolled temperature and light regimescan achieve a controlled level ofhardening, but is expensive and requiresspace in growth chambers for about 30days. A workable compromise is to usenatural field conditions for the first stageof hardening and then transfer the boxesor transplanted plants to growthchambers with a controlled environmentfor the second stage of hardening, whichmay take only 24-130 h. It is importantthat plants be collected before they arecovered by snow.

Freezing response shows significantcultivar x hardening-durationinteractions (Jenkins and Roffey, 1974).Therefore, ideally, the frost resistance ofa genotype should be assessed over arange of hardening and freezing regimes;however, this is not practicable whentesting large numbers of early-generationselections. Several alternative hardeningregimes can be selected, depending onthe breeding requirements:

• natural hardening, which betterreflects the situation in farmers’fields. Many years of testing areneeded to characterize freezingtolerance of a genotype using thismethod;

• an “average hardening” regime,representative of most years in thearea. Various hardening regimes areused by different testing programs(see Table 2);

• striving for maximum level ofhardening, corresponding to“potential freezing tolerance” or“static freezing resistance.”5

5 Ryabchun et al. (1995) recommend adding 36 h at -5ºC, 56 h at -7ºC, 24 h at -9ºC, and 14 h at -10ºC of artificial hardening to the level achieved throughnatural hardening in the field by 14-25 November under conditions in Kharkov, Ukraine.


119

Table 2. Main characteristics of methods used for assessing freezing tolerance in cereals.

Author Planting Hardening Exposure Freezing Recovery Assessment

Jenkins and In paper pots, In growth Pots placed Solution chilled Solution ElectricalRoffey (1974) 1.9 cm chambers at in glass tubes, 2ºC/h to -4.5 heated resistance of

diameter and 8/5 ºC for immersed in At -4.5 ºC to +1 ºC in 7 h leaves, by clipping6.4 cm deep around 30 a refrigerated for 7 1/2 h; 2 platinum

days bath with 40% Decreased electrodes 2 cmethylene glycol 2ºC/h to -9 apart through

At -9 ºC for 11 h lamina of first leafFowler et al. Plants from In the field Crowns above Equilibrated At 0 ºC for 15 h Plant survival(1981) the field (trimmed 3 cm 12 h at -3 ºC Planted in

and 0.5 cm Decreased 2ºC/h soil-perlite-peatbelow the to 5 test temp At 15 ºC, 3crown) placed separated by weeksin aluminum 2 ºC intervalsdishes with Dishes removedmoist sand when test temp.

is reachedLarsson In plastic boxes In growth Seedlings in Decreased 1ºC/h Foliar damage(1986) with mixture of chambers plastic boxes to 5 test temp. on primary leaves

peat and sand at +1ºC, separated byGrown 2 weeks 20-30 days 1.3ºC intervalsin glasshouse

Poltarev Plants from In the field Plants in boxes Decreased Increased Plant survival(1990) the field or trays 2-3 ºC/h to 2 2-3 ºC/h

Transferred in test temp. at 15-16 daysboxes or trays 2 ºC interval at 20-22 ºC

or3 days at24-26 ºC

O’Connor et al. Plants from In the field Plants in Decreased 2ºC/h Thawing at Plant survival(1993) the field rootrainers to 8 test temp. 4 ºC for 15-20 h

transplanted to separated by Recovery atfolding 2 ºC intervals 17 ºC for 3 weeksrootrainers.

Ryabchun et al. Plants from In the field + Plants in boxes Decreased 1ºC/h Increased Plant survival(1995) the field artificial 36h or trays Exposed for 2-3 ºC/h to

Transferred in at -5ºC 56h 24 h at -16, -18, -2ºC, then 1ºC/hwooden boxes at -7ºC 24h -20 and -22 ºC Crowns plantedor special trays at -9ºC 14h in soil at +20ºC

at -10ºC for 15-16 daysFedoulov In wooden In natural Plants in Decreased 1ºC/h Increased 1ºC/h Plant survival(1997) boxes field wooden boxes Exposed for 24 h 24 h at +5 ºC

conditions at -17 to -20 ºC 21 days in+ artificial greenhouse24h at -5ºC

Tischner et al. In wooden Artificial Plants in 24 h at -15ºC In phytotron Plant survival(1997) boxes 7 days at wooden boxes

(38x26x11 cm) +3 to -3ºC4 days at -4ºC

Dencic et al. In pots 20 In the field Plants in pots 24h at -15ºC 120h at +5 Plant survival(1997) cm deep + 24h at 0ºC 96h at -17 ºC to +7ºC + leaf damage

gradual decrease in temperature (by 1-3ºC/h), but direct exposure to the testtemperature can also be used (Dencic,1997; Tischner et al., 1997).

Difficulty in reproducing coldacclimation conditions severely limitsthe resolution of controlled-freeze teststhat employ a single minimum (test)temperature. Therefore, it is best to useof a series of test temperatures, usuallyseparated by 2 ºC intervals, todetermine the LT 50, i.e., the lowest testtemperature at which 50% or more ofthe plants of a wheat genotype survivefreezing (Fowler and Limin, 1997).

As stated above, cold tolerance ofwinter cereals is reduced by prolongedexposure to sub-lethal temperaturesand, consequently, both minimumtemperature and exposure time areimportant variables in controlled-freezetest procedures. For economic reasons,most methods prefer shorter exposuresto lower temperatures, but longerexposures might be advantageous ifthermal inertia is large or if freezingcabinets have limitations in reachinglower temperatures. Thomas et al.(1988) recommended prolonged

Figure 3. Crown prepared for artificialfreezing test by trimming the upper part ofthe plant to 2-3 cm above the crown and theroots to 0.5-1.0 cm.

COLD TOLERANCE

120

EC) x 100. Lethal temperatures aredetermined by fitting data with asigmoidal response curve and usingthe inflection point of the sigmoidalresponse curve to predict the lethaltemperature (Fry et al., 1993);

• electrical resistance of plant tissue(Jenkins and Roffey, 1974);

• chlorophyll fluorescence analysis israpid, sensitive, non-destructive tothe plant, relatively cheap, and ableto detect injury before visiblesymptoms occur. Onset of chillinginjury is accompanied by a decreasein chlorophyll fluorescence in vivo; ifthe chilling treatment is prolonged,chlorophyll fluorescence eventuallydeclines to zero (Wilson andGreaves, 1990). Unfortunately, theneed for costly equipment limits theuse of this type of analysis;

• tissue viability, estimated by usingchemicals (such as tetrazolium oracid fuxine) that change color in thepresence of oxidation reactions. Thisapproach is very time consuming andlabor intensive (Poltarev, 1990);

• enzyme activity (Bolduc et al., 1978).

Figure 4. Grading freezing injuries in seedlings.Source: Larsson (1986).

freezing of dark-hardened seedlings forrating and selecting winter wheats forwinter survival.

There are small variations amongmethods for the recovery procedures.Most authors recommend a gradualincrease in temperature until thawing,followed by a 2-3 week recovery period at15-22 ºC. To reduce this long period whengreenhouse facilities are used, Poltarev(1990) recommends transferring theplants at higher temperatures (24-26 ºC)where survival can be observed afteronly 3 days, provided great care is takento avoid desiccation.

After the recovery period, freezingdamage is usually assessedeither by plant survivalcounts or by visually scoringleaf damage (Pictures 5 and 6).However, such indices are tosome extent subjective, havehigh experimental errors, andinvolve a considerable delaybetween freezing andsurvival assessment. Manytechniques have been

proposed to evaluate the effect offreezing on plants by measuring othertraits, such as:

• electrical conductivity of solutionsresulting from exosmosis afterfreezing. Electrical conductivity,which depends on electrolyte content,is measured using a solution analyzerafter freezing (initial EC) and againthe next day, after killingtissue by immersing testtubes in a 80 ºC water bathfor 1 h (final EC). Percentelectrolyte leakage at eachtemperature is defined as:EL (%) = (initial EC / final

Picture 6. Differential recovery of crownssubmitted to artificial freezing.

Picture 5. Differential damage in plants grown in boxes,after artificial freezing.

4 3 2 1 0


121

These methods for assessing freezingdamage have lower coefficients ofvariation and produce results withoutdelay after freezing, but require specialequipment and more laboratory work.This probably explains why mostbreeding programs still rely ondetermining plant survival.

Table 2 summarizes the maincharacteristics of methods used fordirectly evaluating freezing tolerance inwheat. This could serve as inspiration foradapting a method fitted to availableequipment and conditions.

Cultivars can be compared not only fortheir hardening potential or maximumfreezing tolerance, but also for thestability of their freezing tolerance and theability to reharden (Prasil et al., 1994),which, in some areas, may be equallyimportant. Three approaches have beenused for evaluating these traits:

1. repeating freezing tests several timesduring the winter, based on theassumption that plants are naturallysubjected to variable conditionsleading to dehardening andrehardening;

2. exposing plants to controlled thawingand rehardening before the freezingtest. For example, at the OdessaInstitute, hardened plants are subjectedto thawing at 10-12ºC for 120 h withcontinuous light, rehardened at -2 to -4ºC for 24 h and then frozen at -12ºCfor 24 h (Litvinenko and Musich,1997);

3. estimating the stability of thehardening condition based on the timeneeded to fulfill the vernalizationrequirements. To estimate the timeneeded to complete vernalization,Poltarev et al. (1992) recommendedplanting the genotypes in several boxesin the field and then transferring themto a greenhouse at about 20ºC andcontinuous light, at 47, 55, 62, 68, 74,and 82 days after planting. Growingpoint development is evaluated afterone month in the greenhouse, and thenumber of plants that show spikeletdifferentiation is recorded.

The authors have established a closecorrelation between the length of time tocomplete vernalization and the stabilityof freezing tolerance. Even if thiscorrelation is not general (see Gusta etal., 1997a), it can be useful in selectingfor stability of freezing tolerance,especially when segregants with low andvery low vernalization requirements arecommon in a breeding program.

Indirect freezing testsMany scientists have tried to avoidproblems related to direct freezing ofplants (expensive freezing cabinets, highexperimental error) by suggesting indirectmethods that estimate the level ofhardening instead of freezing damage.

Water content in plants is reduced duringhardening, especially in hardiergenotypes. Water content after hardeningwas found to be correlated with wintersurvival (Fowler et al., 1981).

Proline is thought to play a protectiverole in plants subjected to severalstresses, including frost. Considerableamounts of free proline accumulate inleaves and shoots during cold hardening,and proline accumulation is positivelycorrelated to genotype-specific frosttolerance (Dörffling et al., 1990).Measuring proline content afterhardening can therefore provideinformation about potential freezingtolerance of cultivars.

There is a close correlation between thedegree of freezing tolerance and theaccumulation of a specific cold-regulated(COR) wheat protein (WCS120). Thecorresponding antibody discriminatesbetween frost-resistant and frost-susceptible wheat cultivars (Houde et al.,1992). Therefore this protein can be usedas a molecular marker to select forfreezing tolerance.

Tissue electrical resistance of eight-day-old seedlings was found to be correlatedwith freezing tolerance. Used as a

selection tool, it is very convenient inbreeding schemes that employ artificialclimate for accelerating generations(Musich, 1987; Litvinenko and Musich,1997).

Restricted fragment lengthpolymorphisms (RFLPs) and othermolecular markers may also be used todetect the presence of alleles havingpositive effects on winter hardiness.

Although very attractive, indirectmethods generally describe only some ofthe mechanisms involved in the controlof genotypic differences in freezingtolerance, and therefore can probablyexploit only part of the genetic potentialavailable in a breeding program.Besides, they are generally moreexpensive and therefore restricted tostronger research programs.

Conclusions

Little progress has been made inbreeding for increased tolerance to lowtemperature stress since the introductionof the winter wheat variety Minhardi atthe beginning of this century (Grafius,1981). However, this statement refers tothe absolute minimum temperaturewheat plants can survive. Most of thewinter wheat growing areas in the worlddo not require wheat varieties with sucha high level of winter hardiness.Consequently, the main breedingobjective in many winter wheat breedingprograms is not to lose the winterhardiness level present in commercialcultivars, rather than to increase it.

This target is often reached throughroutine field screening. The costs relatedto screening in controlled environmentsor using other indirect methods areprobably one of the reasons why themeasurement of winter survival in thefield is still the standard procedure formost winter wheat breeding programs.With the identification of genes that

COLD TOLERANCE

122

control frost resistance and thedevelopment of markers, it is likely thatsome of the problems related to fieldtesting and/or controlled environmentscreening will be overcome. However,field testing will remain for some time tocome the final measure of the winterhardiness of a wheat cultivar.

To increase the efficiency of breeding forwinter hardiness in wheat, werecommend:

• identifying priorities among stressfactors involved in winterkill acrossthe target area. Evaluate long-termdata on temperature fluctuation, snowcover, diseases, etc.;

• estimating the minimum freezingtolerance needed in the target area toreduce the risk of significant winterdamage to an acceptable level;

• establishing a set of check cultivars,preferably with a long growinghistory in the region, representing themaximum and minimum winterkillrisk assumed;

• taking every opportunity to select inthe field to reduce the frequency ofwinter-tender lines;

• adapting an artificial freezingprocedure suitable to the availablefacilities and potential. Standardizeplanting, hardening, freezing,recovery, and assessment proceduresto increase reproducibility; and

• creating a database on winterhardiness of potential parents andadvanced lines. Avoid crosses wherenone of the parents has the desiredlevel of hardiness.

References

Allen, S.G., G.A. Taylor, and J.M. Martin. 1986.Agronomic characterization of “Yogo” hardred winter wheat height isolines. AgronomyJournal 78:63-66.

Andrews, C.J., M.K. Pomeroy, and I.A. de laRoche. 1974. Changes in cold hardiness ofoverwintering winter wheat. Can. J. PlantSci. 54:9-15.

Blum, A. 1988. Plant breeding for stressenvironments. CRC Press, Boca Raton,Florida.

Bolduc, R., L. Rancourt, P. Dolbec, and L.Chouinard-Lavoie. 1978. Mesure del’endurcissment au froid et de la viabilitedes plantes exposee au gel par le dosage desphosphatases acides libres. Canadian J. PlantSci. 58:1007-1018.

Braun, H.-J. 1997. Winter hardiness of breadwheats derived from spring x winter crosses.Acta Agronomica Hungarica 45(3):317-327.

Briggle, L.W., and B.C. Curtis. 1987. Wheatworldwide. In Wheat and WheatImprovement. E.G. Heyne (ed.). 2nd edition,ASA, CSSA, SSSA, Madison, WI, USA.pp. 1-32.

Dencic, S., N. Przulj, N. Mladenov, B. Kobiljski,and L. Vapa. 1997. Cold tolerance, earlinessand stem height in wheat genotypes ofdifferent origin. Proc. Int. Symp. CerealAdapt. to Low Temp. Stress, Martonvasar,Hungary, June 2-4, 1997. pp. 216-220.

Dexter, S.T. 1956. The evaluation of crop plantsfor winter hardiness. Advances in Agronomy8:203-239.

Dörffling, K., S. Schulenburg, S. Lesselich, andH. Dörffling. 1990. Abscisic acid andproline levels in cold hardened winter wheatleaves in relation to variety-specificdifferences in freezing resistance. J. Agron.Crop Sci. 165:230-239.

Fedoulov, Y. 1997. Investigation of frostresistance structure in winter cereals. Proc.Int. Symp. Cereal Adapt. to Low Temp.Stress, Martonvasar, Hungary, June 2-4,1997. pp. 183-189.

Fowler, D.B., L.V. Gusta, and N.J. Tyler. 1981.Selection for winterhardiness in wheat. III.Screening methods. Crop. Sci. 21:896-901.

Fowler, D.B., A.E. Limin, A.J. Robertson, andL.V. Gusta. 1993. Breeding for lowtemperature tolerance in field crops. pp.357-362 in: D.R. Buxton, R. Shibles, R.A.Forsberg, B.L. Bad, K.H. Asay, G.M.Paulsen, and R.F. Wilson (eds.).International Crop Science I. CSSA,Madiso, WI, USA.

Fowler, D.B., and A.E. Limin. 1997. Breeding forwinter hardiness in cereals. ActaAgronomica Hungarica 45(3):301-309.

Fowler, D.B., L.P. Chauvin, A.E. Limin, and F.Sarhan. 1996a. The regulatory role ofvernalization in the expression of low-temperature-induced genes in wheat and rye.Theor. Appl. Genet. 93:554-559.

Fowler, D.B., A.E. Limin, S-Y. Wang, and R.W.Ward. 1996b. Relationship between low-temperature tolerance and vernalizationresponse in wheat and rye. Can. J. Plant Sci.76:37-42.

Freyman, -S., and M.S. Kaldy. 1979. Relationshipof soil fertility to cold hardiness of winterwheat crowns. Can. J. Plant Sci. 59:853-855.

Freyman, -S.O., and W.M. Hamman. 1979. Effectof phenoxy herbicides on cold hardiness ofwinter wheat. Can. J. Plant Sci. 59:237-240.

Fry, J.D., N.S. Lang, R.G.P. Clifton, and F.P.Maier. 1993. Freezing tolerance andcarbohydrate content of low-temperatureacclimated and nonacclimated centipedes.Crop Sci. 33:1051-1055.

Grafius, J.E. 1981. Breeding for winter hardiness.In: Analysis and Improvement of Plant ColdHardiness. Olien, C.R., Smith, M.N. (eds.).CRC Press, Boca Raton, FL. pp. 161-174.

Gusta, L.V., M. Boyachek, and D.B. Fowler. 1978.A system for freezing biological materials.Hort Science 13:171-172.

Gusta, L.V., M.J. Burke, and N.J. Tyler. 1982.Factors influencing hardening and survival inwinter wheat. In: P.H. Li, and A. Sakai (ed).Plant cold hardiness and freezing stress. VolII. Academic Press, New York. pp. 23-40.

Gusta, L.V., and T.H.H. Chen. 1987. Thephysiology of water and temperature stress.In Wheat and Wheat Improvement. E.G.Heyne (ed.). 2nd edition, ASA, CSSA,SSSA, Madison, WI, USA. pp. 115-150.

Gusta, L.V., R. Willen, P. Fu, A.J. Robertson, andG.H. Wu. 1997a. Genetic and environmentalcontrol of winter survival of winter cereals.Acta Agronomica Hungarica 45(3):231-240.

Gusta, L.V., B.J. O’Connor, and M.G.MacHutcheon. 1997b. The selection ofsuperior winter-hardy genotypes using aprolonged freeze test. Can. J. Plant Sci.77:15-21.

Hincha, D.K., and J.M. Schmitt. 1994.Cryoprotection of thylakoid membranes. I.Protection by sugars. In: Crop Adaptation toCool Climates. Workshop October 12-14,1994, Hamburg, Germany. pp. 151-165.

Hömmö, L.M. 1994. Winterhardiness of wintercereal species in Finnish conditions, withspecial reference to their frost and snowmould resistance. In: Crop Adaptation toCool Climates. Workshop October 12-14,1994, Hamburg, Germany. pp. 65-73.

Houde, M., R.S. Dhindsa, and F. Sarhan. 1992. Amolecular marker to select for freezingtolerance in Gramineae. Mol. Gen. Genet.243:43-48.

Jenkins, G., and A.P. Roffey. 1974. A method ofestimating the cold hardiness of cereals bymeasuring electrical conductance afterfreezing. J. Agric. Sci. Camb. 83:87-92.

Kacperska, A. 1994. Primary events in acclimationof herbaceous plants to low temperature. In:Crop Adaptation to Cool Climates. WorkshopOctober 12-14, 1994, Hamburg, Germany.pp. 25-36.

Larsson, S. 1986. New screening methods fordrought resistance and cold hardiness incereals. In: Svalöf 1886-1986, Research andresults in plant breeding. G. Olsson (ed.). LTsFörlag, Stockholm, Sweden. pp. 241-251.

Levitt, J. 1972. Responses of plants toenvironmental stresses. Academic Press,New York. pp. 697.


123

Limin, A.E., and D.B. Fowler. 1994. Relationshipbetween guard cell length and cold hardinessin wheat. Can. J. Plant Sci. 74:59-62.

Litvinenko, N.A., and V.N. Musich. 1997.Employment of artificial climate in thecourse of development of frost-hardy winterbread wheat cultivars. Proc. Int. Symp.Cereal Adapt. to Low Temp. Stress,Martonvasar, Hungary, June 2-4, 1997.pp. 177-182.

Marshall, H.G. 1965. A technique of freezingplant crowns to determine the coldresistance of winter oats. Crop Sci. 5:83-85.

Marshall, H.G. 1976. Genetic changes in oat bulkpopulations under winter survival stress.Crop Sci. 16:9-11.

Marshall, H.G., C.R. Olien, and E.H. Everson.1981. Techniques for selection of coldhardiness in cereals. In: Plant Breeding - Acontemporary basis. Vose, P.B. (ed.).Pergamon Press, Oxford. pp. 139-159.

Metcalfe, E.L., E.L. Cress, C.R. Olien, and E.H.Everson. 1970. Relationship between crownmoisture content and killing temperature forthree wheat and three barley cultivars. CropSci. 10:362-365.

Musich, V.N. 1987. Pokazatel electrosoprotivleniaprorostkov v otbore ozimoj pshenitsy namorozostojkost. J. Genetika, Phiziologija iSelektsija Zernovyh Kultur, Nauka, Moskva.pp. 58-65.

Musich, V.N., and B.M. Kornelli. 1982. Anevaluation of frost hardiness in winter wheatvarieties according to the electric resistanceof radicles. Soviet Agriculture Sciences5:27-29.

Nam, J.H., B.H. Hong, and B.K. Kim. 1982.Evaluation of winter hardiness in fieldgrown winter wheat. Korean J. Breeding14:314-318.

O’Connor, B.J., M.J.T. Reaney, and L.V. Gusta.1993. A practical method of assessing thefreezing tolerance of large populations offield grown winter cereals. Can. J. Plant Sci.73:149-153.

Olien, C.R. 1967. Freezing stresses and survival.Ann. Rev. Plant Physiol. 18:387-392.

Olien, C.R. 1969. Freezing stresses in barley. In:Barley Genetics II. Proceedings of SecondInt. Barley Genetics Symp. R.A. Nilan (ed.).Pullman, WA: WSU. pp. 356-363.

Olien, C.R., M.N. Smith, and P.K. Kindel. 1986.Production of freeze inhibitorpolysaccharides of rye and barley in tissueculture. Crop Sci. 26:189-191.

Paliwal, Y.C., and C.J. Andrews. 1979. Effects ofyellow dwarf and wheat spindle streakmosaic viruses on cold hardiness of cereals.Can. J. Plant Pathology 1:71-75.

Poltarev, E.M. 1990. Razrabotka metodovdiagnostiki zimostojkosti ozimyh kultur(Metodicheskje rekomendatsij)Vsesoiuznaja Akademija S.-h. nauk,Harkov. 63 p.

Poltarev, E.M., L.P. Borisenko, and N.I.Ryabchun. 1992. Diagnosis of winter wheatresistance to thawing and ice encasement aspart of the complex evaluation ofwinterhardiness (methodologicalrecommendations). (In Russian.) UkrainianAcademy of Agronomy Sciences,Harkov. 33 pp.

Prasil, I., P. Prasilova, K. Papazisis, and J. Valter.1994. Evaluation of freezing injury anddynamics of freezing resistance in cereals.In: Crop Adaptation to Cool Climates.Workshop October 12-14, 1994, Hamburg,Germany. pp. 37-48.

Qian, C.M., Aili Xu, and G.H. Liang. 1986.Effects of low temperature and genotypes onpollen development in wheat. Crop Sci.26:43-46.

Rizza, F., C. Crosatti, A.M. Stanca, and L.Cattivelli. 1994. Studies for assessing theinfluence of hardening on cold tolerance ofbarley genotypes. Euphytica 75:131-138.

Roberts, D.W.A. 1990a. Duration of hardeningand cold hardiness in winter wheat. Can. J.Bot. 57:1511-1517.

Roberts, D.W.A. 1990b. Identification of loci onchromosome 5A of wheat involved incontrol of cold hardiness, vernalization, leaflength, rosette growth habit, and height ofhardened plants. Genome 33:247-259.

Ryabchun, N.I., L.R. Borisenko, E.M. Poltarev,V.N. Ivanova, and B.I. Dolgopolova. 1995.Opredelenie staticheskoj morozostoikostiozimyh zernovyh kultur (metodicheskierekomendatsij). Ukrainskaya AkademiaAgrarnyh nauk, Harkov. 9 p.

Sasek, A., J. Cerny, and A. Hanisova. 1984.Gliadinove bloky - markerymrazuvzdornosti u psenice obecne (Gliadinblocks - markers of frost hardiness incommon wheat). Genetika a Slechteni -UVTIZ 20:199-206.

Sãulescu, N.N., G. Ittu, M. Balota, M. Ittu, and P.Mustatea. 1997. Breeding wheat for lodgingresistance, earliness and tolerance to abioticstresses. In: Wheat: Prospects for GlobalImprovement. H.-J. Braun et al. (eds.).Kluwer Academic Publishers, Netherlands.

Single, W.V., and H. Marcellos. 1974. Studies onfrost injury to wheat. IV. Freezing of earsafter emergence from the leaf sheath. Aust.J. Agric. Res. 25:679.

Snape, J.W., A. Semokhoskii, L. Fish, R.N.Sarma, S.A. Quarrie, G. Galiba, and J.Sutka. 1997. Mapping frost tolerance loci inwheat and comparative mapping with othercereals. Acta Agronomica Hungarica45(3):265-270.

Stushnoff, C., D.B. Fowler, and A. Brule-Babel.1984. Breeding and selection for resistanceto low temperature. In: Plant Breeding - Acontemporary basis. Vose, P.B. (ed.).Pergamon Press, Oxford. pp. 115-136.

Tischner, T., B. Koszegi, and O. Veisz. 1997.Climatic programmes used in theMartonvasar phytotron most frequently inrecent years. Acta Agron. Hung. 45:85-104.

Thomas, J.B., G.B. Schaalje, and D.W.A. Roberts.1988. Prolonged freezing of dark-hardenedseedlings for rating and selection of winterwheats for winter survival ability. Can. J.Plant Sci. 68:47-55.

Thomashow, M.F. 1993. Genes induced duringcold acclimation in higher plants. In:Advances in low-temperature biology. P.L.Steponkus (ed.). JAI Press, London.pp. 183-210.

Warnes, D.D., and V.A. Johnson. 1972. Crown-freezing and natural survival comparisons inwinter wheat. Agron. J. 64:285-288.

Veisz, O. 1997. Effect of abiotic and bioticenvironmental factors on the frost toleranceof winter cereals. Acta AgronomicaHungarica 45(3):247-256.

Wilson, J.M., and J.A. Greaves. 1990. Assessmentof chilling sensitivity by chlorophyllfluorescence analysis. In: Chilling injury ofhorticultural crops. C.Y. Wang (ed.). CRCPress, London. pp. 130-139.

Wisniewski, K., B. Zagdanska, and M. Pronczuk.1997. Interrelationship between frosttolerance, drought and resistance to snowmould (Microdochium nivale). ActaAgronomica Hungarica 45(3):311-316.

COLD TOLERANCE

124

Summer Crop Winter Crop Environment

Rice Wheat Asia

Maize Wheat Africa KenyaTanzania

Asia South AsiaChinaThailand

Soybean Wheat Latin AmericaCentral America

Cotton Wheat Africa EgyptSudan

Asia PakistanIndia

Heat ToleranceM.P. Reynolds,1 S. Nagarajan,2 M.A. Razzaque,3 and O.A.A. Ageeb4

Temp oC30

25

20

15

10Nov. Dec. Jan. Feb. Mar. Apr.

Figure 2. Typical average temperaturesduring the wheat cropping cycle for threetypes of hot environments (Wad Medani,Sudan; Dharwad, India; Dinajpur,Bangladesh) and one temperateenvironment (Ciudad Obregon, Mexico).

Sudan India

Obregon

Bangladesh

Figure 1. Wheat in tropical cropping systems.

C HAPTER 1 0

1 CIMMYT Wheat Program, Apdo. Postal 6-641, Mexico, D.F., Mexico, 06600.2 Indian Council of Agricultural Research, P.O. Box 158, Kunjpura Road, Karnal, Haryana 132001, India.3 Bangladesh Agricultural Research Council, Fermgate, Dhaka 1215, Bangladesh.4 Sudan Agricultural Research Corporation, P.O. Box 126, Wad Medani, Sudan.

Wheat is the most widely grown cerealin temperate environments, and is alsocultivated in many tropical croppingsystems, where it is often grown as thewinter season crop in rotation with anumber of other crops—for example,with maize in Africa, rice in Asia, andsoybean in Latin America (Figure 1).Among the numerous advantages ofcultivating wheat in this niche are that itis stress tolerant, relatively highyielding, and easy to transport andstore.

There are also disadvantages linked togrowing wheat in tropical areas;foremost among them are the differenttypes of high temperature stress that

affect the crop. Perhaps the greatestchallenge to understanding thephysiological problems associated withheat stress is to encompass the diversityof hot environments all over the world(Figure 2). Continual heat stress affectsapproximately 7 million ha of wheat indeveloping countries, while terminalheat stress is a problem in 40% oftemperate environments, which cover36 million ha. Continual heat stress isdefined by a mean daily temperature ofover 17.5°C in the coolest month of theseason (Fischer and Byerlee, 1991), andover 50 countries (importing more than20 million tons of wheat per year)experience this type of stress throughoutthe wheat cycle.

When consulted, representatives ofnational agricultural research systems(NARSs) from the major wheat-growingregions in the developing worldidentified heat stress as one of their topresearch priorities (CIMMYT, 1995).

CIMMYT / NARSCollaboration on HeatTolerance

Breeding efforts by a number of nationalwheat breeding programs has resulted inthe release of germplasm adapted towarm growing environments, such as inEgypt and Sudan (AbdElShafi andAgeeb, 1994), India (Tandon, 1994),

125

The information emanating from theIHSGE may be useful in establishingindirect selection criteria for heattolerance. The application of some ofthese traits to breeding will bediscussed in subsequent sections, butfirst we will present a brief review ofsome of the physiological traitsassociated with heat stress.

Physiological TraitsAssociated with HeatTolerance

Genetic diversity for heat tolerance incultivated wheat is well established(Midmore et al., 1984; Rawson, 1986;Wardlaw et al., 1989; Al-Khatib andPaulsen, 1990; Reynolds et al., 1994).Photo-assimilation is more likely to beyield-limiting under heat stress than intemperate environments, especially asstress typically intensifies duringgrainfilling, when demand for assimilatesis greatest. This is borne out by theobservation that under stress, total above-ground biomass typically shows astronger association with yield than withpartitioning, i.e., harvest index (Table 1);the situation is usually reversed undertemperate conditions.

Hence traits affecting radiation useefficiency (such as early ground cover,stay-green, and photosynthetic rate) couldbe expected to be important under heatstress. Although early ground coverseems to be important in an agronomiccontext (Rawson, 1988; Badaruddin etal., 1999), variation in this trait amonggenotypes does not seem to be associatedwith heat tolerance (Table 1). The stay-green trait has been used widely inbreeding for heat tolerance, partly as anindicator of disease resistance (Kohli etal., 1991). Physiological evidenceindicates that loss of chlorophyll duringgrainfilling is associated with reducedyield in the field (Reynolds et al., 1994).Studies in controlled environments haverevealed genetic variability inphotosynthetic rate among wheatcultivars when exposed to hightemperatures (Wardlaw et al., 1980;Blum, 1986).

Such differences in photosynthesis underheat stress have been shown to beassociated with a loss of chlorophyll anda change in the a:b chlorophyll ratio dueto premature leaf senescence (Al-Khatiband Paulsen, 1984; Harding et al., 1990).Studies at CIMMYT comparing 16diverse semi-dwarf wheats demonstratedgenetic variability for photosynthetic rateunder heat-stressed field conditions

Table 2. Genetic correlations (Rg) forphysiological parameters measured inTlaltizapan, Mexico, and wheat yields for 10varieties averaged over 16 low relativehumidity environments, IHSGE 1990-94.

Physiological trait R(g)

Canopy temperature depression 0.86**Membrane thermostability 0.81**Leaf chlorophyll (grainfilling) 0.72**Leaf conductance (heading) 0.63*Photosynthesis (heading) 0.63*

* Denotes significance at ≤ 0.05, ** significance at ≤ 0.01.

Bangladesh (Razzaque et al., 1994), andUruguay (Pedretti and Kohli, 1991).CIMMYT has been actively involved inmany of these regions (Kohli et al.,1991; Ortiz-Ferrara et al., 1994).Collaboration between CIMMYT andNARS on physiological aspects of heattolerance in wheat started in 1990, withthe establishment of a network involvingwheat scientists in Bangladesh, Brazil,Egypt, India, Nigeria, Sudan, andThailand.

Collaborative experiments conducted bynetwork scientists were named theInternational Heat Stress GenotypeExperiment (IHSGE) (Reynolds et al.,1992; 1994; 1997; 1998; Reynolds,1994). The IHSGE was grown in wheat-growing areas classified by CIMMYT asheat stressed, i.e., CIMMYT mega-environment 5 (ME5). The mainobjectives of the IHSGE were toestablish the degree of genotype byenvironment interaction (GxE) in ME5,evaluate potential physiologicalscreening techniques by observinggenetic diversity for traits and theirassociation with heat tolerance, andimprove our understanding of thephysiological and genetic basis of heattolerance.

There were three main outcomes of thestudy. First, cluster analyses of over 40hot sitex year combinations indicatedthat most interaction between sites andgenotypes was accounted for by relativehumidity (RH). Hence low RH sites(e.g., Sudan, Mexico, and India) andhigh RH sites (e.g., Bangladesh andBrazil) showed less Gx E within RHgroups than when comparison was doneacross RH groups (Reynolds et al., 1998;Vargas et al., 1998). This kind ofanalysis indicates that breeding for thesetwo broad environments should beundertaken as separate objectives.Second, data collected on IHSGE linesin the low RH sites showed consistentassociation between yield and a numberof morphological traits (Table 1). Third,physiological data collected in Mexicoshowed that several parameters wereassociated with performance atinternational low RH sites (Table 2).

Table 1. Genetic correlations betweenmorphological traits and wheat yields for 10varieties averaged over 16 low relative humidityenvironments in ME5, IHSGE 1990-94.

Trait Genetic correlation

Final biomass (above ground) 0.88**Grains/m2 0.77**Grains/spike 0.67*Harvest index 0.51Kernel weight -0.10Spikes/m2 0.0

Days to anthesis 0.83**Days to maturity 0.81**Plant height 0.20

% ground cover (anthesis) 0.67*Biomass at anthesis 0.35Plant dry weight (5-leaf stage) -0.45% ground cover (5-leaf stage) -0.30Plants/m2 -0.15


HEAT TOLERANCE

126 M.P. REYNOLDS ET AL.

(Reynolds et al., 2000). In addition,both canopy temperature depression(CTD) and flag-leaf stomatalconductance, as well as photosyntheticrate, were highly correlated with fieldperformance at a number ofinternational locations (Reynolds et al.,1994). Besides being a function ofstomatal conductance (Amani et al.,1996), CTD itself is a mechanism ofheat escape, as suggested, for example,by Cornish et al. (1991) in cotton.

Conductometric measurement of soluteleakage from cells was used in severalstudies to estimate heat damage to theplasma membranes. Genetic variation inmembrane thermostability (MT) hasbeen inferred using conductometricmeasurements in various field-growncrops, including spring wheat (Blumand Ebercon, 1981). Shanahan et al.(1990) obtained a significant increase inyield of spring wheat in hot locations byselection of membrane-thermostablelines, as determined by measurementson flag leaves at anthesis. By applyingthe MT test on winter wheat seedlings,Saadalla et al. (1990) found a highcorrelation in MT between seedlingsand flag leaves at anthesis for genotypesgrown under controlled environmentalconditions. Measurements of MT of 16spring wheat cultivars were comparedinternationally with performance atseveral heat-stressed locations.Variation in MT of both field-acclimated flag leaves and seedlingsgrown in controlled conditions wasassociated with heat tolerance in warmwheat-growing regions (Reynolds et al.,1994). Other studies have confirmedgenetic variation of these materials andindicated high heritability for the trait(Fokar et al., 1998).

Although the physiological basis for theassociation of MT with heat tolerancehas not been determined, plasmamembranes are known to be more heattolerant than the photosyntheticthylakoid membranes, for example(Berry and Bjorkman, 1980). While lossof membrane integrity may be the causeof ion leakage from the cell, thisphenomenon could also be caused by

thermally induced inhibition ofmembrane-bound enzymes responsiblefor maintaining chemical gradients in thecell. Direct evidence for a biochemicallimitation to heat tolerance in wheatcomes from studies of the enzymesinvolved in grainfilling, specificallysoluble starch synthase, which isdeactivated at high temperatures(Keeling et al., 1994). If conversion ofsucrose to starch is a limitation to yieldunder heat stress, this would explain theincreased levels of carbohydrates invegetative tissue of wheat observedwhen grainfilling was limited by heatstress (Spiertz, 1978).

Several other processes are clearlyaffected by high temperatures, but notdiscussed in depth here, since eithergenetic variation has not been shown tobe associated with performance, or theydo not lend themselves to simplescreening. There is some evidence thatmeiosis is adversely affected at hightemperatures (Saini et al., 1983).Respiration costs are higher astemperature increases, leadingeventually to carbon starvation becauseassimilation cannot keep pace withrespiratory losses (Levitt, 1980).However, this apparently wastefulprocess would seem unavoidable, at leastin current germplasm, as evidenced bypositive associations observed betweendark respiration at high temperatures andheat tolerance of sorghum lines (Gerikand Eastin, 1985). On the other hand,high rates of dark respiration in grainsmay be severely detrimental to yield(Wardlaw et al., 1980).

Heat shock proteins are synthesized atvery high rates under high temperaturestress and are thought to have aprotective role under stress; nonetheless,their role in determining geneticdifferences in heat tolerance has notbeen established. Chlorophyllfluorescence may be more promising asa screening trait, given that associationsbetween heat tolerance and lowerfluorescence signals have been reportedin a number of crops, including wheat(Moffat et al., 1990). Although screeningprotocols are yet to be thoroughly

evaluated, preliminary evidence withCIMMYT lines has indicated thatfluorescence parameters may lendthemselves to screening for heattolerance (Balota et al., 1996).

While there is still no definitive pictureof the physiological basis of reducedgrowth rates under heat stress, manydrought-adaptive traits may be usefulunder heat stress. Examples wouldinclude leaf glaucousness to reduce theheat load, awn photosynthesis whenhigh temperatures reduce assimilationrate of the leaves, and early escape fromheat stress. Heat stress is almostcertainly a component of drought stress,since one of the principal effects ofdrought is to reduce evaporative coolingfrom the plant surface. Nonetheless, notall traits conferring heat tolerance arealso associated with genetic variabilityfor drought tolerance, a good examplebeing membrane thermostability (Blum,1988). In addition, wheat germplasmthat typically performs well under heatstress is not necessarily useful underdrought (S. Rajaram, pers. comm.).

PhysiologicalApproaches toBreeding for HeatTolerance

Different physiological mechanismsmay contribute to heat tolerance in thefield—for example, heat tolerantmetabolism as indicated by higherphotosynthetic rates, stay-green, andmembrane thermostability, or heatavoidance as indicated by canopytemperature depression. Breedingprograms may measure such traits toassist in the selection of heat tolerantparents, segregating generations, oradvanced lines (Figure 3). Based onfield data collected in the IHSGE, anumber of physiological traits that hadbeen presented in the literature wereevaluated as potential selection criteria(Table 3). While useful as indicators,these conclusions are by no meansdefinitive for two important reasons.

127

For one thing, the results cannot beextrapolated with any certainty toenvironments outside the test site. Also,the data were for the most partmeasured on unrelated fixed lines andas such do not necessarily imply thatselection for these traits would result ingenetic gains in yield among theprogeny of a cross. To establish thepotential genetic gains associated withindirect selection criteria, similarexperiments need to be conducted withrandomly derived sister lines using anumber of relevant crosses andheritability of traits established, asoutlined in the introduction of this book.

Evidence for applying three traits(namely, canopy temperaturedepression, leaf conductance, andmembrane thermostability) in selectingfor heat tolerance is presented in thefollowing sections; sufficient evidencehas been collected on these traits tosuggest their potential as breeding tools.Nonetheless, if these techniques havenot been evaluated in a given breedingenvironment, they should first beevaluated, as outlined in chapter one,before being applied to mainstreambreeding operations.

Canopy temperaturedepressionAs discussed earlier, experimental datahave shown a clear association of CTDwith yield in both warm and temperateenvironments. CTD shows high geneticcorrelation with yield and high valuesof proportion of direct response toselection (Reynolds et al., 1998),indicating that the trait is heritable andtherefore amenable to early generationselection. Since an integrated CTDvalue can be measured almostinstantaneously on scores of plants in asmall breeding plot (thus reducing errornormally associated with traitsmeasured on individual plants), workhas been conducted to evaluate itspotential as an indirect selectioncriterion for genetic gains in yield. CTDis affected by many physiologicalfactors, which makes it a powerful

Table 3. Summary of heat stress mechanisms previously reported for wheat and theirassociation with yield in the IHSGE.

Reported heat stress mechanism Accounting for genetic variation in yield in IHSGE

Accelerated development Yes, lateness associated with higher yield in many(Midmore et al., 1984) environments

Stand establishment (Rawson, 1988) No, poor correlation with early growth

Evaporative cooling (Idso et al., 1984) Yes, strong correlation of CTD with yield

Inhibition of meiosis No, sterility not observed. Grain:spikelet ratio not(Saini et al., 1983; Zeng et al., 1985) correlated with yield

Sensitive growth phase Partial least squares analysis confirmed spike growth(Fischer, 1985; sensitivity, especially to high night temperatures (VargasShpiler and Blum, 1991) et al., 1988)

Photosynthesis/chlorosis Yes, high association of photosynthesis and stay-green(Al-Khatib and Paulsen, 1990; with yield in field plotsShpiler and Blum, 1991)

Thylakoid thermostability Preliminary data on IHSGE lines confirms association of(Moffatt et al., 1990) chlorophyll fluorescence with yield (Balota et al., 1996)

Membrane thermostability (MT) Yes, MT measured on seedlings and flag leaves associated(Shanahan et al., 1990) with yield at several sites

Inhibition of starch synthase No clear evidence, but yield not associated with TGW(Bhullar and Jenner, 1986; Rijven, 1986)

BATT CHG O-1V

OUT

EMISSIVITY 2:3

Figure 3. Potential use of canopy temperature depression in a breeding program.

HEAT TOLERANCE

128

integrative trait, but its use may belimited by its sensitivity toenvironmental factors (Figure 4).

Factors affecting expression of CTD.Leaf temperatures are depressed belowair temperature when water evaporatesfrom their surface. One of the factorsdetermining evapotranspiration isstomatal conductance, which itself isregulated by the rate of carbon fixation.To maintain high rates of photosynthesis,a genotype must have an effectivevascular system for transpiration ofwater, as well as for transport ofnutrients and assimilates. Since CTD isdirectly or indirectly affected by anumber of physiological processes, it isa good indicator of a genotype’s fitnessin a given environment. CTD also seemsto be affected by the ability of agenotype to partition assimilates to yield,indicated by the fact that CTD frequentlyshows a better association with yield andgrain number than it does with totalabove-ground biomass (Table 4).

For a given genotype, CTD is a functionof a number of environmental factors(Figure 4), principally soil water status,air temperature, relative humidity, andincident radiation. The trait is bestexpressed at high vapor pressure deficit

Figure 4. Factors affecting canopy temperature depression (CTD) in plants.

conditions associated with low relativehumidity and warm air temperature(Amani et al., 1996). For these reasons,CTD is not a useful selection trait ingenerally cool and/or humid conditions,and is quite sensitive to environmentalfluxes. Therefore, it is important tomeasure the trait when it is bestexpressed—that is, on warm, relativelystill, cloudless days. Someenvironmental flux during themeasurement period is inevitable, butcorrecting data against reference plots,spatial designs, use of replication, andrepetition of data collection during thecrop cycle can compensate for this.

When measuring CTD, care should betaken to view the plot so as to avoidincluding soil temperature. If a plot issown in rows, it is best to stand to oneside of it so that the thermometer ispointed at an angle to the rows. Ifground cover is low (e.g., leaf areaindex of less than 2-3), it is best to pointthe thermometer at a low angle to thehorizontal to minimize the likelihood ofviewing soil (Figure 5).

Association of CTD with performance.Measurements of CTD made on 23wheat lines at CIMMYT’s subtropicalexperiment station (Tlaltizapan,

Mexico) showed a high correlation withyield measured on the same plots(Figure 6). Sixteen of the same cultivarswere yield tested at a number of hotwheat-growing regions internationally,and their performance compared withCTD measurements made in Mexico(Table 5). In some cases, CTD wasassociated with over 50% of yieldvariability of the same lines at sites inBrazil, Sudan, India, and Egypt, clearlyindicating the promise of CTD as anindirect selection criterion for yield.

In subsequent work, crosses were madebetween parents contrasting in CTD togenerate homozygous sister lines. Thesewere evaluated for both CTD and yieldin warm and temperate environments.Populations of randomly derived F5sister lines from two crosses showed aclear association of CTD with yieldpotential in both warm and temperateenvironments (Figure 7; Table 6), withCTD explaining up to 50% of yieldvariation.

While heritability of CTD has not beenthoroughly evaluated, preliminary datasuggest moderate heritability values forthe trait. When comparing traitsmeasured on F2:5 bulks with subsequentyields in F5:7 lines, performance was

Biological

Partitioning

Clouds

Environmental

(TºC)

Radiation

H2O

evapotranspiration

Metabolism

Vasculartransport

H2O (soil water availability)

Wind

Table 4. Association of CTD with traits of60 advanced lines, Ciudad Obregon (March-sown), Mexico, 1995.

Correlation coefficientTrait with CTD

Yield 0.60**Biomass 0.40**Harvest index 0.14Kernel weight -0.32*Grains m-2 0.62**Spikes/m2 0.33*Grains/spike 0.40**Days to maturity 0.10Days to flowering 0.42**Height 0.10



129

cycle at 15 warm environments: 4 inMexico, 4 in Sudan, 3 in Bangladesh, 3in India, and 1 in Nigeria. Physiologicaltraits were measured on yield plots andon smaller 3-row plots in the selectionenvironment, i.e., a March-sown trial inObregon. Yield and CTD in theselection environment were comparedwith performance of ALs averaged

Grain yield (t ha-1)5.55.04.54.03.53.02.52.01.5

5.0 6.0 7.0 8.0 9.0Canopy temperature depression (C)

Figure 6. Relationship of mean grain yieldto mean CTD for 23 genotypes, averagedover two sowings, Tlaltizapan, Mexico,1992-93.Source: Amani et al. (1996).

Grain yield (t ha-1)5

4

3

2

16.0 7.0 8.0 9.0

Canopy temperature depression (C)

Figure 7. Regression of yield on CTDmeasured after heading for 40 recombinantinbred lines from a cross between linescontrasting in heat tolerance (Seri 82* SieteCerros 66), Tlaltizapan, Mexico, 1995-96.Source: Reynolds et al. (1998).

Table 5. Correlation coefficients between yield, averaged over two cycles at six locations ofthe IHSGE (1990-92), and CTD of 16 wheat lines measured at different stages ofdevelopment, December and February sowings, Tlaltizapan, Mexico, 1992-93.

CTD December CTD February

Location Pre-anthesis Anthesis Post-anthesis Pre-anthesis Anthesis Post-anthesis

Brazil 0.45 0.60* 0.50* 0.68** 0.52* 0.68**Egypt 0.73** 0.91** 0.91* 0.82** 0.79** 0.78**India 0.33 0.56* 0.62** 0.60** 0.37 0.64**Sudan 0.71** 0.91** 0.88** 0.77** 0.75** 0.71**Tlaltizapan 0.66** 0.84** 0.78** 0.50* 0.53* 0.43

Average 0.58 0.76 0.74 0.67 0.59 0.65correlation

* Denotes significance at ≤ 0.05, ** significance at ≤ 0.01.Source: Reynolds et al. 1994.

Figure 5. How to view a plot to avoid including soil temperature when measuring canopytemperature depression with an infrared thermometer (IRT).

When crop is sown inrows, point thermometerat an angle to rows, notparallel to them.

At low leaf areaindex, point IRT atoblique, NOT acute,angles to avoidintercepting soil.

Rows of plants

IRT

better predicted by CTD than it was byyield, when both were measured onbulks (Reynolds et al., 1997).

CTD as an efficient means ofevaluating advanced lines. In additionto the work on early and intermediategeneration breeding lines, experimentswere also conducted at CIMMYT withadvanced lines to assess the power of

CTD as a tool for predictingperformance (Reynolds et al., 1997;1998). Sixty advanced lines (ALs) ofdiverse genetic backgrounds wereselected for superior performance underhot conditions using late sowings inCiudad Obregon, Mexico. The 60 ALswere multiplied and grown as replicatedyield trials in the 1995-96 spring wheat

HEAT TOLERANCE

130

across the 15 environments. CTDmeasured in the selection environmentexplained at least as much of thevariability in performance across allwarm sites as yield itself (Table 7).

In this study, several other physiologicaland morphological traits were evaluatedalong with CTD. While some alsoshowed significant association withyield (e.g., leaf chlorophyll, leafconductance, spike number, biomass,and flowering date), no other single traitwas consistently as well associated withperformance as CTD (Reynolds et al.,1997; 1998). Data also indicated thatCTD measured in 3-row plots was asgood a predictor of yield as CTDmeasured in yield plots, suggesting thatthe technique would be amenable toselection in smaller plots.

Stomatal conductanceCanopy temperature depression is highlysuitable for selecting physiologicallysuperior lines in warm, low relativehumidity environments where highevaporative demand leads to leaf coolingof up to 10 °C below ambienttemperatures. This permits differencesamong genotypes to be detectedrelatively easily using infraredthermometry. However, such differencescannot be detected in high relativehumidity environments because theeffect of evaporative cooling of leaves isnegligible. Nonetheless, leaves maintain

their stomata open to permit the uptakeof CO2, and differences in the rate ofCO2 fixation may lead to differences inleaf conductance that can be measuredusing a porometer.

Porometry can be used to screenindividual plants, unlike CTD, whichcan only be estimated on a canopy. Theheritability of stomatal conductance isreasonably high, with reported valuestypically in the range of 0.5 to 0.8(Vilhelmsen et al., 2001; Rebetzke, pers.comm.); genetic correlation with yield isalso high (Table 2). Plants can beassessed for leaf conductance using aviscous flow porometer that is newlyavailable on the market (Thermoline andCSIRO, Australia). This instrument cangive a relative measure of stomatalconductance in a few seconds (Rebetzkeet al., 1996), making it possible toidentify physiologically superiorgenotypes from within bulks.

For reliable results, more than onereading of stomatal conductance shouldbe taken per plot or per plant. Single-leaf readings always have associatederrors that may be caused byenvironmental fluxes, leaf position, andthe fact that leaves may show diurnaland cyclical patterns in stomatalbehavior. When crops are irrigated, it isbest to take measurements shortly afterirrigation to avoid effects of soilheterogeneity that may affect wateravailability. It is advisable in

Table 7. Phenotypic correlations betweenmean yield of 60 advanced lines atinternational sites and CTD and yieldmeasured in Ciudad Obregon (March-sown),Mexico, 1995-96.

Average yield

Trait n=11† n=15

Yield 0.62** 0.59**CTD 3-row plot 0.66** 0.56**CTD 5-row plot 0.65** 0.58**

** Denotes significance at ≤ 0.01.† 11 locations with least G*E determined by cluster

analysis for crossover interaction.

preliminary studies to measure leafconductance at different times of dayand during different stages of the cropcycle to ascertain when differencesbetween genotypes are best expressed.

Since CTD and leaf conductance showan association with each other and withyield (Amani et al., 1996), thepossibility of combining selection forboth traits is attractive. For example,CTD could be used to select amongearly generation bulks that areheterogeneous and may still besegregating. Porometry can be used toidentify the best genotypes from amongthe plants making up the bulk (Figure 8).Work in Mexico where leaf conductancewas measured on individual plants in aF2:5 bulk indicated the utility of thisapproach (Figure 9; Gutiérrez-Rodríguez et al., 2000).

Membrane thermostabilityAlthough resistance to hightemperatures involves several complextolerance and avoidance mechanisms,the membrane is thought to be a site ofprimary physiological injury by heat(Blum, 1988), and measurement ofsolute leakage from tissue can be usedto estimate damage to membranes.Since membrane thermostability isreasonably heritable (Fokar et al., 1998)and shows high genetic correlation withyield (Table 2), it has potentialapplication in breeding, but does requirea laboratory methodology to makemeasurements.

Laboratory methodology. Membranethermostability (MT) can be measuredon leaf tissue taken at almost anyphenological stage, from 10-day-oldseedlings to grainfilling. Plants must beheat-acclimated either in situ if growingconditions are warm enough, or byputting them in a controlledenvironment for 48 hours atapproximately 35/15°C max/min. Atleast four leaves should be sampled perplot to ensure that tissue isrepresentative, and 10 or more if the

Table 6. Association of CTD with yieldpotential of homozygous sister lines from twocrosses, sown in warm (1995-96) andtemperate environments (1996-97).

Correlation coefficient ofCTD with yield

Cross 1 Cross 2

Seri 82* Seri 82*Site Siete Cerros Fang 60

Tlaltizapan (warm) 0.64** 0.39*Ciudad Obregon (warm) - 0.55**Ciudad Obregon (temperate) 0.64** 0.51**



131

plot contains segregating lines or linesthat are genetically heterogeneous.Leaves should be randomly collectedand placed with their cut ends immersedin water in stoppered glass jars. All jarsshould be placed in a cold box fortransportation from the field to thelaboratory.

In the laboratory, the middle portions ofleaves can be isolated, quickly washedwith de-ionized water, and completelyre-hydrated by keeping them in de-ionized water overnight in arefrigerator. To measure MT, 1-cmsections of each leaf can be cut for boththe control and heat-shock treatments.To measure MT on seedlings, fungicide-treated seed should be germinated onmoistened paper and grown in anenvironmental growth chamber at 10-20°C. The oldest leaves of 10-day-oldseedlings can be used; however,seedlings must be acclimated beforemeasuring MT. For this purpose,approximately 10 seedlings are placedin a covered water bath with their rootsimmersed in water maintained at 35°Cfor 48 h.

Once acclimated, plant material (flagleaves or seedlings) should be washedwith de-ionized water and divided intovials containing 17 mL de-ionizedwater. Half of the vials are maintained

F 5:7 yield (t ha-1)6

5

4

3

2

112 14 16 18 20 22 24 26

An of individual F5 plants (µmol m-2 s-1)

6

5

4

3

2

10.0 0.1 0.2 0.3 0.4 0.5 0.6

gs of individual F5 plants (µmol m-2 s-1)

Figure 9. (a) Relationship between F5:7 grainyield and leaf photosynthesis rate (An) ofindividual F5 plants. (b) Relationshipbetween F5:7 grain yield and stomatalconductance (gs) of individual F5 plants.* significant at p = 0.05.

a

b

r2 = 0.28*y = 185x– 424

F7A cycleF7B cycle

r2 = 0.28*y = 4116x+ 2057

F7A cycleF7B cycle

Figure 8. Using canopy temperature depression (CTD) and leaf conductance in earlygeneration selection.

Selected bulks with high CTD Selected plants for high leaf conductance

hot cool hot

hot cool

e.g. F3 bulks

at 46.5°C (flag leaves) or 49°C(seedlings) for 60 min in a water bath.The second set of vials is used ascontrols and maintained at roomtemperature for the same time periods.After the treatment periods, the heat-treated and control samples are held at6°C overnight. A first conductometricreading is made at 25°C and a second(also at 25°C) after autoclaving for 20min at 120°C and 0.10 MPa. MT isexpressed as relative injury (RI) usingthe following:

RI% = (1-(1-T1/T2)/(1-C1/C2)) x 100,

where T is treatment, C is control, and 1and 2 refer to the first and secondreadings of conductance, i.e., before andafter autoclaving.

Measuring MT on seedlings vs flagleaves. At CIMMYT experiments wereconducted on 16 lines of the IHSGEusing both seedlings and flag leaves(Reynolds et al., 1994). The MT trait wasfavorably correlated with yield in anumber of heat stressed internationalenvironments, using both methodologies(Table 8). When comparing MT forseedlings versus field-grown flag leaves,there was a significant positivecorrelation (r2 = 0.67, n = 16) indicatingthat the MT determined at the two

Table 8. Spearman correlation coefficientsbetween yield, averaged over two cycles ateach of six locations of the IHSGE (1990-92), and membrane relative injury of 16wheat genotypes measured using twodifferent methods.

Heat- Flag leaf Seedlingsstressed (field (chamberlocation grown) grown)

Tlaltizapan Dec. -0.65** -0.40Tlaltizapan Feb. -0.31 -0.01Brazil -0.59* -0.57*Egypt -0.69** -0.64**India -0.66** -0.57*Sudan -0.69** -0.58*

Average correlation -0.60 -0.46

* and ** refer to P < 0.05 and P < 0.01, respectively.

HEAT TOLERANCE

132

development stages was wellassociated. These data support the ideathat using seedlings raised underartificial conditions for screening MTmay be a viable alternative to usingfield-grown tissue. The use of seedlingsis preferable logistically because theconditions of plant acclimation can becontrolled, which is not possible in thefield. The importance of this point isillustrated indirectly by the data.

Using the seedling procedure, the threerepetitions of the experiment weremeasured for MT on three subsequentdays. While the interaction of genotypewith repetition was not significant, themain effect of repetition (i.e., day ofexperiment) was highly significant(data not shown). Even undercontrolled conditions, unintentionaldiscrepancies either in procedure orday-to-day variability of conditionsinfluenced absolute values of MT.Since it would not be practical for abreeding program to assess MT on allthe germplasm of interest in oneexperimental run, a methodologyinvolving controlled conditions wouldseem preferable.

Another advantage of using seedlingsrather than more mature tissue is thatMT is unlikely to be affected byphenology at such an early stage ofdevelopment. In these experimentsthere was a range in anthesis andmaturity dates among the genotypes.Instead of measuring MT on the precisedate of anthesis for each genotype, MTvalues were measured on all flag leaveson the same calendar date andsubsequently adjusted using the numberof days between measurement of MTand anthesis as a covariate.

Genetic Diversity forHeat Tolerance Traits

While genetic diversity for heattolerance has been shown to existamong conventional wheat cultivars(Rawson, 1986; Wardlaw et al., 1989;Al-Khatib and Paulsen, 1990; Reynolds

et al., 1994), progress would be limited ifnew sources of genetic diversity werenot exploited. Materials that could beexploited fall into two broad categories:landraces that can be used directly inconventional breeding efforts and wildspecies with compatible genomes fromwhich genes can be introduced intocultivated wheats using wide crossingapproaches.

Genetic diversity for heat tolerance hasbeen shown to exist in wild Triticum andAegilops species by Edhaie and Waines(1992), who tested accessions fromAfghanistan, Iran, Iraq, Israel, Jordan,Syria, Lebanon, Turkey, and the USSR.Interestingly, all of the heat tolerantaccessions came from only three regions:eastern Israel, western Jordan, andsouthwestern Syria. The authors suggestthat a search among the bread and durumwheat landraces from these regions mayprovide genotypes with a high degree ofheat tolerance that could be incorporatedinto modern wheat backgrounds.

Some work has been conducted toidentify new sources of heat tolerancetraits among accessions in the CIMMYTwheat genebank. For example, high leafchlorophyll content has been identifiedin Mexican landrace collections wherethe best genotypes showed substantiallygreater leaf chlorophyll concentrationthan the check Seri-M82. While high leafchlorophyll content does not guaranteeheat tolerance, the stay-green trait hasbeen associated with heat tolerance infixed lines (Reynolds et al., 2000), andhigh chlorophyll was associated withheat tolerance of sister lines in somewheat crosses (Reynolds et al., 1997).

High stomatal conductance (which maypermit leaf cooling throughevapotranspiration) has started to beexamined in accessions fromCIMMYT’s genebank collections, underheat stressed conditions. For reasonsdiscussed earlier, measuring stomatalconductance as an indication of heattolerance/escape is more suitable thanmeasuring CTD, since it can beevaluated relatively easily on individualplants, a necessary efficiency when

screening very large numbers ofaccessions from a germplasm bank.Apart from identifying genetic diversityfor the trait, preliminary work alsoindicated reasonable levels of broad-sense and realized heritability (60-75%)for the trait (Vilhelmsen et al., 2001).

Molecular approaches may be helpfulfor identifying useful genetic diversityexpressed in the progeny of widecrosses. Genetic diversity from wheatwild relatives has already beenexploited through wide crossing tointroduce disease resistance (e.g.,Villareal et al., 1995). Potential existsfor identifying the loci encoding otherquantitatively inherited traits associatedwith abiotic stress tolerance using QTLanalysis in mapping of delayedbackcross generations (Tanksley andNelson, 1996).

Agronomic Strategiesfor Ameliorating theEffects of Heat Stress

Optimal crop growth requires a non-limiting supply of water, nutrients, andradiation; as temperatures rise, thedemand for growth resources increasesdue to higher rates of metabolism,development, and evapotranspiration(Rawson, 1988). When growthresources are limited by heat stress, thesize of plant organs such as leaves,tillers, and spikes is reduced (Fischer,1984). The apparent sensitivity ofmetabolic processes to heat stress in thefield (Reynolds et al., 1998; 2000),coupled with the reduced length of lifecycle at high temperature (Midmore etal., 1984), explains why grain yield isstrongly associated with total plantbiomass in hot environments. Theseinteractions make crop managementpractices critical to sustaining wheatyields in warm environments.

A few studies have shown the benefitsof specific management practices understress. For example, the application offarmyard manure (FYM) has beenreported to improve soil physical and


133

chemical conditions, and to helpconserve soil moisture (Sattar andGaur, 1989; Gill and Meelu, 1982;Tran-Thuc-Son et al., 1995). A one-time application of FYM (10-15 t ha-1)increased wheat yields for up to threesuccessive crop cycles, when appliedin conjunction with inorganic Nfertilizers, and for up to four years withthe addition of P fertilizers under hotand humid conditions in Bangladesh(Mian et al., 1985). Under high-temperature conditions, volatilizationof N fertilizers such as NH3 is morelikely, and further decreases wheatyields compared with the application ofequivalent N in organic forms such asFYM (Tran-Thuc-Son et al., 1995).

Straw mulch is another agronomicinput with the potential to amelioratestress by reducing evaporation of soilmoisture and increasing infiltration rate(Lal, 1975). Straw mulch has also beenreported to lower soil temperature(Benoit and Kirkhoun, 1963) and toimpede seedling emergence, a negativeeffect (Chopra and Chaudhary, 1980).Surface soil temperatures can exceedair temperature by 10 to 15°C if thesoil surface is bare and radiationintensity is high; straw mulch in suchconditions may increase seedlingemergence and survival (Fischer,1984). Given that wheat growth underwarm conditions is highly sensitive tomanagement, judicious combinationsof management practices couldsubstantially benefit performance byimproving crop establishment and theavailability of water and nutrientsduring subsequent growth stages.

A collaborative study was conductedby CIMMYT and the national wheatresearch programs of Sudan andBangladesh to provide informationfrom warm environments on theresponse of wheat to managementfactors such as mulching andapplication of FYM, and to elevatedlevels of inorganic fertilizer andincreased irrigation frequency(Badaruddin et al., 1999). The researchwas conducted to determine whethermodifications to recommended crop

management practices couldsignificantly improve grain yield.Control treatments representedrecommended practices and gave yieldsof 3.6 t ha-1, averaged across allenvironments. Considering main effects,FYM (10 t ha-1) gave the highest yieldresponse (14%), and approximatelyequivalent levels of NPK the lowest(5.5%), suggesting that organic fertilizerprovided growth factors in addition tonutrient content. Mulch and extrairrigation increased yield in the hot, lowrelative humidity environments (i.e.,Sudan and Mexico), but not inBangladesh, which is hot and humid.

In Mexico, extra inputs were morebeneficial under hotter, spring-sownconditions than in winter sowings.Comparison of heat tolerant (Glennson81) and heat sensitive (Pavon 76)genotypes showed that the heat tolerantgenotype was generally more responsiveto additional inputs. Improvedperformance in response to inputs wasgenerally associated with better standestablishment and with significantincreases in plant height, grain m-2, andabove-ground biomass; in Mexico it wasalso related to higher canopy temperaturedepression and light interception.

These results clearly indicate that wheatyields in warm environments can beraised significantly by modifyingagronomic practices. Overall, theapplication of animal manure had thelargest and most consistent effect onyield. Some of the benefits associatedwith extra organic matter may also beprovided by practicing residue retentionand reduced tillage. Such integratedapproaches to crop and soil managementin abiotically stressed environments arebecoming increasingly relevant in lightof diminishing water supplies in manyagro-ecosystems.

This study did not attempt to analyze theeconomic basis of management factors,only to establish their biological value.Nonetheless, data indicate thatrecommended levels of fertilizer,whether organic or otherwise, were notgenerally sufficient to meet crop

requirements. Average yield responses toNPK and FYM at a given site were asmuch as 17% and 24%, respectively,suggesting that in hot regions eveneconomic yields might be improvedthrough better crop nutrition.

The economic basis of increasingirrigation frequency is more complex fortwo reasons. First, irrigation schemessuch as the one in the Gezira of centralSudan lack the flexibility to permitfarmers to irrigate at will. Water isusually available only at set times in agiven area as water is passedsystematically through the wholeirrigation scheme. Second, wateravailability is declining in many regionsof the world, so the expectation ofraising economic returns throughincreased irrigation may not be fulfilledif water prices rise dramatically. Asmentioned previously, it may be possibleto obtain the benefits of mulching and,perhaps, increase soil organic matterthrough a combination of residueretention and reduced tillage practices.Nonetheless, significant investment willbe required on the part of nationalagricultural research systems and theirgovernments, or agriculturaldevelopment agencies sponsored byindustrialized countries, if such practicesare to become a reality in the developingworld.

Conclusions

While patterns of heat stress may varywidely between wheat growing regions,a major factor explaining genotype byenvironment interaction has been shownto be relative humidity (RH). In low RHenvironments, lack of physiological heattolerance is the major yield constraint,while in high RH environments, diseasepressure may be an additional andpossibly more serious limitation. Canopytemperature depression seems to be apotentially powerful indirect selectioncriterion in low RH environments, whilestomatal conductance and membranethermostability may be applied in all hotenvironments. However, genetic gains to

HEAT TOLERANCE

134

selection should be tested in any newenvironment using locally adaptedgermplasm (as outlined in theintroductory chapter) before the use ofphysiological traits as indirectselection criteria is incorporated intomainstream breeding.

Where germplasm collections areavailable, accessions from abioticallystressed regions should be screened forheat tolerance characteristics as ameans of introducing new sources ofgenetic diversity into the breedingprogram. In addition to geneticimprovement, agronomic strategies(such as residue retention to lower soilsurface temperatures and increase soilorganic matter) are also a means ofincreasing productivity in warmenvironments.

ReferencesAbdElShafi, A.M., and O.A.A. Ageeb. 1994.

Breeding strategy for developing heat-tolerant wheat varieties adapted to upperEgypt and Sudan. In: D.A. Saunders andG.P. Hettel (eds.). Wheat in Heat StressedEnvironments: Irrigated, Dry Areas andRice Farming Systems. Proceedings of theInternational Conference, Wheat in Hot,Dry, Irrigated Environments. Mexico, D.F.:CIMMYT.

Al Khatib, K., and G.M. Paulsen. 1984. Mode ofhigh temperature injury to wheat duringgrain development. Plant Physiol. 61:363-368.

Al-Khatib, K., and G.M. Paulsen. 1990.Photosynthesis and productivity duringhigh temperature stress of wheat cultivarsfrom major world regions. Crop Sci.30:1127-1132.

Amani, I., R.A. Fischer, and M.P. Reynolds.1996. Evaluation of canopy temperature asa screening tool for heat tolerance in springwheat. J. Agron. Crop Sci. 176:119-129.

Badaruddin, M., M.P. Reynolds, and O.A.A.Ageeb. 1999. Wheat management in warmenvironments: effect of organic andinorganic fertilizers, irrigation frequency,and mulching. Agronomy J. 91.

Balota, M., D. Rees, and M.P. Reynolds. 1996.Chlorophyll fluorescence parametersassociated with spring wheat yield under hot,dry, irrigated conditions. Proceedings of theInternational Drought Stress Congress.Belgrade.

Benoit, G.R., and R.J. Kirkhoun. 1963. The effectof soil surface conditions on evaporation ofsoil water. Soil Sci. Soc. Am J. 27:495-498.

Berry, J.A., and O. Bjorkman. 1980. Photosyntheticresponse and adaptation to temperature inhigher plants. Ann. Rev. Plant Physiol.31:491.

Bhullar, S.S., and C.F. Jenner. 1986. Effects oftemperature on the conversion of sucrose tostarch in the developing wheat endosperm.Aust. J. Plant Physiol. 13:605-615.

Blum, A. 1986. The effect of heat stresss on wheatleaf and ear photosynthesis. J. ExperimentalBotany 37:111-118.

Blum, A. 1988. Plant Breeding for StressEnvironment. CRC Press, Inc., Boca Raton,Florida.

Blum, A., and A. Ebercon. 1981. Cell membranestability as a measure of drought and heattolerance in wheat. Crop Sci. 21:43-47.

Chopra, U.K., and T.N. Chaudhary. 1980. Effect ofsoil temperature alternation by soil covers onseedling emergence of wheat (Triticumaestivum L.) sown on two dates. Plant Soil57(1):125-129.

CIMMYT. 1995. CIMMYT/NARS Consultancy onME1 Bread Wheat Breeding. Wheat SpecialReport No. 38. Mexico, D.F.

Cornish, K., J.W. Radin, E.L. Turcotte, Z.-M Lu,and E. Zeiger. 1991. Enhanced photosynthesisand gs of pima cotton (Gossypium barbadenseL.) bred for increased yield. Plant Physiol.97:484-489.

Edhaie, B., and Waines, J:G. 1992. Heat resistancein wild Triticum and Aegilops. J. Genetics andBreeding 46:221-228.

Fischer, R.A. 1984. Physiological limitations toproducing wheat in semi-tropical and tropicalenvironments and possible selection criteria.In: Wheats for More Tropical Environments.A Proceedings of the InternationalSymposium. Mexico, D.F.: CIMMYT. pp.209-230.

Fischer, R.A. 1985. Number of kernels in wheatcrops and the influence of solar radiation andtemperature. J. Agric. Sci. (Cambridge)105:447-61.

Fischer, R.A., and D.B. Byerlee. 1991. Trends ofwheat production in the warmer areas: Majorissues and economic considerations. In: D.A.Saunders (ed.). Wheat for Nontraditional,Warm Areas. Mexico, D.F.: CIMMYT. pp. 3-27.

Fokar, M., H.T. Nguyen, and A. Blum. 1998. Heattolerance in spring wheat. I. Geneticvariability and heritability of cellularthermotolerance. Euphytica 104:1-8.

Gerik, T.J., and J.D. Eastin. 1985. Temperatureeffects on dark respiration among diversesorghum genotypes. Crop Sci. 25:957-961.

Gill, H.S., and O.P. Meelu. 1982. Studies on thesubstitution of inorganic fertilizers withorganic manure and their effect on soilfertility in rice-wheat rotation. FertilizerRes. (Netherlands). 3(4):303-314.

Gutiérrez-Rodríguez, M., M.P. Reynolds, and A.Larqué-Saavedra. 2000. Photosynthesis ofwheat in a warm, irrigated environment. II:Traits associated with genetic gains in yield.Field Crop Res. 66:51-62.

Harding, S.A., J.A. Guikema, and G.M. Paulsen.1990. Photosynthetic decline from hightemperature stress during maturation ofwheat. I. Interaction with senescenceprocess. Plant Physiol. 92:648-653.

Idso, S.B., R.J. Reginato, K.L. Clairson, and M.G.Anderson. 1984. On the stability of stressbaselines. Agric. Forest Meteorol.32:177-182.

Keeling, P.L., R. Banisadr, L. Barone, B.P.Wasserman, and A. Singletary. 1994. Effectof temperature on enzymes in the pathwayof starch biosysthesis in developing maizeand wheat grain. Aust. J. Plant Physiol.21:807-827.

Kohli, M.M., C.E. Mann, and S. Rajaram. 1991.Global status and recent progress inbreeding wheat for the warmer areas. In:D.A. Saunders (ed.). Wheat forNontraditional, Warm Areas. Mexico, D.F.:CIMMYT. pp 96-112.

Lal, R. 1975. Role of mulching techniques intropical soil and water management. IITATech. Bull. I. Ibadan, Nigeria.

Levitt, J. 1980. Response of plants toenvironmental stresses. Chilling, Freezingand High Temperature Stresses. I. AcademicPress New York. 497 p.

Mian, M.I.A., M.A. Rouf, M.A. Rashid, M.A.Mazid, and M. Eaqub. 1985. Residualeffects of triple super phosphate (TSP) andfarmyard manure (FYM) under renewedapplication of urea on the yield of crops andsome chemical properties of soil.Bangladesh J. Agric. Sci. 10(2):99-109.

Midmore, D.J., P.M. Cartwright, and R.A. Fischer.1984. Wheat in tropical environments. II.Crop Growth and Grain yield. Field CropsRes. 8:207-227.

Moffat, J.M., G. Sears, T.S. Cox, and G.M.Paulsen. 1990. Wheat high temperaturetolerance during reproductive growth. I.Evaluation by chlorophyll fluorescence.Crop Sci. 30:881-885.

Ortiz-Ferrara, G., S. Rajaram, and M.G. Mosaad.1994. Breeding strategies for improvingwheat in heat-stressed environments. In:D.A. Saunders and G.P. Hettel (eds.). Wheatin Heat Stressed Environments: Irrigated,Dry Areas and Rice Farming Systems.Proceedings of the International Conference,Wheat in Hot, Dry, Irrigated Environments.Mexico, D.F.: CIMMYT.


135

Pedretti, R.R., and M.M. Kohli. 1991. Wheatproduction in Paraguay: Trends, majorconstraints, and potential. In: D.A. Saunders(ed.). Wheat for Nontraditional, WarmAreas. Mexico, D.F.: CIMMYT.

Rawson, H.M. 1988. Effect of high temperatureson the development and yield of wheat andpractices to reduce deleterious effects. p. 44-62. In: A.R. Klatt (ed.). Wheat ProductionConstraints in Tropical Environments.Mexico, D.F.: CIMMYT.

Rawson, H.N. 1986. High temperature-tolerantwheat: A description of variation and asearch for some limitations to productivity.Wild Crops Res. 14:197-212.

Razzaque, M.A., A.B.S. Hossain, S.M. Ahmed,A.K. Sarkar, M. Rahman, C.D. Barma, andR. Amin. 1994. Wheat breeding strategies inthe rice farming system of Bangladesh. In:D.A. Saunders and G.P. Hettel (eds.). Wheatin Heat Stressed Environments: Irrigated,Dry Areas and Rice Farming Systems.Proceedings of the International Conference,Wheat in Hot, Dry, Irrigated Environments.Mexico, D.F.: CIMMYT.

Rebetzke, G.J., A.G. Condon, and R.A Richards.1996. Rapid screening of leaf conductancein segregating wheat populations. In: R.A.Richards, C.W. Wrigley, H.M. Rawson, J.L.Davidson, and R.I.S. Brettell (eds.). Proc.Eighth Assembly, Wheat Breeding Societyof Australia. pp. 130-134.

Reynolds, M.P. 1994. Summary of data from the1st and 2nd International Heat StressGenotype Experiment. In: D.A. Saundersand G.P. Hettel (eds.). Wheat in HeatStressed Environments: Irrigated, Dry Areasand Rice Farming Systems. Proceedings ofthe International Conference, Wheat in Hot,Dry, Irrigated Environments. Mexico, D.F.:CIMMYT.

Reynolds, M.P., E. Acevedo, O.A.A. Ageeb, S.Ahmed, M. Balota, L.J.B. Carvalho, R.A.Fischer, E. Ghanem, R.R. Hanchinal, C.Mann, L. Okuyama, L.B. Olugbemi, G.Ortiz-Ferrara, M.A. Razzaque, and J.P.Tandon. 1992. Results of the FirstInternational Heat Stress GenotypesExperiment. Wheat Special Report No. 14.Mexico, D.F.: CIMMYT.

Reynolds, M.P., M. Balota, M.I.B. Delgado, I.Amani, and R.A. Fischer. 1994.Physiological and morphological traitsassociated with spring wheat yield underhot, irrigated conditions. Aust. J. PlantPhysiol. 21:717-30.

Reynolds, M.P., M.I. Delgado, M.Gutiérrez-Rodríguez, and A. Larqué-Saavedra. 2000.Photosynthesis of wheat in a warm, irrigatedenvironment. I: Genetic diversity and cropproductivity. Field Crops Res. 66:37-50.

Reynolds, M.P., R.P. Singh, A. Ibrahim, O.A.A.Ageeb, A. Larqué-Saavedra, and J.S. Quick.1998. Evaluating physiological traits tocomplement empirical selection for wheat inwarm environments. Euphytica 100:84-95.

Reynolds, M.P., S. Nagarajan, M.A. Razzaque,and O.A.A. Ageeb (eds.). 1997. Usingcanopy temperature depression to select foryield potential of wheat in heat-stressedenvironments. Wheat Special Report No. 42.Mexico, D.F.: CIMMYT.

Rijven, A.H.G. 1986. Heat inactivation of starchsynthase in wheat endosperm. Plant Physiol.81:448-453.

Saadalla, M.M, J.F. Shanahan, and J.S. Quick,1990. Heat tolerance in winter wheat. I.Hardening and genetic effects on membranethermstability. Crop Sci. 30:1243-1247.

Saini, H.S., M. Sedgley, and D. Aspinall. 1983.Effect of heat stress during floraldevelopment of pollen tube growth andovary anatomy in wheat (Triticum aestivumL.) Aust. J. Plant Physiol. 10:137-144.

Sattar, M.A., and A.C. Gaur. 1989. Effect of VA-mycorrhiza and phosphate dissolvingmicroorganism on the yield and phosphorusuptake of wheat (Triticum vulgare) inBangladesh. Bangladesh J. Agric. Res.14(3):233-239.

Shanahan, J.F., I.B. Edwards, J.S. Quick, and R.J.Fenwick. 1990. Membrane thermostailityand heat tolerance of spring wheat. CropSci. 30:247-251.

Shpiler, L., and A. Blum. 1991. Heat tolerance foryield and its components in different wheatcultivars. Euphytica 51:257-263.

Spiertz, J.H.J. 1978. Grain production andassimilate utilization of wheat in relation tocultivar characteristics, climatic factors andnitrogen supply. Versl. Land-bouwkund,Onderz 881:1-35.

Tandon, J.P. 1994. Wheat cultivation, researchorganization, and production technology inthe hot dry regions of India. In: D.A.Saunders and G.P. Hettel (eds.). Wheat inHeat Stressed Environments: Irrigated, DryAreas and Rice Farming Systems.Proceedings of the International Conference,Wheat in Hot, Dry, Irrigated Environments.Mexico, D.F.: CIMMYT.

Tanskley, S.D., and J.C. Nelson. 1996. Advancedback-cross QTL analysis: A method for thesimultaneous discovery and transfer ofvaluable QTLs from unadapted germplasminto elite breeding lines. Theor. Appl. Genet.92:191-203.

Tran-Thuc-Son, U. Singh, J.L. Padilla, and R.J.Buresh. 1995. Management of urea anddegraded soils of Red River Delta (Vietnam):Effect of growing season and culturalpractice. p. 161-175. In G.L. Denning and Vo-Tong-Xuan (eds.). Vietnam and IRRI, APartnership in Rice Research, Proceedings ofa Conference. Los Baños, Laguna (Phil.).IRRI.

Vargas, M., Crossa, J., Sayre, K.D., Reynolds, M.P.,Ramírez, M.E., and Talbot, M. 1998.Interpreting genotypes by environmentinteraction in wheat by partial least squareregression. Crop Sci. 38:679-689.

Vilhelmsen, A.L., M.P. Reynolds, B. Skovmand, D.Mohan, K.N. Ruwali, S. Nagarajan, and O.Stoelen. 2001. Genetic diversity andheritability of heat tolerance traits in wheat.Wheat Special Report (in preparation).

Villareal, R.L., G. Fuentes-Davila, and A. Mujeeb-Kazi. 1995. Synthetic hexaploids x Triticumaestivum advanced derivatives resistant toKarnal bunt (Tilletia indica Mitra). CerealRes. Com. 27:127-132.

Wardlaw, I.F., I.A. Dawson, P. Munibi, and R.Fewster. 1989. The tolerance of wheat to hightemperatures during reproductive growth. ISurvey procedures and general responsepatterns. Aust. J. Agric. Res. 40:1-13.

Wardlaw, J.F., I. Sofield, and P.M Cartwright. 1980.Factors limiting the rate of dry matteraccumulation in the grain of wheat grown athigh temperature. Aust. J. Plant Physiol.7:387-400.

Zeng, Z.R., J.M. Morgan, and R.W. King. 1985.Regulation of grain number in wheat:Genotypic difference and responses to appliedAbscisic acid and to high temperature. Aust.J. Plant Physiol. 12:609-619.

HEAT TOLERANCE

136

More than one third of the world’sirrigated areas suffer occasional or morefrequent waterlogging (Donmann andHouston, 1967). Waterlogging has beenshown to limit wheat yields in manyregions of the world; an area estimatedat 10 million ha is waterlogged eachyear in developing countries (Sayre etal., 1994). Waterlogging occurs whenrainfall or irrigation water collects onthe soil surface for prolonged periodswithout infiltrating the soil. Soilcharacteristics that contribute towaterlogging include soil physicalproperties that allow formation of acrust on the soil surface or of a pan inthe subsoil. Waterlogging can also occurwhen the amount of water addedthrough rainfall or irrigation is morethan what can percolate into the soilwithin one or two days.

Waterlogging occurs in many wheatgrowing regions around the world,especially irrigated and high rainfallenvironments. In irrigated regions, themain culprit seems to be the lack ofproper drainage systems. Irrigationfacilities do not allow easy drainage ofexcess water, sometimes due to poorlykept irrigation canals from which waterseeps out. Major examples are the IndianSubcontinent, certain river basins inChina, and the Nile River Delta in Egypt.In the northern Indo-Gangetic Plains ofIndia alone, 2.5 million ha of wheat areaffected by irregular waterlogging(Sharma and Swarup, 1988).

Waterlogging ToleranceA. Samad,1 C.A. Meisner,2 M. Saifuzzaman,3 and M. van Ginkel4

1 Principal Scientific Officer, Wheat Research Centre, Bangladesh Agriculture Research Institute, Joydebpur, Gazipur, Bangladesh (same mailingaddress as CIMMYT below).

2 CIMMYT Natural Resources Group, P.O. Box 6057, Gulshan, Dhaka-1212 Bangladesh.3 Senior Scientific Officer, Wheat Research Centre, Bangladesh Agriculture Research Institute, Joydebpur, Gazipur, Bangladesh (same mailing address

as CIMMYT above).4 CIMMYT Wheat Program, Apdo. Postal 6-641, Mexico, D.F., Mexico, 06600.

The effects of waterlogging are mostwidespread in the irrigated rice-wheatregions of South and Southeast Asia(i.e., China, Vietnam, Thailand,Bangladesh, Nepal, India, and Pakistan)and in the southern United States(i.e., Georgia, Mississippi, andLouisiana). A common denominator inthese countries is that rice rotations arepracticed on much of the land. Soils aregenerally puddled to restrict waterpercolation and create floodedconditions for rice cultivation. Due tosoil puddling, wheat that follows rice inthe drier season is planted under lessthan optimal soil physical conditions.The soil pan that was createdintentionally for rice cultivation is oftenleft undisturbed and may create abarrier for water movement, causingwaterlogging when excessive irrigationor rainfall occurs.

In South Asia, wheat is a relatively newoption within rice rotation schemes.Some farmers, accustomed to applyinggenerous amounts of water to rice, tendto over-irrigate their wheat crop.Additionally, many rice-wheat soils aresilt or loam and susceptible to crusting,which creates waterlogging byrestricting percolation from the surface.Declines in organic matter in the topsoilof South Asia are well documented andalso contribute to poor soil physicalquality (FAO, 1994; Hobbs and Morris,1996; Nagarajan, 1998).

Waterlogging can affect other irrigatedareas in Asia besides the rice-wheatgrowing regions. Wheat-producingareas in Egypt, Sudan, and Nigeria alsosuffer regularly from waterlogging. Inparts of Africa and Latin America,heavy rainfall combined with heavyclay soils creates waterlogging thatlimits wheat production. In thetraditional wheat-growing regions inthe Ethiopian highlands, downpours areheavy and prolonged during the rainyseason. Hence waterlogging is acommon occurrence at the beginning ofthe wheat cycle. The situation is furtherexacerbated by the black vertisols inEthiopia, consisting of heavy clays thatinhibit infiltration, swell, and crackseverely. Waterlogging limits wheatyields in Australia due to risinggroundwater (Grieve et al., 1986;McDonald and Gardner, 1987; Meyerand Barrs, 1988).

Conditions andSymptoms Associatedwith Waterlogging

Except at sowing or during earlygermination, waterlogging will notgenerally destroy wheat plants noraffect plant establishment (Musgrave,1994). The major morphological andbiochemical effects will be discussed indetail later, but under mildwaterlogging wheat plant growth is

C HAPTER 1 1

137

usually stunted, bottom leaves senesce,tiller survival is reduced, and florets maybecome sterile.

High temperatures tend to exacerbate thenegative effects of waterlogging. Whenaerobic soil conditions re-occur, plantgrowth resumes slowly. Consequently,wheat yields are affected.

An entire field will rarely bewaterlogged; waterlogging is usuallyrestricted to the lower lying areas of afield (Picture 1). Waterlogging occurswhen the soil is fully saturated, andstanding water replaces the air in the soilpore spaces. There is a lack of oxygen inthe soil, restricting aerobic respiration bygrowing roots and other livingorganisms. Soil chemical propertieschange when anaerobic conditionspersist for several days, increasing theavailability of some major or minorelements while decreasing theavailability of others. Plant transpirationis affected until wheat roots recover(when soil aerobic conditions recur) oradapt to the anaerobic conditions.However, extended waterlogging willresult in root death. Waterlogging alsolimits the wheat plant’s nutrient uptakeby reducing plant transpiration anddiminishing root function.

Another effect of waterlogging is tostimulate the production of certain planthormones. In anaerobic conditions thesehormones are released from the roots ingreater concentrations and may affectleaf and root responses. Ethylene isproduced both by the roots and bymicroorganisms in waterlogged soils.The hormonal effects of ethylenereleased under waterlogging areattracting a great deal of interest. Wateracts as a barrier to the escape of ethyleneproduced in roots and other submergedtissue. Ethylene is known to be a trigger(not a promoter) of leaf senescence(Dong et al., 1983).

Waterlogging during sowing orgermination generally kills the seed orseedling. The seedling’s radicle androots do not adapt readily towaterlogging or are more susceptible toseedling diseases that may follow

(Belford et al., 1985). Generally, thewheat plant’s tolerance to waterloggingincreases as it ages, and the detrimentaleffect on yield decreases (Meyer andBarrs, 1988). Once the wheat crop isestablished, many genotypes canwithstand waterlogging up to 10 dayswith no yield loss, if the wheat leavesare not submerged. Wheat crops canmake an amazing recovery followingearly waterlogging stress, if suppliedwith extra nitrogen.

In waterlogged soils and in the roots ofplants growing in them, exceptionallyhigh levels of ethylene may build up,given that ethylene diffuses more slowlyin water than in aerated soil. The firstresponse of a wheat plant to anaerobicconditions involves its biochemicalpathways as a response to the lack ofrespiration by root cells. Varioushormones are stimulated andtransported to the leaves, causing earlysenescence of older leaves within days(Dong et al., 1983; Dong and Yu, 1984).Seminal roots are generally killed ortheir growth greatly restricted (Huangand Johnson, 1995).

However, some wheat genotypes havenodal or adventitious roots that beginaerenchyma cell formation.Aerenchyma is tissue that can carryoxygen from the leaves to the rootsunder anaerobic conditions to maintain

root respiration, though on a morelimited basis than in aerobic conditions.The process is accelerated if temperaturesare elevated. Genetic variability for thistrait has been documented in the literature(Cao et al., 1995).

Winter wheat areas may also be proneto waterlogging. Winter wheats aresometimes grazed and allowed to re-grow for grain production in the spring.Trampling of saturated pasture soils bycattle can cause restricted watermovement and waterlogging.

The literature documents sometolerance of winter wheats towaterlogging (Musgrave, 1994). Yet thismay not be true tolerance, since thecolder soil temperatures associated withwaterlogging in winter-wheat-growingareas reduce the amount of oxygenrequired for root respiration. Thus yieldreductions associated with waterloggingin colder areas are not as great as thosein the more temperate and tropical areasof the world. On the other hand, somestudies show soil oxygen decline underwaterlogging is rapid at mosttemperature ranges (Trought and Drew,1982). It should also be noted thatwinter wheats are longer maturing andhence less sensitive to waterloggingthan the earlier maturing spring wheats(Gardner and Flood, 1993).

Picture 1. Non-uniform waterlogging in a wheat field in Bangladesh.

WATERLOGGING TOLERANCE

138

sources of waterlogging tolerance (Caoand Cai, 1991; Taeb et al., 1993; Cai etal., 1994); however, there may be othersources of tolerance within wheat. Boru(1996) concluded that there were fourgenes involved in waterloggingtolerance: one major gene, twointermediate ones, and one minor gene.Triticale has proved to be superior tobread wheat in tolerance towaterlogging (Johnson et al., 1991a).The Chinese have reported considerablymore work on breeding waterloggingtolerant lines in the literature than anyother country.

Screening techniques in the laboratoryor the field are well documented in theliterature. While waterlogging toleranceis directly related to the ability toquickly form roots with aerenchymacells under anaerobic conditions, theremay be concurrent tolerance to Mntoxicity (Wagatsuma et al., 1990).Tolerance to Mn toxicity seems to besecondary to the formation ofaerenchyma cell in roots for extendingtolerance. Wagatsuma et al. (1990) alsodetermined that when any tolerance wasexpressed, it was not due to the abilityof plant roots to tolerate low O2 levels.

One study showed differences in nodalroots and aerenchyma cell formationamong wheat and triticale varieties(Thomson et al., 1992). Those lines withincreased nodal and aerenchymaforming abilities endured waterloggingwith fewer detrimental effects. Datasuggest that waterlogging tolerance maybe related to the ability to produce morecrown roots and more aerenchyma inthose roots, to maintain stomatalopening, and to more quickly resumeseminal root growth and stomatalopening when aerobic conditions recur(Huang et al., 1994). Boru (1996)showed that aerenchyma cell formationand yield were highly correlated in linesthat survived severe waterlogging. Theircortical tissue had dissolved to form theaerenchyma; in contrast, sensitivegenotypes expressed little or noaerenchyma formation afterwaterlogging.

Techniques for screening forwaterlogging toleranceThe authors feel that screening forwaterlogging is best done in the field,using simple designs, rather than in lessrealistic laboratory conditions.

Table 1. Soil chemical responses to waterlogging as reported in the literature.

Chemical response Reference

Increased Mn concentration that Sparrow and Uren, 1987; could be toxic to plant growth Wagatsuma et al., 1990

Decreased soil oxygen; Belford et al., 1985 generally greater at warmer temperatures

Decreased Mo availability; Mo application Salcheva et al., 1984 in waterlogged, acid soils retained plastid pigments, cyclic phosphorylation, and CO2 fixation within wheat plants

Denitrification of both organic Feigenbaum et al., 1984; Singh et al., 1988; and inorganic soil N Mascagni and Sabbe, 1991; Humphrey et al., 1991

Mineral (Fe) coating of epidermal surface Ding and Musgrave, 1995 of roots under waterlogging

Volatile fatty acids and phenolic compounds Lynch, 1978; Jackson and St. John, 1980 accumulated in soils high in organic matter affect root metabolism and growth

The literature contains many referenceson the possible genetic variability fortolerance of wheat to waterlogging,hypoxia, or anoxia. This chapter willreview the physiological andbiochemical causes of wheat yieldreductions due to waterlogging. It willalso explore the different options forscreening wheat for waterlogging, plusthe advantages of incorporatingwaterlogging tolerance into a breedingprogram. Agronomic practicesdeveloped through research or beingused by farmers to alleviate thedetrimental effects of waterlogging arealso included in this chapter.

Effect of Waterloggingon Soil Chemistry

Decreases in yield brought about bywaterlogging may be caused bynumerous factors acting upon the wheatplants, such as changes in soilchemistry. As an example,denitrification of soil nitrogen as aresult of waterlogging may affect theamount of nitrogen that concentratesand accumulates in the upper leaves ofthe plants, which will eventually have anegative effect on grain yield. Table 1shows a list of soil chemical responsesand the corresponding bibliographicreferences that can be consulted forfurther information on each.

Genetic Improvementof WaterloggingTolerance

Some studies have suggested that thewaterlogging tolerance trait is highlyheritable (Cao et al., 1995; Boru, 1996);others demonstrated there is littlevariability for waterlogging toleranceamong durum wheat lines (Tesemma etal., 1991). Some authors have foundthat the trait is controlled by a singlegene (Cao et al., 1992; Cao et al., 1995),while others maintain it is polygenic(Hamachi et al., 1989; Boru, 1996).Closely related species of wheat may be

A. SAMAD ET AL.

139

11. Reduced uptake of N, P, K, Ca, Mg, and Zn whileincreasing Na, Fe, and Mn absorption underalkaline soil conditions (Sharma and Swarup,1989; Stieger and Feller, 1994a).

12. Reduced root respiration (Wu et al., 1992; Wanget al., 1996b).

13. In wheat oxygen concentrations between 33 and66 µg m2 s-1 were categorized as deficient and< 33 µg m2 s-1 as critical. Roots weresignificantly reduced by the small amount ofoxygen available, especially at lower depths.Temperature also influenced root reduction, with15o C appearing to be the best soil temperaturefor root growth (Box et al., 1991).

14. Decreases in wheat yields of 37-45% due towaterlogging have been observed (Musgrave,1994; Wu et al., 1992; Cai et al., 1994; vanGinkel et al., 1992; Boru, 1996). Wheat yielddepression was due to reduced kernel numberand weight rather than to an effect on standestablishment.

15. Waterlogging was shown in one study to causeonly slight suppression of flag-leafphotosynthesis and leaf conductance inwaterlogging intolerant wheat lines (Musgrave,1994). Other studies showed overall loweredrates of plant photosynthesis, stomatalconductance, and transpiration (Dong and Yu,1984).

16. Root carbohydrate supply was shown in somestudies not to be a limiting factor for root growthand respiration (Huang and Johnson, 1995).

17. Anoxia (waterlogging) inhibited the transport ofsugars from the shoots to the roots by more than79% in seedlings. However, there areinteractions between temperature and other

Specific PhysiologicalResponses of Wheat toWaterlogging as Reportedin the Literature

1. Chlorosis of lower leaves (Sparrow and Uren,1987; van Ginkel et al., 1992) (Picture 2).

2. Early senescence of lower leaves (Dong et al.,1983; Dong and Yu, 1984).

3. Decreased plant height (Sharma and Swarup,1989; Wu et al., 1992).

4. Delayed ear emergence (Sharma and Swarup,1989).

5. Reduced root and shoot growth (Huang andJohnson, 1995).

6. Lower number of spike-bearing tillers (Belford etal., 1985; Sharma and Swarup, 1989; Wu et al.,1992) (Picture 3).

7. Fewer grains per spikelet and reduced kernelweight (Belford et al., 1985; Musgrave, 1994;van Ginkel et al., 1992).

8. Reduced diameter of metaxylem andprotoxylem vessels of the nodal roots (Huang etal., 1994).

9. Enhanced formation of aerenchyma cells in thecortical tissue of both seminal and nodal roots(Huang et al., 1994; Boru, 1996).

10. Leakage of cell electrolytes (Wang et al.,1996a).

environmental factors that could affectinterpretation of data on tolerance of wheat toanoxia, which explains the lack of consistentresults in the literature (Waters et al., 1991).

18. Data collected on wheat under waterloggedconditions (i.e., deficient in oxygen) in the fieldand glasshouse showed that the biosynthesis ofnew tissue was more inhibited than the supplyof substrates for growth (Attwell et al., 1985).

19. Flower sterility associated with waterlogging islinked to lower transpiration and, hence, to lessuptake of boron (and other nutrients) (Somrith,1988; Saifuzzaman and Meisner, 1996; Rawsonet al., 1996; Misra et al., 1992; Kalidas, 1992;and Subedi, 1992).

20. Ethylene production increases and acts as atrigger (not promoter) of accelerated wheatplant senescence (Dong et al., 1983).Exogenous cytokinins applied to wheatseedlings at the onset of waterlogging delayeddegradation of chlorophyll and otherbiochemical processes (Dong and Yu, 1984).Enhancement of ACC (1-aminocyclopropane-1-carboxylic acid), its precursor, and ethylene wasmore pronounced in older leaves than inyounger ones during waterlogging (Dong et al.,1986).

21. Less nitrogen concentrates and accumulates inthe upper leaves of waterlogged wheat,probably due to the denitrification of soilnitrogen (McDonald and Gardner, 1987).

22. Nitrogen remobilization from lower leaves isaccelerated on flooded soils and explains theirchlorosis (Stieger and Feller, 1994b).

23. Reduced rooting depth and increased rootporosity (Yu et al., 1969).

Picture 3. Waterloggingreduces the number ofspike-bearing tillers.Picture 2. Lower leaf chlorosis.


140

Retaining water on the surface is easierto achieve on heavy soils than onlighter soils. To administerwaterlogging stress on heavy soils,wheat lines can be irrigated such thatwater is retained at or slightly above thesoil surface from emergence to the bootstage (van Ginkel et al., 1992; Sayre etal., 1994). In the former study, carriedout using extreme waterlogging stress,only three genotypes were shown to betolerant out of a total of 1,344 lines. Inlighter soils, waterlogging may be moredifficult to induce.

Our experience shows that even “over-watering” (i.e., keeping the soil slightlyat or above field capacity at variousgrowth periods) can inducewaterlogging that is adequate forscreening wheat lines.

Evaluating differences among varietiesin leaf chlorosis or withering after 15days of waterlogging has been shown tobe a quick method for assessingtolerance. Using this method thenumber of green leaves remaining onthe main stem was correlated with thenumber of fertile grains in the main earand grain weight per plant (Cai andCao, 1990; van Ginkel et al., 1992).Field studies in Mexico and Bangladeshon hundreds of CIMMYT wheat linesover years have shown clear evidenceof variability to waterlogging tolerance.Fields were kept flooded fromemergence to boot stage. Percentagefoliar chlorosis at heading and simpleagronomic scores during grainfillingappeared to be highly correlated withyield in large plots. Many lines can bescreened rapidly using this simplemethodology (van Ginkel et al., 1992).

Studies in Japan showed that assessingleaf senescence in early generationswas useful for screening forwaterlogging tolerance (Hamachi et al.,1989). Wiengweera et al. (1997) used a“stagnant” nutrient solution in agarclosely resembling waterlogged soil forrapid screening of wheat seedlings inthe lab. Studies in China indicated thatan index based on the number of grainsper ear and one thousand grain weight

was effective for evaluatingwaterlogging tolerance (Lin et al.,1994). Musgrave (1994) found that flagleaf photosynthesis in winter wheatcorrelated well with grain weight underwaterlogging.

Although waterlogging during earlyseed germination and seedling growthis very detrimental to the wheat crop,studies have shown there are geneticdifferences in the ability of wheatgenotypes to withstand earlywaterlogging stress (Johnson et al.,1991b). Mineral (Fe) coating of riceroots (showing oxygen release from theroots) is highly correlated to rice yields,but the trait was negatively correlatedto wheat yields in 12 cultivars grownunder waterlogged conditions (Dingand Musgrave, 1995).

Since tiller production decreases duringwaterlogging, tiller production, shootdry matter, and root penetration wereused for screening Triticeae species fortolerance. When these criteria wereused, many wild species expressed alevel of tolerance to waterlogging thatwas better than that of wheat (Taeb etal., 1993).

Agronomic PracticesKnown to ReduceWaterlogging

Setting planting dates to coincide withreduced rainfall patterns is one way toavoid waterlogging (Aggarwal et al.,1987). However, this may not alwaysbe possible due to rotation restrictions,and may be associated with loweryields due to sub-optimal climaticconditions.

Application of nitrogen fertilizer afterwaterlogging has been shown to reducethe detrimental effects of this stress(Trought and Drew, 1980a; Swarup andSharma, 1993; de Oliveira, 1991).Waterlogging under optimum soilnutrient (N) supply conditions resultedin less growth restriction than under asub-optimal nutrient supply (Guyot and

Prioul, 1985). Further studies providedevidence that doubling the concentrationof nutrients supplied to the plants underwaterlogging reduced the rate of declinein photosynthetic rate, chlorophyllcontent, and number of nodal roots,while improving shoot N status andgrowth (Huang et al., 1994).

Singh et al. (1992) found that the use ofgreen manures, straw, and animalmanures increased the availability of Feand Mn several fold under floodedconditions. Organic manures can alsoimprove soil physical factors and reducesoil surface crusting, enhance plantrooting, and alleviate the effects of panformation on yields. Therefore the useof manures is considered beneficial inwaterlogging-prone environments.

Seed treatments such as calciumperoxide (Thomson et al., 1983) weretried with mixed results for alleviatingthe detrimental effects of waterloggingduring germination or early seedlinggrowth.

Several cultivation and sowingtechniques have been shown to giveyield increases under waterloggedconditions. For example, Rasmussen(1988) found that direct drilled wheatwas more sensitive to waterloggingbetween germination and emergencethan conventionally plowed wheat. Thefurrow or bed planting system hassignificant yield advantages, even whenthere is no waterlogging. Furrows alsomake it possible to drain fields or keep alarge portion of the root system out ofwaterlogged soils (Abebe et al., 1991;Tedia et al., 1994).

Shifting from basin flooding to furrowor sprinkler irrigation on waterlogging-prone soils has been shown to reducethe problem significantly (Melhuish etal., 1991). Surface seeded wheat (sownon top of uncultivated, saturated soil)showed the least sensitivity towaterlogging compared to wheat sownin conventionally plowed and chiseledsoil (Table 2).

A. SAMAD ET AL.

141

Table 3. Characteristics of selected waterlogging tolerant and sensitive wheat genotypesgrown under varying waterlogging conditions, WRC, Nashipur, Dinajpur, Bangladesh, 1994.

Avg. grain Avg. Avg. Visual Leaf Plantyield TGW grains sterility yellowing† vigor‡

Genotypes (kg ha-1) (g) spikelet-1 (%) (1-5) (1-5)

Waterlogging tolerantMOZ-2 (Bangladesh) 4,333 49.8 1.80 0 1 5

BAW-451 (Bangladesh) 4,233 30.1 2.67 0 1 5

BR-16 (Brazil) 3,767 42.1 1.90 17 1 5

IAS58/4/KAL/BB/CJ/3/ALD/5/VEECM88971-9Y-0M-0Y-3M-0Y 3,700 49.9 1.86 34 3 5

MOZ-1 (Bangladesh) 3,533 49.1 1.64 0 1 5

Waterlogging sensitiveHD 22629 (India) 1,167 42.6 1.92 68 3 2

BAW-905 = K 9162 (Bangladesh) 1,233 42.8 1.93 12 4 1

K 8962 1,233 38.4 2.20 16 4 1

Aestivum Roelz W9047 1,300 35.3 2.24 0 3 3

FLN/ACC/ /ANA/3/DOVECM65720-3Y-1M-1Y-1M 1,367 38.6 1.85 74 3 2† Recorded in the field at 65 DAS using 1 to 5 scale, where 1 = yellowing of lower leaves and 5 = of flag leaves.‡ Judged using a 1 to 5 scale at 65 DAS, where 1 = very poor growth and 5 = excellent plant vigor.

Screening forWaterlogging Toleranceunder BangladeshConditions

Identification of waterlogging tolerantwheat genotypes began during the 1993wheat season in Bangladesh. In thenorthwestern part of the country,screening of wheat genotypes forwaterlogging tolerance spanned fourseasons. The soil type at the DinajpurWheat Research Centre experimentstation is deep, sandy loam. Over threeseasons (1993-95), 162 wheat genotypeswere subjected to waterlogging byirrigating at 10-day intervals from 10 to100 days after sowing (DAS). The fieldwas flooded and the land submerged for24-36 hours in each of the 10 irrigations.

However, under those soil conditions(sandy loam), our treatments were closerto “over-irrigation” than truewaterlogging, since the water percolatedquickly within 36 hours of waterlogging.Five replications were used during thefirst two seasons, and three replicationswere practiced in the other two seasonsto collect data on 2.5-m plots consistingof three rows, 20 cm wide. Theexperiments were sown at normalsowing time (third week of November).Seeding rate was 120 kg/ha, andfertilizer rates were as recommended(100: 60: 40: 20: kg/ha of NPKS).

Additional N fertilizer was top-dressedjust after the second irrigation, asrecommended in Bangladesh (33:0:0).

In contrast to previous years, in whichwaterlogging resembled “over-watering,” in the 1996 season 64 wheatgenotypes were subjected to truewaterlogging as in a rice field; the fieldwas irrigated consecutively for threedays at three growth stages: crown rootinitiation, booting, and grainfilling.Crop growth was severely affectedduring this season, and most genotypesdid not produce sufficient spikes forsampling to record spikelet/spike,grains/spike and one thousand grainweight (TGW). Some genotypesproduced only a few small spikes withhardly any grain.

Twenty-one waterlogging tolerant andtwenty waterlogging sensitive wheatgenotypes were identified from linessubjected to various modes ofwaterlogging over the years. As an

example, the scores of lines tolerant andsensitive to waterlogging in 1994 arepresented in Table 3.

Another experiment with eightwaterlogging treatments (including acontrol) was conducted in centralBangladesh during the 1996 growingseason. Soils were heavy 2:1montmorillanitic. Waterlogging treatmentswere imposed at 10 (T2), 20 (T3), 30 (T4),40 (T5), 50 (T6), 60 (T7), and 70 (T8)DAS, which correspond to Zadoks’ growthstages 12, 21, 31, 42, 52, 63, and 73.Control was normal irrigation (T1). Waterleft standing for four days in the treatmentplots was considered waterlogging in thesesoils. The control plot received threenormal irrigations. The objective of thisexperiment was to observe the effect ofwaterlogging on seed set and yield, as wellas to determine which crop growth stagesare critically related to poor seed set andyield in wheat under simulatedwaterlogging conditions compared withthe other years and locations.

Table 2. Wheat plant population underwaterlogged conditions at sowing and earlygermination in different tillage systems.

Tillage and Wheat plantsowing system population (plants m-2)

Conventional tillage:broadcast sowing 136a†

Chisel tillage: broadcastsowing 142a

Zero tillage:surface seeding 225b† LSD among the rows are designated by letters.Source: Unpublished field data from Bangladesh

(Badaruddin, 1997).


142

There was no significant influence ofwaterlogging on spikes per unit area(Figure 1), which is consistent with theliterature. Grains m-2 was used as anindicator of wheat seed set (Meisner etal., 1992). Waterlogging affected seedset. Misra et al. (1992) also reportedthat waterlogging affected seed set inwheat in Nepal. Seed set was mostaffected when waterlogging wasimposed at 30 DAS (T4), followed by10 DAS (T2). The highest number ofgrains m-2 was obtained in the controltreatment (T1), followed bywaterlogging treatments T6, T5, and T8(Figure 2). Waterlogging stress duringZadoks’ 31 was identified as being mostcritical for seed set in wheat, followedby Zadoks’ 12. This is consistent withthe data of van Ginkel et al. (1992). Thewheat crop was found to be sensitive towaterlogging stress, though to a lesserdegree, during Zadoks’ 21 and 63.

One thousand grain weight was notaffected by imposing waterloggingtreatments at different growth stages(Figure 3) in our experiment.Differences in grain weight did notoccur because waterlogging treatmentswere applied at and before anthesis butnot from grainfilling onward. Otherstudies show contrasting results (vanGinkel et al., 1992).

Wheat grain yields differed withwaterlogging treatments (Figure 4).Luxmoore et al. (1973) also observednegative effects on wheat grain yieldswhen waterlogging was imposed for 30days during grainfilling at 15 and 250Csoil temperatures, which reduced grainyield by 20 and 70%, respectively.Waterlogging reduced wheat yields dueto poor seed set and fewer spikes perunit area.

Conclusions

Realistic but cautiously optimisticconclusions can be drawn based on theabove review of the literature and ondata from the case study in Bangladesh.

Waterlogging is a widespread problemin the irrigated and high rainfall wheat-growing regions of the world. Despitethe breadth of the problem,understanding of the basic soil and plantprocesses involved in waterloggingtolerance is improving. The good newsis that there is genetic variability for

Grains m-2 (000s)10

8

6

4

2

0Control 10 20 30 40 50 60 70

Waterlogging treatments (DAS)

Figure 2. Waterlogging at different growthstages (days after sowing-DAS) affectedgrains per unit area in 1995-96 atJoydebpur, Bangladesh.

CV= 10.0% LSD (5%) = 1367

Spikes m-2

400

300

200

100

0Control 10 20 30 40 50 60 70


Figure 1. Waterlogging at different growthstages (days after sowing-DAS) affectedspikes per unit area in 1995-96 atJoydebpur, Bangladesh.

CV= 16.8% NS

1000-grain weight (g)

50

40

30

20

10

0Control 10 20 30 40 50 60 70

Waterlogging treatments

Figure 3. Waterlogging at different growthstages (days after sowing-DAS) affected1000-grain weight in 1995-96 atJoydebpur, Bangladesh.

CV= 3.5 % NS

Grain yield (t ha-1)5

4

3

2

1

0Control 10 20 30 40 50 60 70


Figure 4. Waterlogging at different growthstages (days after sowing-DAS) affectedgrain yield in 1995-96 at Joydebpur,Bangladesh.

LSD = (5%) = 629.2 CV = 10.1%a

bcab

c

ab ababc ab

waterlogging tolerance within wheat,and that the genetics of waterloggingtolerance appears to be relativelysimple, with medium to highheritabilities. Therefore, the prospectsare good that varieties suitable for areassuffering from waterlogging stress canbe bred and/or identified.

A. SAMAD ET AL.

143

References andSuggested ReadingAbebe, M., T. Mamo, M. Duffera, and S. Kidanu.

1992. Durum wheat response to improveddrainage of Vertisols in the central highlandsof Ethiopia. In D.G. Tanner and W. Mwangi,(eds.). Seventh Regional Wheat Workshop forEastern, Central and Southern Africa. Nakuru,Kenya: CIMMYT. pp. 407-414.

Aggarwal, P.K, S.P. Liboon, and R.A. Morris. 1987.A review of wheat research at theInternational Rice Research Institute. IRRIResearch Paper Series No. 124.

Attwell, B.J., H. Greenway, and E.G. BarrettLennard. 1985. Root function and adaptiveresponses in conditions of oxygen deficiency.In W.A. Muirhead, E. Humphreys, (ed.). Rootzone limitations to crop production on claysoils: symposium of the Australian Society ofSoil Science Inc., Riverina Branch.Melbourne, Vic., Australia. CommonwealthScientific and Industrial ResearchOrganization. pp. 65-75.

Badaruddin, M. 1997. Review of 1996/97 WheatResearch. Internal Review of the WheatResearch Centre, Bangladesh AgricultureResearch Institute, Gazipur Bangladesh.

Belford, R.K, R.Q. Cannell, and R.J. Thomson.1985. Effect of single and multiplewaterlogging on the growth and yield ofwinter wheat on a clay soil. Journal of theScience of Food and Agriculture36:142-156.

Boru, G. 1996. Expression and Inheritance ofTolerance to Waterlogging Stresses in Wheat(Triticum aestivum L.). Ph.D Thesis, OregonState University, Corvallis, OR. 88 pp.

Box, Jr., J.E., B.L. McMichael (ed.), and H.Persson. 1991. The effects of waterlogging onrooting of soft red winter wheat plant rootsand their environment. In Proceedings of anISSR 21-226 August, 1998, Uppsala, Sweden.pp. 418-430.

Cai, S.B., and Y. Cao. 1990. Study of waterloggingtolerance in wheat at booting. Institute ofFood Crop Research, Jiangsu Academy ofAgricultural Sciences, Xiaolinwei, Nanjing,China. Crop Genetic Resources 4:27-28.

Cai, S.B., Y. Cao, J.M. Yan, X.W. Fang, and W.Zhu. 1994. Genotypes in response to hypoxiaand subsequent resumption of aeration. CropScience 34:1538-1544.

Cao, Y., and S.B. Cai. 1991. Some waterloggingtolerant wheat varieties. Crop GeneticResources 2:25-26.

Cao, Y., S.B. Cai, W. Zhu, and X.W. Fang. 1992.Genetic evaluation of waterlogging resistancein the wheat variety Nonglin 46. Crop GeneticResources 4:31-32.

Cao, Y., S.B. Cai, Z.S. Wu, W. Zhu, X.W. Fang, andE.H. Xion. 1995. Studies on genetic featuresof waterlogging tolerance in wheat. JiangsuJournal of Agricultural Sciences 11:11-15.

Ding, N., and M.E. Musgrave. 1995. Relationshipbetween mineral coating on roots and yieldperformance of wheat under waterloggingstress. Journal of Experimental Botany46:939-945.

Dong, J.G., Z.W. Yu, and S.W. Yu. 1983. Effect ofincreased ethylene production duringdifferent periods on the resistance of wheatplants to waterlogging. ActaPhytophysiologia Sinica 9:383-389.

Dong, J.G, and S.W. Yu. 1984. Effect of cytokininon senescence and ethylene production inwaterlogged wheat plants. AetaPhytophysiologia Sinica 10:55-62.

Dong, J.G., S.W. Yu, and Z.G. Li. 1986. Changesin ethylene production in relation to 1-aminocyclopropane-1-carboxylic acid and itsmalonyl conjugate in waterlogged wheatplants. Acta Botanic Sinica 28:396-403.

Donmann, W.W., and C.E. Houston. 1967.Drainage related to irrigation management.In: Drainage of Agricultural Lands. R.W.Hagan, H.R. Haise, and T.W. Ediminster(eds.). Am Soc Agronomy pp. 974-987.

Feigenbaum, S., N.G. Seligman, and R.W.Benjamin.1984. Fate of nitrogen-15 appliedto spring wheat grown for three consecutiveyears in a semiarid region. Soil ScienceSociety of America Journal 48:838-843.

FAO. 1994. Sustainability of Rice-WheatProduction Systems in Asia. Regional Officefor Asia and the United Nations, Bangkok.R.S. Paroda, T. Woodhead, R.B. Singh(eds.). RAPA Publication 1994/11.

Gardner, W.K., and R.G. Flood. 1993. Lesswaterlogging damage with long seasonwheats. Cereal Research Communications21(4):337-343.

Grieve, A.M., E. Dunford, D. Marston, R.E.Martin, and P. Slavich. 1986. Effects ofwaterlogging and soil salinity on irrigatedagriculture in the Murray Valley: A review.Aust J Exp 26:761-777.

Guyot, C., and J.L. Prioul. 1985. Alleviation ofwaterlogging effects in winter wheat bymineral fertilizer application. II Experimentsin Hydroponic Culture. Agronomia5:751-759.

Hamachi, U.Y., M. Furusho, and T. Yoshida. 1989.Heritability of wet endurance in maltingbarley. Japanese Journal of Breeding39:195-202.

Harper, S.H.T., and J.M. Lynch. 1982. The role ofwater-soluble components in phytotoxicityfrom decomposing straw. Plant Soil65:11-17.

Harrison, S.A., K.M. Tubajika, J.S. Russin, andC.A. Clark. 1997. Effect of managementinputs on hybrid wheat production inLouisiana. Annual Wheat Newsletter.

Hobbs, P., and M. Morris. 1996. Meeting SouthAsia’s future food requirements from rice-wheat cropping system. Priority issuesfacing researchers in the post-greenrevolution era. NRG Paper 96-01. Mexico,D.F.: CIMMYT. 46 p.

Huang, B.R., J.W. Johnson, S. Nesmith, and D.C.Bridges. 1994. Growth, physiological andanatomical responses of two wheatgenotypes to waterlogging and nutrientsupply. Journal of Experimental Botany45(271):193-202.

Huang, B.R., and J.W. Johnson. 1995. Rootrespiration and carbohydrate status of twowheat genotypes in responses to hypoxia.Annals of Botany 75(4):427-432.

Humphreys, E., F.M. Melhuish, R.J.G. White,W.A. Muirhead, and X. Zhen Bang. 1991.Flood irrigation of wheat on a transitionalred-brown earth. 2. Effect of duration ofponding on availability of soil and fertilizernitrogen. Australian Journal of AgriculturalResearch 42(7):1037-1051.

Hunt, P.G., R.B. Champbell, R.E. Sojka, and J.E.Parsons. 1981. Flooding-induced soil andplant ethylene accumulation and water statusresponse of field-grown tobacco. Plant Soil59:427-439.

Jackson, P.C., and J.B. St. John. 1980. Changes inmembrane lipids of roots associated withchanges in permeability. 1. Effect ofundissociated organic acids. Plant Physiol.66:801-804.

Johnson, J.W., J.E. Box, Jr., J.B. Manandhar, E.L.Ramseur, and B.M. Cunfer. 1991a. Breedingfor rooting potential under stress conditions.In E. Acevedo, A.P. Conesa, P. Monneveux,and J.P. Srivastava (eds.). Physiologybreeding of winter cereals for stressedMediterranean environments. Montpelier,France. pp. 307-317.

Johnson, J.W., B.M. Cunfer, and J. Manandhar.1991b. Adaptation of triticale to soils of thesoutheastern USA. In Proceedings of theSecond International Triticale Symposium.Mexico, D.F.: CIMMYT. pp. 659-662.

Kalidas, S. 1992. Wheat sterility in Nepal: Areview. In: Boron deficiency in wheat.Wheat Special Report No. 11. C.E. Mannand B. Rerkasem (eds.). Mexico, D.F.:CIMMYT. pp. 57-64.

Lin, Y., Y. XiaoYan, L. FentLan; Y.B. Lin, X.Y.Yang, and F.L. Liu. 1994. Study onevaluation of waterlogging tolerance in 50wheat varieties (Triticum aestivum L.). ActaAgriculture Shanghai 10:79-84.

Luxmoore, R.J., R.A. Fischer, and L.H. Stolzy.1973. Flooding and soil temperature effectson wheat during grain filling. Agron J65:361-364.

Lynch, J.M. 1978. Production and phytotoxicityof acetic acid in anaerobic soils containingplant residues. Soil Biol Biochem10:131-135.

Mascagni, H.J. Jr, and W.E. Sabbe. 1991. Latespring nitrogen applications on wheat on apoorly drained soil. Journal of PlantNutrition (USA) 14(10):1091-1103.

McDonald, G.K., and U.K. Gardner. 1987. Effectof waterlogging on the grain yield responseof wheat to sowing date in southwesternVictoria. Australian Journal of ExperimentalAgriculture 27:661-670.

Meisner, C.A., E. Acevedo, D. Flores, K.D. Sayre,I. Ortiz-Monasterio, D. Byerlee, and A.Limon. 1992. Wheat production and growerpractices in the Yaqui Valley, Sonora,Mexico. Wheat Special Report No. 6.Mexico, D.F.: CIMMYT.


144

Melhuish, F.M., E. Humphreys, W.A. Muirhead,and R.J.G. White. 1991. Flood irrigation ofwheat on a transitional red-brown earth. I.Effect of duration of ponding on soil water,plant growth, yield and N uptake. AustralianJournal of Agricultural Research 7:1023-1035.

Meyer, U.S., and H.D. Barrs. 1988. Response ofwheat to single short-term waterloggingduring and after stem elongation. AustralianJournal of Agricultural Research 39:11-20.

Misra, R., R.C. Munankarmi, S.P. Pandey, andP.R. Hobbs. 1992. Sterility work in wheat atTarahara in the Eastern Tarai of Nepal. InC.E. Mann and B. Rerkasem (eds.). Borondeficiency in wheat. Mexico, D.F.:CIMMYT. pp. 65-71.

Musgrave, M.E. 1994. Waterlogging effects onyield and photosynthesis in eight winterwheat cultivars. Crop Sci 34(5):1314-1318.

Nagarajan, S. 1998. Understanding the issuesinvolved and steps needed to increase wheatyields under rice/wheat system: A case studyof the Karnal area, Haryana, India.Submitted to the Alexander von HumboldtFoundation; Institute for Plant Disease,University of Bonn, Germany. 74 pp.

Oliveira, M.R.G. de. 1991. Performance of somewheat cultivars under waterloggedconditions. Revista de Ciencias Agrarias14:53-58.

Rasmussen, K.J. 1988. Ploughing, direct drillingand reduced cultivation for cereals.Tidsskrift for Planteavl 92:233-248.

Rawson, H.M., H.M. Rawson, and K.D. Subedi.1996. Hypothesis for why sterility occurs inwheat in Asia. In Sterility in wheat insubtropical Asia: extent, causes andsolutions. Proceedings of a workshop. H.M.Rawson (ed.). ACIAR Proceedings No. 72.pp. 132-134.

Saifuzzaman, M., and C.A. Meisner. 1996. Wheatsterility in Bangladesh: An overview of theproblem, research and possible solutions. InSterility in wheat in subtropical Asia: extent,causes and solutions. Proceedings of aworkshop. H.M. Rawson (ed.). ACIARProceedings No. 72. pp. 104-108.

Salcheva, G., G. Ignatov, D. Georgieva, and S.Doncheva. 1984. Effect of molybdenum onthe functional and structural state ofchloroplasts and mitochondria from leves ofwinter wheat and rye grown on awaterlogged acidic soil. Fiziologiya naRasteniyata 10:12-21.

Sayre, K.D., M. van Ginkel, S. Rajaram, and I.Ortiz-Monasterio. 1994. Tolerance towaterlogging losses in spring break wheat:Effect of time on onset of expression. In:Annual Wheat Newsletter 40:165-171.

Sharma, D. P., and A. Swarup. 1988. Effect ofshort term waterlogging on growth, yieldand mineral composition of wheat in sodicsoils under field condition. Plant and Soil107:137-143.

Sharma, D.P., and A. Swarup. 1989. Effect ofnutrient composition of wheat in alkalinesoils. Journal of Agricultural Science (UK)112:191-197.

Singh, B., Y. Singh, U.S. Sadana, and O.P. Meelu.1992. Effect of green manure, wheat strawand organic manures on DTPA extractableFe, Mn, Zn and Cu in a calcareous sandyloam soil at field capacity and underwaterlogged conditions. Journal of theIndian Society of Soil Science 40:114-118.

Singh, Y., B. Singh, M.S. Maskina, and O.P.Meelu. 1988. Effect of organic manures,crop residues and green manure (Sesbaniaaculeata) on nitrogen and phosphorustransformations in a sandy loam at fieldcapacity and under waterlogged conditions.Biology and Fertility of Soils 6:183-187.

Somrith, B. 1988. Problems associated with soilmanagement issues in rice-wheat rotationareas. In: Wheat production constraints intropical environments. Proceedings of theinternational conference. A.R. Klatt (ed.).Mexico, D.F.: CIMMYT. pp. 63-70.

Sparrow, L.A., and N.C. Uren. 1987. The role ofmanganese toxicity in crop yellowing onseasonally waterlogged and strongly acidicsoils in northeastern Victoria. AustralianJournal of Experimental Agriculture27:303-307.

Stieger, P.A., and U. Feller. 1994a. Nutrientaccumulation and translocation in maturingwheat plant grown on waterlogged soil.Plant and Soil 160(1):87-96.

Stieger, P.A., and U. Feller. 1994b. Senescenceand protein re-mobilization in leaves ofmaturing wheat planes grown onwaterlogged soil. Plant and Soil 166(2):173-179.

Subedi, K. 1992. Wheat sterility in Nepal: Areview. In C.E. Mann and B. Rerkasem(eds.). Boron deficiency in wheat. Mexico,D.F.: CIMMYT. pp. 57-64.

Swarup, A., and D.P. Sharma. 1993. Influence ofto dressed nitrogen in alleviating adverseeffects of flooding of growth and yield ofwheat in a sodic soil. Field Crops Research35:93-100.

Taeb, M., R.M.D. Koebner, and B.P. Forster.1993. Genetic variation for waterloggingtolerance in the Triticeae and thechromosomal location of genes conferringwaterlogging tolerance in Thinopyrumelongatum. Genome 36:825-830.

Tedia, A., J. Sherington, and M.A. MohamedSaleem. 1994. Integration of forage and foodcrops grown sequentially on Vertisols underrainfed conditions in the mid-altitudeEthiopian highlands. ExperimentalAgriculture 30:291-298.

Tesemma, T., G. Belay, and D. Mitiku. 1992.Evaluation of durum wheat genotypes fornaturally waterlogged highland vertisols ofEthiopia. In D.G. Tanner and W. Mwangi(eds.). Seventh Regional Wheat Workshopfor Eastern, Central and Southern Africa.Nakuru, Kenya: CIMMYT. pp. 96-102.

Thomson, C.J., T.D. Colmer, E.L.J. Walkin, andH. Greenway. 1992. Tolerance of wheat(Triticum aestivum cvs. Gamenya and Kite)and triticale (Triticosecale cv. Muir) towaterlogging. New Phytologist 120:335-344.

Thomson, R.J., R.K. Belford, and R.Q. Cannell.1983. Effect of a calcium peroxide seedcoating on the establishment of winter wheatsubjected to pre-emergence waterlogging.Journal of the Science of Food andAgriculture 34:1159-1162.

Trought, M.C.T., and M.C. Drew. 1982. Effects ofwaterlogging on young wheat plants(Triticum aestivum L.) and on soil solutes atdifferent soil temperatures. Plant and Soil69:311-326.

Trought, M.C.T., and M.C. Drew. 1980a. Thedevelopment of waterlogging damage inwheat seedlings (Triticum aestivum L.) I.Shoot and root growth in relation to changesin the concentration of dissolved gases andsolutes in the soil solution. Plant Soil54:77-94.

Trought, M.C.T., and M.C. Drew. 1980b. Thedevelopment of waterlogging damage inwheat seedlings (Triticum aestivum L.) II.Accumulation and redistribution of nutrientsby the shoot. Plant Soil 56:187-199.

van Ginkel, M., S. Rajaram, and M. Thijssen.1992. Waterlogging in wheat: Germplasmevaluation and methodology development. InD.G. Tanner and W. Mwangi (eds.). SeventhRegional Wheat Workshop for Eastern,Central and Southern Africa. Nakuru, Kenya:CIMMYT. pp. 115-124.

Wagatsuma, T., T. Nakashima, K. Tawaraya, S.Watanbe, A. Kamio, and A. Ueki. 1990.Relationship between wet tolerance,anatomical structure of aerenchyma and gasexchange ability among several plant species.Agricultural Science (Japan) 11(1):121-132.

Wang, S., H. LiRen, L. ZhengWei, Z. JinGuo, C.YouRong, H. Lei, S.G. Wang, L.R. He, Z.W.Li, J.G.Zeng, Y.R. Chi, and L. Hou. 1996a. Acomparative study on the resistance of barleyand wheat to waterlogging. Acta AgronomicaSinica 22:228-232.

Wang, C.Y., M. YuanXi, Z. Sumei, Z. YunJi, L.JiuXing, W. HuaCen, C. Wang, Y.X. Ma,S.M. Zhou, Y.J. Zhu, J.X. Li, and H.C. Wang.1996b. Effects of waterlogging on themetabolism of active oxygen and thephysiological activities of wheat rootsystems. Acta Agronomica Sinica22:712-719.

Waters, I., P.J.C. Kuiper, E. Watkin, and H.Greenway. 1991. Effects of anoxia on wheatseedlings. I. Interaction between anoxia andother environmental factors. Journal ofExperimental Botany 42:1427-1435.

Wiengweera, A., H. Greenway, and C.J. Thomson.1997. The use of agar nutrient solution tosimulate lack of convection in waterloggedsoils. Annals of Botany 80(2):115-123.

Wu, J.G., S.F. Liu, F.R. Li, and J.R. Zhou. 1992.Study on the effect of wet injury on growthand physiology winter wheat. ActaAgriculture Universitatis Henanensis26: 31-37.

Yu, P.T., L.H. Stolezy, and J. Letey. 1969. Survivalof plants under prolonged flooded conditions.Agron J 61:844-847.

Zadoks, J.C., Chang, T.T., and Konzak, C.F. 1974.A decimal code for the growth stages ofcereals. Weed Res.14:415-421.

A. SAMAD ET AL.

145

Figure 2. Bread produced from sprouted grain (left) and from sound grain (right).

Rainfall during or just prior to harvestcan cause wheat grain to germinatewhile still on the spike (Figure 1). Thisphenomenon, called preharvestsprouting (PHS), reduces yield, lowerstest weight, and adversely affects themilling and baking quality of harvestedgrain. Farmers receive lower prices forsprouted grain and, in severe cases,their harvests may be downgraded toanimal feed. The occurrence of PHS isgenerally erratic and as difficult topredict as rainfall in most wheat-growing areas. Researchers have,however, been able to provide farmersin areas prone to PHS with a degree ofprotection. This chapter attempts tooutline the role of physiology andplant breeding in the wider effort todevelop strategies to combat thisintractable problem.

Preharvest Sprouting ToleranceR.M. Trethowan1

Extent of the Problem

Rainfall can cause extensive PHSdamage in most wheat-growing regions;however, some areas are more pronethan others to its occurrence. Althoughmany of these regions are found indeveloped countries, significant areas ofthe developing world are also affected.Northern Europe, the Pacific Northwestof the United States, and the wheatgrowing areas of central Canada andnortheastern Australia periodicallysuffer PHS damage. In developingcountries the Southern Cone of SouthAmerica, encompassing parts of Chile,Argentina, and Brazil, and the wheat-growing areas of eastern Africa areprone to PHS. The damage is morepronounced in regions where white-grained wheat is grown. Red-grainedwheat is more tolerant to the problem,since there is an association between

Damage Caused

As its name indicates, preharvest sproutingbegins before the grain is harvested, whileit still on the spike. The process is set inmotion by rainfall, during which the seedimbibes water. This causes germination tobegin as starch reserves in the grainendosperm are hydrolized through theaction of germinative enzymes calledamylases. The embryo swells and grows asit consumes the hydrolized carbohydratereserve.

The test weight and flour milling yield ofsprouted grain are considerably lowerthan those of non-sprouted wheat. Breadproduced from sprouted grain has poorloaf volume and crumb structure, and isunsuitable for marketing (Figure 2). Thequality of flat breads and chapatis is lessaffected by the use of sprouted grain, buttheir texture is nevertheless impaired,resulting in a less favorable product.Sprouted grain causes discoloration ofChinese noodles and spaghetti, loweringthe value of these products as well.

1 CIMMYT Wheat Program, Apdo. Postal 6-641, Mexico, D.F., Mexico 06600.

Figure 1. Two spikes damaged by pre-harvestsprouting (left) and a sound spike (right).

C HAPTER 1 2

grain color and graindormancy, the primarymechanism of PHStolerance (Gale, 1989).

146

Tolerance Mechanisms

The primary mechanism of PHStolerance is grain embryo dormancy.Dormant seeds will imbibe water and yetnot germinate. Grain dormancy is,however, influenced greatly byenvironmental conditions prior to andduring grain maturation. Hightemperatures during this period canreduce the expression of dormancy, ascan cool conditions following rainfall(Plett and Larter, 1986; Trethowan,1995). The expression of dormancy isalso linked to seed-coat color (Gale,1989). Red-seeded wheats are generallymore dormant than white-seeded types.Crosses between dormant red and non-dormant white-seeded wheat canproduce dormant white-seeded progeny;however, the level of dormancy in theseprogeny is always lower than that of theoriginal red-seeded parent (DePauw andMcCaig, 1987). This suggests that theexpression of dormancy is governed byepistasis between the seed color anddormancy loci.

The seed coat or husk (in the case ofbarley) may also influence PHS becauseof the differences in water permeability(Trethowan et al., 1993). Thismechanism involves a physical barrierthat keeps water from entering the seed,thereby reducing the effects of PHSfollowing light rains.

Similarly, the bracts or floral structure ofthe wheat spike may physically impedethe entry of water into the grain. Thebracts may also chemically inhibitgermination through the release of water-soluble inhibitory chemicals (Trethowanet al., 1993). Some evidence suggeststhat awnless wheat has an advantageunder rainfall pressure because awnlessspikes shed water quickly (King, 1989).In contrast, awns collect water, therebymaintaining a higher level of humidity inthe spike.

The combination of grain dormancy withseed- or bract-based physical orchemical tolerances will greatly enhancethe overall tolerance of wheat to PHS.

Screening Methods andPhysiological Tools

It is difficult to effectively screen forPHS tolerance in the field because ofrainfall variability in mostenvironments. Variable maturity datescharacteristic of genetic materials inmost plant breeding programs alsoconfound interpretation of dormancy insegregating and advanced lines undernaturally occurring rainfall. Rain-simulation facilities have beendeveloped by some researchers toremove the confounding effects of theenvironment after physiologicalmaturity (Mares, 1989). When using rainsimulation, it is critical that all materialsbe harvested at the same stage ofdevelopment (harvest ripeness),stabilized at the same moisture content(usually 12%), and stored at lowtemperatures (-20ºC or lower) prior toevaluation in the rain simulator. Lowtemperature storage ensures that allenzymatic activity in the seed is halted.Spikes from plants with differentmaturity dates can then be evaluatedtogether in the rain simulator.

Temperature, humidity, and spikewetting are strictly controlled in thesimulator, and spikes are evaluated forvisible germination after a fixed numberof days. This method is very effectivefor identifying dormant progeny whendormancy is present in thephysiologically mature grain. However,the expression of dormancy at maturityis often suppressed by rainfall during thethree weeks prior to maturity. Someresearchers use rain shelters during thisperiod to protect field-grown plantsfrom the confounding effects of rainfall(Trethowan, 1995). However,temperature fluctuations during the laterstages of grainfilling cannot becontrolled in the field.

Rain simulators and rain shelters areeffective in controlling some of theenvironmental factors that influence theexpression of grain dormancy; however,such equipment is expensive to build or

buy, and many scientists cannot affordit. Grain dormancy, the primarymechanism of tolerance, can bemeasured more simply by hand-threshing mature grain and measuringgermination in a petri dish using filterpaper as the water-adsorbent medium.The rate of germination is thencompared to that of non-dormantremnant seed germinated in the sameway. Differences between the twotreatments indicate the possiblepresence or absence of dormancy. Seedshould be washed in a surface sterilantsuch as 20% chlorox solution and rinsedin distilled water prior to placement inthe petri dish (Trethowan et al., 1993).The effect of environmental fluctuationsbefore maturity can be minimized byplanting and evaluating the samematerial on several dates.

CIMMYT’s bread wheat breedingprogram, based in Mexico, uses a fieldscreening technique to evaluate manythousands of lines. Materials are sownin the field during the dry season inJanuary, which causes plants to ripen atthe height of the rainy season in July/August. Physiological maturity (PM) isscored, and spikes are harvested fromeach plot a fixed number of days postPM. Grains are then threshed andevaluated visually for germination. Thecorrelation between this technique andthe rain simulator is very high(Trethowan et al., 1996). Thismethodology is very much dependentupon the stability of the screeningenvironment. The CIMMYT test site issituated at high altitude (2600 masl),and rainfall is a daily event during cropmaturation time.

Overhead sprinkler irrigation of field-grown materials may also be effectivein inducing high levels of PHS damage(Trethowan et al., 1994). This methodhas the advantage of providing spikewetting at the most desirable times;however, maintaining effectivehumidity in the canopy to induce thedesired levels of germination postwetting is again a function of the testenvironment.

R.M. TRETHOWAN

147

Molecular markers linked to graindormancy are being developed. Onceavailable, these markers will greatlyfacilitate the development of PHStolerant wheat. The most obviousbenefit will be the ability to trackdormancy genes away from theconfounding effects of the environment.Furthermore, genetic engineeringtechniques may in future provide a wayof silencing the a-amylase genes,thereby providing an immediatesolution to this problem (Gale, 1989).The new “terminator” technology, nowsubject to a US patent, may also providea comprehensive solution to theproblem of PHS. This technology, stillin development, produces non-viableseed through the insertion of a lethal-gene and promotor which causes seeddeath during late embryogenesis(AgBiotech Reporter, 1998). Thismethod ensures that farmers will returnto the seed company each year to buytheir seed. The hidden benefit forfarmers in sprouting-prone areas is that,regardless of rainfall, the grain will notgerminate.

Some researchers use the Hagbergfalling number test (AACC, 1983) tomeasure enzymatic activity in theharvested grain. This test measures thelevel of starch degradation caused bygerminative enzymes and is thereforestrongly correlated to PHS. The test is arelatively fast and easy one; however,grain must be harvested in sufficientquantity and milled to produce aminimum of 7 g of flour before the testcan be implemented. Other dyed-substrate tests are available which linkcolor change in a starch substrate toamylase activity (Meredith andPomeranz, 1985). These tests are quickand easy to perform and require grainhalves or at most 1 g of ground wheat.

Other tools that can be used to evaluatePHS damage include the Amylograph(Brabender OHG, Germany: AACCMethod No. 22-10) and the Rapid ViscoAnalyzer (Dengate, 1984). These toolsare used by the cereal chemist tomeasure dough and starch pastingproperties of harvested wheat. The testsprovide the most comprehensiveassessments of PHS available; however,they require large quantities of flour,particularly in the case of theAmylograph, and are relatively time-consuming.

Conclusions

The methods for assessing PHStolerance described in this chapter havea wide range of potential uses. Theserange from breeders’ early generationtesting for grain dormancy usinggermination tests in petri dishes, to thecereal chemist’s comprehensivequantification of the effects of sproutingon physical dough properties using theAmylograph. Assessments can becarried out using expensive equipmentsuch as rain simulators and rain shelters,or can be more cheaply conductedthrough germination tests or dyed-substrate assays for detecting thepresence or absence of germinativeenzymes. The strong environmentalinfluence on the expression of graindormancy makes molecular markers afavorable option, particularly whenscreening for the more subtle variationsin dormancy expression found in white-grained wheats.

There are, therefore, a variety of optionsavailable to the researcher looking tosolve this intractable problem. Theseoptions will allow plant breeders toprogress towards developing new andbetter cultivars, regardless of thepractical limitations imposed upon themby their current resource base.

ReferencesAACC. 1983. Falling number determination.

Method 56-81B. 8th Edn. AmericanAssociation of Cereal Chemists:St. Paul, MN.

AgBiotech Reporter. 1998. Two Technologies inOpposite Directions. Vol. 15, No. 4. April,1998.

Dengate, H.N. 1984. Swelling, pasting and gellingof wheat starch, Adv. Cereal Sci. Technol.6:49.

DePauw, R.M., and T.N. McCaig. 1987. Recoveryof sprouting resistance from red-kerneledwheats in white-kerneled segregants. Proc.4th Int. Symp. Pre-harvest Sprouting inCereals. D.J. Mares (ed.). Westview Press:Boulder, CO.

Gale, M.D. 1989. The genetics of preharvestsprouting in cereals, particularly in wheat. InPreharvest Field Sprouting in Cereals. N.F.Derera (ed.). pp. 85-110. CRC Press, Inc.

King, R.W. 1989. Physiology of sproutingresistance. In Preharvest Field Sprouting inCereals. N.F. Derera (ed.). pp. 27-60. CRCPress, Inc.

Mares, D.J. 1989. Preharvest sprouting damageand sprouting tolerance: Assay methods andinstrumentation. In Preharvest FieldSprouting in Cereals. N.F. Derera (ed.). pp.129-170. CRC Press, Inc.

Plett, S., and E.N. Larter. 1986. The influence ofmaturation temperature and stage of kerneldevelopment on sprouting tolerance ofwheat and triticale. Crop Sci 26:804-7.

Trethowan, R.M., W.H. Pfeiffer, R.J. Pena, andO.S. Abdalla. 1993. Preharvest sproutingtolerance in three triticale biotypes. Austr Jof Agric Res 44:1789-1798.

Trethowan, R.M., R.J. Pena, and W.H. Pfeiffer.1994. Evaluation of preharvest sprouting intriticale compared to wheat and rye using aline source rain gradient. Austr J of AgricRes 45:65-74.

Trethowan, R.M. 1995. Evaluation and selectionof bread wheat (Triticum aestivum L.) forpreharvest sprouting tolerance. Austr J ofAgric Res 46:463-474.

Trethowan, R.M., S. Rajaram, and F.W. Ellison.1996. Preharvest sprouting tolerance ofwheat in the field and rain simulator. Austr Jof Agric Res 47:708-716.

PREHARVEST SPROUTING TOLERANCE

F1

F1

F1

F1

F1F1

F1

F1

F1

F1

F1

F1

F1

F1

F1

F2

F2

F2

F2

F2

F2

F2

F2

F2

F2

F2

F2

F3

F3

F3

F3

F3

F3

F3

F3

F4

F4

F4

F3

183

Genotypic Variation for ZincEfficiencyI. Cakmak1 and H.-J. Braun2

C HAPTER 1 6

Zinc deficiency is a commonmicronutrient deficiency in wheat andother cereals. It is estimated that about50% of soils used for cereal productionin the world have low levels of plantavailable Zn (Graham and Welch, 1996).Wheat shows substantial decreases ingrowth and grain yield under Zn-deficient field conditions (Graham et al.,1992; Cakmak et al., 1996a). Zincdeficiency in soils also reduces Znconcentration in wheat grain anddiminishes its nutritional quality.Approximately 40% of world’spopulation suffers from micronutrientdeficiencies (the so-called “hiddenhunger”), including Zn deficiency(Bouis, 1996; Graham and Welch, 1996).High consumption of cereal-based foodswith low Zn content is considered to beone of the major reasons for thewidespread occurrence of Zn deficiencyin humans, especially in developingcountries.

In bread and durum wheat, there is awide range of genotypic variation inresponse to Zn deficiency (Graham et al.,1992; Cakmak et al., 1997a; 1998). Suchlarge variation is promising fordeveloping wheat genotypes that makemore efficient use of Zn, whichconstitutes a sustainable solution to theproblem of Zn deficiency. Zinc-efficientgenotypes show better growth and highergrain yield under Zn-deficient conditionsthan other genotypes.

Breeding wheat for Zn efficiency is aprocess that involves taking into account

many inter-related and complex issues.This chapter gives a broad review ofthese issues by concentrating on: 1) theoccurrence, distribution, prediction, andcorrection of Zn deficiency, 2) the extentof genotypic variation for Zn efficiency,3) the description of mechanismsinvolved in the expression of Znefficiency, 4) methods useful in screeninggenotypes for Zn efficiency, 5) thegenetics of Zn efficiency, and 6) Zn andphytic acid bioavailability in grain.

Zinc efficiency, as used in this paper, isdefined as the ability of a genotype togrow and yield better than othergenotypes in soils deficient in Zn(Graham, 1984).

Distribution of Zinc-Deficient Soils

Zinc deficiency is widespread throughoutthe world, occurring in different climateregions and almost all countries(Sillanpää, 1982; Sillanpää and Vlek,1985). Usually, Zn deficiency in cropplants occurs either in acid and highlyleached soils with low Zn content or incalcareous soils with Zn that for the mostpart is not available to plants. Zincdeficiency occurs more frequently incalcareous soils with high pH such asthose found in arid and semiarid regions,and in the Mediterranean region. Based

on the analysis of 298 (Sillanpää andVlek, 1985) and 1511 (Eyüpoglu et al.,1994) soil samples, Zn deficiency hasbeen identified as the most widespreadmicronutrient deficiency in Turkey,particularly in Central Anatolia. Soils inthis region are highly alkaline, organicmatter content is low (mostly below1.5%), and annual precipitation is 300-400 mm (Cakmak et al., 1996a). Zincdeficiency has also been reported inAustralia (Graham et al., 1992) and India(Takkar et al., 1989) in a wide range ofsoil types and crop plants.3

Soil and ClimaticFactors Causing ZincDeficiency in Plants

Soil pH is the most decisive factoraffecting availability of Zn to plant roots.Increases in soil pH stimulate Znadsorption to surfaces of various soilconstituents, such as metal oxides andclay minerals; this results in substantialdecreases in solubility and, hence,reduced availability of Zn to plants(Brümmer et al., 1988; Barrow, 1993).High pH decreases desorption of Zn fromsoil surfaces, which also limitsavailability of Zn to plants (Dang et al.,1993). The desorption rate of Zn fromsoil surfaces is essential for maintaining acontinuous supply of Zn to plants.

1 Department of Soil Science and Plant Nutrition, Cukurova University, Adana, Turkey.2 Winter Wheat Breeding, CIMMYT, PK 39 Emek, 06511 Ankara, Turkey.3 For detailed information on the distribution of Zn deficiency throughout the world, see Welch et al.

(1991) and Takkar and Walker (1993).

184

Precipitation of Zn in high pH soils alsolowers Zn availability to plant roots. Athigh pH, Zn can precipitate in the formof ZnCO3, Zn(OH)2, and Zn2SiO4 (Maand Lindsay, 1993). Zinc concentrationin soil solution is, therefore, largelydependent on pH. At pH 5.0, theconcentration of Zn in soil solution isabout 10-4 M, and at pH 8.0 it is 10-10 M(Lindsay, 1991). Asher (1987) reportedthat adequate Zn concentration in soilsolution for a number of crop speciesranges between 6x10-8 M and 8x10-6 M.However, in most cases calcareous soilscannot maintain this level of Znconcentration in solution, whichincreases the risk of Zn deficiency.

Liming of acid soils to increase pHraises the risk of Zn deficiency in plants.An increase in soil pH from 5.2 to 6.8 byliming results in about a 10-folddecrease in Zn concentration in plants(Parker and Walker, 1986).

Other factors that contribute to Zndeficiency in plants are low organicmatter, low soil water regime, low soiltemperature, and high light intensity(Moraghan and Mascagni, 1991;Marschner, 1993). Zinc transport fromsoil solution to roots occurs mostly bydiffusion (Wilkinson et al., 1968). Sincediffusion of Zn in soils is highly

dependent on soil moisture content, Znnutrition of plants may be at risk in aridand semiarid regions where soils,particularly topsoil, are usually deficientin water for long periods during thegrowing season. Accordingly, it hasbeen shown that wheat yield reductionsin Zn-deficient calcareous soils aremore severe under rainfed than irrigatedconditions (Table 1; Ekiz et al., 1998).

Zinc diffusion rate is also influenced byorganic matter. An adequate level oforganic matter is believed to increasesolubility and Zn diffusion rate in soils(Sharma and Deb, 1988; Moraghan andMascagni, 1991). In field experimentswith wheat, Zn uptake was found to bepositively correlated with the level ofsoil organic matter (Sillanpää, 1982;Hamilton et al., 1993). In addition, theoccurrence of Zn deficiency in plantscan be accelerated in areas wheretopsoil is removed by land leveling,erosion, or terracing (Martens andWestermann, 1991).

Increased phosphorus fertilization caninduce Zn deficiency in wheat. Abroadcast application of 160 kg P ha-1

significantly reduced Zn concentrationin wheat (Wagar et al., 1986). Highsupply of P can reduce colonization ofroots by vesicular-arbuscular (VA)mycorrhizae and, hence, Zn uptake by

roots in wheat (Singh et al., 1986). Therole of VA mycorrhizae in Zn uptake inwheat plants is well documented(Swamvinathan and Verma, 1979;Marschner, 1993). High P supply canalso induce Zn deficiency by decreasingthe physiological availability of Zn atthe cellular level (Cakmak andMarschner, 1987).

Zinc deficiency is more prevalent underlow soil temperature conditions. Thiseffect is due to reductions in rootgrowth, VA mycorrhizal infection, Znuptake by roots, and Zn translocationinto shoots (Moraghan and Mascagni,1991). High light intensity is anotherclimatic factor that promotes thedevelopment of Zn deficiency symptomsin wheat, particularly in durum wheat(Cakmak et al., 1996b). This light effectcan be explained by Zn deficiency-induced photooxidation in leaves, andtherefore it is pronounced at low airtemperatures (Cakmak et al., 1995).

Predicting ZincDeficiency through SoilAnalyses

As reviewed by Sims and Johnson(1991), several soil testing proceduresare available for predicting Znavailability in soils and response ofplants to Zn fertilization. Among thesetests, the DTPA (diethylenetriaminepentaacetic acid) method is widely usedfor predicting plant-available Zn in soils,particularly in calcareous soils (Lindsayand Norvell, 1978). Increases in DTPA-extractable Zn in soils from 0.1 mg kg-1

soil to 0.6 mg kg-1 soil are associatedwith increases in grain yield and Znconcentration in plants (Sillanpää, 1982;Cakmak et al., 1996a). Fieldexperiments at six different locations inCentral Anatolia have shown that wheatgrown in calcareous soils containing less

Table 1. Effects of different Zn applications on grain yield of wheat cultivars grown ina Zn-deficient calcareous soil under rainfed and irrigated conditions. +Zn refers to the meanof the grain yields obtained with Zn applications of 7, 14 and 21 kg Zn ha-1, as thedifferences between the Zn doses were not significant.

Rainfed IrrigatedGrain yield Increase Grain yield Increase

Cultivars - Zn +Zn in yield - Zn +Zn in yield

kg ha-1 % kg ha-1 %Bread wheatGerek 79 3100 4460 44 4070 5250 29Bezostaja-1 1900 4150 118 3190 5080 59Durum wheatKunduru 1149 330 2470 648 1550 3580 130Cakmak 79 170 760 347 630 2870 355

Source: Ekiz et al. (1998).

I. CAKMAK AND H.-J. BRAUN

185

than 0.4 mg kg-1 DTPA-extractable Zncan respond to soil Zn applications (23 kgZn ha-1) with increases in grain yield(Cakmak et al., 1996a). In 20 fieldexperiments on farmers’ fields in India,Zn fertilization of wheat at the rate of 5kg Zn ha-1 enhanced grain yield inlocations where DTPA-extractable Znwas below 0.6 mg kg-1 soil (Dwivedi andTiwari, 1992). In contrast, Zn fertilizationhad no significant effect on grain yield ofwheat and barley in Saskatchewan insoils containing less than 0.5 mg kg-1 soilof DTPA-extractable Zn (Singh et al.,1987).

These contrasting results may be due todifferences in soil conditions or genotypicdifferences in root-induced changes in Znsolubility and uptake (see “Screening forZinc Efficiency” on p. 188). According toBrennan (1996), the critical DTPA-extractable Zn concentration for wheat-growing acidic soils in Australia is 0.25mg kg-1; Zn application to soils havingmore than 0.25 mg extractable Zn perkilogram of soil is not effective forincreasing wheat yields.

Correcting ZincDeficiency

Soil applications of zincZinc deficiency can be corrected byapplying Zn to soils or plant foliage aswell as by treating seeds with Zn. Bothorganic and inorganic Zn fertilizers areused to correct Zn deficiency. Zincsulphate (ZnSO4) is the most commonsource of Zn fertilizer because of its highsolubility in water, existence in bothcrystalline and granular forms, and itslow cost compared to synthetic Znchelates such as ZnEDTA (Martens andWestermann, 1991; Mordvedt and Gilkes,1993). In most instances, Zn deficiency inplants can be corrected by broadcastapplications of Zn at 4.5 to 34 kg ha-1 as

ZnSO4 (Martens and Westermann, 1991).Differences in rates of Zn applicationdepend mainly on variations in DTPA-extractable Zn in soils and the severity ofZn deficiency symptoms. Ekiz et al.(1998) found that broadcast applicationsof ZnSO4 at the rate of 7 kg Zn ha-1 toseverely Zn-deficient calcareous soilssignificantly enhanced wheat yield, butthat increasing applied Zn from 7 to 21 kgha-1 did not result in an additionalincrease in grain yield. Compared to ZnO,ZnSO4 seems to be more effective inimproving Zn nutrition of wheat plantsunder deficient supply of Zn (Sharma atal., 1988).

Foliar applications of zincFoliar applications of Zn are effective incorrecting Zn deficiency in plants. Asreported by Martens and Westermann(1991), 0.5 to 1.0 kg Zn ha-1 as ZnSO4 or0.2 kg Zn ha-1 as ZnEDTA is often used tocorrect Zn deficiency in plants.Depending on growth stage of wheat,foliar applications of ZnEDTA and ZnSO4at the rate of 400-450 g Zn ha-1 are eitherequally effective or ZnEDTA is superiorto ZnSO4 in correcting Zn deficiency infield-grown wheat (Brennan, 1991).

However, soil applications of Zn aremore effective than foliar applications incorrecting Zn deficiency (Yılmaz et al.,1997). In field experiments with fourwheat cultivars grown in Zn-deficient soil(DTPA-Zn: 0.1 mg kg-1 soil) different Znapplication methods were tested for theireffectiveness in correcting Zn deficiency(Table 2) (Yılmaz et al., 1997). Thehighest increase in grain yield wasobtained from soil, soil+leaf, andseed+leaf applications, while seed onlyand leaf only applications were lesseffective. Soil application alone wasconsidered the most economical methodon a long-term basis. Soil+leafapplications can be effective when bothhigh grain yield and high Znconcentration in grain are desired(Yılmaz et al., 1997). Since Zn applied tosoils has a great residual effect, it is notnecessary to apply Zn every year(Martens and Westermann, 1991).

Sowing seeds with higherzinc contentThere is increasing evidence that sowingseeds containing higher amounts of Zncan be a practical solution to alleviatewheat yield depression under Zn-deficient conditions. In greenhouse

GENOTYPIC VARIATION FOR ZINC EFFICIENCY

Table 2. Effects of different zinc application methods on grain yield of bread wheat cultivarsGerek-79, Dagdas-94, and Bezostaja-1 and durum wheat cultivar Kunduru-1149 grown in aZn-deficient calcareous soil.

Application Increase by Znmethods Gerek-79 Dagdas-94 Bezostaja-1 Kunduru-1149 Mean application

kg ha-1 %Control† 738 633 805 56 558 -Soil‡ 2700 2225 2340 903 2042 265Seed§ 2052 1997 1958 772 1695 204Leaf# 1472 1365 1555 617 1253 124Soil + Leaf†† 2712 1955 2330 818 1954 250Seed + Leaf‡‡ 2768 2100 2380 987 2059 268

† Control (no Zn application).‡ 23 kg Zn ha-1 as ZnSO4.§ 1 liter of 30% ZnSO4 for 10 kg seed.# 2x220 g Zn ha-1 as ZnSO4 in 450 liters during tillering and stem elongation.†† Combination of methods 2 and 4.‡‡ Combination of methods 3 and 4.Source: Yılmaz et al. (1997).

186

experiments with low seed-Zn (250 ngZn per seed) and high seed-Zn (700 ngZn per seed), wheat plants derived fromthe high seed-Zn have better seedlingvigor and grain yield (Rengel andGraham, 1995a,b). In a field with Zn-deficient soils, wheat seeds with low Zncontent (355 ng Zn seed-1) had asignificantly lower grain yield comparedto seeds with medium (800 ng Zn seed-1)and high (1465 ng Zn seed-1) Zn content(Table 3).

Grain yield differences of 18% and 116%are of such magnitude that genotypicdifferences in Zn efficiency are maskedwhen seeds with great differences in Zncontent are used. It is thereforeabsolutely crucial to use seeds from thesame source or at least with similar Zncontent in experiments related toscreening genotypes for Zn efficiency.However, high seed-Zn content could notmatch the effect of soil Zn application ongrain yield (Table 3). Nevertheless, theuse of seed with high Zn content couldprovide a practical solution to theproblem, especially where farmers arenot aware of Zn deficiency, and Znapplications are not practiced.

Cellular Functions ofZinc

In Zn-deficient plants, a number ofcritical cellular functions and processesare impaired, leading to severe reductionsin growth and development (Brown et al.,

1993). The adverse effects of Zndeficiency occur more distinctly inmeristematic tissues, where intensivecell division and elongation take place.There is a high specific requirement forZn in meristematic tissues, especiallyfor maintenance of protein synthesis(Kitagishi et al., 1987; Cakmak et al.,1989).

More than 300 enzymes are Zndependent (Coleman, 1992). Zinc playscatalytic, co-catalytic, and structuralroles in these enzymes. The enzymesuperoxide dismutase (SOD) hasattracted much attention in recent years.An SOD isoenzyme, copper-zinc SOD(CuZn-SOD), is required for protectingplant cells against toxic superoxideradical (O2

.-); activity of this enzyme isvery low in plants under Zn-deficientconditions (Cakmak and Marschner,1988b,c; 1993). Measurement of CuZn-SOD is suggested as a tool forscreening cereals for Zn efficiency (see“Internal utilization” on p. 192).

Zinc plays a decisive role in DNA andRNA metabolism, chromatin structure,and gene expression (Vallee andFalchuk, 1993). Several Zn-containingproteins are known to be involved inDNA replication and gene transcriptionprocesses (Klug and Rhodes, 1987).The role of Zn in regulating geneexpression is currently an excitingresearch area.

The structural and functional integrity ofcellular membranes is greatly affected bythe lack of Zn, which plays bothstructural and protective roles inmembrane integrity (Welch et al., 1982).Zinc can stabilize membranes by bindingto phospholipids and sulphydryl groupsof cell membranes, thereby protectingthese compounds against oxidativedamage (Marschner, 1995; Cakmak andMarschner, 1988b,c; Welch and Norwell,1993; Rengel, 1995a). When thesecompounds undergo oxidative damage,membrane integrity of Zn-deficient plantcells is impaired, and leakage of ionsfrom Zn-deficient root cells is increased(Welch et al., 1982). Several plantspecies, including wheat, showed adramatic increase in the exudation ofseveral organic compounds, such ascarbohydrates and amino acids, under Zndeficient conditions (Table 4; Cakmakand Marschner, 1988a). Due to theincreased leakage of carbon containingcompounds in the rhizosphere, Zn-deficient plants may be susceptible toroot diseases such as Fusariumgraminearum (Sparrow and Graham,1988), Gaeumannomyces graminis(Brennan, 1992), and Rhizoctonia solani(Thongbai et al., 1993).

Zinc deficiency also affects metabolismof phytohormones, especiallyindoleacetic acid (IAA). Theconcentration of IAA is reduced by Zndeficiency, and this reduction isaccompanied by decreases in shootelongation (Skoog, 1940; Cakmak et al.,

Table 4. Effect of zinc nutritional status ofwheat plants on root exudation of aminoacids, sugars, and phenolics.

AminoZn supply acids Sugars Phenolics

µ g g-1 root dry wt.6h-1

- Zn (deficient) 48±3 615±61 80±6+ Zn (sufficient) 21±2 315±72 34±6

Source: Cakmak and Marschner (1988a).


Table 3. Effect of seed zinc content on grain yield in two cropping seasons under rainfedconditions in a Zn-deficient soil with (+Zn=23 kg Zn ha-1) and without Zn (-Zn) fertilization.

1994-1995 cropping season 1995-1996 cropping season

Seed Zn content -Zn +Zn -Zn +Zn

ng seed-1 kg grain ha-1

355 480 2720 2490 3180800 920 3170 2930 31801465 1040 2840 2950 3220

Source: Yılmaz et al. (1998)

187

1989). According to Suge et al. (1986), besidesIAA, concentration of gibberelin-like substancesare also decreased in Zn-deficient plants.

Diagnosing Zinc Deficiency

Visible symptoms of zinc deficiencyDiagnosing Zn deficiency based on visiblesymptoms is not always easy. Symptomdevelopment is highly dependent on climaticconditions and varies greatly among plantspecies. In some instances, diagnosis of Zndeficiency can be complicated by thesimultaneous occurrence of toxicity of a nutrientsuch as P (Webb and Loneragan, 1988), intenselight-induced chlorophyll damage, (Marschnerand Cakmak, 1989; Cakmak et al., 1995), andvirus infection (Bergmann, 1992).

The first and most characteristic reaction of wheatplants to Zn deficiency is a decrease in shootelongation and leaf size. These symptoms arefollowed by development of whitish-brownpatches and then necrotic lesions on the leafblades, predominantly in the middle and olderleaves (Picture 1; Cakmak et al., 1996a,b). Asseverity intensifies, necrotic lesions on leavesspread, and leaf blades often collapse in themiddle and take on a “scorched” appearance. Inmost cases, the basal part of those same leavesremains green. Young leaves are small in size andbecome yellowish green in color, but show nonecrotic lesions. Similar visible symptoms of Zndeficiency in wheat have been reported by Danget al. (1993) and Rengel and Graham (1995c).

The development period and severity of Zndeficiency symptoms vary greatly between durumand bread wheats (Picture 2). Decreases in shootelongation and appearance of necrotic patches onleaves occur more rapidly and severely in durumwheats than in bread wheats (Cakmak et al.,1994, 1997a; Rengel and Graham, 1995 c,d).

Critical zinc concentrations in plantsCritical Zn concentration is influenced by plantdevelopment stage and leaf age (Table 5). In mostcases, the critical Zn concentration in leaves or

Picture 1. Development of Zn deficiency necrotic lesions on wheat leaves.

Picture 2. Growth of different bread and durum wheat cultivars in nutrientsolution (above) or Zn-deficient calcareous soil (below) without Zn supply.


188

whole shoots varies considerably amongcereals and among cultivars of a givencereal, such as wheat. Under Zn-deficientconditions, wheat cultivars show distinctdifferences in sensitivity to Zn deficiencyand in yield reductions caused by thedeficiency. However, such genotypicdifferences are not related to Znconcentration in leaves or in shoots(Graham et al., 1992; Cakmak et al.,1997a; 1998). Total Zn concentration istherefore not always a reliable parameterfor diagnosing Zn deficiency in wheat.Apparently, genotypes containing similartotal Zn concentration under Zn-deficientconditions may differ in use of Zn at thecellular level. Therefore, measuring theactivity of Zn-containing enzymes can bea better tool for diagnosing the Znnutritional status of plants, as shown inwheat by measuring the activity ofcarbonic anhydrase (Rengel, 1995b) andsuperoxide dismutase (Cakmak et al.,1997b) in leaves.

Measuring Zn in grains is suggested to beindicative of Zn nutritional status ofplants. According to Rashid and Fox(1992) the critical Zn concentration inwheat is 17 mg kg-1 dry weight forrecently maturated leaves of seedlingsand 15 mg kg-1 dry weight for grains.Viets (1966) suggested that the criticallevel of Zn in grains is 15 mg kg-1 dryweight. Several authors (Graham andRengel, 1993; Cakmak et al., 1997a)

have found that Zn concentration in grainis not always a good indicator of the Znnutritional status of plants under Zn-deficient conditions.

Screening for ZincEfficiency

Currently, the development of genotypeswith high Zn efficiency is attractingincreasing interest worldwide. Zinc-efficient genotypes may provide anumber of benefits, such as reductions inthe use of fertilizers, improvements inseedling vigor, resistance to pathogens,minimization of yield losses, increasedyields, and enhancement of grainnutritional quality (Graham and Rengel,1993; Bouis, 1996; Graham and Welch,1996).

Genotypes are usually selected for Zn useefficiency based on the expression of Zndeficiency symptoms in leaves anddecreases in shoot dry matter or grainyields. Zinc efficiency can be calculatedas the ratio of yield (shoot dry matter orgrain yield) produced under Zndeficiency (-Zn) to yield produced withZn fertilization (+Zn), as follows(Graham, 1984):

[Zn efficiency = (yield at –Zn/yield at +Zn) x 100]

When comparing Zn-efficiency ratios ofcultivars with similar yield capacity

under Zn-deficient conditions, it shouldbe kept in mind that cultivars with thesmallest response to Zn fertilizer willhave the highest Zn-efficiency ratio.Cultivars with low response to Znapplication are often landraces (M.Kalayci et al., unpublished).

Various mechanisms have been studiedto explain differences in Zn efficiency.The results reported in the literaturesuggest that more than one mechanismis involved in the expression of Znefficiency in plants (Graham andRengel, 1993; Cakmak et al., 1998). Abetter understanding of themorphological, physiological, andgenetic bases of Zn efficiency isrequired for developing rapid andreliable screening procedures to be usedin identifying and breeding genotypeswith high Zn efficiency (Table 6).

Leaf symptomsZinc efficiency of genotypes can bedetermined based on phenotypicmanifestations such as severity of visibledeficiency symptoms on leaves. Zincdeficiency symptoms seem to be auseful criterion to identify Zn-efficientgenotypes. The severity of zincdeficiency symptoms (i.e., intensity ofwhitish-brown necrotic patches onleaves, Picture 1) can be evaluated usinga scale from 1 (severe symptoms) to 5(slight or no symptoms). Actually, agood correlation between the severity(score) of Zn deficiency symptoms andcalculated Zn efficiency ratios has beenfound in cultivars of wheat, barley, oat,and rye (Cakmak et al., 1997b; Schlegelet al., 1998). However, caution shouldbe exercised when assessing Znefficiency based on leaf symptoms (see“Visible symptoms of zinc deficiency”on p. 187), and using them together withother criteria such as the Zn efficiencyratio and total Zn content isrecommended (see below). Scoring Zndeficiency symptoms is quick and easy,

Table 5. Critical concentrations of zinc in different tissues and development stages of wheat.

Leaf Stage Zn concentration Referencemg kg-1 dry wt

Youngest leaf blade tillering 11 Reuter and Robinson (1986)Youngest leaf blade post anthesis 7 Reuter and Robinson (1986)Youngest leaf anthesis 16 Dong et al. (1993)Youngest leaf tillering 17 Riley et al. (1992)Youngest leaf milk stage 7 Riley et al. (1992)Mature leaf - 17 Rashid and Fox (1992)Whole shoots tillering 10-15 Graham et al. (1992)Whole shoots tillering 10-15 Cakmak et al. (1997a)Grains mature 15 Viets et al. (1966)Grains mature 15 Rashid and Fox (1992)


189

and allows screening large numbers ofgenotypes with no special equipment.

Shoot and root growthIn Zn-deficient nutrient solution, Zn-efficient wheat cultivars are characterizedby higher shoot dry matter production anda better shoot:root dry weight ratiocompared to Zn-inefficient cultivars(Rengel and Graham, 1995c; Cakmak etal., 1996b). However, we found that Zn-efficient genotypes have lower shoot:rootdry weight ratios than Zn-inefficientgenotypes in Zn-deficient calcareous soils(Cakmak et al., unpublished results).

Genotypic differences in zincuptakeIn general, Zn-efficient wheat genotypesshow either enhanced Zn uptakeefficiency by roots, or enhanced Znutilization efficiency within the plant, orboth. Enhanced Zn uptake might beaffected by root surface area, rootcolonization by VA mycorrhizae,reduction in rhizosphere pH, release ofZn-mobilizing compounds such asphytosiderophores (PS) from roots, andinduction of polypeptides involved in Znuptake and transport across the plasmamembranes.

In field experiments on Zn-deficientcalcareous soils, Zn efficiency ispositively correlated with total amount(uptake) of Zn in shoots (Graham et al.,1992; Cakmak et al., 1997a). Totalamount of Zn per plant can becalculated as:

Total plant dry weight (kg) x Zn concentration (mg Znkg–1 dry weight)

Under deficient supply of Zn in nutrientsolution, bread wheat cultivars withfewer Zn deficiency symptomscontained about 42% more Zn per shoot(or 29% more Zn per plant) than durumwheat cultivars showing very severe Zndeficiency symptoms (Cakmak et al.,1996b). Despite their higher Znaccumulation capacity per plant, Zn-efficient genotypes do not contain moreZn per unit dry weight of plants (see“Zinc concentration in tissues” on p.190). However, under sufficient Znsupply, Zn-efficient and Zn-inefficientcultivars are not different in total Znuptake (Cakmak et al., 1997a). Recently,Rengel and Wheal (1997b) clearlydemonstrated in short-term uptakeexperiments that differences in Znefficiency between wheat cultivars are

closely related to differences in Znuptake capacity (Table 7). Increases in Znuptake capacity of Zn-efficient genotypescan also be demonstrated by measuringuptake of labelled Zn (65Zn) per root dryweight in nutrient solution experiments(Cakmak et al. 1997b; Rengel et al.,1998).

Root-to-shoot zinctranslocationUnder Zn-deficient conditions, Zn-efficient genotypes translocate more Znfrom roots to shoots than Zn-inefficientgenotypes, but not under Zn-sufficientconditions (Rengel and Graham, 1995d;Cakmak et al., 1996b). The ability ofgenotypes to translocate Zn from roots toshoots can be estimated as mg Zn shoot-1/mg Zn whole plant-1 or µg Zn shoot-1/groot dry wt.

In rye, which is very Zn-efficient, shootgrowth and grain yield are not affectedby Zn deficiency or Zn fertilization(Cakmak et al., 1997a; 1998). The highZn efficiency of rye is closely related toits ability to absorb Zn and translocate itto shoots at much greater rates than othercereals (Cakmak et al., 1997a, b).

These results indicate that enhancementsof Zn uptake by roots and itstranslocation into shoots under deficientsupply of Zn are of major importance forexpression of Zn efficiency. Determiningtotal Zn uptake per shoot or per plant is

Table 6. Plant traits used in screening wheat genotypes for zinc efficiency.

GrowthPlant trait conditions† Reference

Scoring Zn deficiency symptoms N, G, F Cakmak et al. (1997a, b), Schlegel et al. (1997)Shoot/root dry weight N Rengel and Graham (1995c); Cakmak et al. (1996b)Fine roots with diameter ≤0.02 N, G Dong et al. (1995); Rengel and Wheal (1997a)Synthesis of a 34-kDa polypeptide N Rengel and Hawkesford (1997)Zn amount (content) per shoot F Graham et al. (1992); Cakmak et al. (1997a)Zn uptake (content) per shoot G Cakmak et al. (1997b, c)Zn uptake (content) per plant N Cakmak et al. (1996b)Zn uptake per root dry weight N Cakmak et al. (1997b); Rengel and Wheal (1997b)Zn translocation into shoot N Rengel and Graham (1995d); Cakmak et al. (1996b)65Zn translocation into shoot N Cakmak et al. (1997b)SH-groups of plasma membranes N Rengel (1995b)Release of phytosderophores N Cakmak et al. (1994); Walter et al. (1994)Carbonic anhydrase N Rengel (1995a)Superoxide dismutase G Cakmak et al. (1998)

† N: Nutrient solution, G: Greenhouse, F: Field.

Table 7. Maximum rate of net zinc uptake(Imax) by wheat cultivars Excalibur (Zn-efficient bread wheat), Gatcher (moderatelyZn-efficient bread wheat), and Durati (Zn-inefficient durum wheat) grown in nutrientsolution.

Cultivar Zn uptake rate, Imax

µg Zn g-1 root dry wt.h-1

Excalibur 0.36±0.04Gatcher 0.24±0.03Durati 0.10±0.02

Source: Rengel and Wheal (1997b).


190

recommended in screening studies. Zincuptake capacity of genotypes is notalways a good parameter for determiningZn efficiency when calculated per rootdry weight instead of per plant (Rengeland Graham, 1995d).

Zinc concentration in tissuesAlthough Zn-efficient genotypes possesshigher Zn uptake capacity, they do notnecessarily have higher Zn concentration(amount of Zn per unit dry weight) inleaf or shoot tissue, or grain (Graham etal., 1992). Zinc-inefficient wheatgenotypes may even contain higher Znconcentrations in leaves or grains thanZn-efficient genotypes (Rengel andGraham, 1996; Cakmak et al., 1997a, b;1998). Enhanced Zn uptake by efficientgenotypes under Zn deficiency improvesdry matter production and results incorresponding decreases of Znconcentration in tissues to concentrationssimilar to those present in Zn-inefficientgenotypes (“dilution by growth,”Marschner, 1995). In Zn-deficient soil,the most Zn-efficient rye has lower Zntissue concentration (mg Zn kg-1 tissue)at heading than a Zn-inefficient durumwheat. However, total amount of Zn pershoot is about four times higher in ryethan in durum wheat, since rye producesfour times more shoot dry matter yieldunder Zn deficiency (Cakmak et al.,1997a). Therefore, we do not recommendusing Zn concentration in leaves, shoots,or grains as criteria to select for Zn-efficient wheat cultivars.

Root morphologicalpropertiesMycorrhizae play an important role in Znuptake (Marschner, 1993; 1995).However, to our knowledge, genotypicdifferences in mycorrhizal colonizationunder Zn-deficient conditions are notreported in the literature. We recentlyshowed that Zn-efficient and Zn-inefficient wheat cultivars grown in Zn-deficient soils do not differ in

mycorrhizal colonization of roots atflowering at 0-30 cm soil depth(unpublished results).

Dong et al. (1995) and, more recently,Rengel and Wheal (1997a) showed thatZn-efficient wheat cultivars possess moreroot surface area and a greater proportionof fine roots (≤0.02 mm in diameter) thanZn-inefficient wheat cultivars (Table 8).Obviously, a greater proportion of fineroots in the total root biomass wouldallow more efficient contact with soilconstituents, which in turn wouldcontribute to efficient Zn uptake. Dong etal. (1995) and Rengel and Wheal (1997a)used the same bread wheat and durumwheat cultivars in their studies. Testingmore bread and durum wheat cultivarsand alien species is necessary before anyroot properties can be recommended foruse as selection criteria in an appliedbreeding program.

Membrane effects on zincuptakeIncreased Zn uptake capacity of Zn-efficient genotypes has been related toinduced de novo synthesis of the 34-kDapolypeptide in plasma membranes of rootcells (Rengel and Hawkesford, 1997).Synthesis of this polypeptide occurs onlyunder Zn deficiency conditions in Zn-efficient bread wheats, but not in Zn-

inefficient durum wheats. There is somecorrelation between increase in Zn uptakeand increase in 34-kDa polypeptidesynthesis in the presence of Zndeficiency. Rengel and Hawkesford(1997) suggest that this polypeptidemight be a structural or regulatory unit ofa putative plasma membrane Zntransporter and may therefore beconnected with the Zn uptake process.There is promise in the use of suchspecific polypeptides for screeninggenotypes for Zn efficiency.

The higher Zn uptake capacity of Zn-efficient genotypes might be attributableto a greater amount of sulfhydryl groupsin root-cell plasma membranes,particularly in ion transport-relatedproteins (Rengel, 1995a; Rengel andWheal, 1997b). Maintenance of higherlevels of sulphydryl groups in iontransport proteins or ion channels ofplasma membranes is consideredimportant for Zn absorption into cells(Kochian, 1993; Welch, 1995).

PhytosiderophoresRelease of phytosiderophores (PS) maybe an adaptive response of graminaceousspecies to Zn and Fe deficiency.Phytosiderophores, also calledphytometallophores (Welch, 1995), arenon-proteinogenic amino acids releasedfrom roots of graminaceous species underdeficiency of Fe (Takagi, 1976;Marschner et al., 1986) and Zn (Zhang etal., 1989). Phytosiderophores releasedfrom roots are highly effective incomplexing and mobilizing Zn incalcareous soils (Treeby et al., 1989);they are involved in mobilizing Zn fromroot apoplast of wheat plants (Zhang etal. 1991) and, possibly, in solubility andlong-distance translocation of Zn withinplants (Figure 1; Welch, 1995).

Under Fe deficiency, the rate of PSrelease differs between and within cerealspecies. This release is closely related to

Table 8. Length of roots (diameter ≤0.2mm) and average root surface area of 24-day-old wheat cultivars Excalibur (Zn-efficient bread wheat), Gatcher (moderatelyZn-efficient bread wheat), and Durati (Zn-inefficient durum wheat) grown in nutrientsolution.

Average rootCultivar Root length surface area

m plant-1 mm2 plant-1

Excalibur 2.1 3350Gatcher 1.6 2300Durati 1.3 1950

Source: Data calculated from Rengel and Wheal (1997a).


191

genotypic differences in tolerance to Fedeficiency in cereals (Takagi et al., 1984;Marschner et al., 1986, Jolley and Brown,1989). Differences in Zn efficiencybetween durum and bread wheat cultivarsare closely correlated to the rate of PSrelease (Table 9; Figure 2; Cakmak et al.,1994; Walter et al., 1994). Under Zndeficiency, the rate of PS release is about6-8 times lower in Zn-inefficient durumwheats than in Zn-efficient bread wheats(Table 9; Cakmak et al., 1996c). Zinc-efficient cultivars also contain higheramounts of PS (mainly deoxymugineicacid) in root tissues under Zn deficiency(Figure 2). Enhanced PS synthesis andincreased PS release from roots havebeen suggested as mechanisms that helpwild grasses to adapt to severely Zn-deficient calcareous soils (Cakmak et al.,1996d).

Based on these results, selection ofgenotypes with high capacity tosynthesize and release PS appears to be apromising screening method. However,PS release rate is not always positivelyrelated to differences in Zn efficiencyamong bread wheat genotypes (Erenogluet al., 1996; Cakmak et al., 1997b). Inaddition, the rate of PS release of ryecultivars is not significantly higher thanthat of Zn-efficient or Zn-inefficientbread wheat cultivars (Table 9). Wheatreleases PS 2’-deoxymugineic acid(DMA) and rye releaseshydroxymugineic acid (HMA) (Mori,1994; Cakmak et al., 1996d).Considering the exceptionally high Znefficiency of rye (Cakmak et al., 1997a),it can be argued that compared to otherPSs, HMA may be more efficient in Znsolubilization and mobilization both inthe rhizosphere and within plants. Thispoint needs to be further investigated.

Despite insignificant differences in theircapacity to release PS, Zn-efficient andZn-inefficient genotypes may differ intheir capacity to absorb either Zn-complexed PS or Zn from Zn-PScomplexes (Figure 1). Recently it wasshown that the rate of PS release of twomaize genotypes did not differ, while theiruptake and root-to-shoot translocation ofZn-complexed PS (Zn-DMA) differedgreatly (von Wiren et al., 1996).

The importance of PS in enhancing Znuptake has been questioned, since PSsshow greater affinity for Fe than Zn, andFe exists in soils in larger amounts thanZn (Murakami et al., 1989; Ma andNomoto, 1996). However, it should bestressed that in many instances smallchanges in Zn concentration in planttissues (around 1 to 2 mg Zn kg-1 dryweight) seem to be decisive in improvingplants under Zn-deficient conditions(Jones, 1991; Cakmak et al., 1997a). Thus

Table 9. Effect of zinc deficiency on the rate ofphytosiderophore release from roots of variouscereals grown for 14 days in nutrient solutionwithout Zn supply.

Leaf symptoms Release ofCereals of Zn deficiency† phytosiderophores

µmol 48 plants-1 3h-1

Secale cerealeAslım 5 11.4±2.3

TriticalePresto 5 8.0±3.3

Triticum aestivumDagdas-94 3 9.3±1.8Gerek-79 3 9.2±1.1BDME-10 2 8.1±1.8Partizanka Niska 2 5.5±1.0Bul-63-68-7 2 7.2±2.4

Triticum durumKızıltan-91 1 1.7±0.1Kunduru-1149 1 1.2±0.1

† Leaf symptoms of Zn deficiency noted in Zn deficientcalcareous soils in Central Anatolia: 1= very severe,2=severe, 3= mild, 4= slight and 5= very slight or absent.

Source: Cakmak et al. (1997b).

Soil Root

Shoot

Cellmembrane

TR: Transporter protein

CytoplasmPhytosiderophores(PS)

ZnII-PS

ZnII-PS

ZnII-PS

Zn+2

ZnII-PSZn+2

Zn

ZnZn

Zn

Zn

Zn Zn

TRFigure 1. Model for release ofphytosiderophores (PS) and uptakeof Zn in graminaceous species.Source: Based on von Wiren et al. (1996) andCakmak et al. (1998).

Figure 2. High pressure liquid chromatograms of root exudates and root extracts ofZn-deficient bread wheats Gerek-79 and Kırac-66 and Zn-deficient durum wheatsKızıltan-91 and Kunduru-1149.Source: Cakmak et al. (1996b).

Root Extracts

Root Exudates


192

just a small contribution of PS to Znuptake by plants can greatly affect plantgrowth under Zn deficiency. Asdiscussed below, PS can also influenceZn utilization at the cellular level.

In conclusion, measuring PS release inroot exudates and also Zn uptake by theroots from Zn-PS complexes or directlyin the form of the Zn-PS complex shouldbe considered in evaluating genotypesfor Zn efficiency. Collecting rootexudates and measuring PS in rootexudates are easy to do and can beapplied to plants grown in nutrientsolution. The rate of PS release can bemeasured either directly by using highpressure liquid chromatography (HPLC)or indirectly using Zn loaded resin (formethods see Zhang et al., 1989; Cakmaket al., 1994, 1996c; Walter et al., 1994).

Internal utilizationAs mentioned before, the amount of Znper unit dry weight of shoots or leaves isnot different between Zn-efficient andZn-inefficient genotypes under Zn-deficient conditions. Therefore, it can beargued that genotypic differences in Znefficiency might also be related todifferences in the internal utilization ofZn. Zinc-efficient genotypes mightcontain a higher proportion of Zn thatreadily participates in metabolicreactions and binds to critical cellcompounds, such as Zn-requiringenzymes. Thus under Zn-deficientconditions and at same or similar Znconcentrations in leaves, Zn-efficientwheat genotypes demonstrate higheractivity of Zn-requiring enzymes thanZn-inefficient wheat genotypes, asshown for carbonic anhydrase (Rengel,1995b) and superoxide dismutase (SOD)(Cakmak et al., 1997b). These resultsindicate better Zn utilization at thecellular level in Zn-efficient genotypes.Higher SOD activity in Zn-efficientgenotypes may better protect them

against oxidative attack by toxic O2radicals. Despite time- and equipment-related constraints, enzyme analysescan contribute valuable information togaining a better understanding ofefficiency mechanisms and identifyinggenotypes with high Zn efficiency.

The reasons for higher Zn utilization inZn-efficient genotypes are not known.Zinc-efficient genotypes may possesshigher amounts of chelators that bindZn and increase its physiologicalavailability at the cellular level.Examples of such Zn chelators in planttissues are PS, nicotianamine, and S-containing amino acids, such ashistidine and methionine (Stephan etal., 1994; Welch, 1995; Cakmak et al.,1998). These compounds may beimportant in Zn translocation fromroots into shoots or in Znretranslocation from older tissues toapical shoot meristems. Further studiesare needed to clarify the relevance ofthese chelators to the expression of Znefficiency.

Considerations for field andpot screeningSelection under field conditions fortolerance to low levels of availablemicronutrients is in general moredifficult than selection for tolerance tomicronutrient toxicities. This is becausetoxicities are likely to occur every year,whereas micronutrient deficiencies—including Zn—are not due to theabsolute lack of a micronutrient, butrather to its poor availability in soils.Screening under field conditions isfurther complicated by the irregulardistribution of Zn within a plot. Alsoimportant is that soils low in availableZn cannot be used for several yearsafter Zn fertilizer has been applied.These factors influence year-to-yearvariability, as well as spatial variabilitywithin a field.

In areas where the climate is relativelysimilar over years and field conditionsare relatively homogeneous over largeareas, field screening is an efficient andrelatively cheap way of screening largenumbers of genotypes. Yield tests underlow Zn field conditions are the final testof the Zn efficiency of a given genotype.However, both deficiency symptoms andyield are influenced by many factors,which may cause high experimentalerror or increase genotype x environmentand genotype x environment x yearinteractions.

Alternatively, screening in pots is fast,cheap, and overcomes problems of soilheterogeneity. Using 23 wheat cultivarswe recently demonstrated a closerelationship between Zn efficiencies infield and greenhouse (Figure 3). Mostcultivars grown in the same Zn-deficientsoil in field and greenhouse behavedsimilarly in both environments.However, in soils having both Zndeficiency and boron toxicity thecorrelation between Zn efficiencies infield and greenhouse was poor (Figure3). In a greenhouse experiment withKonya soil, surface soil (0-30 cm) withlower B toxicity (around 9 mg soluble Bkg-1 soil) was used, while in the fieldroots were exposed to higher B toxicityin deeper soil (around 20-30 mg solubleB kg-1 soil at 60-90 cm). These resultsindicate the suitability of screeningcultivars for Zn efficiency in potexperiments using Zn-deficient soilwithout additional nutrient stress (Figure3). However, growing conditions are lessrealistic, and the ranking is not alwaysclosely correlated to grain yield obtainedin field trials. In addition, pots should bebig enough to avoid root binding, whichcan affect results (Graham and Welch,1996). Greenhouse experiments can,however, act as a primary screening toreduce the number of genotypes that willbe tested under field conditions.


193

grow in Zn-deficient soils. Littleinformation is available on how Znefficiency is inherited.

Several mechanisms that are most likelycontrolled by several genes may affect Znuptake. In contrast, micronutrient useefficiency seems to be controlled by asingle, major gene (Graham and Welch,1996). Results from a diallel analysis inrice suggested that the genes controllingZn efficiency are additive and, to a lesserdegree, dominant (Majumder et al., 1990).In maize, four additive genes werereported to affect Zn concentration in theear leaf (El-Bendary et al., 1993).However, we are not aware of reports onthe inheritance of Zn efficiency inhexaploid wheat. Our unpublished datafrom a diallel experiment comparingseven wheat cultivars differing in Znefficiency with their F1 derivativessuggest that in bread wheat, genescontrolling Zn efficiency are dominant(Table 10). Although we are far from fullyunderstanding how Zn efficiency isinherited in wheat, progress has beenmade through phenotypic selection.

Although Zn deficiency in soils iscommon in Turkey, it was recognized as awheat production constraint less than adecade ago. Data from two-year yieldtrials in the Central Anatolian Plateaushowed a strong correlation between grainyield and Zn efficiency (M. Kalaycı et al.,unpublished). Bread wheat cultivarsderived from landrace populations (e.g.,Yayla 305, Sertak 52, and Ak 702) arewell adapted to Zn-deficient conditions(Figure 4). However, recently developedcultivars Gerek 79, ES 90-3, and Kirgizoutyielded Yayla 305 by 6%, 12%, and15% respectively. All widely grown and/or recently released winter wheat cultivars(e.g., Gerek 79, Dagdas, Kirgiz, and Gun91) are Zn efficient. However, it should bestressed that modern cultivars have muchhigher yield potential in soils with no Zndeficiency (M. Kalayci et al.,unpublished).

Figure 3. Relationship between Zn efficiency values obtained in greenhouse and field fortwo locations. Soils (0-30 cm) in Eskisehir have Zn deficiency (DTPA-Zn: 0.09 mg kg-1 soil)and no B toxicity (soluble B: lower than 1 mg kg-1 soil). Soils (0-30 cm) in Konya haveboth Zn deficiency (DTPA-Zn: 0.10 mg kg-1 soil) and B toxicity (soluble B: 9 mg kg-1 soil).

Konya (Center)50

45

40

35

3045 60 75 25 50 75 100

1993-94 1994-95Field efficiency

Eskisehir (Sultanonu)90

80

70

60

50

4060 75 90 105 60 75 90

1993-94 1994-95Field efficiency


Gree

nhou

se ef

ficien

cy (%

)Gr

eenh

ouse

effic

iency

(%)

Inheritance of ZincEfficiency in BreadWheat

Wheat genotypes vary in their ability toproduce grain in Zn-deficient soils; thiscan be exploited by breeders to improve

wheat for Zn efficiency (Shukla and Raj,1974; Graham et al., 1992; Cakmak etal., 1997a). Most research related to Zndeficiency in wheat and other crops hasconcentrated on the physiological aspectsof Zn uptake, or has compared wheatgenotypes in their relative efficiency to

Table 10. Mean, minimum, and maximum score for zinc deficiency symptoms of seven breadwheat cultivars and sixteen F1 lines derived from a diallel cross with seven parents.

Parent cultivarsZn deficiency scores† 1 2 3 4 5 6 7

of parents Zinc deficiency scores ofParents Mean§ Min Max F1 obtained from a diallel‡

1. Arapahoe 1.4 1.0 2.3 1.4#

2. Sn64/ /SKE/2*ANE/3/SX/4/BEZ/5/JUN 1.5 1.2 2.2 1.5†† 1.5

3. Dagdas 2.4 1.5 3.5 2.5 2.0 2.44. F134/Nac 3.2 2.5 4.0 3.5 2.5 - 3.25. F4105W-2-1 3.3 2.8 4.0 - 3.0 2.5 2.5 3.36. Gun 3.4 2.5 4.0 - 3.5 3.0 3.5 4.5 3.47. Katia 3.4 3.0 4.0 4.0 3.0 - 4.5 - 4.5 3.4

† 1 = severe deficiency symptoms, 5 = no symptoms. ‡ Only 16 of the 21 possible combinations were obtained.§ Mean of 8 replications. # Diagonal gives mean score of respective parent.†† Mean of 2 replications.

194

concentrate on studies dealing with theeffect of rye chromosomes on Znefficiency in bread wheats. Using wheat/rye addition lines, Graham (1988) foundthat several loci on chromosomes 2R, 3R,and 7/4R enhanced Zn-efficiency.Cakmak et al. (1997c) reported that 1R,2R, and 7R had positive effects on Znefficiency, but 3R was found to have anegative effect on Zn uptake. Based onavailable results, rye chromosomes 1R,2R, and 7R are most likely to carry genesthat enhance Zn efficiency, with genes on1RS and 7RS being most effective(Cakmak et al., 1997c; Schlegel andCakmak, 1997).

Genes controlling Zn efficiency in rye areexpressed in triticale to a similar extent asin rye (Cakmak et al., 1997c). In studieswith wheat amphiploids, Schlegel et al.(1997; 1998) found that chromosomes L1

from Agropyron intermedium and V7from Haynaldia villosa, both belongingto homoeologous group 7, increased Znefficiency in wheat. This suggests that inscreening alien species for new sourcesof Zn efficiency, researchers should startwith chromosomes belonging tohomoeologous group 7.

Results from wheat/rye translocationlines indicate that particulartranslocations involving the short arm of1R enhance Zn efficiency. Thesetranslocations would be an ideal sourceof Zn efficiency because they are presentin many wheat cultivars. However,several lines carrying the 1B/1Rtranslocation, e.g., Seri 82 and BDME10= Ctk/Vee, were found to be sensitive toZn deficiency (Schlegel at al., 1997),suggesting that epistatic effects areimportant. The 5RL/4A translocation,

Table 11. Leaf symptoms and total zinc content (per shoot) of the ten best and worstaccessions among 155 entries in the crossing block of the International Winter WheatImprovement program, 1994-95. Plants were grown in a Zn-deficient calcareous soil undergreenhouse conditions.

Cross Origin Symptom severity† Zn contentµg shoot-1

F4549-W1-1 Romania 3 4.73602-156-22 Bulgaria 4 4.27JUWELL/LLV32/ /2*FL80 Romania 3 4.16F134.71/NAC Mexico 4 3.97AGRI/BJY/ /VEE Mexico-Turkey 2 3.94KATIA1 Bulgaria 4 3.75KSK/ /INIA/LFN/3/CALIBASAN Turkey 3 3.704206/3/911B8.10/K351/ /SAD1/MEXIPAK Bulgaria 4 3.69SADOVA-1 Bulgaria 4 3.69DAGDAS Turkey 4 3.50

1D13.1/MLT Mexico 2 1.84F362 K 2.121 Romania 3 1.82NE7060/VG3 N89L771 USA 1 1.82NZT/BEZ/ /ALD/4/NAD/ /TMP/CI12406/3/ USA-Mexico 1 1.77ES 14 USA 3 1.76KS73H530/VEE USA 1 1.7463-122-77-2/NO66/ /LOV2/3/KVZ/HYS/4/ USA 1 1.72KS82142 USA 3 1.66PYN/ /TAM101/AMIGO USA 1 1.63SU92/CI13465/ /PGFN/3/PHO/4/YMH/TO USA-Mexico 1 1.63

† Symptom severity: 1: very severe - 5: no symptoms.Source: Torun (1997).

Figure 4. Grain yield and zinc efficiency of40 wheat cultivars across two years atEskisehir, Turkey.Source: M. Kalaycı (unpublished data).

Grain yield (kg/ha-1)3500

3000

2500

2000

150050 60 70 80 90 100

Zinc efficiency (%)

LandraceTurkish cultivarIntroduced cultivarDurum wheat

Besides Turkish winter wheat cultivars,winter wheat accessions from Bulgariaand Romania were found to be highlytolerant to Zn deficiency, whereas winterwheats from the Great Plains of theUnited States are mostly sensitive. Table11 shows the 10 best- and 10 worst-performing accessions among 155 entriestested for Zn efficiency in the winterwheat crossing block of the InternationalWinter Wheat Improvement Program inthe 1994/95 season (Torun, 1997). Thegenerally poor Zn efficiency of cultivarsfrom the Great Plains may explain why,despite similar climatic conditions, onlyone cultivar (Bolal) of the thousands ofintroductions tested from the Great Plainswas released for the Central AnatolianPlateau. This suggests that whenintroductions from areas with similarenvironmental conditions lack adaptation,the possible cause may be micronutrientdisorders.

In experiments comparing wheat and ryecultivars for Zn efficiency, rye cultivarswere always more Zn efficient than thebest bread wheat (Graham, 1988; Cakmaket al., 1997a,b; Schlegel et al., 1997,1998). The existence of large numbers ofwheat/rye translocation, substitution, oraddition lines has led researchers to


195

which increases the copper use efficiencyin wheat, does not effect Zn efficiency(Graham et al, 1992), indicatingindependent gene action for Zn and Cuefficiency.

Few results are available on Znefficiency of alien species. Substantialgenotypic variation was observed amongAegilops spp. and T. dicoccoidesaccessions, but no species showed Znefficiency comparable to that of rye(Cakmak et al., unpublished). Noinformation is available regardinginheritance of Zn deficiency in durumwheat. Accessions of all tetraploidspecies evaluated so far, including T.durum, T. dicoccoides, and T. polonicum,are much more sensitive to Zn deficiencythan T. monococcum and bread wheats(Cakmak et al., unpublished). This wouldsuggest that genes controlling Znefficiency might be located in the Agenome of T. monococcum; it alsoindicates the possible existence ofsuppressor genes in the B genome oftetraploid species. Whether these genesare suppressed by genes in the D genomeof bread wheat or whether the D genomealso carries genes enhancing Znefficiency is not known at this time.

Although the genetic variability for Znefficiency in bread wheats and itspresumably simple inheritance shouldallow progress towards Zn efficiency, abreakthrough is more likely to come fromintrogressing genes from rye. Schlegel etal. (1997) pointed out several advantagesof this approach. Wheat/ryetranslocations have been widely used inwheat breeding (Rabinovich, 1998) andare genetically and meiotically stable. Ifdeleterious genes can be removed fromthe translocated rye segment, the genesencoding for Zn efficiency will likely bemaintained in the wheat genome withoutundergoing recombination duringcrossing and backcrossing, since ryesegments introduced into the wheatgenome do not normally recombine with

the wheat chromosomes. Molecularmarkers for the genes controlling Znefficiency would greatly enhanceselection efficiency, but no marker hasbeen reported so far. The realbreakthrough in increasing Zn efficiencyof wheat may come when genes arefound in alien species and cloned.Transformation systems to introgressthese genes into wheat are now in place.

Genetic variability for Zn concentrationin grain is relatively low, and we are notaware of any study on the inheritance ofthis trait in wheat.

Phytic Acid andBioavailability of Zinc

Improved Zn efficiency is usually notaccompanied by higher Zn concentration(density) in grain; in fact, Zn-inefficientdurum genotypes may have higher Znconcentrations in grain than Zn-efficientones (Graham et al., 1992; Cakmak et al.,1997a). Lack of higher Zn concentrationin grains of Zn-efficient genotypes canbe attributed at least in part to a dilutioneffect as the result of greater shoot drymatter production and higher grain yield.

Wheat grains are inherently rich incompounds, such as fiber and phyticacid, that depress utilization/absorptionof Zn in human cells (Welch, 1993).Phytic acid (or its salt, phytate) is themain storage form of phosphorus incereal grains. Around 75% of total P inwheat grains is stored as phytic acid(Raboy et al., 1991), particularly in thegerm and aleurone layers (O’Dell et al.,1972). Because phytic acid has high Zn-binding and -complexing ability, ithampers zinc’s biological availability indiets (Welch, 1993). In animalexperiments, the addition of phytate todiets reduced Zn bioavailability andanimal growth, while removing phytatefrom diets by the addition of phytate-

degrading enzymes improved Znbioavailability and animal growth (Lei etal., 1993). In other studies, widespreadoccurrence of Zn deficiency in humans inrural regions of Iran was attributed tohigh consumption of cereal-based foodsrich in phytic acid (Halsted et al., 1972;Prasad, 1984).

The phytate:Zn molar ratios in foods canbe used for predicting Zn bioavailabilityin diets and the risk that Zn deficiencywill occur. Based on animal experiments,ratios above 20 are associated withreduced Zn absorption and increased riskfor Zn deficiency (Oberleas and Harland,1981; Solomons, 1982). Therefore, toincrease bioavailability of Zn in grains 1)Zn concentration in grains can beincreased, 2) the concentration of phyticacid can be lowered, or 3) concentrationof promoters of Zn bioavailability suchas S-containing amino acids (i.e.,methionine, histidine, and lysine) ingrains can be enhanced (Welch, 1993;Graham and Welch, 1996). For example,adding methionine to animal dietsincreases Zn absorption in animal cells(Hause et al., 1996). Currently, the short-term means of increasing Znbioavailability in grains is through theapplication of higher amounts of Znfertilizers. The long-term solution is todevelop new genotypes with higher Znconcentration and promoters that affectZn bioavailabilty.

There is large variability for phytic acidconcentration in seed among wheat(Raboy et al., 1991) and triticale (Feiland Fossati, 1997) genotypes. Theseresults indicate that breeding genotypesfor lower phytic acid is feasible and maybe the solution to phytate-inducednutritional problems in humans.However, it is still possible that reducingphytic acid levels in seed may have notonly beneficial but also adverse effectson seed quality and human health (seeRaboy et al., 1991 for more detail; Feiland Fossati, 1997).


196

Conclusions

Given the widespread occurrence of Zndeficiency in plants, particularly wheat,it is of great importance to breed breadwheat genotypes with high Zn efficiency.High Zn efficiency in wheat seems to belinked to various morphological andphysiological plant traits. Particular planttraits involved in expression of Znefficiency are root surface area, releaseof Zn-mobilizing PS from roots, affinityof membranes to absorb and transportZn, and better utilization of Zn at thecellular level.

Great genetic variability for Znefficiency exists within and amongcereals and alien species of wheat, andbreeders are able to enhance Znefficiency in modern bread wheatcultivars. Genes encoding for Znefficiency in bread wheat seem to bedominant. Markers are not available atthis time. Among cereals, rye is mostefficient, followed by triticale, barley,and bread and durum wheat. Genestransferred from rye into bread wheat areexpressed in wheat/rye addition andtranslocation lines. Therefore, breedingefforts should concentrate on transferringrye genes into a wheat background.

Successful breeding for Zn efficiency inwheat depends on the existence of quickand reliable screening methods. Effortsaimed at developing suitable screeningmethods have, so far, concentrated ondifferent morphological andphysiological plant traits. Evaluatingseverity of Zn deficiency symptoms onleaves, together with the Zn efficiencyratio (yield at -Zn/yield at +Zn), appearsto be a reasonable approach for reliablyscreening large numbers of genotypesfor Zn efficiency within a short time.

Acknowledgments

The authors thank Dr. Zed Rengel(Western Australia University) for hisvaluable comments on the manuscriptand corrections of the English text. Thisstudy was supported by NATO’sScientific Affairs Division in theframework of the Science for StabilityProgramme, and by the DANIDAProject coordinated by the InternationalFood Policy Research Institute,Washington, D.C., USA.


Asher, C.J. 1987. Effects of nutrient concentrationin the rhizosphere on plant growth. Proc.XIIIth Congress Int. Soc. Soil Sci.Symposium Vol.5, pp. 209-216.

Barrow, N.J. 1993. Mechanisms of reaction ofzinc with soil and soil components. In: Zincin Soils and Plants. A.D. Robson (ed.).Kluwer Academic Publishers, Dordrecht,The Netherlands. pp. 15-31.

Bergmann, W. 1992. Nutritional Disorders ofPlant Development: Visual and AnalyticalDiagnosis. Fischer Verlag, Jena.

Bouis, H. 1996. Enrichment of food staplesthrough plant breeding: A new strategy forfighting micronutrient malnutrition.Nutrition Rev. 54:131-137.

Brennan, R.F. 1991. Effectiveness of zinc sulphateand zinc chelate as foliar sprays inalleviating zinc deficiency of wheat grownon zinc-deficient soils in Western Australia.Aust. J. Exp. Agric. 31:831-834.

Brennan, R.F. 1992. The effect of zinc fertilizeron take-all and the grain yield of wheatgrown on zinc-deficient soils of theEsperance region, Western Australia. Fert.Res. 31:215-219.

Brennan, R.F. 1996. Availability of previous andcurrent applications of zinc fertilizer usingsingle superphosphate for the grainproduction of wheat on soils of SouthWestern Australia. J. Plant Nutr. 1099-1115.

Brown, P.H., Cakmak, I., and Zhang, Q. 1993.Form and function of zinc in plants. In: Zincin Soils and Plants. A.D. Robson (ed).Kluwer Academic Publishers, Dordrecht,The Netherlands. pp. 93-106.

Brümmer, G.W., Gerth, J., and Tiller, K.G. 1988.Reaction kinetics of the adsorption anddesorption of nickel, zinc and cadmium bygoethite. II. Adsorption and diffusion ofmetals. J. Soil Sci. 39:37-52.

Cakmak, I., and Marschner, H. 1987. Mechanismof phosphorus-induced zinc deficency incotton. III. Changes in physiologicalavailability of zinc in plants. Physiol. Plant.70:13-20.

Cakmak, I., and Marschner, H. 1988a. Increase inmembrane permeability and exudation ofroots of zinc deficient plants. J. PlantPhysiol. 132:356-361.

Cakmak, I., and Marschner, H. 1988b. Zinc-dependent changes in ESR signals, NADPHoxidase and plasma membrane permeabilityin cotton roots. Physiol. Plant. 73:182-186.

Cakmak, I., and Marschner, H.1988c. Enhancedsuperoxide radical production in roots ofzinc-deficient plants. J. Exp. Bot. 39:1449-1460.

Cakmak, I., Marschner, H., and Bangert, F. 1989.Effect of zinc nutritional status on growth,protein metabolism and levels of indole-3-acetic acid and other phytohormones in bean(Phaseolus vulgaris L.). J. Exp. Bot.40:405-412.

Cakmak, I., and Marschner, H. 1993. Effect ofzinc nutritional status on activities ofsuperoxide radical and hydrogen peroxidescavenging enzymes in bean leaves. Plantand Soil 155/156:127-130.

Cakmak, I., Gülüt, K.Y., Marschner, H., andGraham, R.D. 1994. Effect of zinc and irondeficiency on phytosiderophore release inwheat genotypes differing in zinc efficiency.J. Plant Nutr. 1-17.

Cakmak, I., Atlı, M., Kaya, R., Evliya, H., andMarschner, H. 1995. Association of highlight and zinc deficiency in cold inducedleaf chlorosis in grapefruit and mandarintrees. J. Plant Physiol. 146:355-360.

Cakmak, I., Yılmaz, A., Kalaycı, M., Ekiz, H.,Torun, B., Erenoglu, B., and Braun, H.J.1996a. Zinc deficiency as a critical problemin wheat production in Central Anatolia.Plant and Soil 180:165-172.

Cakmak, I., Sarı, N., Marschner, H., Kalaycı, M.,Yılmaz, A., Eker, S., and Gülüt, K.Y. 1996b.Dry matter production and distribution ofzinc in bread and durum wheat genotypesdiffering in zinc efficiency. Plant and Soil180:173-181.

Cakmak, I., Sarı, N., Marschner, H., Ekiz, H.,Kalaycı, M., Yılmaz, A., and Braun, H.J.1996c. Phytosiderophore release in breadand durum wheat genotypes differing in zincefficiency. Plant and Soil 180:183-189.

Cakmak, I., Öztürk, L., Karanlık, S. Marschner,H., and Ekiz, H. 1996d. Zinc-efficient wildgrasses enhance release ofphytosiderophores under zinc deficiency. J.Plant Nutr. 19:551-563.

Cakmak, I., Ekiz, H., Yılmaz, A., Torun, B.,Köleli, N., Gültekin, I., Alkan, A., and Eker,S. 1997a. Differential response of rye,triticale, bread wheat and durum wheats tozinc deficiency in calcareous soils. Plant andSoil 188:1-10.


197

Cakmak, I., Oztürk, L., Eker, S., Torun, B., Kalfa,H.I., and Yılmaz, A. 1997b. Concentrationof zinc and activity of copper/zinc-superoxide dismutase in leaves of rye andwheat cultivars differing in sensitivity tozinc deficiency. J. Plant Physiol. 151:91-95.

Cakmak, I., Derici, R., Torun, B., Tolay, I., Braun,H.J., and Schlegel, R. 1997c. Role of ryechromosomes in improvement of zincefficiency in wheat and triticale. Plant andSoil 196:249-253.

Cakmak, I., Torun, B., Erenoglu, B., Oztürk, L.,Marschner, H., Kalaycı, M., and Ekiz, H.1998. Morphological and physiologicaldifferences in cereals in response to zincdeficiency. Euphytica 100 (1-3): 349-357.

Coleman, J.E. 1992. Zinc proteins: enzymes,storage proteins, transcription factors, andreplication proteins. Annu. Rev. Biochem.61:897-946.

Dang, Y.P., Edwards, D.G., Dalal, R.C., and Tiller,K.G.1993. Identification of an index tissueto predict zinc status of wheat. Plant andSoil 154:161-167.

Dong, B., Rengel, Z. and Graham, R.D. 1995.Root morphology of wheat genotypesdiffering in zinc efficiency. J. Plant Nutr.18:2761-2773.

Dwivedi, B.S., and Tiwari, K.N. 1992. Effect ofnative and fertilizer zinc on dry matter yieldand zinc uptake by wheat (Triticumaestivum) in Udic Ustochrepts. Trop. Agric.69:357-361.

Ekiz, H., Bagcı, S.A., Kıral, A., Eker, S.,Gültekin, I., Alkan, A., and Cakmak, I. 1998.Effects of zinc fertilization and irrigation ongrain yield and zinc concentration of variouscereals grown in zinc-deficient calcareoussoils. J. Plant Nutr. 21(10):2245-2256.

El-Bendary, A.A., El-Fouly, M.M., Raksa, F.A.,Omar, A.A., and Abou-Youssef, A.Y. 1993.Mode of inheritance of zinc accumulation inmaize. J. Plant Nutr. 16:2043-2053.

Erenoglu, B., Cakmak, I., Marschner, H.,Römheld, V., Eker, S., Daghan, H., Kalaycı,M., and Ekiz, H. 1996. Phytosiderophoresrelease does not relate well to zinc efficiencyin different bread wheat genotypes. J. PlantNutr. 19:1569-1580.

Eyüpoglu, F., Kurucu, N., and Sanısag, U. 1994.Status of plant available micronutrients inTurkish soils. In: Soil and FertilizerResearch Institute Annual Report. ReportNo. R-118. Ankara, Turkey (in Turkish). pp.25-32.

Feil, B., and Fossati, D. 1997. Phytic acid intriticale grains as affected by cultivar andenvironment. Crop Sci. 37:916-921.

Graham, R.D. 1984. Breeding for nutritionalcharacteristics in cereals. In: Advances inPlant Nutrition. Vol.1. P.B. Tinker and A.Läuchli (eds.). Praeger, NY. pp.57-102.

Graham, R.D. 1988. Development of wheats withenhanced nutrient efficiency: Progress andpotential. In: Wheat production constraints intropical environments. A.R. Klatt (ed.).Mexico, D.F.: CIMMYT. pp. 305-320.

Graham, R.D., and Webb, M.J. 1991.Micronutrients and plant disease resistanceand tolerance in plants. In: Micronutrients inAgriculture. J.J. Mortvedt, F.R. Cox, L.M.Shuman, and R.M. Welch (eds.). SSSA BookSeries No.4. Madison, WI. pp. 329-370.

Graham, R.D., Ascher, J.S., and Hynes, S.C. 1992.Selecting zinc-efficient cereal genotypes forsoils of low zinc status. Plant and Soil146:241-250.

Graham, R.D., and Rengel, Z. 1993. Genotypicvariation in zinc uptake and utilization byplants. In: Zinc in Soils and Plants. A.D.Robson (ed.). Kluwer Academic Publishers,Dordrecht, The Netherlands. pp. 107-118.

Graham, R.D., and Welch, R.M.1996. Breedingfor staple-food crops with high micronutrientdensity. Working Papers on AgriculturalStrategies for Micronutrients, No. 3.International Food Policy Research Institute,Washington, D.C.

Halsted, J.A., Ronaghy, H.A. and Abadi, P. 1972.Zinc deficiency in man: The Shirazexperiment. Am. J. Med. 53:277-284.

Hamilton, M.A., Westermann, D.T., and James,D.W. 1993. Factors affecting zinc uptake incropping systems. Soil Sci. Am. J.57:1310-1315.

Hause, W.A., Van Campen, D.R., and Welch, R.M.1996. Influence of dietary sulfur-containingamino acids on the bioavailability to rats ofzinc in corn kernels. Nutr. Res. 16:225-235.

Jolley, V.D., and Brown, J.C. 1989. Iron efficientand inefficient oats. I. Differences inphytosiderophore release. J. Plant Nutr.12:423-435.

Jones, J.B. Jr. 1991. Plant tissue analysis inmicronutrients. In: Micronutrients inAgriculture. J.J. Mordvedt, F.R. Cox, L. M.Shuman, and R.M. Welch (eds.). SSSA BookSeries No.4. Madison, WI. pp. 477-521.

Kitagishi, K., Obata, H., and Kondo, T. 1987.Effect of zinc defciency on 80S ribosomecontent of meristematic tissues of rice plant.Soil Sci. Plant Nutr. 33:423-430.

Klug, A., and Rhodes, D. 1987. “Zinc fingers”: Anovel protein motif for nucleic acidrecognition. Trends Biochem. Sci.12:464-469.

Kochian, L.V. 1993. Zinc absorption fromhydroponic solutions by plant roots. In: Zincin Soils and Plants. A.D. Robson (ed.).Kluwer Academic Publishers, Dordrecht,The Netherlands. pp. 45-57.

Lei, X., Ku, P.K., Miller, E.R., Ullrey, D.E., andYokoyama, M.T. 1993. Supplementalmicrobial phytase improves bioavailabilityof dietary zinc to weanling pigs. J. Nutr.123:1117-1123.

Lindsay, W.L., and Norvell, W.A. 1978.Development of a DTPA soil test for zinc,iron, manganese and copper. Soil Sci. Soc.Am. J. 42:421-428.

Lindsay, W.L. 1991. Inorganic equilibria affectingmicronutrients in soils. In: Micronutrients inAgriculture. J.J. Mordvedt, F.R. Cox, L.M.Shuman, and R.M. Welch (eds). SSSA BookSeries No. 4. Madison, WI. pp. 89-112.

Ma, Q.Y., and Lindsay, W.L. 1993. Measurementof free zinc activity in uncontaminated andcontaminated soils using chelation. Soil Sci.Soc. Am. J. 57:963-967.

Ma, J.F., and Nomoto, K. 1996. Effectiveregulation of iron acquisition ingraminaceous plants. The role of mugineicacids as phytosiderophores. Physiol. Plant97:609-617.

Majumder, N.D., Rakshit, S.C., and Borthakur,D.N. 1990. Genetic effects on uptake ofselected nutrients in some rice (Oryza sativaL.) varieties in phosphorus deficient soil.Plant and Soil 123:117-120.

Marschner, H., Römheld, V., and Kissel, M. 1986.Different strategies in higher plants inmobilization and uptake of iron. J. PlantNutr. 9:695-713.

Marschner, H., and Cakmak, I. 1989. High lightintensity enhances chlorosis and necrosis inleaves of zinc, potassium, and magnesiumdeficient bean (Phaseolus vulgaris). Plantand Soil 159:89-102.

Marschner, H. 1993. Zinc uptake from soils. In:Zinc in Soils and Plants. A.D. Robson (ed.).Kluwer Academic Publishers, Dordrecht,The Netherlands. pp. 59-77.

Marschner, H. 1995. Mineral Nutrition of HigherPlants. Academic Press, London.

Martens, D.C., and Westermann, D.T. 1991.Fertilizer applications for correctingmicronutrient deficiencies. In:Micronutrients in Agriculture. J.J. Mortvedt,F.R. Cox, L.M. Shuman, and R.M. Welch(eds.). Soil Sci. Soc. Am. Madison, WI. pp.549-592.

Moraghan, J.T., and Mascagni Jr, H.J. 1991.Environmental and soil factors affectingmicronutrient deficiencies and toxicities. In:Micronutrients in Agriculture. J.J. Mordvedt,F.R. Cox, L.M. Shumann, and R.M. Welch(eds.). SSSA Book Series No.4. Madison,WI. pp. 371-425.

Mordvedt, J.J., and Gilkes, R.J. 1993. Zincfertilizers. In: Zinc in Soils and Plants. A.D.Robson (ed.). Kluwer Academic Publishers,Dordrecht, The Netherlands. pp. 33-44.

Mori, S. 1994. Mechanisms of iron acquisition bygraminaceous (strategy II) plants. In:Biochemistry of Metal Micronutrients in theRhizosphere. J.A. Manthey, D.E. Crowley,and D.G. Luster (eds.). CRC Press Inc.,Boca Raton, FL. pp. 225-250.


198

Murakami, T., Ise, K., Hayakawa, M., Kamei, S.,and Takagi, S. 1989. Stabilities of metalcomplexes of mugineic acids and theirspecific affinities for iron (III). Chem. Lett.pp. 2137-2140.

Oberleas, D., and Harland, B.F. 1981. Phytatecontent of foods: Effect on dietary zincbioavailability. J. Am. Diet. Assoc.79:433-436.

O’Dell, B.L., de Boland, A.R., and Koirtyohann,S. 1972. Distribution of phytate andnutritionally important elements among themorphological components of cereal grains.J. Agric. Food Chem. 20:718-721.

Parker, M.B., and Walker, M.E. 1986. Soil pH andmanganese effects on manganese nutritionof peanut. Agron. J. 78:614-620.

Prasad, A.S. 1984. Discovery and importance ofzinc in human nutrition. Fed. Proc.43:2829-2834.

Rabinovich, S.V. 1998. Importance of wheat-ryetranslocations for breeding modern cultivarsof Triticum aestivum. Euphytica100(1-3):323-340.

Raboy, V., Noaman, M.W., Taylor, G.A., andPickett, S.G. 1991. Grain phytic acid andprotein are highly correlated in winterwheat. Crop Sci. 31:631-635.

Rashid, A., and Fox, R.L. 1992. Evaluatinginternal zinc requirements of grain crops byseed analysis. Agron. J. 84:469-474.

Rengel, Z. 1995a. Sulfhydryl groups in root-cellplasma membranes of wheat genotypesdiffering in zinc efficiency. Physiol. Plant.95:604-612.

Rengel, Z. 1995b. Carbonic anhydrase activity inleaves of wheat genotypes differing in zincefficiency. J. Plant Physiol. 147:251-256.

Rengel, Z., and Graham, R.D. 1995a. Importanceof seed zinc content for wheat growth onzinc-deficient soil. I. Vegetative growth.Plant and Soil 173:259-266.

Rengel, Z., and Graham, R.D. 1995b. Importanceof seed Zn content for wheat growth on Zn-deficient soil. II. Grain Yield. Plant and Soil173:267-274.

Rengel, Z., and Graham, R.D. 1995c. Wheatgenotypes differ in zinc efficiency whengrown in the chelate-buffered nutrientsolution. I. Growth. Plant and Soil176:307-316.

Rengel, Z., and Graham, R.D. 1995d. Wheatgenotypes differing zinc efficiency whengrown in the chelate-buffered nutrientsolution. II. Plant and Soil 176:307-316.

Rengel, Z., and Graham, R.D. 1996. Uptake ofzinc from chelate-buffered nutrient solutionsby wheat genotypes differing in zincefficiency. J. Exp. Bot. 47:217-226.

Rengel, Z., and and Hawkesford, M.J. 1997.Biosynthesis of a 34-kDa polypeptide in theroot-cell plasma membrane of a Zn-efficientwheat genotype increase upon Zndeficiency. Aust. J. Plant Physiol.24:307-315.

Rengel, Z., Marschner, H., and Römheld, V. 1998.Uptake of zinc and iron by wheat genotypesdiffering in zinc efficiency. J. Plant Physiol.154(4-5):433-438.

Rengel, Z., and Wheal, M.S. 1997a. Herbicidechlorsulfuron decreases growth of fine rootsand micronutrient uptake in wheatgenotypes. J. Exp. Bot. 48:927-934.

Rengel, Z., and Wheal, M.S. 1997b. Kineticparameters of zinc uptake by wheat areaffected by the herbicide chlorsulfuron. J.Exp. Bot. 48:935-941.

Reuter, D.J., and Robinson, J.B. 1986. PlantAnalysis: An Interpretation Manual. InkataPress Ltd., Melbourne, Australia.

Riley, M.M., Gartrell, J.W., Brennan, R.F.,Hambin, J., and Coates, P. 1992. Zincdeficiency in wheat and lupins in WesternAustralia is affected by the source ofphosphate fertilizers. Aust. J. Exp. Agric.32:455-460.

Schlegel, R., Cakmak, I., Torun, B., Eker, S.,Tolay, I., Ekiz, H., Kalaycı, M., and Braun,H.J. 1998. Screening for zinc efficiencyamong wheat relatives and their utilisationfor an alien gene transfer. Euphytica100(1-3):281-286.

Schlegel, R., Cakmak, I., Torun, B., Eker, S., andKöleli, N. 1997. The effect of rye geneticinformation on zinc, copper, manganese andiron accumulation in wheat shoots. CerealRes. Comm. 25:177-184.

Schlegel, R., and Cakmak, I. 1997. Physicalmapping of rye genes determiningmicronutritional efficiency in wheat. In:Plant Nutrition for Sustainable FoodProduction and Environment. T. Ando, K.Fujita, T. Mae, H. Matsumoto, S. Mori, andJ. Sekiya (eds.). Developments in Plant andSoil Sciences, Vol. 78. Kluwer AcademicPublishers, Dordrecht, The Netherlands. pp.287-288.

Sharma, B.D., Singh, Y., and Singh, B. 1988.Effect of time of application on theeffectiveness of zinc sulphate and zinc oxideas sources of zinc for wheat. Fert. Res.17:147-151.

Sharma, K.N., and Deb, D.L. 1988. Effect oforganic manuring on zinc diffusion in soilsof varying texture. J. Indian Soc. Soil. Sci.36:219-224.

Shukla, U.C., and Raj, H. 1974. Influence ofgenetic variability on zinc response in wheat(Triticum spp.) Soil Sci. Soc. Am. Proc.38:477-479.

Sillanpää, M. 1982. Micronutrients and thenutrient status of soils: A global study. FAOSoils Bulletin 48. Food and AgricultureOrganization of the United Nations, Rome.pp. 75-82.

Sillanpää, M., and Vlek, P.L.G. 1985.Micronutrients and the agroecology oftropical and Mediterranean regions. Fert.Res. 7:151-167.

Sims, J.T., and Johnson, G.V. 1991. Micronutrientsoil tests. In: Micronutrients in Agriculture.J.J. Mordvedt, F.R. Cox, L.M. Shuman, andR.M. Welch (eds.). SSSA Book Series No.4.Madison, WI. pp. 427-476.

Singh, J.P., Karamanous, R.E., and Stewart,J.W.B.1986. Phosphorus-induced zincdeficiency in wheat on residual phosphorusplants. Agron. J. 78:668-675.

Singh, J.P., Karamanos, R.E., and Stewart, J.W.B.1987. The zinc fertility of Saskatchewansoils. Can. J. Soil Sci. 67:103-116.

Skoog, F. 1940. Relationship between zinc andauxin in the growth of higher plants. Am. J.Bot. 27:939-950.

Solomons, N.W. 1982. Biological availability ofzinc in humans. Am. J. Clin. Nutr.35:1048-1075.

Sparrow, D.H., and Graham, R.D. 1988.Susceptibility of zinc-deficient wheat plantsto colonization by Fusarium graminearumSchw. Group I. Plant and Soil 112:261-266.

Stephan, U.W., Schmidke, I., and Pich, A.1994.Phloem translocation of Fe, Cu, Mn and Znin Ricinus seedlings in relation to theconcentrations of nicotianamine, anendogenous chelator of divalent metal ionsin different seedlings parts. Plant and Soil165:181-188.

Suge, H. Takahashi, H., Arita, S., and Takaki, H.1986. Gibberelin relationships in zinc-deficient plants. Plant Cell Physiol.27:1010-1012.

Swamvinathan, K., and Verma, B.C. 1979.Response of three crop species to vesiculararbuscular mycorrhizal infections in zincdeficient Indian soils. New Phytol.82:481-487.

Takagi, S. 1976. Naturally occurring iron-chelating compounds in oat and rice rootwashing. I. Activity measurement andpreliminary characterization. Soil Sci. PlantNutr. 22:423-433.

Takagi, S., Nomoto, K., and Takemoto, T. 1984.Physiological aspect of mugineic acid, apossible phytosiderophore of graminaceousplants. J. Plant Nutr. 7:469-477.

Takkar, P.N., Chibba, I.M., and Mehta, S.K. 1989.Twenty Years of Coordinated Research ofMicronutrients in Soil and Plants (1967-1987). Indian Institute of Soil Science,Bhopal, IISS, Bull. I.

Takkar, P.N., and Walker, C.D. 1993. Thedistribution and correction of zincdeficiency. In: Zinc in Soils and Plants. A.D.Robson (ed.). Kluwer Academic Publishers,Dordrecht, The Netherlands. pp. 151-166.

Thongbai, P., Hannam, R.J., Graham, R.D., andWebb, M.J. 1993. Interaction between zincnutritional status of cereals and Rhizoctoniaroot rot severity. Plant and Soil 153:207-214.

Torun, B. 1997. Differences in zinc efficiencyamong and within cereals and wheats (inTurkish). PhD Dissertation. CukurovaUniversity.


199

Treeby, M., Marschner, H., and Römheld, V.1989. Mobilization of iron and othermcronutrient cations from a calcareous soilby plant-borne, microbial and syntheticmetal chelators. Plant and Soil 114:217-226.

Vallee, B.L., and Falchuk, K.H. 1993. Thebiochemical basis of zinc physiology.Physiol. Rev. 73:79-118.

Viets, F.G. 1966. Zinc deficiency in the soil plantsystem. In: Zinc Metabolism. A.S. Prasad(eds.). C.C. Thomas, Springfield, IL.pp. 90-127.

von Wiren, N., Marschner, H., and Römheld, V.1996. Roots of iron-efficient maize (Zeamays L.) take up also phytosiderophore-chelated zinc. Plant Physiol. 111:1119-1125.

Wagar, B.I., Stewart, J.W.B., and Henry, J.L.1986.Comparison of single large broadcast andsmall annual seed-placed phosphorustreatments on yield and phosphorus and zinccontents of wheat on chernozomic soils.Can. J. Soil Sci. 66:237-248.

Walter, A., Römheld, V., and Marschner, H. 1994.Is the release of phytosiderophores in zinc-deficient wheat plants’ response to impairediron utilization? Physiol. Plant. 92:493-500.

Webb, M.J., and Loneragan, J.F. 1988. Effect ofzinc deficiency on growth, phosphorusconcentration, and phosphorus toxicity ofwheat plants. Soil Sci. Soc. Am. J.52:1676-1680.

Welch, R.M., Webb, M.J., and Loneragan, J.F.1982. Zinc in membrane function and itsrole in phosphorus toxicity. In: Proceedingsof the Ninth Plant Nutrition Colloquium. A.Scaife (ed.). Warwick, UK. CommonwealthAgricultural Bureau, Farnham Royal, Bucks.pp. 710-715.

Welch, R.M., Allaway, W.H., House, W.A., andKubota, J.1991. Geographic distribution oftrace element problems. In: Micronutrientsin Agriculture. J.J. Mortvedt, F.R. Cox, L.M.Shuman, and R.M. Welch (eds.). SSSABook Series No.4. Madison, WI. pp. 31-57.

Welch, R.M., and Norvell, W.A. 1993. Growthand nutrient uptake by barley (Hordeumvulgare L. cv. Herta): Studies using an N-(2-hydroxyethyl) ethylenedinitrilotriaaceticacid-buffered nutrient solution technique.Plant Physiol. 101:627-631.

Welch, R.M. 1995. Micronutrient nutrition ofplants. Crit. Rev. Plant. Sci. 14:49-82.


Wilkinson, H.F., Loneragan, J.F., and Quick, J.P.1968. The movement of zinc to plant roots.Soil Sci. Soc. Amer. Proc. 32:831-833.

Yılmaz, A. Ekiz, H., Torun, B., Gültekin, I.,Karanlık, S., Bagcı, S.A., and Cakmak, I.1997. Effect of different zinc applicationmethods on grain yield and zincconcentration in wheat grown on zinc-deficient calcareous soils in CentralAnatolia. J. Plant Nutr. 20:461-471.

Yılmaz, A., Ekiz, H., Gültekin, I., Torun, B.,Karanlık, S., and Cakmak, I. 1998. Effect ofseed zinc content on grain yield and zincconcentration of wheat grown in zinc-deficient calcareous soils. J. Plant Nutr.21:2257-2264.

Zhang, F., Römheld, V., and Marschner, H. 1989.Effect of zinc deficiency in wheat on therelease of zinc and iron mobilizing exudates.Z. Pflanzenernaehr. Bodenk. 152:205-210.

Zhang, F.S., Römheld, V., and Marschner, H.1991. Diurnal rhythm of release ofphytosiderophores and uptake rate of zinc iniron-deficient wheat. Soil Sci. Plant Nutr.37:671-678.

200

Nitrogen and Phosphorus UseEfficiency1

J.I. Ortiz-Monasterio,2 G.G.B. Manske,3 and M. van Ginkel2

C HAPTER 1 7

The improvement of nutrient useefficiency in wheat cropping systems canbe achieved through two main strategies:by adopting more efficient cropmanagement practices (such as nutrientrate, timing, source, and placement) andbreeding more nutrient use efficientcultivars. Although both are important,this paper will focus on improvingnutrient use efficiency (specifically, ofnitrogen and phosphorus) through plantbreeding. More detailed guidelines onhow to improve nitrogen use efficiencyin wheat through crop management havebeen described elsewhere (Ortiz-Monasterio, 2001).

It is important to clearly define nutrientuse efficiency before describing methodsfor improving it. We have found thedefinition proposed by Moll et al. (1982)useful in looking at genetic differences innitrogen use efficiency among wheatcultivars. (Though the concept wasdeveloped using nitrogen as an example,it can also be applied to phosphorus.)These authors define nitrogen andphosphorus use efficiency in wheat asgrain yield per unit of nutrient supplied(from the soil and/or fertilizer). Theydivide nutrient use efficiency into twocomponents: uptake, or the ability of theplant to extract the nutrient from the soil,and utilization efficiency, or the ability of

the plant to convert the absorbed nutrientinto grain yield. Hence

Nutrient use efficiency = Uptake efficiency xUtilization efficiency

Gw Nt Gw (1)= x

Ns Ns Nt

where Gw = grain dry weight, Nt = totalabove-ground plant nutrient at maturity,and Ns = nutrient supplied. All units arein g m-2. Utilization efficiency can alsobe subdivided into two components, assuggested by Ortiz-Monasterio et al.,1997a, and expressed as follows:

Utilization efficiency = Harvest index x Nutrientbiomass production efficiency

Gw Gw Tw (2)= x

Nt Tw Nt

where Tw = total above-ground plant dryweight at maturity.

Utilization efficiency can also beexpressed as:

Utilization efficiency = Harvest index x Inverse oftotal nutrient concentrationin the plant

Gw Gw 1 (3)= x

Nt Tw Nct

where Nct = total nutrient concentrationin the plant as a percentage.

The definition for nitrogen use efficiencyproposed by Moll et. al. (1982) can beused for both low and high inputsituations. However, there are othernutrient efficiency classification systemsthat take into account performance bothin the presence and in the absence ofnutrient stress as, for example, the systemproposed by Gerloff (1977), whichseparates cultivars into four groups basedon their response to P. The groups are 1)efficient, responder; 2) inefficient,responder; 3) efficient, non-responder,and; 4) inefficient, non-responder. Anefficient cultivar has higher yield than theother cultivars under low nutrient supply,while a responder cultivar has higheryield under high nutrient supply. Thisclassification groups cultivars based onperformance under low (efficient vs.inefficient) and high (responder vs. non-responder) nutrient supply, and allowsthe identification of those cultivars withadaptation to a range of soil nutrientconditions.

CIMMYT and its predecessor have beengenerating wheat germplasm for thedeveloping world since the 1940s.CIMMYT bread wheats were first andmost rapidly adopted in irrigated areas ofthe developing world (e.g., the YaquiValley in Mexico, the Indian Punjab, andthe Pakistani Punjab) (Byerlee, 1996).Fertilizer is widely applied (sometimes atsub-optimal levels) by farmers in thoseareas as a way to correct nutrientdeficiencies. However, in other targetenvironments farmers do not apply

1 This chapter does not attempt to make an exhaustive review of the literature but rather presentspractical information based on CIMMYT Wheat Program experience working on nitrogen andphosphorus use efficiency.

2 CIMMYT Wheat Program, Apdo. Postal 6-641, Mexico, D.F., Mexico, 06600.3 Center for Development Research (ZEF), Universität Bonn, Walter-Flex-Str. 3, 53113 Bonn,

Germany.

201

fertilizers because they cannot affordthem or because inputs are simply notavailable. It is thus essential thatCIMMYT wheats be widely adapted andable to grow in different (low and high)soil nutrient situations.

In this chapter we will discuss howstudying the individual components ofnutrient use efficiency (uptake vs.utilization) under different nutrient levelscan help us gain a better understanding ofthe opportunities and limitations ofbreeding for nitrogen and phosporus useefficiency.

Nitrogen

With the adoption of the input-responsiveand lodging tolerant semidwarf wheatcultivars that launched the greenrevolution in the 1960s, the use ofnitrogen fertilizer rapidly increased, asdid yields. Thanks to the introduction ofthe new genetic material, the amount ofgrain produced per unit of N applied hasincreased significantly (Figure 1).

We have documented the changes in thenitrogen use efficiency of CIMMYTbread wheats developed between 1950and 1985 under medium to high levels of

N fertility. Results show that morerecent CIMMYT cultivars outyield bothearlier semidwarfs and old tall cultivarsat all nitrogen levels (Ortiz-Monasterioet al., 1997a). This indicates that thecurrent strategy of selecting andevaluating under medium to high Nlevels has resulted in germplasm thatproduces higher yield when grown underlow or high levels of N fertility.CIMMYT bread wheats from 1950 to1985 gradually became not only moreresponsive to N inputs, but also moreefficient in their use, according toGerloff’s classification (1977). As aresult, CIMMYT bread wheats do notrequire more N than the old tallcultivars; in fact, they often need less Nto produce the same yield. In addition,since CIMMYT bread wheats are moreresponsive to N application, theoptimum economic rate is higher thanthat for the old tall cultivars (Ortiz-Monasterio et al., 1997a).

Although our current breeding strategyhas been successful in addressing theneeds of both low input and high inputwheat-producing environments, we areinterested in identifying alternativeselection methods that might be evenmore successful. To that end, wecharacterized relevant CIMMYTgermplasm for two main components ofnitrogen use efficiency: nitrogen uptakeand utilization efficiency. We found thatthere is genetic diversity for both traits.

Our work and that of others has shownthat the level of N in the soil plays avery important role in the expression ofuptake and utilization efficiency(Dhugga and Waines, 1989; Ortiz-Monasterio et al. 1997a). However, theeffect of different soil N levels on theexpression of a given component ofnitrogen use efficiency in spring wheatmay be affected by genotype and/orlocation. Dhugga and Waines (1989)found better expression of uptakeefficiency under high soil N and better

expression of utilization efficiency underlow N. In contrast, Ortiz-Monasterio etal. (1997a) found better expression ofuptake efficiency under low Nconditions and better expression ofutilization efficiency under high Nconditions. These findingsnotwithstanding, available informationhas shown that the level of soil N maybe manipulated together with geneticvariability to develop cultivars withimproved performance under both lowand high input conditions (Ortiz-Monasterio et al., 1997a; van Ginkel etal., 2001).

Nitrogen uptake vs.utilization efficiencyIn view of the above, an importantaspect of our current research is toidentify the best selection strategies fordeveloping genotypes that producehigher grain yields as a result of theirimproved uptake and/or utilizationefficiency. The question is whichcomponent to emphasize.

Utilization efficiency has ecologicalappeal, since it means either higheryields with the same nutrient levels inthe plant or the same yield with lowernutrient levels in the plant, whichrequires fewer resources. As indicatedearlier, utilization efficiency can bebroken down into harvest index andbiomass production efficiency. If weanalyze which component has been mostassociated with utilization efficiencygains in the past, we find that mostprogress has been associated withimprovements in harvest index (HI)rather than in biomass productionefficiency (Eq. 2). However, Fischer(1981) and Calderini et al. (1995)suggest that the possibilities of furtherimproving HI as a way to increase grainyield are limited.

There are two main routes for makingfurther progress in grain yield throughbetter utilization efficiency: 1) to

Figure 1. Response of tall (Yaqui 50) andsemidwarf spring wheat cultivars toincreasing levels of nitrogen fertilizer.

Grain yield (kg/ha)9000

8000

7000

6000

5000

4000

3000

20000 75 150 225 300

Nitrogen (kg/ha)

Yaqui 50Siete Cerros 66Genaro 81

NITROGEN AND PHOSPHORUS USE EFFICIENCY

202

increase grain yield while maintaining orreducing nutrient concentration in theplant, and 2) to reduce total nutrientconcentration in the plant whileincreasing or maintaining grain yield(Eq. 2). Most CIMMYT high yieldingwheats grown under a wide range of Nlevels tend to have, on average, anitrogen harvest index of around 75%. Inother words, 75% of the plant’s total N isfound in the grain at maturity. Thismeans that cultivars with higherutilization efficiency, which is notassociated with HI (assuming a constantHI), will have lower proteinconcentration in the grain. This cannegatively affect the grain’s breadmaking quality and nutritional value,unless the percent protein reduction iscompensated for by a proportionalimprovement in protein quality.

We should point out that bread makingquality, which is a key issue for breedingprograms in the developed world, is nowgaining significance for breedingprograms in developing countries. Theoriginal focus of many wheat breedingprograms in developing countries—i.e.,generating sufficient yield increases tofeed their rising populations—hasexpanded to include fulfilling farmers’need to produce high quality grain thatcompetes well on the market.

The nutritional value of wheat grain isanother issue that is gaining insignificance due to its perceived potentialto better the nutrition of developingcountry populations. The nutrient contentof wheat grain is negatively affected bylower protein concentration in the grain.Studies in Mexico and Argentina haveshown that protein concentration in thegrain has decreased as grain yield hasincreased throughout the history ofbreeding (Ortiz-Monasterio et al., 1997b;Calderini et al., 1995). This reduction inprotein N has been associated withhigher utilization efficiency. Thus animportant challenge for breeding

programs in both developed anddeveloping countries will be to continueto improve nitrogen use efficiency and, atthe same time, maintain or improve thebread making quality and/or nutrientcontent of wheat grain.

A similar dilemma arises when uptakeefficiency, the other component ofnutrient use efficiency, is implemented asa strategy to improve grain yield. Forresource poor farmers who cannot affordfertilizers and grow wheat under lowinput conditions, the development ofcultivars with high N uptake efficiencymay not be desirable because it mayaccelerate soil nutrient mining. Incontrast, in high input environments, highuptake efficiency is a very desirable traitbecause residual soil N (soil N notabsorbed by the crop) may either leachthrough the soil to pollute waterwayswith soil nitrates or escape into theatmosphere as N2, N2O, NOx, or NH3.

Nitrate leaching has been welldocumented in many developed countries(CAST, 1985; Keeney, 1986). Theproblem tends to be associated with theapplication, especially in sandy soils, ofmore nitrogen than is required forproducing maximum yield. Until recently,total N fertilizer use in the world wasalmost evenly divided between developedand developing countries from a total of80 Tg y-1 (FAO, 1990). However, the useof N fertilizer has been accelerating in thedeveloping world. Of the 60-90%increase in global application of Nfertilizer estimated to take place by 2025,two thirds will occur in developingcountries (Galloway, 1995).

There are wheat production systems inthe developing world where very highrates of N fertilizer are already beingapplied—for example, in certain wheat-growing areas of Mexico and Egypt. Inthe high input wheat systems ofnorthwestern Mexico, where farmersapply an average of 250 kg N/ha,researchers have recorded large N

leaching losses (Riley et al., 2000) andhigh emissions of greenhouse gases intothe atmosphere (Matson et al., 1998). Ifcultivars and crop management systemsremain as they are now, as N ratesincrease, the problems of N leaching andgreenhouse gas emissions (N2O),common in many industrializedcountries, will also become widespread inthe high input areas of developingcountries.

Strategy for improvingnitrogen use efficiencyGrain yields of CIMMYT bread wheatsdeveloped between 1950 and 1985 havegradually increased. We evaluated thesewheats at N levels commonly applied byfarmers in irrigated areas of thedeveloping world (75-150 kg N/ha) andfound that 50% of the yield gains wasassociated with higher nitrogen uptakeefficiency and the other 50% with betterutilization efficiency (Ortiz-Monasterio etal., 1997a). This clearly shows thatimprovements in both uptake andutilization efficiency have been importantin the past and most likely will continueto be in the future.

Hence, it is important to select andevaluate for nitrogen use efficiency underboth low and high nutrient conditions;this allows the researcher to identifygenotypes that perform well undernutrient stress (low input) (efficient) andgenotypes that respond well to high inputconditions (responder) (Picture 1).

In a study that evaluated five N selectiontreatments (low, medium, high,alternating high-low, and alternating low-high N levels), we found that the highestyielding germplasm in medium or highinput environments was obtained byalternately selecting (from F2 to F7) underhigh and low N conditions. Nodifferences between N selectiontreatments were observed when theresulting lines were evaluated in low Nenvironments (van Ginkel et al., 2001).

J.I. ORTIZ-MONASTERIO, G.G.B. MANSKE, AND M. VAN GINKEL

203

We conclude that the relative importanceof both uptake and utilization efficiencywill vary according to the needs ofdifferent production systems. Given thatwide adaptation is a primary objective inbreeding CIMMYT germplasm, we willcontinue to improve both components.

Phosphorus

Many soils have large reserves of totalphosphorus, but low levels of “available”phosphorus. Al-Abbas and Barber (1964)reported that total soil P is often 100times higher than the fraction of soil Pavailable to crop plants. Our objective inbreeding for P efficient and responsivecultivars has been to identify wheatcultivars that can access P not usuallyavailable to the average cultivar underlow P conditions (P efficiency), but alsorespond to P applications (Presponsiveness).

As in the case of N, CIMMYT has beenbreeding under medium to high levels ofP in the soil. Preliminary results suggestthat phosphorus use efficiency inCIMMYT bread wheat cultivars between1950 and 1992 has improved under lowas well as high levels of P fertility (Ortiz-Monasterio et al., unpublished data).Again using Gerloff’s (1977) definition,CIMMYT bread wheat germplasm has

become more efficient as well as moreresponsive to P applications during thattime period.

There is little information on thecontribution of uptake and utilization tototal P use efficiency in wheat. In arecent CIMMYT study, the relativeimportance of uptake and utilization inspring wheat was evaluated in twodifferent environments: a rainfed areawith Andisols in the central highlands ofMexico and an irrigated, low-altitudearea with Vertisols in northwesternMexico. Uptake and utilization werecharacterized in a set of CIMMYT lines.Results showed that in an acid Andisolwith no Al toxicity, uptake was moreimportant than utilization in explaining P

use efficiency. In contrast, in the samegroup of genotypes utilization efficiencywas more important when evaluated in analkaline Vertisol (Manske et al., 2000a).In these two different environments itwas shown that there was geneticdiversity for both uptake and utilizationefficiency in the CIMMYT materialtested.

This study shows that, as in the case of N,the environment where a given set ofgenotypes is evaluated plays a veryimportant role in the expression of Puptake and utilization efficiency.However, in the case of P, whatinfluenced the expression of uptake vs.utilization was not low P vs. high P, butrather the effect of location. At this pointit is not clear how much of the locationeffect is due to soil effects and how muchto above-ground effects (radiation,temperature, etc.) (Manske, 1997). Alsoto be determined is why the same geneticmaterial expresses genetic diversity foruptake efficiency in some environmentsbut not in others.

Evaluating germplasm under both lowand high nutrient conditions allows theidentification of genotypes that performwell under nutrient stress (low input) andgenotypes that are responsive to highinput conditions (Picture 2). Preliminarydata suggest that evaluating advanced

Picture 1. Varieties with and without nitrogen application, Ciudad Obregon, Sonora, Mexico.(Photo: J.I. Ortiz-Monasterio.)


Picture 2. Screening plots for phosphorus use efficiency, Patzcuaro, Michoacan, Mexico. In bothpictures, plants on the right received 80 kg P / ha, while the ones on the left received none.(a) P use efficient genotype. (b) P use inefficient genotype. (Photos: J.I. Ortiz-Monasterio.)

a

b

204

genetic materials under low P conditionsis useful for identifying exceptionalgermplasm for P stress conditions. Whenadvanced genetic materials are evaluatedonly under high input conditions, thissometimes results in genotypes that areoutstanding under low P conditions, butintermediate under high input conditions.This germplasm might be discarded if itis tested only under high input conditions(since only the top 10-15% of the linesare selected) and its performance underhigh input conditions is intermediate(Trethowan et al., unpublished data).Hence the importance of selecting andevaluating under both low and highnutrient conditions. More definite datawill be available once a CIMMYT studyis completed in which germplasm isselected under low vs. high and underalternating low and high P levels, as wasdone in the N study.

In acid soils P deficiency is oftenaccompanied by Al and Mn toxicity,especially when soil pH is below 5.4.Evidence available so far indicates thatgenes controlling adaptation to Al andMn toxicity and tolerance to P deficiencyappear to be independently inherited andrecombinable (Polle and Konzak, 1990).Therefore the recommendation is thatscreening for P use efficiency be donefirst in soils without Al or Mn toxicity, ifpossible. Once elite materials have beenselected for P use efficiency in the field,they can be screened for Al and/or Mntoxicity either in the field or inhydroponics (see chapter by Hede andSkovmand).

We suggest that screening for P uptakeefficiency under nutrient cultureconditions be avoided until a satisfactorycorrelation between performance in thefield and in nutrient cultures has beenshown. This is particularly important forP, given that very little of the crop’s Prequirement is provided by mass flow

(transpiration flow). Diffusion is moreimportant, but difficult to simulate insolution culture. It is generallyrecognized that nutrient culture shouldbe limited as a screening environmentprimarily because of the low correlationof the results with those of field tests.Nutrient cultures cannot simulate thesoil-plant interface properly.

Phosphorus uptake vs.utilization efficiencyPhosphorus utilization efficiency (grainyield per unit P in the plant) is dependenton the plant’s internal P requirement.Increased harvest index, P harvest index,and low P concentration in grain mayimprove P utilization efficiency (Jones etal., 1989; Batten, 1992).

Most CIMMYT high yielding wheatshave a P harvest index of about 80%under irrigated conditions. As in the caseof N, breeding for higher P utilizationefficiency, given the small margin tobreed for higher HI, will result in lowerP concentration in the grain. Selectionfor wheat genotypes that remove smallamounts of P from the soil due to theirlow P grain concentration can contributeto sustainable land use (Schulthess et al.,1997). Genotypic differences in grain Pconcentration are fairly consistent acrossenvironments (Schulthess et al., 1997). Ifbreeders in Australia, which is a majorexporter of wheat grain but has soils thatare poor in P availability, can reduce theP concentration in the grain of wheatcultivars, farmers will have to purchasesubstantially less P to replace the Pexported with the grain.

However, the strategy of reducing Pconcentration in the grain has a limit.There is evidence that excessively lowvalues of P concentration in the grainaffects seed vigor, particularly in Pdeficient soils. A study on a set ofhistorically important CIMMYT

semidwarf bread wheats showed that Pconcentration in the grain decreasedsignificantly over the years as a result ofbreeding (Manske, 1997). Similarinformation is available from a wheatbreeding program in Argentina (Calderiniet al., 1995). As in the case of N, thisreduction in P concentration in the grainis associated with gains in utilizationefficiency.

Most nitrogen absorbed by plants comesfrom mass flow (i.e., soil water movestowards the roots as the plant loses waterthrough transpiration), but phosphorus isabsorbed mainly by diffusion throughgradients created by root absorption.Phosphate concentrations in soil solutionare small (<0.05 µ g-1) compared tonitrate-N concentrations (100 µ g-1), andvery little phosphate is moved to the rootsby capillary water movement. Theamount of P extracted is limited by Pconcentration at the root-soil interface,which means that wheat roots have togrow to come into contact with new soilfrom which they can extract phosphate.Root length is thus a major determinant ofthe absorbing surface area.

Wheat genotypes with greater root lengthdensity are able to take up morephosphorus (Manske et al., 2000b). WhenP supply is low, the correlation betweenroot length density and P uptake or grainyield is usually 0.50-0.60, but withadequate P supply this correlation islower. In some environments, P uptakecan be more important than utilizationefficiency. In areas where uptake is themain component associated with P useefficiency, P uptake efficiency holds greatpromise for improving P use efficiency,since soils with relatively high levels oftotal P in the soil often have low levels ofavailable P.


205

Strategies for improvingphosphorus use efficiencyDifferent approaches can be used toenhance P uptake (Polle and Konzak,1990; Johansen et al., 1995).

• Increasing the root surface/soilcontact area. This can be achievedby modifying root morphology. For aconstant level of root biomass, rootswith higher specific root length (i.e.,roots with smaller diameter) cancover a larger surface area. A secondapproach for achieving the sameobjective is through increased hairroot development. Root fineness orbranching is an importantdeterminant of P uptake efficiency inwheat (Jones et al. 1989). This routeseems promising given that there isevidence of large genetic variabilityfor this trait in wheat. However, thetime consuming and labor intensivemethodologies currently in use limitits application in breeding programswhere large numbers of genotypesneed to be screened.

• Increasing the effective root area.Root symbiosis with arbuscularmycorrhizal fungi (AMF) has beenshown to enhance P absorption byincreasing the effective root area(Hayman and Mosse, 1971). AMFinfection improves P influx (P uptakeper unit root length). On the otherhand, the information availablediscussing the genetic diversitypresent among wheat cultivars toassociate with vesicular-arbuscularmycorrhiza (VAM) is not consistent(Vlek et al., 1996). There are reportsthat show differences in mycorrhizalassociation among wheat cultivars(Vlek et al. 1996). In contrast,extensive screening of CIMMYT’sspring wheat cultivars formycorrhizal association found verysmall differences among genotypes;the differences were not stronglyassociated with higher P absorption(Manske et al., 2000b).

• Increasing nutrient availabilitythrough rhizosphere modification.Root exudates, ranging from protonsto complex organic molecules, caninfluence nutrient availability anduptake. Phosphatases have beenreported to transform poorly availableorganic phosphorus, which usuallyaccounts for 40-50% of a plant’s totalP supply, into inorganic formsavailable to the plant (Randall, 1995).There are genotypic differences inroot phosphatase excreted or bound atthe root surface (McLachlan, 1980).Our work in an Andisol showed anassociation between acidphosphatases and P uptake in differentwheat and triticale cultivars (Portilla-Cruz et al., 1998).

As in the case of N, most opportunitiesfor breeding for higher utilizationefficiency probably lie in improvingbiomass production efficiency (BPE)rather than HI. In this case biomassproduction will have to either increasewith the current levels of P in the plant orbe maintained with a lower concentrationof P in the plant. Utilization efficiency isassociated with the efficiency with whichplants use absorbed P; this, in turn, is afunction of 1) how efficiently P isdistributed to the functional sites and 2)the P requirement of the cells at thosesites (Loneragan, 1978).

Calculating NutrientUptake Efficiency

As defined earlier, uptake efficiencyrefers to the ability of the crop to extractor absorb nutrients from the soil.

Uptake efficiency = Total above-ground nutrient in theplant at maturity (Nt)/Nutrient supplied (Ns)

Uptake efficiency can be measured at anystage of development, but particularlyuseful information can be collected at

anthesis and physiological maturity.Follow the steps described below tomeasure uptake efficiency.

First, a biomass sample is collected byeither harvesting all the above-groundbiomass in a given area (a minimum of0.5 m2 is suggested) or harvesting apredetermined, representative number ofplants (a minimum of 50 stems issuggested) at random. Detailed methodsfor doing this type of sampling atdifferent stages of development aredescribed by Bell and Fischer (1994).

If the sample is collected right before orshortly after anthesis, there is no need toseparate the grain from the rest of theplant for N or P analysis. However, if thesample is collected at or aroundphysiological maturity, it is important toseparate the grain from the rest of thebiomass for N analysis. This is becausethere is a large difference in % nutrientconcentration between the grain andnon-grain biomass (leaves, stems, chaff).In well fertilized spring wheat cropsunder irrigated conditions, we haveobserved values of approximately 2% Nin the grain and 0.8% N in non-grainbiomass. Therefore it is best to take aweighted average to calculate totalnutrient in the plant, using the followingformula:

Total above-ground nutrient in the plant at maturity(Nt) =[Grain weight at 0% moisture (g m-2) x Nutrientconcentration in the grain (%)] +[Non-grain biomass at 0% moisture (g m-2) x Nutrientconcentration in non-grain biomass (%)]

Total nutrient in the plant is then dividedby the amount of nutrient supplied(g m-2) as fertilizer. If soil samples arecollected and the amount of soilavailable nutrient is known, this can beadded to the amount supplied asfertilizer.


206

Nutrient absorption is dependent on rootcharacteristics, especially for immobileplant nutrients in the soil, such asphosphorus. Methods for measuring roottraits in wheat are explicitly explained inthe chapter by Manske et al.

Calculating NutrientUtilization Efficiency

Nutrient utilization efficiency is definedas a crop’s ability to convert theabsorbed nutrients into grain yield.

Utilization efficiency = Tw/Nt

where Tw = total above-ground plant dryweight at maturity and Nt = total above-ground plant nutrient at maturity. Tomeasure uptake efficiency, certaininformation needs to be collected. First,calculate the harvest index (HI), asfollows:

HI = Gw/Tw

where Gw = grain weight at 0% moistureand Tw = total plant biomass at 0%moisture.

This can be done either on an area or aplant basis, as suggested by Bell andFischer (1994). Finally, biomassproduccion efficiency (BPE) iscalculated as:

BPE = Gw/Nt

Conclusions

Bread wheat breeding work at CIMMYThas shown that selection and evaluationof genetic material under medium tohigh nitrogen levels results in geneticgains expressed when this material istested under low, medium, or highnitrogen levels. In other words, selectingfor high yield potential under optimumconditions has resulted in germplasm

with higher nitrogen use efficiency underlow, medium, or high nitrogen fertilityconditions. Now there is evidence thatbreeding under alternating low and highnitrogen levels may produce germplasmthat is even more efficient andresponsive to nitrogen.

It is clear that nutrient use efficiency andnutrient responsiveness are under geneticcontrol. Some researchers consider thesetraits as two different breedingobjectives, but it has been shown thatthey are not incompatible. One of thebest pieces of evidence for this are theresults achieved by bread wheat breedersat CIMMYT. During the last severaldecades, CIMMYT has been breedingwheat under medium to high levels ofnitrogen and phosphorus and hasdeveloped cultivars that are not onlymore responsive to nitrogen andphosphorus, but also more efficient intheir use.

To characterize and better understand themechanisms associated with higher Nand P use efficiency:

• Use the definition of N and P useefficiency suggested by Moll et al.(1982).

• Distinguish between efficiency andresponsiveness. This will require thatall germplasm be evaluated underlow as well as high N and Pconditions.

• Establish the importance of uptakevs. utilization efficiency in the targetenvironment.

• Understand the mechanismsassociated with higher uptake (moreroots, phosphatases, etc.) orutilization efficiency (biomassproduction efficiency vs HI). If thesemechanisms are well understood,they can be used as selection criteria.

• Once genetic markers for genescontrolling these traits are identified,selection for these traits could bedone in the laboratory.

ReferencesAl-Abbas, A.H., and Barber, S.A. 1964. A soil test

for phosphorus based upon fractionation ofsoil phosphorus. I. Correlation of soilphosphorus fractions with plant-availablephosphorus. Soil Science Society of AmericaProceedings 28:218-221.

Barber, S.A. 1984. Soil Nutrient Bioavailability. AMechanistic Approach. John Wiley: NewYork.

Barber, S.A. 1979. Growth requirement fornutrients in relation to demand at the rootinterface. In: The Soil-Root Interface. Harley,J.L., and Scott-Russell, R. (eds.). London:Academic Press. pp. 5-20.

Batten, G.D. 1992. A review of phosphorusefficiency in wheat. Plant and Soil146:163-168.

Bell, M.A., and R.A. Fischer. 1994. Guide toPlant and Crops Sampling: Measurementsand Observations for Agronomic andPhysiological Research in Small GrainCereals. Wheat Special Report No. 32.Mexico, D.F.: CIMMYT.

Byerlee, D. 1996. Modern varieties, productivity,and sustainability: Recent experience andemerging challenges. World Development24(4):697-718.

Calderini, D.F., S. Torres-Leon, and G.A. Slafer.1995. Consequences of Wheat Breeding onNitrogen and Phosphorus Yield, GrainNitrogen and Phosphorus Concentration andAssociated Traits. Annals of Botany76:315-322.

Council for Agricultural Science and Technology.1985. Agricultural and ground-water quality.Report No. 103. CAST, Ames, IA.

Dhugga, K.S., and J.G. Waines. 1989. Analysis ofnitrogen accumulation and use in bread anddurum wheat. Crop Sci. 29:1232-1239.

Engels, C., and H. Marschner. 1995. Plant uptakeand utilization of nitrogen. In: NitrogenFertilization and the Environment. P.E.Bacon (ed.). Marcel Dekker: New York.pp. 41-83.

FAO Fertilizer Yearbook. 1990. United NationsFood and Agricultural Organization, Rome.

Fischer, R.A. 1981. Optimizing the use of waterand nitrogen through breeding of crops. Plantand Soil. 58:249-278.

Fischer, R.A., and P.C. Wall. 1976. Wheatbreeding in Mexico and yield increases. J.Aust. Inst. Agric. Sci. 42(3):139-148.

Galloway, J.N., Schlesinger, W.H., Levy, H.,Michaels, A., and Schnoor, J.L. 1995.Nitrogen fixation – anthropogenicenhancement-environmental response. GlobalBiogeochemical Cycles 9:235-252.

Graham, R.D. 1984. Breeding for nutritionalcharacteristics in cereals. In: Advances inPlant Nutrition. P.B. Tinker and A. Lauchli(eds.). New York: Praeger Publishers.pp. 57-102.

Gerloff, S. 1977. Plant efficiencies in the use ofN, P and K. In: Plant adaptation to mineralstress in problem soils. M.J. Wright (ed.).Cornell Univ. Press: New York. pp. 161-174.


207

Haymann, D.S., and Mosse, B. 1971. Plantgrowth response to vesicular-arbuscularmycorrhiza. I. Growth of Endogoneinoculated plants in phosphate deficient soils.New Phytol. 70:19-27.

Johansen, C., Subbarao, G.V., Lee, K.K., andSharma, K.K. 1995. Genetic manipulation ofcrop plants to enhance integrated nutrientmanagement in cropping systems: The caseof phosphorus. In: Genetic manipulation ofcrop plants to enhance integrated nutrientmanagement in cropping systems. 1.Phosphorus: Proceedings of an FAO/ICRISAT Expert Consultancy Workshop.Johansen, C., Lee, K.K., Sharma, K.K.,Subbarao, G.V., and Kueneman, E.A. (eds.).Andhra Pradesh, India: International CropsResearch Institute for the Semi-Arid Tropics.pp. 9-29.

Jones, G.P.D., Blair, G.J., and Jessop, R.S. 1989.Phosphorus efficiency in wheat – a usefulselection criteria? Field Crops Res.21:257-264.

Jones, G.P.D., Jessop, R.S., and Blair, G.J. 1992.Alternative methods for the selection ofphosphorus efficiency in wheat. Field CropsRes. 30:29-40.

Keeney, D.R. 1982. Nitrogen management formaximum efficiency and minimumpollution.. In: Nitrogen in agricultural soils.F.J. Stevenson (ed.). Agron. Monogr. 22.ASA, CSSA and SSSA, Madison, WI. pp.605-649.

Loneragan, J.F. 1978. The physiology of planttolerance to low P availability. In: CropTolerance to Suboptimal Land Conditions.G.A. Jung (ed.). Am. Soc. Agron. Spec. Publ.No. 32. Madison, WI. pp. 329-343.

McLachlan, K.D. 1980 Acid phosphatase activityof intact roots and phosphorus nutrition ofplants. II. Variation among wheat roots. Aust.J. Agric. Res. 31:441-448.

Manske, G.G.B. 1997. Utilization of thegenotypic variability of VAM-symbiosis androot length density in breeding phosphorusefficient wheat cultivars at CIMMYT. FinalReport of Special Project No. 1-60127166,funded by BMZ, Germany.

Manske, G.G.B., J.I. Ortiz-Monasterio R., M. vanGinkel, R.M. Gonzalez, R.A. Fischer, S.Rajaram, and P. Vlek. 2000a. Importance ofP-uptake efficiency vs. P-utilization for wheatyield in acid and calcareous soils in Mexico.European J. Agronomy. 14(4):261-274.

Manske, G.G.B., J.I. Ortiz-Monasterio, M. vanGinkel, R.M. Gonzalez, S. Rajaram, E.Molina, and P.L.G. Vlek. 2000b. Traitsassociated with improved P-uptake efficiencyin CIMMYT’s semidwarf spring bread wheatwheat grown on an acid andisol in Mexico.Plant and Soil 22(1):189-204.

Matson, P.A., Naylor, R., and Ortiz-Monasterio, I.1998. Integration of environmental,agronomic and economic aspects of fertilizermanagement. Science 280:112-115.

Moll, R.H., E.J. Kamprath, and W.A. Jackson.1982. Analysis and interpretation of factorswhich contribute to efficiency of nitrogenutilization. Agron. J. 74:562-564.

Ortiz-Monasterio, J.I. 2001. NitrogenManagement in Irrigated Spring Wheat. B.Curtis (ed.). Wheat. FAO. Rome, Italy. (inpress).

Ortiz-Monasterio, I., K.D. Sayre, and O. Abdalla.1992. Genetic Progress in CIMMYT’s DurumWheat Program in the Last 39 Years.Abstracts of the First International CropScience Congress, 64. Ames, Iowa, July 14-22, 1992.

Ortiz-Monasterio R., J.I., K.D. Sayre, S. Rajaram,and M. McMahon. 1997a. Genetic progress inwheat yield and nitrogen use efficiency underfour N rates. Crop Sci. 37(3):898-904.

Ortiz-Monasterio R., J.I., R.J. Peña, K.D. Sayre,and S. Rajaram. 1997b. CIMMYT’s geneticprogress in wheat grain quality under four Nrates. Crop Sci. 37(3):892-898.

Polle, E.A., and C.F. Konzak. 1990. Genetics andbreeding of cereals for acid soils and nutrientefficiency. In: Crops as Enhancers of NutrientUse. V.C. Baligar and R.R. Dunca (eds.).Academic Press. pp. 81-121.

Portilla-Cruz, I., E. Molina Gayosso, G. Cruz-Flores, I. Ortiz-Monasterio, and G.G.B.Manske. 1998. Colonización micorrízicaarbuscular, actividad fosfatásica y longitudradical como respuesta a estrés de fósforo entrigo y triticale cultivados en un andisol.Terra 16(1):55-61.

Randall, P.J. 1995. Genotypic differences inphosphate uptake. In: Genetic manipulationof crop plants to enhance integratedmanagement in cropping systems. 1.Phosphorus: Proceedings of an FAO/ICRISAT Expert Consultancy Workshop.Johansen, C., Lee, K.K., Sharma, K.K.,Subbarao, G.V., and Kueneman, E.A. (eds.).Andhra Pradesh, India: International CropsResearch Institute for the Semi-Arid Tropics.pp. 31-47.

Riley, W.J., I. Ortiz-Monasterio, and P.A. Matson.2001. Nitrogen leaching and soil nitrate, andammonium levels in an irrigated wheatsystem in northern Mexico. Nutrient Cyclingin Agroecosystems (in press).

Schulthess, U., Feil, B., and Jutzi, S.C. 1997.Yield-independent variation in grain nitrogenand phosphorus concentration amongEthiopian wheats. Agron. J. 89:497-506.

Van Ginkel, M., I. Ortiz-Monasterio, R.Trethowan, and E. Hernandez. 2001.Methodology for selecting segregatingpopulations for improved N-use efficiency inbread wheat. Euphytica 119(1-2):223-230.

Van Sanford, D.A., and C.T. MacKown. 1986.Variation in nitrogen use efficiency amongsoft red winter wheat genotypes. Theor. Appl.Genet. 72:158-163.

Vlek, P.L.G., Lüttger, A.B., and Manske, G.G.B.1996. The potential contribution ofarbuscular mycorrhiza to the development ofnutrient and water efficient wheat. In: TheNinth Regional Wheat Workshop for Eastern,Central and Southern Africa. Tanner, D.G.,Payne, T.S., and Abdalla, O.S. (eds.). AddisAbaba, Ethiopia: CIMMYT. pp. 28-46.


208

Techniques for MeasuringGenetic Diversity in RootsG.G.B. Manske,1 J.I. Ortiz-Monasterio,2 and P.L.G. Vlek1

C HAPTER 1 8

Physiologists, agronomists, and breedershave dedicated relatively little attentionto research on root systems, mainlybecause root studies require much timeand intensive labor. Nonetheless, roottraits have a significant role to play in thedevelopment of new wheat genotypeswith improved input efficiency andadaptability, especially for marginalenvironments.

This chapter describes methods forconducting root research in the field;most have the advantage of not requiringlarge investments in equipment. Themain focus is on: 1) areas of rootresearch in wheat; 2) genetic diversityand heritability of root traits; 3) examplesof successful root research; 4) root traitspotentially useful in geneticimprovement, and 5) screening methodsand trait measurement techniques.

Areas of Root Research

Root research can be divided into threemain fields of study: root ecology, rootphysiology, and selection of root traitsfor genetic improvement. Root ecologystudies deal with environmental factorsthat influence root growth and are nearlyalways combined with other ecologicalresearch. Important ecological factors aresoil bulk density, soil pH, soil water, andnutrient availability in the soil. Root

physiology, an integral part of plantphysiology, deals with physiologicalprocesses in the roots, such as celldivision, nutrient and water uptake, androot-to-shoot transport mechanisms.Measurable root traits for geneticimprovement are architecture,morphology, number, weight, volumeand diameter, root length density, roothair density, root:shoot ratio, infection byvesicular-arbuscular or arbuscularmycorrhizal fungi [(V)AMF], and rootexudates.

Genetic differences in root traits areessential in the following areas of wheatresearch:

• Nutrient uptake efficiency• Drought tolerance• Tolerance to mineral toxicity• Lodging resistance• Simulation of nutrient and water

uptake

Morphology of WheatRoots

Cereals have two types of roots: seminalroots (also called primary roots), whichdevelop in the embryonic hypocotyl ofgerminating seed, and adventitious roots(also called nodal, secondary, or crownroots), which emerge from the base of theapical culm and tillers. The architectureof wheat roots has been described indetail by Manske and Vlek (2002).

Seminal roots constitute 1-14% of theentire root system. They grow andfunction throughout the vegetativeperiod, and penetrate the soil earlier anddeeper than adventitious roots. In theory,the number of seminal roots may be ashigh as 10, but not all root primordiadevelop. In wheat, three to six seminalroots normally emerge from the seed, butthere is genetic variation for this trait(Robertson et al., 1979). Seed size andseminal root number are positivelycorrelated, though not in all genotypes(O’Toole and Bland, 1987).

Adventitious roots grow mostly in theupper soil layers, and their numberdepends mainly on the plant’s tilleringability. The number of adventitious rootsand tillering are thus positivelycorrelated (Hockett, 1986). The ratio ofseminal to adventitious roots is alteredby the degree of tillering and, later, byinterplant competition.

High-input wheats are characterized by alow number of tillers and a high harvestindex, and depend mainly on seminalroots. In contrast, adventitious roots areessential to low-input wheat genotypes,which develop larger root systems toexplore the greatest volume of soilpossible.

Results at CIMMYT showed that thenumber of tillers was positivelycorrelated with phosphorus acquisitionand grain yield among semidwarf breadwheats grown in P-deficient acid soils(Manske et al., 2000a). Phosphorus

1 Center for Development Research (ZEF), Bonn University, Walter-Flex-Str. 3, 53113 Bonn,Germany, [email protected].

2 CIMMYT Wheat Program, Apdo. Postal 6-641.Mexico, D.F., Mexico, 06600.

209

uptake was not affected by tiller numberwhen phosphorus was amply available.

Roots hairs and extramatrical hyphae ofvesicular-arbuscular or arbuscularmycorrhizal fungi [(V)AMF] enlarge theeffective surface of roots considerably.Thin protrusions from epidermal rootcells, root hairs are 0.003–0.007 mm indiameter and 3-13 mm in length, with anormal lifespan of a few days. Theyemerge just behind the root tip, in thearea of root elongation. The (V)AMFhyphae are 10 times finer than roothairs. The average radius of wheat rootsis 0.07-0.15 mm. Root length densityvaries between 2 and 10 cm cm-3 soildepending on plant development stage,soil depth, and environmental factors.

The size of a wheat plant’s completeroot system depends on theenvironment. The roots’ horizontalspread is usually 30-60 cm (Russell,1977). Roots may be abundant at soildepths of more than 100 cm, with somereaching beyond 200 cm. However,about 70% of total root length is foundwithin the top 0-30 cm of soil (Figure1). This is because roots grow towardsareas of higher nutrient and waterconcentrations (but avoid toxic levels),where root branching can increase.

Heritability andGenetic Diversity ofRoot Traits

Knowledge of heritability andinheritance patterns of morphologicalroot traits in wheat is still restricted, butindicates they are controlled by apolygenic system. Root systems arelargely influenced by additive genes thatmay allow breeding progress to be madeby selecting for root quantity and depthof penetration (Monyo and Whittington,1970). Over 32% of total phenotypicvariability for the root:shoot ratio isconditioned by additive gene effects(Kazemi et al., 1979). Although thisadditive portion of total variance is notespecially high, the authors suggest thatlines with favorable root:shoot ratiosmay be obtained by direct selection inearly generations of wheat crosses.Moderate heritability values for totalroot length (0.62) and root branching(0.42) have been reported (Monyo andWhittington, 1970). Narrow senseheritability for root length rangedbetween 0.38 and 0.46. Manyresearchers have found the number ofseminal roots to be highly heritable(Tiwari et al., 1974; Mac Key, 1973).

Genetic diversity for root traits in breadwheat (Mac Key, 1973) and durumwheat (Motzo et al., 1993) is welldocumented. Considerable variation inthe degree of root branching has been

found among wheat cultivars (O’Brien,1979). Many landraces and wild speciesof wheat possess large root systems, buttend to lodge due to their height(Manske, 1989; Vlek et al., 1996). Theextent of rooting can be modified throughselection during breeding, regardless ofdwarfing genes or shoot dry matter (MacKey, 1973; Gale and Youssefian, 1985).McCaig and Morgan (1993) found nosignificant relationship between dwarfinggenes and root dry matter.

At CIMMYT, selecting for tolerance toacid soils, low phosphorus conditions,and aluminum toxicity has resulted insuperior semidwarf bread wheats withhigher grain yield, improved phosphorusuptake, and greater root length densitycompared with outstanding tall cultivarsfrom Brazil (Table 1)(Manske et al.,1996; Egle et al., 1999).

Impact of Root Traitson Wheat Growth

Good root growth is a prerequisite forimproved shoot growth and higher yields,especially in marginal environments(Manske and Vlek, 2002). Complexinteractions between roots and soil takeplace in the rhizosphere, whichcommonly includes 20-30% of topsoilvolume. The extent of soil exploration isoften dependent on water and nutrientavailability. Plants may respond tonutrient and water stress by altering root

Figure 1. Distribution of root length densityin different soil layers; mean of 12semidwarf bread wheats grown on residualsoil moisture in Cd. Obregon, Mexico.

Table 1. Grain yield, P uptake, and root length density, means of 8 semidwarf and 8 tallbread wheats from Brazil, grown without and with P fertilization (0 or 80 kg P2O5 ha-1),with irrigation in a calcareous Aridisol.

High P regime ** Low P regimeSemidwarf Tall Semidwarf Tall

Yield (12% moisture) (kg ha-1) 6240 b 4598 a 4539 b 3169 a

P uptake into total above ground biomass (kg P ha-1) 22.5 b 19.2 a 11.3 b 9.5 a

Root length density (cm/cm3) 10.4 b 8.7 a 7.7 b 6.6 a

** a<b for LSD P=0.05, separately calculated for P levels.Source: Manske (unpublished data).

Root length density (cm/cm3)0.0 0.5 1.0 1.5 2.0 2.5

0-20

20-40

40-60

60-80

80-100

Soil d

epth

(cm)

TECHNIQUES FOR MEASURING GENETIC DIVERSITY IN ROOTS

210

branching and extension rates (Figure 2),rate of uptake per unit root length orweight, partitioning between roots andshoots, root exudates and (V)AMFinfection, and by a lower demand fornutrients and water. Each of theseparameters can be altered by selectionand breeding. Wheat exhibits remarkableplasticity in root growth, which adjusts tosoil nutrient and water status (Cholick etal., 1977; Vlek et al., 1996). Thepartitioning of assimilates between rootsand shoots appears to be highlydependent on genotype (Sadhu andBhaduri, 1984).

Nutrient use efficiencyRoot system geometry is essential toimproving nutrient uptake andmaximizing roots per unit soil volume inareas where nutrients and water areavailable. Root length is a majordeterminant of the absorbing surfacearea. Wheat genotypes with higher rootlength density are able to take up morephosphorus (Manske et al., 2000a). Rootfineness or branching is also an importantdeterminant of P uptake efficiency inwheat (Jones et al., 1989).

Under low phosphorus conditions, thecorrelation between root length densityand P uptake or grain yield is usually0.50-0.60, but decreases when P supply is

adequate. The benefit of high root lengthdensity generally diminishes in irrigatedwheat with high fertilizer input. If thedemand for carbohydrates in large rootsystems is not compensated for byimproved P and water acquisition (ashappens when there is adequate waterand nutrient supply), the rootsthemselves may limit yield.

Under high P and good water supply,root hairs are rudimentary, whereasunder deficient conditions long root hairsare abundant (Foehse and Jungk, 1983).Root hair length and density modify Pdepletion profiles in the rhizosphere(Gahoonia and Nielsen, 1996). Root hairdensity scored in plants derived from potcultures in quartz sand variedconsiderably among 19 bread wheatgenotypes. Root hair density scores werepositively correlated with P uptake atanthesis when plants were grown in a Pdeficient calcareous Aridisol innorthwestern Mexico.

Vesicular-arbuscular or arbuscularmycorrhizal fungi [(V)AMF] areessential for uptaking nutrients that arerelatively immobile in soil, such as P(Hayman and Mosse, 1971), Cu (Gildonand Tinker, 1983), and Zn (Swaminathanand Verma, 1979). Screeningexperiments with wheat have shownconsiderable genotypic variability in

(V)AMF colonization and efficiency(Bertheau et al., 1980; Manske, 1990a;Kalpulnik and Kushnir, 1991; Hetrick etal., 1992). Both the extent of (V)AMFinfection and the degree of benefit from(V)AMF are heritable traits (Manske,1990b; Vlek et al., 1996).

Botanically, mycorrhiza is the symbioticassociation between soil-borne fungi androots of higher plants. Endomycorrhizalfungi are obligate symbiotic fungi thatcannot as yet be cultured on artificialmedia. Extramatrical hyphae grow fromthe roots into the rhizosphere and canexplore an area around the roots that farexceeds that available to root hairs. It isthe ability of these hyphae to absorbrelatively immobile elements (P, Zn, Cu)in this additional soil that enablesendomycorrhizal fungi to benefit plants,especially those with poorly developedroot systems.

Efficiency of (V)AMF infection isinfluenced by soil type. In a P-fixingAndisol, P uptake into wheat shoots andP uptake per unit root length waspositively correlated with (V)AMFinfection rate. In a calcareous Aridisolthese correlations were negative. Therole of (V)AMF in calcareous soils isstill questionable (Bolan, 1991). In bothsoils, P uptake was positively correlatedwith root length density (Table 2).

Figure 2. Wheat root system for dryland andirrigated conditions.Source: Weaver (1926).

Table 2. Correlation between P uptake, root length density, AMF infection rate, and P uptakeper root length of semidwarf bread wheats grown in field-inserted pots with calcareousAridisol from Cd. Obregon and acid Andisol from Patzcuaro, Mexico.

Calcareous AridisolP uptake intoabove-ground Root length AMF infection P uptake per

biomass density rate root length

P uptake intoabove-ground biomass 0.71 -0.73 0.83Root length density 0.86AMF infection rate -0.67P uptake per root length 0.82

Source: Buddendiek (1998).

Acid

Andis

ol

30 cm 30 cm

Dryland conditions

Irrigated conditions

G.G.B. MANSKE, J.I. ORTIZ-MONASTERIO, AND P.L.G. VLEK

211

Wheat genotypes may have differentialcapacities to utilize poorly available soilP. Genotypic differences in rootphosphatase excreted or bound at theroot surface have been identified(McLachlan, 1980). Organic P in thesoil, which commonly accounts for 40-50% of the total P supply of plants, canbe utilized by root phosphatase (Helal,1990). The acid phosphatase activity of42 bread wheat genotypes from theCIMMYT collection that were screenedin nutrient solution was positivelycorrelated with P uptake efficiency ingenotypes grown in the field in a Pdeficient Andisol.

Drought toleranceTraits associated with high water uptakeefficiency are similar to those associatedwith nutrient uptake efficiency.However, soil water status is moredynamic in nature. Thus, a well-developed root system combines theability to reach residual moisture deep inthe soil profile with a high degree ofplasticity that allows it to adjust to rapidchanges in topsoil water status. Wheatgrown under residual moisture dependson deeper roots to access soil moisture

from the deep soil profile (Jordan et al.,1983; Mian et al., 1993). In rainfedwheat, root length densities are muchhigher in drier years (Hamblin et al.,1990). Varietal differences in rootingdepth of wheat were demonstrated byHurd (1968). In a study conducted atCIMMYT, most drought tolerantsemidwarf bread wheats (Pastor,Synthetic 2, Sujata, and Nesser) formedmore roots in deeper soil layers, whereasthe non-tolerant checks (Tevee 2, Pavon,and CRC) had fewer roots in deep soil(Manske et al., 2000b).

Root Parameters andMethods for TheirDetermination

Advances in our knowledge of root traitsdiscussed so far have been due largely tothe use of relatively simple, but time-consuming methods. Recentlydeveloped techniques (Table 3), thoughoften more precise and quicker, areexpensive and/or not suitable for fieldstudies on many genotypes. The oldertechniques were comprehensivelyreviewed by Böhm (1979).

Methods used in root research can betermed as descriptive or quantitative.Many descriptive methods are used toestimate root quantities in a time- andlabor-efficient manner. They are non-destructive and provide only limitedinformation about roots of field-grownwheat.

Descriptive methods used tomeasure root distributionThe core break method. A root auger isused to take soil-root cores in the field.The bi-partite root auger (for example,from FA. Eijkelcamp) (diameter: 8 cm;length: 15 cm; volume: 750 cm3) issuitable for hard, heavy soils. Introducethe auger into the ground using animpact-absorbing hammer with a nylonhead (ø 70 mm; weight: 2.3 kg). Now putthe complete auger cylinder into the soiland then extract by rotating clockwise.Invert the auger and turn the crank handleto force the soil core out of the cylinder(Picture 1a).

The amount of roots in a sample (rootlength density or root mass) is estimatedby breaking the soil core horizontallythrough the middle and counting the rootsexposed on both faces of the breakage(Picture 1b). Every exposed root iscounted, regardless of its length. Thenumber of roots visible on the surface of abroken sample is expressed as a surface-covering figure. Accuracy increases whenthe sample is broken in more than oneplace. Each exposed surface has only onesurface-covering figure. An average of allfigures is calculated for more reliableresults. Counting is greatly facilitated ifthe two breakage faces are wetted with asprayer to make the roots more visible(Köpke, 1979).

The exposed surface must be comparedwith a reference, consisting of a numberof circles having the same diameter as theroot auger, to depict the increasing rootintensity and estimate a constant C for

Table 3. Methods of root research.

Non-sophisticated methods New, sophisticated methods, Methods feasible only forfeasible for root research in expensive and not feasible detailed studies of ecologicalplant breeding for many genotypes and physiology root research

Core break method Minirhizothron Trench profile methodElectrical capacitance method Radioactive tracer methods Profile wall methodMeasuring root-pulling strength Non-radioactive tracers Glass wall methodRoot angle method RhizothronMesh bag, container method Excavation methodSampling methods (spate, root box, Split-root technique

root auger)Root washing techniquesRoot parameters for determination

after washing: Root number,weight, surface, volume, diameter,root length density, root/ shoot ratio

Image analyzer methodsRoot hairs, arbuscular mycorrhizaeRoot studies in nutrient solution


212

root length density (cm root length/cm3

soil). The root length density is thenestimated by multiplying rootintersections by constant C. Most studiesusing the core break method have beenconducted on gramineaceous species.Bland (1989) demonstrated the use ofthe core break method in wheat.

The mesh bag method. The dynamics ofroot growth and root turnover can bestudied in root-free soil placed in meshbags (nylon stockings can be used). Themesh bags are placed in holes in the fieldand removed at established intervals.Roots that have invaded the mesh bagsare measured and used as an index tocalculate root productivity (Fabiao et al.,1985). The disadvantage of this methodis that the soil is disturbed when the bagsare placed in it, which can alter rootgrowth (Fitter, 1982).

The electrical capacitance method. Rootvolume and root length density can beestimated by the electrical capacitancemethod (Dalton, 1995). This in-situmethod is based on measuring theelectrical capacitance of an equivalentparallel resistance-capacitance circuitformed by the interface between soil-water and the plant root surface. Themethod is adequate for measuring the

root surface of field-grown plants quicklywithout using labor-intensive proceduressuch as root sampling, washing, andcounting. However, this technique doesnot measure the spatial distribution ofroots in a soil profile.

Van Beem (1996) developed such atechnique for maize (Figure 3). Thenegative electrode of a capacitance meter(a BK Precision 810A instrument set atthe 200 nF level) was attached to themaize stem exactly 6 cm above the soilsurface. The positive electrode wasattached to a copper ground rod, inserted15 cm deep into the soil, 5 cm from the

base of the maize stem. Capacitancereadings were taken at maximum soilmoisture after irrigation, 5 seconds afterthe meter was turned on.

It is more complicated to take thesemeasurements on wheat because eachplant has several tillers. An apparatusdeveloped at CIMMYT measures thecapacitance of all wheat stems and theirroots growing in a 30-cm planting row,but more research is needed before it canbe used routinely.

Rhizothron and minirhizothron. Rootgrowth can be observed by using a rootperiscope (minirhizothron) to peerthrough a glass or acrylic tube inserted inthe soil. This technique involves the useof cameras coupled with endoscopes orminiaturized color video cameras, andthus is more costly (Upchurch andRitchie, 1983). A large computer systemand special software are needed toanalyze large numbers of root imagesquantitatively. The system can provideimages of small root hairs and may beused for fractal analysis of root systems.

Several authors have compared this newtechnique with the unsophisticated soil-core root-counting method. Root lengthdensities derived from the two methodsare not always well correlated (Majdi etal., 1992; Volkmar, 1993; Box andRamseur, 1993). Usually, the number ofroots encountering the tube (barrier-soilinterface) exceeds the number of rootspassing through an analogous area ofsoil. Roots encountering vertical tubestend to grow along the tube.

The wax layer method. The ability ofcereal roots to penetrate compacted soilshelps them avoid drought stress. Rootpenetration can be screened by usingwax-petrolatum layers with definedresistance strength (Yu et al., 1995).

The root angle method. Deep rooting isassociated with drought tolerance ofwheat grown under residual moisture.

Picture 1. (a) root auger; (b) breakingthe 10-cm long soil core to counting theroots at the breakage faces.

Figure 3. Electrical capacitance measurementof root size in a maize plant.Source: van Beem (1996).

a

b


213

The gravitropic response of roots axesdetermines the shape and spatialdistribution of root systems in cereals.The plagiogravitropic response of rootsdepends on the growth angle of the rootaxes, which is controlled by genetic andenvironmental factors.

Genotypic characteristics of the verticaldistribution of wheat roots in the fieldcan be estimated based on the growthangle of seminal roots of wheatseedlings grown in baskets. The basketscan be buried in the field or used as potsin the greenhouse. A wheat grain is sown1 cm deep with the tip of the embryo atthe center of the soil-filled basket. Afterseveral weeks, the angle between thevertical axis and the line connecting thegrain and the mesh of the basketsthrough which the root appeared isdetermined (Figure 4).

Root growth angle is negativelycorrelated with root length density in thetop 0-10 cm of soil, but positivelycorrelated with root length density at adepth of 10-30 cm (Nakamoto andOyanagi, 1994; Oyanagi et al., 1993).This method thus promises to determinegenetic differences in deep-rootingability, a trait important for droughttolerance. However, more root data fromthe field need to be compared toestablish its reliability.

Root pulling strength. Root traitscontribute to lodging resistance in wheat,as do dwarfness, resistance, and stemelasticity. The spreading angle of root

systems (Pinthus, 1967), root-clumpweight (Thompson, 1968), root tensilestrength (Spahr, 1960) and root pullingstrength (Ortmann et al., 1968) are allrelated to lodging resistance. Simpleroot-pulling machines have beenconstructed for use in maize (Ortmann etal., 1968). Vertical pulling strength ismeasured with a tensiometer.Homogeneous field soil is required forreliable measurements with this method.Despite its simplicity, this method israrely used in field studies.

Quantitative methodsAssessing root parameters from rootssampled in the field. This approach hasbeen widely applied to measure rootlength density, depth, and lateral extentof root systems. Roots are separated fromthe soil to determine root fresh weight,dry weight, (V)AMF infection, and, ifneeded, nutrient content. Recoveringroots from the field involves takingsamples, removing roots from the soil,separating roots from organic debris,storing the roots, and determining andcalculating several root parameters. Withthis approach the root system can only beobserved at one point in time, unlikeexcavation, which allows an accuratedescription of the morphology of thewhole root system.

Sampling roots in the field. Roots from adefined volume of soil can be extractedin the form of monoliths and/or soilcores. Root core samples can be takenwith a root auger, as previously described(Figure 5). Soil monoliths can becollected by driving a root box (metalframe) into the soil or by using spates.

The root box supports the monolith, whichis important if the soil lacks sufficientstructure.

Metal boxes of the required size are easyto build. Root boxes are open at thebottom and the top. An extra sheet ofmetal is added to reinforce the upper rimso it will resist being hammered into thesoil (Picture 2). Although an impact-absorbing hammer with a nylon head (2kg, Δ 70 mm) is preferred for driving theroot box into the soil, a rubber or woodenhammer—but not a metal one—can alsobe used. Heavy loamy soils must have theright soil moisture content for rootsampling (not too dry, not too wet). Sandysoil should be wet; otherwise, parts of themonolith may be lost from the bottom.

The size of the root box depends on thecrop and field design. Cereal crops shouldbe planted in rows, not broadcast, whichcauses high root variation. The box shouldcover the width of one row completely.The sampling area consists of twosymmetrical halves on either side of therow. The ideal depth of a root box is 20-30cm, and its length should be 30-50 cm.Usually, 70-80% of the total root systemis found within the top 20 cm of soil.

Figure 5. Root auger.

Picture 2. Root box with extracted root-soilmonolith together with wheat shoots.

Figure 4. Determination of root growthangle by growing a wheat plant in a basket.

Grain

Root

1

5 6

32 7

4

Root auger1: Cylindric auger body2: Pinion casing3: Crank handle4: Pressure plate5: Pressure plate shaft containing rack6: Pinion7: Lubrication nipple


214

Sampling wheat roots with an auger isanother option (Kumar et al., 1993). Itsadvantage is that the soil is minimallydisturbed, but sample volume is small.Samples are normally taken at random,but how many samples per plot shouldbe taken to obtain reliable rootinformation is difficult to determine.Böhm (1979) recommended five boreholes per plot. The position of thesampling sites within or between thewheat rows is also very important, anddifferent methods are used. Gajri andPrihar (1985) employed two samplingsites, on the plant row and midwaybetween the rows. Gregory et al. (1978)sampled wheat roots between plant rows.Gajri et al. (1994) analyzed horizontalroot distribution between wheat rows toidentify the ideal site for augering.

Excavating the root/soil sample in a rootbox is much faster than using a rootauger (Vepraskas and Hoyt, 1988), butthe auger allows deeper sampling.However, spatial variability rises as soildepth increases. The root box and theroot auger can be combined as follows.The first 0-20 cm are sampled with a rootbox, which gives the best accuracy in thetop soil, and the root auger is used deeperin the profile. At CIMMYT, six boreholes were sufficient for 1-m depth. TheCV for root length density was about27%, and significant differences betweenwheat cultivars were observed.

Samples can be stored for 2-3 weeks ifprotected against the rain and hightemperature. Plastic bags may be used,but they should have holes and must bekept open to allow ventilation and keepthe samples from getting moldy.

Removing roots from the soil. Böhm(1979) reviewed various washing andfloating procedures, as well as chemicaldispersing agents. Especially in the caseof heavy, loamy soil, samples should besoaked overnight in a saturated NaCl orsoap solution to increase buoyancy anddisperse soil aggregates.

The simplest and often the mosteconomic technique of separating rootsfrom soil is to wash them by hand. Firstthe soil-root sample is suspended inwater and poured into sieves where rootsare retained and collected for furthercleaning. The sieve mesh should not betoo fine (or the procedure becomes verytime consuming) or too coarse (or it willnot retain very fine roots). Ordinaryplastic sieves (10-mm mesh size) areadequate. The roots are then washed byhand under a water jet or spray. Thistechnique can be used in remote fieldsites when large amounts of soil/rootsamples cannot be transported to thelaboratory. The roots may even bewashed in irrigation channels or nearbyrivers. In this case, the roots can bewashed in baskets that act as sieves.

A root-washing machine can be effectivewhen manual labor is in short supply, butoften takes longer, especially with largersoil volumes. A machine cannot washroots better than can be done by hand,and using tweezers to separate debrisfrom the roots is still necessary. Smuckeret al. (1982) devised a root washingapparatus, termed a hydropneumaticelutriation device. The device has a high-kinetic-energy first stage in which waterjets erode the soil from the roots and asecond low-kinetic-energy flotationstage in which the roots are deposited ina submerged sieve. The procedure isquick (3-10 min for one core sample),has a high root recovery rate, and doesnot sever laterals. Its biggest advantageis greatly improved consistency insample handling. Root washers with fouror eight tubes that can wash severalsamples at a time are available on themarket; one to four people can use themachine simultaneously.

Separating roots from organic debris.Roots must be separated manually whenthere is an appreciable amount of organicdebris. Distinguishing between living

and dead roots (debris) is usually donesubjectively based on criteria such ascolor, but methods for objectivelydetecting live tissue have beenestablished. For example, a tetrazoliumdye technique suitable for estimating theproportion of live material was reportedby Joslin and Henderson (1984), andWard et al. (1978) used congo red forquantitatively estimating living wheat-root lengths in soil cores. These methodsnotwithstanding, many workers stillprefer to subjectively separate largenumbers of root samples by sortingthrough them manually.

Storing washed roots. Washed roots maybe stored before further processing, forexample, in small plastic bags containing50% alcohol in a refrigerator at 7°C(freezing destroys root structures).

Assessing root parameters afterwashing. Total root length or root lengthdensity are measured on representativesubsamples of the whole (washed) rootsystem, especially for plant species withfine roots, such as wheat. At CIMMYT,the following method was developed forrepresentative subsampling: a washedroot system, free from organic debris, iscut with scissors into 1-2 cm pieces thatare placed in a small bowl of water andmixed. The water is poured through asieve to collect the root pieces, and totalfresh weight is determined. Subsamplesare extracted and weighed, and rootlength measured.

The simplest way of measuring freshweight is by blotting roots with blottingpaper. Since the accuracy of this methodis affected by how the sample is handled,the same person should always do theprocedure. To determine dry weight,clean roots are dried in an oven at 60-70ºC (higher temperatures couldpulverize the roots) for about 24 h.

G.G.B. MANSKE, J.I. ORTIZ-MONASTERIO, AND P.L.G. VLEK*

215

Measuring root length, diameter, andsurface area. Root length per unit ofsoil volume (root length density) is oneof the best parameters for studyingwater and nutrient uptake by plants (Nyeand Tinker, 1969; Claassen and Barber,1974). The line-intercept methoddeveloped by Newman (1966) andmodified by Tennant (1975) counts thetotal number of intersections betweenroots and the vertical and horizontallines of a grid drawn on a plastic petridish (Picture 3). Root length can beestimated by the equation

R = π AN / 2 H,

where R is the total length of the rootsin area A of the petri dish, and N is thenumber of intersections between theroots and random horizontal grid linesof total length H. For a grid ofindeterminate dimensions, theintersection counts can be converted tocentimeters using the equation

Root length (R) = 11/14 x number of intersections(N) x grid unit

As proposed by Tennant (1975), 11/14in the equation can be combined withthe grid unit to obtain a lengthconversion factor. The factor for 1 cm is0.786. If 1/2 inch (1.27 cm) is used as thegrid unit, each intersection countrepresents 1 cm of root length (because

1.270 x 0.786 = 1.00, exactly 0.998),and the equation for calculating the totalroot length per sample is

R = N x fresh weight of total sample/fresh weight of counted subsample

The size of the grid and of thesubsample used for counting depends onthe amount of roots to be measured.According to Köpke (1979), the numberof intersections counted in a root sampleshould not be above 400, to keep theoperator from tiring, or below 50, soaccuracy will not be affected. For wheat,200-300 gives maximum accuracy.Subsample size should be 0.1 g forcoarse roots and 0.05 g for fine roots.Depending on location and soiluniformity in the field, the coefficient ofvariation for root length density rangedbetween 20 and 35%.

Image analysis. Instead of countingroots tediously by hand, one can useimage analysis systems, which havebecome available on the market in thelast few years. An image analysis systemis a multipurpose instrument thatmeasures leaf area, numbers of objects(seeds), different portions of an area(leaf necrosis due to disease), and,through root analysis, root length,diameter, and surface. It usually consistsof a computer, a video image monitor,and a video camera or scanner.

The correlation between image analysisand the manual line-intercept methoddepends on calibration and the type ofimaging system used. The digitalmethod often underestimates root lengthbecause it does not detect small, fineroot systems (Farrell et al., 1993).However, the technology is progressingrapidly, and the latest image analyzersystems have improved resolution andlighting systems that can distinguishbetween fine and coarse roots, andcorrect for overlapping roots (Arsenaultet al., 1995).

Root radius. Average root radius can becalculated from root fresh weight andtotal root length. Assuming that thewheat root contains almost 90% water,the specific weight is almost 1, and theformula for obtaining the average rootradius is

r = (root fresh weight/total root length/π)-1/2

Shoot:root ratio. A common parameterused to study the relation between above-and below-ground plant growth is theroot:shoot ratio. The root length:shootdry weight ratio gives more informationthan root weight:shoot weight, whetherfresh or dry. Relationships between rootlength and plant development stages canbe observed.

Root hair density. Quantifying root hairs(Picture 4) is difficult because rootlength density varies considerably withinthe same root system. Some root units,usually older ones, have no hairs, whileothers, usually younger roots, have densehairs. At CIMMYT, a scoring methodwas used to determine genotypicdifferences in root hair density in wheat:Hundreds of intersections between rootsand grid lines in a petri dish were scoredat random using a microscope (0 = noroot hairs to 5 = very dense). Averageroot hair density for each sample wascalculated. Though the number of roothairs can usually be counted, this is notpossible in wheat, because there are toomany. The diameter and length of thehairs can be measured by means of abinocular or a microscope.

Vesicular-arbuscular or arbuscularmycorrhizal fungi [(V)AMF]. Hyphae ofendomycorrhizal fungi develop mycelia,arbuscules, and/or vesicles in roots. Inaddition, extramatrical hyphae growfrom the root into the rhizosphere soil(Picture 5) and can explore an areaaround the root which far exceeds thatavailable to root hairs. To quantify AMinfection, the root sample (usually a

Picture 3. Plastic petri dish with grid line,grid unit 1/2 inch (1.269 cm).


216

representative subsample) is cleaned andstained (Philipps and Hayman, 1970),and examined under a microscope. Amodified staining method (Giovannettiand Mosse, 1980) may be used thatstains the root without the use of phenol,which is toxic and carcinogenic.

Studies using containers andmesh bagsRoot properties can be assessed onwheat plants grown in containers, but theresulting data should be viewed withcaution, since the container is often toosmall for unrestricted root growth, androots concentrate at the container wall.Root data obtained from mesh bagsburied in the soil give a more reliablepicture of the field situation. However,here also, taking root samples directlyfrom field soil may be a betteralternative.

Root studies in nutrientsolutionRoot systems grown in nutrient solutionscan be observed and studied in situwithout time-consuming washing andcleaning. This method allows a largenumber of plants to be studiedsimultaneously within a short time.However, growth conditions in nutrientsolutions differ greatly from conditionsin soil and water, and results from thesestudies must be compared with results offield experiments. Mian et al. (1994)

found that wheat genotypes with largeroot systems in hydroponic culture mayalso produce more roots in the field.

Aluminum toxicity in acid soils inhibitsroot growth by preventing cell division,which results in reduced root penetrationin the soil and leads to reduced nutrientand water uptake. At CIMMYT, wheatgenotypes are routinely screened fortolerance to aluminum toxicity (Kohliand Rajaram, 1988). Hundreds of wheatseedlings are screened simultaneously innutrient solution with 46 ppm ofaluminum, which irreversibly damagesthe root meristem of susceptiblegenotypes. After four days’ growth, theroots are stained with 0.2% hematoxylinsolution and transferred to an aluminum-free solution. Selection of tolerantgenotypes is based on root regrowthbeyond the hematoxylin-stained celllayers (Rajaram and Villegas, 1990).

Conclusions

Wheat breeders and physiologists whodecide to study wheat roots should beaware that it can be very labor intensive.

This is the reason most breedingprograms have largely avoided rootstudies. Nonetheless, assessing root traitsis essential in some cases (e.g., toimprove nutrient acquisition, droughttolerance, and other adaptationmechanisms).

Methods exist to study root traitsindirectly for different purposes, forexample, to determine genotypicdifferences in phosphorus uptakeefficiency in wheat. Root angle and waxlayer methods may provide knowledgeabout deep rooting of wheat for droughttolerance. Root volume studied innutrient solution may indicate how rootsgrow in the field. Wheat grown innutrient solution is routinely used atCIMMYT to screen for aluminumtolerance. The electrical capacitance ofroot systems is a promising method forindirectly studying root dimensions inthe field. However, this method needs tobe improved before it can be usedroutinely.

Since direct assessment of root traitscannot be avoided in many cases,techniques for conducting such studiesare presented here. Most of them arelabor intensive, but not sophisticated,and easy to use in the field. Actually,sampling roots in the field and washingand counting them by hand have anadvantage in regions where labor costsare low. In other cases, manual labor maybe the best option, and may comparefavorably with many new techniques.

Root washing machines and new,improved image analysis systems areaffordable and practical in most researchsituations. Their biggest advantages aregreatly improved consistency in handlingand measuring samples and higher laborcost effectiveness. However, these fieldmethods do root analysis only on a smallnumber of genotypes, for example, foridentifying parental lines. Screeninglarge numbers of wheat lines in

Picture 4. Under the microscope: (a) roothairs, and (b) extramatrical hyphae of(V)AM fungi and wheat roots.

a b


217

segregating populations is onlypracticable if root analysis is doneindirectly, for instance, in nutrientsolution. More in-depth studies, such as(V)AMF infection, root geometry, rootmorphology, and the minirhizothrontechnique are not practical for use inselection.

References

Arsenault, J.-L., Pouleur, S., Messier, C., andGuay, R. 1995. WinRHIZOTM, a root-measuring system with a unique overlapcorrection method. Hort Sci. 30:906.

Bertheau, Y., Gianinazi-Pearson, V., andGianinazi, S. 1980. Development andexpression of endomycorrhizal associationsin wheat. I. Evidence of varietal effects.Annales de l’Amélioration des Plantes30:67-78.

Bland, W.L. 1989. Estimating root length densityby the core-break method. Soil Sci. Soc. Am.J. 53:1595-1597.

Böhm, W. 1979. Methods of Studying RootSystems. Springer Verlag, New York.

Bolan, N.S. 1991. A critical review on the role ofmycorrhizal fungi in the uptake ofphosphorus by plants. Plant and Soil134:189-207.

Box, J.E., and Ramseur, E.L. 1993.Minirhizothron wheat root data: Comparisonto soil core root data. Agron. J. 85:1058-1060.

Buddendiek, R. 1998. The effect of (V)Amycorrhiza on the P efficiency of differentwheat genotypes. Master Thesis, Institute ofAgronomy in the Tropics (IAT), Universityof Göttingen, Germany.

Cholick, F.A., Welsh, J.R., and Cole, C.V. 1977.Rooting patterns of semidwarf and tallwinter wheat cultivars under dryland fieldconditions. Crop Sci. 17:637-639.

Claassen, N., and Barber, S.A. 1974. A method forcharacterizing the relation between nutrientconcentration and flux into roots of intactplants. Plant Physiol. 54:564-568.

Dalton, F.N. 1995. In-situ root extentmeasurements by electrical capacitancemethods. Plant and Soil 173:157-165.

Egle, K., Manske, G.G.B., Römer, W., and Vlek,P.L.G. 1999. Improved phosphorusefficiency of three new wheat genotypesfrom CIMMYT in comparison with an olderMexican variety. J. Plant Nutri. Soil Sci.162:353-358.

Fabiao, A., Persson, H.A., and Steen, E. 1985.Growth dynamics of superficial roots inPortugese plantations of Eucalyptus globulusLabill. studied with a mesh bag technique.Plant and Soil 83:233-242.

Farrell, R.E., Walley, F.L., Luckey, A.P., andGermida, J.J. 1993. Manual and digital line-intercept methods of measuring root length –a comparison. Agron. J. 85:1233-1237.

Fitter, A.H. 1982. Morphogenetic analysis of rootsystems: application of the technique andinfluence of soil fertility on root systemdevelopment in two herbacaceous species.Plant, Cell Environm. 5:313-322.

Foehse, D., and Jungk, A. 1983. Influence ofphosphate and nitrate supply on root hairformation of rape, spinach and tomatoplants. Plant and Soil 74:359-368.

Gahoonia, T.S., and Nielsen, N.E. 1996. Variationin acqusition of soil phosphorus amongwheat and barley genotypes. Plant and Soil178:223-230.

Gajri, P.R., and Prihar, S.S. 1985. Rooting, wateruse and yield relations in wheat on loamysand and sandy loam soils. Field Crops Res.12:115-132.

Gajri, P.R., Arora, V.K., and Kumar, K. 1994. Aprocedure for determining average rootlength density in row crops by single-siteaugering. Plant and Soil 160:41-47.

Gale, M.D., and Yousefian, S. 1985. Dwarfinggenes in wheat. In: G.E. Russel. Progress inplant breeding. Vol. 1. ButterworthScientific, London. pp. 1-35.

Gildon, A., and Tinker, P.B. 1983. Interactions ofvesicular-arbuscular mycorrhizal infectionand heavy metals in plants. I. The effects ofheavy metals on the development ofvesicular-arbuscular mycorrhizas. NewPhytol. 95:247-261.

Giovannetti, M., and Mosse, B. 1980. Anevaluation of techniques for measuringvesicular-abuscular mycorrhizal infection inroots. New Phytol. 84:489-500.

Gregory, P.J., McGowan, M., Biscoe, P.V., andHunter, B. 1978. Water relations of winterwheat. 1. Growth of the root system. J.Agric. Sci. Camb. 91:91-102.

Hamblin, A., Tennant, D., and Perry, M.W. 1990.The cost of stress: dry matter partitioningchanges with seasonal supply of water andnitrogen to dryland wheat. Plant and Soil122:47-58.

Hayman, D.S., and Mosse, B. 1971. Plant growthresponse to vesicular-arbuscular mycorrhiza.I. Growth of Endogone inoculated plants inphosphate deficient soils. New Phytol.70:19-27.

Helal, H.M. 1990. Varietal differences in rootphosphatase activity as related to theutilization of organic phosphates. Plant andSoil 123:161-163.

Hetrick, B.D.A., Wilson, G.W.T., and Cox, T.S.1992. Mycorrhizal dependence of modernwheat varieties, landraces and ancestors.Can. J. Bot 70:2032-2040.

Hockett, E.A. 1986. Relationship of adventitiousroots and agronomic characteristics inbarley. Can. J. Plant Sci 66:257-266.

Hurd, E.A. 1968. Growth of roots of sevenvarieties of spring wheat at high and lowmoisture levels. Agron. J. 60:201-205.

Jones, G.P.D., Blair, G.J., and Jessop, R.S. 1989.Phosphorus efficiency in wheat – a usefulselection criterion. Field Crops Res.21:257-264.

Jordan, W.R, Dugas, W.A., and Shouse, P.J. 1983.Water Management 7:281-289.

Joslin, J.D., and Henderson, G.S. 1984. Thedetermination of precentages of living tissuein woody fine root samples usingtriphenyltetrazolium chloride. For. Sci.30:965-970.

Kapulnik, Y., and Kushnir, U. 1991. Growthdependency of wild, primitive and moderncultivated wheat lines on vesicular-arbuscular mycorrhiza fungi. Euphytica56:27-36.

Kazemi, H., Chapman, S.R., and McNeal, F.H.1979. Components of genetic variance forroot/shoot ratio in spring wheat. Proc. 5th

Intern. Wheat Genet. Symp., New Delhi.pp. 597-605.

Kohli, M.M., and Rajaram, S. 1988. Wheatbreeding for acid soils: Review of Brazilian/CIMMYT Collaboration 1974-1986.Mexico, D.F.: CIMMYT.

Köpke, U. 1979. in Vergleich von Feldmethodenzur Bestimmung des Wurzelwachstums beilandwirtschaftlichn Kulturpflanzen. Diss.Agric. Göttingen.

Kumar, K., Prihar, S.S., and Gajri, P.R. 1993.determination of root distribution of wheatby auger sampling. Plant and Soil149:245-253.

MacKey, J. 1973. The wheat root. Proc. 4th Int.Wheat Genet. Symp., Columbia, MO.pp. 827-842.

Majdi, H., Smucker, A.J.M., and Persson, H.1992. A comparison between minirhizothronand monolith sampling methods formeasuring root growth of maize (Zea maysL.). Plant and Soil 147:127-134.

Manske, G.G.B. 1990a. Genetical analysis of theefficiency of VA mycorrhiza with springwheat. I. Genotypical differences andreciprocal cross between an efficient andnon-efficient variety. In: Genetic Aspects ofPlant Mineral Nutrition. N. El Bassam et al.(eds.). pp. 397-405.

Manske, G.G.B. 1990b. Genetical Analysis of theefficiency of VA mycorrhiza with springwheat. Agriculture Ecosystem 29:273-280.

Manske, G.G.B. 1989. Die Effizienz einerBeimpfung mit dem VA-MykorrhizapilzGlomus manihotis beiSommerweizengenotypen und ihreVererbung in F1- und R1-Generationen beiverschiedenen Phosphatdüngungsformenund Witterungsbedingungen. DissertationGöttingen.


218

Manske, G.G.B., J.I. Ortiz-Monasterio, S.Rajaram, and P.L.G. Vlek. 1996. Improvedphosphorus use efficiency in semidwarf overtall wheats with and without P fertilization.Second International Crop ScienceCongress, New Delhi, India.

Manske, G.G.B., Ortiz-Monasterio R., J.I., vanGinkel, M., González, R.M., Rajaram, S.Molina, E., Vlek, P.L.G. 2000a. Traitsassociated with improved P-uptakeefficiency in CIMMYT´s semidwarf springbread wheat grown on an acid Andisol inMexico. Plant Soil 212:189-204.

Manske, G.G.B., Tadesse, N., van Ginkel, M.,Reynolds, M.P., and Vlek, P.L.G. 2000b.Root morphology of wheat genotypes grownin residual moisture. In: Sustainable LandUse in Deserts. S.W. Breckle, M. Veste, andW. Wucherer (eds.). Springer-Verlag, Berlin.

Manske, G.G.B., Ortiz-Monasterio, I., van Ginkel,M., González, R.M., Fischer, R.A., Rajaram,S., and Vlek, P.L.G. 2001. Importance of Puptake efficiency vs. P utilization for wheatyield in acid and calcareous soils in Mexico.Eur. J. Agron. 14(4):261-274.

Manske, G.G.B., and Vlek, P.L.G. 2002. Rootarchitecture: Wheat as a model plant. In:Plant Roots: The Hidden Half. 3rd edition. Y.Waisel, A. Eshel, and U. Kafkafi (eds.). NewYork: Marcel Dekker Inc. pp. 249-259.

McCaig, T.N., and Morgan, J.A. 1993. Root andshoot dry matter partitioning in near-isogenicwheat lines differing in height. Can. J. PlantSci. 73:679-689.

McLachlan, K.D. 1980. Acid phosphatase activityof intact roots and phosphorus nutrition ofplants. I. Assay conditions and phosphataseactivity. Aust. J. Agric. Res. 31:429-440.

Mian, M.A.R., Nafziger, E.D., Kolb, F.L., andTeyker, R.H. 1993. Root growth of wheatgenotypes in hydroponic culture and in thegreenhouse under different soil moistureregimes. Crop Sci. 33:129-138.

Mian, M.A.R., Nafziger, E.D., Kolb, F.L., andTeyker, R.H. 1994. Root size anddistribution of field-grown wheat genotypes.Crop Sci. 34:810-812.

Monyo, J.H., and Whittington, W.J. 1970. Geneticanalysis of root growth in wheat. Agric. Sci.Camb. 74:329-338.

Motzo, R., Atenne, G., and Deidda, M. 1993.Genotypic variation in durum wheat rootsystems at different stages of development ina Mediterranean environment. Euphytica66:197-206.

Nakamoto, T., and Oyanagi, A. 1994. Thedirection of growth of seminal roots ofTriticum aestivum L. and experimentalmodification thereof. Annals of Botany73:363-367.

Newman, E.I. 1966. A method of estimating thetotal length of root in a sample. J. Appl.Ecol. 3:139-145.

Nye, P.H., and Tinker, P.B. 1969. The concept of aroot demand coefficient. J. Appl. Ecol.6:293-300.

O’Brien, L. 1979. Genetic variability of rootgrowth in wheat (Triticum aestivum L.).Aust. J. Agric. Res. 30:587-595.

O’Toole, J.C., and Bland, W.L. 1987. Genotypicvariation in crop plant root systems.Advances in Agronomy 41:91-145.

Ortmann, E.E., Peters, D.C., and Fitzgerald, P.J.1968. Vertical-pull technique for evaluationtolerance of corn root systems to Northernand Western corn rootworms. J. Econ.Entomol. 61:373-375.

Oyanagi, A., Nakamoto, T., and Wada, M. 1993.Relationship between root growth angle ofseedlings and vertical distribution of roots inthe field in wheat cultivars. Jpn. J. Crop Sci.62:565-570.

Philipps, J.M., and Hayman, D.S. 1970. Improvedprocedures for clearing roots and stainingparasitic and vesicular-arbuscularmycorrhizal fungi for rapid assessment ofinfection. Trans. Br. Mycol. Soc.55:158-161.

Pinthus, M.J. 1967. Spread of the root system asan indicator for evaluating lodging resistanceof wheat. Crop Sci. 7:107-110.

Rajaram, S., and Villegas, E. 1990. Breedingwheat (Triticum aestivum) for aluminumtoxicity tolerance at CIMMYT. In: Geneticaspects of plant mineral nutrition. N.E.Bassam et al. (eds.). Dordrecht, TheNetherlands: Kluwer Academic Publishers.pp. 489-495.

Robertson, B.M., Waines, J.G., and Gill, B.S.1979. Genetic variability for seedling rootnumbers in wild and domesticated wheat.Crop Sci. 19:843-847.

Russell, R.S. 1977. Plant root systems: Theirfunction and interaction with the soil.McGraw-Hill Book Company, UK.

Sadhu, D., and Bhaduri, P.N. 1984. Variable traitsof roots and shoots of wheat. Z. Acker- undPflanzenbau 153:216.

Smucker, A.J.M., McBurney, S.L., and Srivastava,A.K. 1982. Quantitative separation of rootsfrom compacted soil profiles by the hydro-pneumatic eludration system. Agron. J.74:500-503.

Spahr, K. 1960. Untersuchungen über dieStandfestigkeit von Sommergerste Z. Acker-Pflanzenbau 110:299-331.

Swaminathan, K., and Verma, B.C. 1979.Response of three crop species to vesicular-arbuscular mycorrhizal infection on zinc-deficient Indian soils. New Phytol.82:481-487.

Tennant, D. 1975. A test of a modified lineintersect method of estimating root length. J.Ecol. 63:995-1001 .

Thompson, D.L. 1968. Field evaluation of cornroot clumps. Agron. J. 60:170-172.

Tiwari, D.K., Nema, D.P., and Tiwari, J.P. 1974.Rooting pattern as a selection parameter ofwheat varieties under moisture stress.Madras Agric. J. 61:334-339.

Upchurch, D.R., and Ritchie, J.R. 1983. Rootobservations using a video camera for rootobservations in mini-rhizothrons. Agron. J.75:1009-1015.

Van Beem, J. 1996. Breeding for nitrogen useefficiency in low-input agriculture: alternateselection criteria, inheritance and rootmorphology. Dissertation, CornellUniversity, USA.

Vepraskas, M.J., and Hoyt, D. 1988. Comparisonof the trench profile and core methods forevaluation of root distribution in tillagestudies. Agron. J. 80:166-177.

Vlek, P.L.G., Lüttger, A.B., and Manske, G.G.B.1996. The potential contribution ofarbuscular mycorrhiza to the development ofnutrient and water efficient wheat. In: TheNinth Regional Wheat Workshop forEastern, Central and Southern Africa.Tanner, D.G., Payne, T.S., and Abdalla, O.S.(eds.). Addis Ababa, Ethiopia: CIMMYT.pp. 28-46.

Volkmar, K.M. 1993. A comparison ofminirhizotron techniques for estimating rootlength density in soils of different bulkdensities. Plant and Soil 157:239-245.

Ward, K.J., Klepper, B., Rickman, R.W., andAllmaras, R.R. 1978. Quantitative estimationof living wheat-root lengths in soil cores.Agronomy J. 70:675-677.

Weaver, J.E. 1926. Root development of fieldcrops. New York-London: McGraw-HillBook Co. pp. 291.

Wilson, J.B. 1988. Shoot competition and rootcompetition. J. Appl. Ecol. 25:279-296.

Yu, L.-X., Ray, J.D., O’Toole, J.C., and Nguyen,H.T. 1995. Use of wax-petrolatum layers forscreening rice root penetration. Crop Sci.35:684-687.


219

MicronutrientsJ.S. Ascher-Ellis,1 R.D. Graham,1 G.J. Hollamby,1 J. Paull,1 P. Davies,2 C.Huang,1 M.A. Pallotta,1 N. Howes,2 H. Khabaz-Saberi,1 S.P. Jefferies,1 and M.Moussavi-Nik1

C HAPTER 1 9

Little effort has been made to applymodern breeding techniques to adapt cropplants to soils of poor nutritional status,even though this is genetically feasible.Rather, the use of fertilizers to solve soilnutrient problems agronomically hasencouraged plant breeders to concentrateon other objectives such as yield, climaticadaptation, disease resistance, andquality.

Some nutritional problems, however, donot appear to be easily resolved, and acase for breeding adapted varieties can bemade in those instances. The mostobvious problems are those of nutrienttoxicities, where the cost of removal ismuch greater than that of applyingfertilizer to address deficiencies; hencebreeding solutions are more practical andindeed necessary. Furthermore, fertilizersare ineffective with micronutrientdeficiencies such as iron and manganese,which are induced by high pH;agronomic solutions may thus not besatisfactory, and a genetic solution isnecessary.

Soils with chronic micronutrientdeficiencies are often high-pH,calcareous soils in seasonally dryclimates, but may include deep sands inany climate. Although micronutrientfertilizers are often strikingly beneficial,the plant’s yield potential can only bereached if its roots penetrate thechemically inhospitable subsoils to accesswater stored there. Tolerance to

micronutrient deficiencies (that is,micronutrient efficiency; see definitionbelow) is generally manifested by greateruptake from deficient soil. Roots mustfind micronutrients in their immediateenvironment to continue to grow, toexpress disease resistance, and to accesswater stored in the subsoil.

In this chapter we outline the case forbreeding crop plants for traits that conferadaptation to nutrient deficiencies insoils. In this context a deficient soil isone that contains reasonable amounts ofthe limiting nutrient; however, it isrelatively unavailable to the commoncultivars of the crop in question. We thendiscuss the genetics of micronutrienttraits and the status of screeningtechniques that can be used in breedingprograms.

Definition of NutrientUse Efficiency

We define nutrient use efficiency (foreach element separately) as a genotype’sability to produce high yield in a soilwhose nutrient content is limiting for astandard genotype. This agronomicdefinition is meaningful to a plantbreeder selecting genetic material in thefield. Often a nutrient-efficient genotypein infertile environments will also havehigh yield potential in fertile soils.Nutrient efficiency may be achieved

physiologically by one or moremechanisms:

• better root system geometry;• faster specific rate of absorption at

low concentrations (low Km);• chemical modification of the root-

soil interface to solubilize more ofthe limiting nutrient;

• improved internal redistribution ofthe nutrient;

• superior nutrient utilization, or alower functional requirement for thenutrient in the cell.

In the field the plant breeder cannotreadily identify the operativemechanism(s), but selection would bemore precise if he/she knew which onesthey are. Examples of this are given inthe text, but for the greater part,efficiency is inferred from yieldmeasurements, nutrient content, orsymptom expression.

The Case for aBreeding Program

Plant breeding is a numbers game, andany new objective, such as micronutrientuse efficiency, represents a considerableescalation of the breeder’s work or else adiversion of effort away from traditionaltargets such as quality and diseaseresistance. Thus there must be a strongcase for a new breeding objective beforeit can be added to the research agenda.The lack of compelling arguments is thereason little effort has been made until1 Department of Plant Science, Waite Campus, University of Adelaide, South Australia.

2 Field Crops Pathology Unit, South Australian Research and Development Institute.

220

now to adapt crop plants tomicronutrient-deficient soils. To arguefor breeding to improve nutritionalcharacters, it is necessary to show:

• a need for it as pressing as for otherbreeding objectives;

• there is reasonable genetic potentialto be exploited;

• it is agronomically, economically,and ecologically feasible.

Graham (1984, 1987, 1988a, b)demonstrated that there is geneticdiversity for micronutrient characterswithin wheat, and further argued thatnearly all soils, no matter how poor, hadsufficient amounts of micronutrientsstored in the profile. The problem wasusually their lack of availability duepartly to soil chemistry, but equally topoor genotypic adaptation. This bringsus back to the agronomic arguments.

Twelve years ago, Ascher and Graham(1993) bulldozed off the topsoil and dugup the subsoil from grave-sized pits on10 soil types scattered across SouthAustralia. Various nutrient treatmentswere applied to the subsoils as theywere returned to the pits in their originallayers. The topsoils were then replacedand the sites sown by each farmer aspart of the field. The farmers applied allthe usual treatments and fertilizers,including in some cases micronutrients,without disturbing the treated subsoilbelow. Responses to the nutrients wereimmediate and often spectacular, thoughin general there was little response tophysical disturbance only or to gypsum,which underlines the chemical nature ofthe problem.

It is important to note that theseresponses (at the five most trace-element-deficient sites) continue to thepresent, and with the original nitrogenmost likely lost by now, the residualresponses are probably due to

phosphorus and trace elements (Picture1). Indeed, the micronutrient treatmentseems to have relatively greater residualvalue as time passes. It appears that rootchannels developed to depth in the earlyyears have been re-utilized annually tothe present. In pot studies we haveshown that wheat roots grow poorly insubsoil even when fertilized withnitrogen and phosphorus. Although weare experimenting with methods of deepplacement of micronutrients with highlyunconventional and expensivemachinery, we believe a better approachto this problem is to breed wheats withroots that will penetrate subsoils withlow availability of phosphorus andmicronutrients.

Immediately relevant to this argument isthe picture emerging from physiologicalstudies of roots spanning four decades.From the papers of Haynes and Robbins(1948), Epstein (1972), Pollard et al.(1977), Bowling et al. (1978), Graham etal. (1981), Welch et al. (1982), Nableand Loneragan (1984), Loneragan et al.(1987) and Holloway (1991), it appearsthat the elements phosphorus, zinc,

boron, calcium, and manganese are allrequired in the immediate external rootenvironment for healthy growth,membrane function, and cell integrity. Inparticular, phosphorus and zincdeficiencies in the external environmentpromote leaking of cell contents such assugars, amides, and amino acids(Graham et al., 1981), which arechemotactic stimuli to pathogenicorganisms.

Phosphorus is phloem mobile, but theother elements are not, or are poorly so;this means that the root tips cannot beadequately supplied from elsewhere inthe root system, such as for example,from roots that come into contact with afertilizer band. Moreover, in the case ofzinc, a high internal zinc content did notprevent leakiness due to a deficiency ofzinc external to the membrane (Welch etal., 1982). It follows that the roots ofthose wheat genotypes that have agreater capacity to mobilize nutrientsstrongly bound to soil particles in therhizosphere will be better able topenetrate the infertile, high pH subsoil.It follows too from the above that roots

Picture 1. Grave-sized barley plots in a farmer’s field at Marion Bay, 1992. The areas aroundthe graves represent completely undisturbed subsoil. The middle plot shows a huge responseto subsoil N, P, and micronutrients, even after seven seasons and despite regular topsoilfertilization by the farmer. (Photo: J. Ascher-Ellis.)

J.S. ASCHER-ELLIS ET AL.

221

that are far from a fertilizer band andhave leaky membranes are at greater riskfrom pathogens.

Recent studies have clearly linked traceelement deficiencies with enhancedsusceptibility to particular pathogens(Graham and Webb, 1991). Manganese-deficient wheat plants are moresusceptible to Gaeumannomycesgraminis var. tritici, the take-all fungus(Graham and Rovira, 1984; Huber andWilhelm, 1988), and to the foliarpathogen, Erysiphe graminis (Graham,1990). Zinc deficiency decreased theresistance of wheat to Fusariumgraminearum, the crown rot fungus(Sparrow and Graham, 1988) and toRhizoctonia solani (Thongbai et al.,1993a, b), the causal agent of bare patch.The implication is that nutrient-efficientgenotypes growing in deficient soil,enjoy better nutrient status and shouldhave greater resistance to thesepathogens; this has been confirmed byWilhelm et al. (1990), Pedler (1994) andRengel et al. (1993) for the manganeseefficiency /take-all system and byGrewal et al. (1996) for the zincefficiency/crown rot system.Collectively, these results suggestcausality in the concurrence in SouthAustralia of some of the world’s mostsevere root disease and micronutrientdeficiency problems.

Topsoil drying also affects wheatproduction on infertile soils of theseasonally humid zone by causing lossof fertilizer efficiency. Mostmicronutrients are in the topsoil byvirtue of fertilizer additions and nutrientcycling. Leaching of the heavy metals isnegligible (Jones and Belling, 1967).When the topsoil dries as a result of aweek or two of dry weather in spring,roots in the nutrient zone are largelydeactivated and the plant must rely ondeeper roots or retranslocation forfurther nutrition. With phloem-immobile

micronutrients and inefficient genotypes,deficiency can result. This occurs in thefield (Grundon, 1980), where copperdeficiency from topsoil drying at earlyboot stage can cause severe sterilityproblems; similarly in a pot trial Ascherand Graham (see Table 5, Graham,1990). These authors showed that Cudeficiency could be induced in wheatplants growing in tall pots if anotherwise adequate Cu supply in thetopsoil was rendered unavailable by soildrying and there was insufficientavailable Cu in the subsoil. Thisproblem may, however, be overcome byusing copper-efficient genotypes such astriticale (Grundon and Best, 1981).

Two further advantages accrue tomicronutrient efficient varieties if byvirtue of their efficiency they alsoaccumulate more of the limiting nutrientin the grain: 1) they improve humannutrition if consumed (iron, zinc,especially), and 2) seedling vigor ismarkedly better when the seed is resownon deficient soils. A final advantage: thedegree of micronutrient efficiencycurrently known to exist in wheatgermplasm, if deployed in modernvarieties, would overcome subclinicaldeficiency commonly unrecognized byfarmers or their advisers.

The Case for HighNutrient Reserves inSeed

Reserves of nutrients in the seed must besufficient to sustain growth until the rootsystem has developed sufficiently tosupply nutrients from the growthmedium. During plant establishmentnutrients are supplied partly from seedreserves and partly from the soil. Highlevels of seed nutrients are particularlyimportant in soils with low nutrient

availability, since a larger root system isrequired for the soil to supply the needsof the crop. Low nutritional status ofseeds has been reported to reduce plantgrowth under conditions of low nutrientavailability (Marcar and Graham, 1986;Asher, 1987; Rengel and Graham, 1995a,b; Moussavi-Nik, 1997).

The nutrient status of seed has beenshown to affect both seed viability andseedling vigor. There are reports in theliterature of minimum seedconcentrations of nutrients below whichseedlings will not grow normally (Ascherand Graham, 1993). Seedling nutritionhas been shown to be an important factorin plant susceptibility to pathogens(Graham, 1983; Graham and Webb, 1991;Pedler, 1994; Streeter, 1998); seednutrient levels that fail to maintainadequate nutritional status of seedlings ininfertile soils may reduce the plant’sresistance to some seedling diseases. Theimportance of seed nutrient status inconfounding the screening procedures formicronutrient use efficiency traits is dealtwith in a later section.

Mineral NutrientQuality of Grain

The mineral content of grain contributesnot only to seedling vigor in the nextgeneration but—equally important—tothe nutrition of humans and animals. Thisis particularly so for micronutrients, sinceover half of the world’s population isdeficient in micronutrients, and cerealgrains, the major component of the diet ofthose at risk, also contribute most of theminerals in their diet. Of primary concernin human nutrition are iron and zinc, andsecondary concerns are provitamin A andthe minerals iodine, calcium, seleniumand copper (Graham and Welch, 1996).

Zinc is an element of special interest as itis commonly deficient in soils (50%globally), plants, animals, and humans.

MICRONUTRIENTS

222

Much can be achieved through the wholefood chain by fertilizing the soil withzinc (Graham and Welch, 1996), butbecause of topsoil drying and the role ofsubsoil fertility, the case has been madeelsewhere in this chapter for breedingzinc-efficient wheats, probably involvingthree or four genes per genome.

Where zinc efficiency translates intohigher zinc concentration in the seeds, asit can, we have a win-win benefit forproducers and consumers, and anadditional mechanism for better yields insubsequent crops on deficient soils: zinc-dense seed. In some cases, however,zinc-efficient cultivars may enhanceyield so much that the concentration ofzinc in grain is diluted by extra drymatter (Graham et al., 1992). Theevidence available indicates there aregenes (independent of those controllingthe agronomic zinc efficiency trait) thatcontrol the transport of zinc to the grainfrom vegetative parts during grainfilling. Since the heavy metal nutrientcations iron, zinc, copper, manganese,cobalt, and nickel are insoluble at thehigh pH of the phloem sap, thesetransport genes probably code for naturalchelators that hold these metals insolution in the phloem sap so they maybe transported to the grain. One suchgene, identified in tomato, codes for thesynthesis of nicotianamine, which isessential for the transport of iron in thatplant. In a nicotianamine-deficientrecessive mutant, Chloranova, this traitis controlled at a single locus (Rippergerand Schreiber, 1982).

Genetic variation for iron and zincconcentration in grain has been exploredin wheat by Ortiz-Monasterio andassociates at CIMMYT. The range inconcentration from the lowest to thehighest for both elements is 3-5, andthere is a potential advantage overcurrent high yielding varieties of a factorof 1.5-3 (Graham et al., 1997). Although

high levels of iron and zinc are not co-inherited, there are genotypes higher inboth, and among them are high yieldingadvanced lines.

Household resource allocation studies(Bouis, 1994) suggest that doubling theiron content in grain would significantlyimprove iron deficiency in humans,provided the iron in iron-dense varietieswas bioavailable. The bioavailability ofthe iron in high-iron beans (and rice) hasalready been tested on iron-deficient ratsand proven to be as available (percent oftotal) as that in low or standard types(Welch et al., 1999). Tests are under wayin humans. Since beans are high inabsorption inhibitors such as phytate andtannins, it is likely that an equallyfavorable result will be established forwheat in the near future.

Genetics ofMicronutrient Traits

The first genetic study of a micronutrientefficiency factor was conducted by Weiss(1943). He showed that iron efficiency insoybeans was due to a single dominantgene that controls the reducing power ofthe root surface. Since this pioneeringstudy, several minor additive genes havebeen discovered to contribute to ironefficiency in soybeans (Fehr, 1982). Thissituation of one major and several minorgenes is generally the case with themicronutrients. Reports to about 1970were reviewed by Epstein (1972), whonoted that boron efficiency wasapparently under simple genetic controlin tomato and celery, as were ironefficiency in maize and tomato, andmagnesium efficiency in celery. Morerecently iron efficiency in tomato hasbeen shown to be controlled by a majorgene, coding for an iron-transportingamine, nicotianamine, and a string ofminor genes (Brown and Wann, 1982;Ripperger and Schreiber, 1982).

Copper efficiencyCopper efficiency in rye appears to be adominant trait controlled at a single locuson the long arm of chromosome 5R(Graham, 1984). Copper efficiency hasbeen transferred from rye to wheat bytranslocating part of 5RL to achromosome of wheat. The translocated5RL chromosome segment carrying thecopper-efficiency trait Ce-1 onto the 4Aβ(now 4Bβ) chromosome confers onplants a much greater ability to mobilizeand absorb copper ions tightly bound tothe soil (Graham, 1984). Several suchtranslocations exist but the 5RL/4Atranslocation appears to be the mostsatisfactory agronomic type (Picture 2);it has been successfully incorporated into

Picture 2. The copper-efficient 5R / 4A plantson the right grow well and show good seedset in soil that is too copper-deficient for thecontrol genotype (2R / 4A).


223

cultivars adapted to South Australia(Graham et al., 1987). The 5RLchromosome arm also confers copperefficiency in triticale, unless a copper-inefficient rye is used in the cross.Triticales generally show agronomicallyuseful copper efficiency, intermediatebetween wheat and rye, and have beenused for this reason on many sandy orpeaty copper-deficient soils in SouthAustralia (Graham, 1987), and ondeficient clayey soils in Queensland tocounter the effects of topsoil drying(Grundon, 1980).

Work with these 5R materials hasshown that copper efficiency in rye isnot clearly linked to zinc or manganeseefficiency. Thus independent andrelatively specific genes are involved,and neither root system geometry norsize appears to be critical (Holloway,1996).

Although rye has a much longer andfiner root system than wheat, triticalesgenerally do not, yet they are usuallymore efficient for all three elements(Graham, 1984; Graham et al., 1987;Harry, 1982; Cooper et al., 1988).Studies of rye addition lines suggest that6R contributes a little to efficiency forall three elements, perhaps by way of aroot geometry feature, but the majorgenes are elsewhere. Manganeseefficiency is located on 2R, a conclusionsupported by the poor performance onmanganese-deficient soils of cv.Coorong, an Armadillo-type triticalelacking 2R. By way of contrast, zincefficiency in rye does not appear to beclear-cut from our studies of ryeaddition lines of wheat, and may bespread across four or fivechromosomes: 2R, 3R, 7/4R and, to alesser extent, as we’ve said, 5RL and 6R(Table 6, Graham, 1988a). Cakmak etal. (1997) have additionally linked Znefficiency to 1R.

Zinc efficiencyStudies of addition lines have shown thatCu, Zn, and Mn efficiency in rye areindependent traits, carried on differentchromosomes (Graham, 1984). Copperand Mn efficiency in rye and Mnefficiency in barley appear to becontrolled by single major genes(Graham, 1984; McCarthy et al., 1988),as is Fe efficiency in soybeans (Weiss,1943). Boron and Mg efficiency in celery(Pope and Munger, 1953a,b) and Befficiency in tomato (Wall and Andrus,1962) likewise appear to be controlled ata single locus (all reported to bedominant).

However, less is known of the geneticsof Zn efficiency. Several loci on as manydifferent chromosomes are involved inZn efficiency in rye, and a few genes areinvolved in Zn efficiency in rice. Thelargest single screening exercise wasdone on 3,703 lines of paddy rice(Ponnamperuma, 1976; IRRI, 1979); 388lines were judged to be tolerant and asimilar range of responses was observed.Following diallel analysis, a recent reportsuggested that the genetic effects

responsible for the Zn efficiency trait inrice are mostly additive, and to a lesserextent dominant (Majumder et al., 1990).

Soybean varieties differ in their responseto Zn fertilizer (Rao et al., 1977; Rose etal., 1981; Saxena and Chandel, 1992).This may be a consequence ofdifferential efficiency of Zn absorption;the distribution of F3 lines from thecross between Zn-efficient and Zn-inefficient genotypes (330 F3 linestested) suggested that only a few genescontrol the Zn efficiency trait (Hartwiget al., 1991).

The various mechanisms of Znefficiency in wheat are likely to beadditive (as shown for rice, Majumder etal., 1990), which suggests that in abreeding program stepwisecompounding of genetic informationshould be greatly emphasized (seeRengel and Jurkic, 1992). Suchpyramiding into one locally adapted cropcultivar of a number of Zn efficiencymechanisms that are expressed atdifferent levels of the plant organism(molecular, physiological, structural, ordevelopmental; see Rengel, 1992) might

Picture 3. Varying degrees of tolerance to zinc deficiency in paired plots of barley linesgrowing in zinc-deficient soil, Horsham, 1998. One plot of each pair was treated with zincfertilizer granules drilled with the seed and, later, a foliar spray. (Photo: J. Lewis.)

MICRONUTRIENTS

224

follow the approaches of Yeo andFlowers (1986). In such a breedingprogram, genotypes having genescontrolling a particular mechanism of Znefficiency may be very important eventhough they themselves may not showphenotypically high overall Znefficiency. It would be advantageous touse local genotypes because this wouldexpedite the development of cultivarswith improved Zn efficiency withoutseverely disrupting the broad adaptationalready achieved (Picture 3).

Durati, a very sensitive durum wheat inthe heavy black clay soils of New SouthWales was also poor in our light sandysoils. It is therefore a valuable indicatorline. Kamilaroi is a derivative of Durati(Durati x Leeds) which not onlyincorporates yield, quality and diseaseresistance from Leeds but also zincefficiency on the heavy black earths ofNew South Wales. However, in SouthAustralia on light sands, Kamilaroiappears worse than Durati for zincefficiency. Zinc deficiency in the blackearths is a complex phenomenoninvolving very high levels of native soilphosphorus and manganese that appearto aggravate the low zinc status. Thuszinc efficiency on these soils may not beso much a “foraging capacity of theroots” but a better discrimination forzinc over manganese and phosphate. InSouth Australia phosphorus andmanganese are relatively low. Wetherefore recognize different types ofzinc efficiency.

Besides diversity for yielding ability onzinc-deficient soils, there may be geneticcontrol over zinc concentrations in tissueand grain (Graham et al., 1992).Excalibur and Warigal 5RL are generallysuperior to the other lines (30 tested inall) in zinc concentration in leaves andzinc uptake at tillering. However,Excalibur had low grain concentration, acondition apparently linked to its high

yield since grain zinc content (g/ha) washigh. There is a distinct trend, as withgrain nitrogen, for a lower grain zincconcentration with increasing yield(across genotypes). This is undesirableboth because of lower seedling vigorwhen low zinc seed is used for resowingand because wheat is generallyconsidered to be too low in zinc foradequate human nutrition when it formsa high proportion of the diet (Welch andHouse, 1983; Graham and Welch, 1996).However, Warigal stood out as havingthe highest grain zinc concentration andcontent under zinc-deficient conditions.It is highly likely that grain zincconcentrations could be improved bybreeding. It should be noted, however,that grain zinc concentration respondsdramatically to fertilization under theseconditions.

Manganese efficiencyRemarkable diversity for manganeseefficiency exists within wheat, especiallyin hexaploids and durums and this maybe further supplemented, if warranted,with efficiency genes from rye.

Manganese efficiency in barley appearsto be simply inherited, taking theevidence of the cross of Weeah(efficient) and Galleon (inefficient)(Graham, 1988b) and similar evidence ofthe cross between Weeah (efficient) andWI2585 (inefficient, McCarthy et al.,1988). This major gene has recently beenmapped to 4H (see later section).However, another line WA73S276 thathas common parentage with Weeah, hasmarkedly more efficiency than Weeah,suggesting that other loci may beinvolved. Moreover, an important parentin the barley breeding program at theWaite Institute, CI3576 from Alexandria,is exceptionally susceptible tomanganese deficiency, and so is a highpercentage of its progeny. At this stagewe are not sure whether this is some type

of semi-dominant manganeseinefficiency trait, or an effect of tightlinkage to another trait.

A low percentage of wheat cultivars alsohave exceptional sensitivity tomanganese deficiency, the genetic basisof which is also unclear. However, recentstudies (Khabaz-Saberi et al., 1998) hasshown that Mn efficiency in durumwheat is contolled by two additive genes.It is likely that these are in homeologouspositions in the two genomes, and fromour experience with barley (single), rye(single) and durum wheat (2 genes), wecan safely predict three major loci inbread wheat. Minor genes are almostcertain to be involved in all species, butin the southern Australian breedingprograms, much progress still needs tobe made with these major genes.

The rankings for zinc and manganeseefficiencies are somewhat inverselycorrelated (see Graham, 1990). Zinc-efficient Excalibur has poor manganeseefficiency, and Bayonet, Millewa, andTakari are reasonably zinc efficient butacutely manganese inefficient. Othersare reasonably efficient for both(Aroona, Machete) and some quiteinefficient for both elements (Durati,Kamilaroi, Songlen, Gatcher). Durati

Figure 1. Frequency diagram of shootmanganese content (mg / pot) of F2individuals from the cross of Stojocri 2 xHazar durum wheats, when grown inmanganese-deficient Wangary soil for 4weeks.Source: Khabaz-Saberi et al. (1998).

Number of individuals40

30

20

10

01 2 3 4 5 6

Shoot Mn content (µg/pot)

Hazar Stojocri 2


225

and Kamilaroi are poor for manganese,zinc, and copper. Indeed, most durumstested are poor for micronutrientefficiencies. Zinc efficiency in wheat, asearlier discussed for rye, appears also tobe independent of copper efficiency.

Boron efficiency and toxicitytoleranceBoth boron deficiency and boron toxicityare common nutritional imbalances ofmany crops, including wheat. Borondeficiency occurs mainly in highlyleached soils of the humid zones whileboron toxicity is more common in lowerrainfall regions where limited leachingresults in boron accumulating in thesubsoil. Deficiency and toxicity affectdifferent physiological processes anddifferent tissues in wheat. Borondeficiency primarily affects pollen

development and pollination, whileboron toxicity affects the growth of alltissues at all stages of development. Ahigh level of genetic variation has beenidentified in wheat in response to bothboron deficiency (Rerkasem andJamjod, 1997) and boron toxicity(Moody et al., 1988). As the twoimbalances are expressed at differentstages of development contrastingscreening methods have been developedto assist selection of required genotypes(Picture 4).

Identification of boron deficiency andboron toxicity can be undertaken, withvarying degrees of success, byrecognition of plant symptomsindicative of the disorders and by plantand soil analysis. As mentioned above,the major effect of boron deficiency ofwheat is on development and function ofpollen, thus sterility may occur in plantswithout foliar symptoms. The symptomsof boron toxicity of wheat consist ofregions of chlorosis and necrosisdeveloping from the tips and along themargins of the oldest leaves. While thesymptoms for wheat are not readilydistinguishable from symptoms of otherstresses, boron toxicity symptoms ofbarley are very distinctive and consist ofblack spots developing within thenecrotic lesions at the tips and marginsof leaves. If boron toxicity is suspected,including a few plots of barley within awheat trial will provide a rapid, low-costmeans of diagnosing the problem.

Tissue analysis, using tissues such asyoungest emerged leaf blades (YEBs),will give an indication of the boronstatus of the plant, and can be used tocompare between genotypes in a highboron situation, but does not readilydifferentiate between efficient andinefficient genotypes under low boronsupply. Boron concentrations in YEBsof less than 2 mg/kg might indicate

deficiency, while concentrations ofgreater than about 20 mg/kg in YEBsand 3 mg/kg in mature grain indicate alikelihood of boron toxicity.

Boron can be extracted from soil by anumber of methods and the amountextracted will depend on the method andsoil type. The most common methodsused are hotwater and hot 0.01M CaCl2extractions. As the amounts extractedvary according to time, soil type andmethod, the absolute values only providea rough guide to the amount of boronavailable to plants. A hot-waterextractable boron concentration in soil ofless than 0.5 mg/kg might indicate borondeficiency, while a concentration greaterthan 15-20 mg/kg is associated withtoxicity. As high concentrations of boronin low rainfall environments aregenerally found in the sub-soil, the soilprofile should be sampled to 1 m.

Additional methods of diagnosing thenutritional problem include applicationof the nutrient, in the case of adeficiency, and inclusion of probegenotypes in trials. Many Australianwheat varieties have been characterizedfor response to high concentrations ofboron and several, including Halberd,Frame, and Spear that are tolerant, whileHartog is very sensitive. Appropriateprobe genotypes for boron deficiency,identified in Thailand, include SW41(inefficient) and Fang 60 (efficient).

Severe boron deficiency can result incomplete sterility of an inefficient wheatgenotype that produced healthyvegetative growth with a similarconcentration of boron in the flag leafand the ear as an efficient genotype witha high level of seed set. As no seedlingor vegetative response has beenidentified that is correlated with seed set,it is necessary to select for efficiency atlow boron supply during thereproductive stage.

Picture 4. Symptoms of boron toxicity onwheat and barley leaves. Barley is generallymore sensitive and the symptoms moredistinctive, making a sensitive variety likeStirling a good indicator line. (Photo: J. Coppi.)

MICRONUTRIENTS

226

Boron deficiency can be induced in sandculture in pots or in field trials. In bothsituations the response of the genotypesunder test can be compared to anadequate boron control (application of 1kg boron/ha, as borax), or to wellcharacterized check genotypes (e.g.,SW41 and Fang 60). The response isdescribed by the grain set index, whichis calculated as the number of grains inthe primary and secondary florets of the10 middle spikelets of the primary spike,expressed as a percentage of thepotential (i.e., 20 grains). Restricting thenumber of grains to the primary andsecondary florets, rather than countingtotal grain set, minimizes confoundingeffects such as moisture stress and othernutritional imbalances that influence thedevelopment of grain in the higher orderflorets.

Screening Techniques

Undoubtedly, the ideal is to understandthe mechanisms involved and to selectfor the desired characters by means oftheir gene products. This may bephytosiderophore release as in the caseof iron efficiency in wheat (Marschner etal., 1986), binding affinity in themembrane (Km), root geometry, orcomposition of simple root exudates thatmay control nutrient availability in therhizosphere. If iron efficiency were aproblem in wheat, it would appear asimple matter to select for greater abilityof the roots to release deoxymugeneicacid under standard conditions of ironstress (as Marschner et al., 1986, havedefined). The all-important advantage ofthis approach is that we are no longerdependent on measuring yield with all itspotential for interactions (as theintegration over time of gene expressionon the limiting factor), but are measuringdirectly the intensity of expression of theefficiency alleles present in thatgenotype.

Field testingSelection in terms of yield is alwaysimprecise and fraught with difficulties:almost everything in the genomecontributes to yield either directly orindirectly. If the selection pressure isgreat enough (that is, deficiency issevere and the primary limiting factor),then efficient genotypes will be selectedbased on yield, but the possibility thatstrong interactions will cloud selection isalways there. Graham et al. (1992)reported inconsistent results betweentwo sites for two lines of barley, one ofwhich, Schooner, was able to respond tolate (October) rains by virtue of its latermaturity and, under warm soilconditions, it benefited from improvedavailability of native zinc in the soil. Inthis case, we believe the results fromLameroo, the other site, are more typicaland reflect better the true zinc efficiencyof the lines. Mid-season harvests help tosupport such interpretations.

Site selection. One of the most difficultproblems with field studies is selectingan even site with a level of deficiencythat will differentiate varieties forefficiency. Appropriate extractiontechniques are unavailable, and soilanalysis is poor in predicting traceelement deficiencies. Field history is avaluable resource, but analyzing planttissue taken from the wheat crop prior tothe experiment is perhaps the mostreliable tool for selecting a site.However, trace element deficiency isextremely dependent on environmentalconditions (e.g., temperature, lightintensity, and rainfall) and carefulattention to tissue concentrations inprevious crops may not ensure adeficiency in the experiment in a givenyear.

It is important that adequate levels of allother nutrients be applied as a basalfertilizer to all plots to avoid nutrientimbalances and interactions of other

nutrient deficiencies with genotype. Oneof the most difficult interactions to avoidhas been that of boron toxicity in our zincdeficiency trials. Genotypes that are moresusceptible to zinc deficiency exhibitgreater symptoms of boron toxicity.

Two-level assessment. Our mainapproach to screening is to use plus-and-minus plot pairs to calibrate theperformance of a genotype in deficientsoil against its own potential with thelimiting element supplied. Our primaryefficiency index then becomes:

GY- veg Y-100 ⋅––– or sometimes 100 ⋅ –––

GY+ veg Y+

where GY = grain yield, and veg Y =vegetative yield.

This is the parameter we callmicronutrient use efficiency. However,we do not rely on this quotient alone,given that for various purposes, the lineswith the highest GY- may be of interestwhen they have outstanding yieldpotential but still respond significantly tozinc (e.g., Excalibur; see Graham et al.,1992). Also of independent interest isGY+, the potential yield for a belt-and-braces approach, that is, a combination ofnutrient and genotype that frequentlyoutyields either alone (Graham, 1988a).We argue that 100 ⋅ GY-/GY+ isprobably the best basis for identifyingparental material for a breeding program,whereas GY- may be of most immediateinterest to producers when the problem issubsoil deficiency or topsoil drying, orwhen they are not willing to usemicronutrients or are unaware of thedeficiency.

It is obvious that a genotype can becharacterized better by a yield responsecurve generated by increasing rates offertilizer than by simple plus-and-minustreatments. However, in such studies onlya few lines can be reasonably handledbefore the task becomes too large.


227

Particularly in field experiments, theincreased area means increased spatialvariability, a serious problem withmicronutrients; this variability appearsto be intrinsically greater than that formacronutrients (see Table 8, Graham,1990). Moreover, the criticalcomparison between any given tworates (whether it be the plus-and-minuspair or any other pair) will be spread byrandomization requirements overgreater distances, with other treatmentsin between. Therefore, provided the sitechosen has the degree of deficiency tobe targeted in the breeding program, thepaired-plot system has the advantagethat the extreme proximity of the twotreatments allows the most precisedetermination of 100 ⋅ GY-/GY+ andthat minimal size permits thecomparison of the greatest number oflines.

This is the system we have used most inSouth Australia, but its justificationdepends on both the selection ofrelevant sites and on a perception of thenumber and nature of the genesinvolved. Paull (1990) found a numberof genes involved in resistance to borontoxicity and proposed a model (to whichFigure 2 is analogous for deficiencies)

to account for the variable expression ofdominance at different selectionpressures. In his work, to select for/against all possible segregants in a tetra-genic system, he selected at threedifferent levels of stress.

In deficiency work, it is common todiscuss the relative merits of thegenotypes with yield-fertilizer rateresponses like those in Figure 2.Genotype A is desirable because itreaches its potential at the lowest level ofsupply. But with micronutrients, it iscommon to add 10-100 times as muchfertilizer as is absorbed by the crop,perhaps for several years (Fe, Mnexcepted). Thus, the lowest rate ofnutrient to achieve the yield potential isnot determined by the paired-plot system;it can be argued that this is not criticaland we need only a point safely on theyield plateau and a point at nil nutrientsupply, provided the latter results in ameaningful degree of deficiency for theregion targeted in the breeding program(for example, y in Figure 2). In this waywe have justified our paired-plotapproach in the past, and until such timeas we have a better idea of the genes andmechanisms involved, it seems the mostpractical approach for our purposes.

Using the paired-plot system. A typicalfield experiment (using zinc as anexample) currently consists of 36genotypes x ± zinc x 4 replications, withborder plots, laid out as a split-plotrandomized block design. The experimentis sown with all main plots entered inpairs; one of each pair (chosen atrandom) receives zinc. Our seed andfertilizer drill uses a magazine system fordelivering seed to a cone seeder, and zincgranules are delivered along with the seedvia the magazine. The fertilizer box thuscontains only basal nutrients (all othernecessary nutrients, especially in ourenvironment, nitrogen, phosphorus,sulphur, copper, manganese,molybdenum, cobalt) and does not need

to be changed. The zinc granules arecommercial zinc oxysulphate (~ 30%zinc) used at several times thecommercial rates (11-14 g per 4.5 m2

plot) because its effectiveness is lessthan when coated on macronutrientgranules (ammonium phosphate), thecurrent commercial practice. The soil-applied zinc is supplemented with afoliar spray of zinc sulphate at tillering(equivalent to 200 g zinc/ha). (Formanganese studies, manganeseoxysulphate granules are used followedby one or two foliar sprays at 1 kgmanganese/ha. For copper experiments,we have used copper sulphate granulesand foliar sprays at 0.2 kg copper/ha).

Mid-season harvests are taken from 0.5m2 quadrants, partly to guard againstloss of information due to end-of-seasonevents (drought, hail, heavy late rains,sheep or cattle getting through thefence), and given that we escape theabove, grain yield is measured by asmall-plot harvester.

Because of the large spatial variation inthe soils we study, advantage is taken ofmodern spatial statistical analyses, whichhave proved to be more efficient, that is,higher F, for the genotype x zincinteraction. Generally, a higher F isfound in this analysis than from thesimple split-plot randomized blockfactorial analysis.

The spatial ANOVA process is iterativeand can identify non-treatment variationassociated with the position of a plot orsubplot in the field-plan array. We haveidentified variation due to the effect oftractor-wheel compaction on some plotsand not on others, direction of sowingand harvest in relation to slope (up-down, left-right), prevailing wind(direction of head-bending), and depth ofsowing. Removing variance of this typefrom the residual increases thesignificance of that due to treatment(Gilmour et al., 1997).

Figure 2. Model of the response to addedmicronutrient of two parents and their F1progeny showing how if screening them at asingle level of stress, the geneticinterpretation could be: at Z, A is dominant;at Y, partial dominance; at X, B is dominant(sensitive parent).Source: On analogy with Paull (1990).

Yield

Nutrient added

AF1 B

x y z

MICRONUTRIENTS

228

In addition to grain yield data analysesand the similar treatment of mid-seasonvegetative yields, a further index ofconsiderable value is generated, afterchemical analysis of the tissues andgrain, by calculating the total uptake ofzinc into vegetative growth and/orgrain. The most efficient genotypes areso because they absorb more zinc andmaintain higher zinc concentrations invegetative tissues and often, but notalways, in grain. The same is true forCu and Mn. Uptake, being the productof yield x zinc concentration, frequentlyshows greater variation amonggenotypes than yield. (Compared withyield, variation in concentration isrelatively small but usually significant).Uptake reflects in some measure theexpression of the nutrient use efficiencytrait integrated over time from sowingto harvest (Picture 5).

Single-level assessment. Single-levelassessment is conducted without controlplots supplied with the limiting nutrient,which are replaced for purpose ofinterpretation by plots of a checkgenotype. A typical design that proveduseful in our manganese program(Graham et al., 1983) involved 196plots and only two completereplications of the 72 barley linestested. The remaining 52 plots werecheck plots of one cultivar locatedevery fourth plot in a regular array.

The purpose of this experiment was toidentify genotypes as nutrient-efficientas the check cultivar (or better) byusing an acutely manganese-deficientsite. The results can be analyzed byspatial analysis techniques (ASREML)and by comparing test plot yield to thecheck genotype yield surface for thesite generated from the array of checkplot yields. More simply, plot yields canbe compared to the arithmetic meanyield of adjacent check plots. The checkplot yield array has proved highlyefficient at defining site variability and

Picture 5. Paired-plot screening of wheat breeding material for zinc efficiency, Horsham,1998-99 (Photo J. Lewis). Plus and minus zinc plots of the one entry are sown side by side,with zinc applied to one plot chosen at random. The trial consists of 30 lines sown in fourreplications in a randomized complete block design, with the data subjected later to spatialANOVA (Graham et al., 1992).

Picture 6. Single-plot screening of barley breeding material for manganese efficiency,Wangary, 1981. No Mn fertilizer was used but a Mn-efficient check was sown every fourthcolumn throughout, with single plots of 72 test lines sown in between (Graham et al., 1983).Two replicates. In the column to the left of center, the four plots are (starting from the front)efficient, inefficient, inefficient, efficient. A column of an efficient check runs the length ofthese plots on the left; parts of three other check plot columns can be seen to the far left andright. (Photo: R. Graham.)


229

fertility trends. Consequently, where awide range of performance is expectedamong the entries, efficient selectionmay be achieved with entries appearingonly twice in the matrix. Quite a numberof borderline genotypes may not beconvincingly placed in either theefficient or inefficient group, but usuallythis is not important. The efficiency ofselection in this simple design wasunderlined by the meaningful geneticrelationships discovered in both theefficient and inefficient groups (Grahamet al., 1983) (Picture 6).

Measurements. In field workmeasurements of grain yield should besupported by observations that assessefficiency at seedling and mid-seasongrowth stages. Atypical weather canconfound measurements at a single stageof growth (such as late rains that favorlate-maturing genotypes). Chlorosis,vigor, and delayed maturity (headingdate) can be scored by eye. However,quadrat sampling (or 1 m of row) at mid-season can quantify vegetative yield and,equally important, provide plant materialfor analysis. Total uptake (dry matteryield x concentration) is a valuable indexof nutrient efficiency that integrates rootsystem vigor and efficiency with shootrequirement. Genotypic differences incritical concentration are recognized(Ulrich and Ohki, 1966) but in ourexperience are relatively small comparedwith the total uptake for defining nutrientefficiency.

Collecting and analyzing plant samples.A reliable measurement in assessing ourfield experiments has been to analyzewhole plants and youngest expanded leafblades (YEBs) by inductively coupledplasma (ICP) spectroscopy (Zarcinas etal., 1987). It is important that plantmaterial be as free from contamination(for example, dust, soil particles,galvanized products such as gates andtools, cigarettes, and some paper bags) aspossible. We recommend the use of

plastic gloves for handling both YEBand whole plant collection. Whencollecting whole plants, plants are cutapproximately 1 cm above ground levelto minimize contamination with soil.They are then stored in suitable paperbags and dried overnight at 80 ºC.

In our experience manganese efficiencymay vary 10-42%, with grain yields of-Mn plots from 0.1 to 0.8 t/ha. Durati,Takari, and Millewa are so inefficientand the +Mn fertilizer treatment (soil +foliar) so ineffective that the +Mn plotswere still deficient (young leaves 12 mg/kg Mn) and perhaps yielded barely halftheir potential (Graham, 1990).Although not a great problem, thisresults in higher efficiency indices thanthe inefficient lines warrant, making theextent of diversity for this characteractually greater than measured.

Screening in controlledenvironmentsSoil cultures. Using soil in pots requiresless effort to set up and maintain thansolution cultures, and the work is lesslabor-intensive than field sites. The samesoil requirements apply to screening infield or in pots, but in pot work theexperimenter obtains, by thoroughmixing, a uniform soil for each genotypetested, albeit in a quite atypicalenvironment. The latter is particularlyimportant. Nutrient stresses arefrequently associated with low soiltemperatures, and uncontrolled soiltemperatures in the glasshouse can beextremely unrealistic. For example, it isoften difficult to produce manganesedeficiency in glasshouse conditions, evenwith soil that is severely deficient in thefield, and low temperature baths or

Picture 7. Paired-plot screening of parents and breeding linesfor manganese efficiency, Wangary, 1987. Manganese wasdelivered as Mn oxysulfate granules with the seed and 1-2foliar sprays of Mn sulfate solution applied mid-season asrequired. (Photo: R. Graham.)

At the manganese-deficient site, theresistance of Aroona,Machete, and Millewa tothe take-all fungus was inline with their manganeseefficiency at the vegetativestage (Graham, 1990;Pedler, 1994). Macheteimproved its ranking from19th at tillering to 7th atgrain harvest, whileGj*Wq faded from 2nd to21st. While only a fewlines changed rankingmarkedly through theseason, this meant thatselecting the top five attillering would have nettedonly three of the top fiveat maturity, a point infavor of selection in thefield (Picture 7).

MICRONUTRIENTS

230

controlled-environment rooms arenecessary to keep temperatures below 15ºC. The results of Fox (1978) are anexcellent example of the genotype xclimate x nutrient interaction preventingthe correct interpretation of nutrientefficiency tests.

The size of pots has also been shown tobe an important consideration whenscreening for nutrient efficiency. While itis tempting to use small pots in thegrowth chamber to allow the screeningof as many lines as possible, it has beenshown that for manganese efficiency, potcapacity should not be less than 0.5 kgsoil for two wheat seedlings grown for28 days. Pot size should roughly doublefor each week of growth beyond 28 days(Huang et al., 1996).

We have done considerable work in pots,usually vegetative growth studies, butwe have occasionally found the rankingsquite disparate with field-based grain-yield rankings, which suggests mid-season or later effects can be important.Examples found in Marcar and Graham(1987), Rerkasem et al. (1990) and inour recent screening for zinc efficiencyin wheat (R.D. Graham et al.,unpublished) suggest that field screeningis important. In screening for manganeseefficiency in barley, only one of twomajor gene loci contributing half of thetrait each could be screened for in pots.The other, it seems, much be screenedfor in the field.

Screening for copper, zinc, andmanganese efficiency in pots seems to besatisfactory (Graham and Pearce, 1979;Grewal and Graham, 1997; Rengel andGraham, 1995a; Huang et al., 1996),except as mentioned above. Whenscreening for manganese efficiency inpots, both the storage, temperature, andmoisture effects on manganeseavailability and the screening technique(Longnecker et al., 1991; Webb et al.,1993a; Huang et al., 1996) are critical.

The mechanism of manganese efficiencyhas proved quite elusive, but empiricalscreening has been effective enough tolead to the development of a molecularmarker for the major gene involved (seelater).

Screening for boron toxicity anddeficiency in pots has been effective.High boron supply results in chloroticand necrotic lesions at the tips of theolder leaves, reduced plant vigor,restricted tillering, poor root elongation,and elevated concentrations of boron inall plant tissues. Tolerant genotypesmaintain lower boron concentrations intissues, develop less severe symptoms oftoxicity, and produce more vigorousshoot and root growth than sensitivegenotypes. These differences in responsehave been used to develop efficientscreening systems that have assisted inbreeding boron tolerant wheat varieties(Moody et al., 1988) and enabled theidentification of major genes controllingboron tolerance and their locations onchromosomes 4A and 7B (Paull, 1990;Chantachume et al., 1994).

There is a strong correlation between theresponse of wheat to high boronconcentrations during seedling growthand at later stages of development. Thishas enabled the development of severalrapid seedling assays to identify borontolerant genotypes. One method involvesgrowing seedlings in a glasshouse.Additional boron in the form of boricacid is uniformly mixed through fertileclay-loam soil to give an extractableboron concentration in the range of 50-80 mg/kg. Plants are watered well forseveral weeks, until they are established,and then less frequently. During thissecond phase, roots of the more tolerantplants will be able to extract moisturefrom deeper in the soil profile, and theseplants will continue to grow. Roots ofthe sensitive genotypes do not grow in

the high boron soil, and the plantsappear stunted. Tolerant genotypes alsodevelop less severe symptoms of borontoxicity.

Solution cultures. Simple solutioncultures can rarely be used to selectnutrient efficiency factors that operateon some feature of the root-soilinterface. Efficiency factors operatingwithin the root surface may be screenedfor in solution: characteristics of theabsorption isotherm, xylem loading,short- and long-distance translocation,and efficiency of nutrient utilization (forexample, carbon fixed per unit ofnutrient absorbed). Such micronutrientefficiency factors operating internally(e.g., boron translocation in tomato;Wall and Andrus, 1962) are also dealtwith by soil techniques.

Special adaptations that operate in soilmay be successfully observed in solutioncultures under certain conditions. Brownand Ambler (1973) used strong ironchelates in solution to study the reducingpower of the root, since the reductionFe3+ to Fe2+ was necessary to break theligand-Fe3+ bond and free ionic iron forabsorption. Clark et al. (1982) usedsolutions modified with high phosphate,nitrate, and calcium carbonate to induceiron stress in susceptible sorghumgenotypes.

Iron and manganese efficiencies may bestudied in solutions containingsuspensions of insoluble iron hydroxideor manganese dioxide, both of whichrequire reduction for dissolution; thisprocess is promoted by proton extrusion(see Brown, 1978; Romheld andMarschner, 1981; Uren, 1982). Theseinsoluble higher oxides/hydroxides mayalso be precipitated on chromatographypaper (Uren, 1982) and roots made togrow along the wet paper surface, asystem that is a compromise betweensoil and solution culture. Effectivereduction and dissolution of the dark-


231

colored oxide by the root is detected by awhite depletion zone on either side of theefficient root. For effective screening ofgenotypes, differences must bequantified, requiring good qualitycontrol over the precipitation processand test conditions.

Higher iron concentrations are onoccasions found in chlorotic tissues thanin green tissues (Brown, 1956), and up to100 times more iron may be found inroots than in shoots of iron-deficientplants (Brown, 1978). Theseobservations show that there areefficiency mechanisms operating withinthe plants that may be tested for insolution cultures.

Flowing culture systems add anotherdimension to solution cultureapproaches. With this technique, it ispossible to define for each genotype thelowest solution-phase concentration ofan element that can sustain maximalgrowth rates. This is clearly a geneticallycontrolled character (Asher, 1981). Suchsystems, while too expensive for routinescreening, provide information aboutphysiological mechanisms to aid indeveloping rapid screening tests.

Solution techniques have been mostwidely and successfully used inscreening for tolerance to mineraltoxicities and in elucidating themechanisms and genetics involved.Screening and selecting for aluminumtolerance, for example, has beenefficiently carried out in solutioncultures (Furlani et al., 1982; Foy et al.,1978; Reid, 1976).

The relatively new technique of chelate-buffered nutrient solution culture haspotential in screening for micronutrienttraits (Webb et al., 1993b; Huang et al.,1994a,b; Rengel and Graham, 1996).However, caution and more developmentare needed in terms of activities of ionsin solution and interpretation of results.

Boron tolerance. The effect of boron onroot growth has been utilized to developa rapid, objective assay to identifygenotypes tolerant to boron toxicity(Chantachume et al., 1994). Seeds areimbibed in petri dishes at 2-4 ºC for 2days, then at 18-20 ºC for 1 day. Large,rectangular filter papers or absorbentpaper towels are soaked in a solution ofboric acid (concentration in the range 5-10 µM), 0.0025 µM zinc sulphate, 0.5µM calcium nitrate, and basal nutrients,and then allowed to drain for 1 min. Theseeds are placed in a row across the topthird of the paper, with the embryosfacing the bottom. The towels are rolledup into a cylinder, enclosed in aluminumfoil and stood on end with the embryosfacing down and stored at 15 ºC for 12days. The towels are then unrolled andthe lengths of the longest roots aremeasured. Tolerant genotypes developlonger roots than sensitive genotypes.Tolerant and sensitive controls should beincluded in each filter paper, andreference should be made to a controlfilter paper without boron to comparerelative root lengths.

In a modification of this method byCampbell et al. (1998), seeds were sownon a fine mesh over a “lunchbox”containing the boron solution, withaeration. Tolerant and sensitivegenotypes are again distinguished on thebasis of root growth. This method has theadvantage of being less labor-intensiveand less expensive because filter paperand aluminum foil are not required;however, correlations with other screensmay not be quite as good. Nevertheless,the ranking of well characterizedgenotypes is consistent among thealternative screening methods and withthe concentration of boron in shoots andgrain when grown in fields with highconcentrations in the sub-soil. The rootlength screen can be combined with ameasure of coleoptile length where this isimportant (Picture 8).

Importance of seed quality inscreening for micronutrientefficiencySeed with high mineral content has beenshown to be associated with earlyseedling vigor; this early advantage dueto nutrient content of seed may still beobserved at maturity as increased grainyield, larger seed size, and more grainsper plant (Longnecker et al., 1991;Rengel and Graham, 1995a, b), evenwhere nutrient supply is non-limiting forplant growth. The nutrient content ofseed is dependent on soil type, nutrientavailability, species, and to a lesserextent, variety and season. Longneckerand Uren (1990) showed for both barleyand white lupin that seasonal effects hadless influence on seed Mn content thanchoice of site.

It has already been demonstrated thatthere is genetic potential for wheatgenotypes to grow at sub-optimal levels

Picture 8. Rapid screening for boron toleranceon wheat seedlings in solution culture.†(Photo: ETU, University of Adelaide.)† Boron concentration: 100 mg/ liter, as boric acid

(Campbell et al., 1998).

MICRONUTRIENTS

232

of micronutrient supply. Since theaddition of micronutrients has in manycases decreased the incidence of diseasesin wheat plants (Graham, 1983; Grahamand Webb, 1991) and since differences innutrient content of seed are also undergenetic control (Moussavi-Nik, 1997),there exists a large overlap of factors thatmay confound the performance ofgenotypes.

Nutrient content of seed has also beenshown to affect the susceptibility ofwheat to diseases, for example, McCay-Buis et al. (1995) demonstrated thatwheat plants grown from seed withhigher manganese content (1.83-2.28 µgseed-1) were more vigorous, had less takeall, and produced more grain than plantsgrown from seed with lower manganesecontent (1.26-1.86 µg seed-1). J.L. Cooke(thesis in preparation) reported lessdamage due to Rhizoctonia in wheatplants that had been grown from seed

with high zinc content (0.66-0.74 µgseed-1) than in wheat plants grown fromseed with low zinc content (0.24-0.26 µgseed-1) (Picture 9).

For durum wheat cultivars differing inmanganese efficiency, Khabaz-Saberi etal. (2000) developed a correlationbetween seed manganese content and theamount of added Mn required inWangary soil to assess the Mn efficiencyin a pot bioassay. This correlationdiffered for the manganese-efficientdurum wheat cultivar Stojocri 2compared to the manganese-inefficientHazar.

When comparing genotypes for drymatter production, grain yield, grainquality, disease resistance, and nutrientefficiency, it is important that the qualityof seed sown be similar for all genotypesto avoid confounding effects. However,seed size and nutrient distribution withinthe seed has been shown to be under

Moussavi-Nik (1997) demonstrated thatgenotype effects accounted for the largestportion of treatment variance, while seedsource effects accounted for from almostnone to 44% of the treatment variationand were significant in 9 of 16 of theexperiments conducted. Generally theinteraction between genotype and seedsource contributed less to treatmentvariation than seed source main effects.Significant seed source or genotype xseed source effects occurred over thewhole range of environments tested, fromthe highest to the lowest yielding. Theseed selected from all locations hadadequate nutrient content for all elementsinvestigated; the eight sites used forassessing the effects of seed source wereall adequate for trace elements. In spiteof this, there were four significantassociations of increasing zinc content ofseed resulting in increased grain yieldand one association of sodium seedcontent decreasing grain yield.

A Breeder ’s Approach

Knowledge of the mechanisms andinheritance of trace element efficiencysimplifies the breeding and selection ofimproved lines because the choice ofparents and the screening methods can bemore targeted and objective. However,lack of this information must not deter abreeder when trace element deficienciesare known problems in his target area.Significant advances can be made whilethe genetics is still being investigated.

Potential parents for trace elementefficiency can be sourced from theliterature, by reputation and, particularly,from cultivars that continue to be grownby farmers in trace element deficientareas, despite the fact that cropevaluation trials indicate that thesecultivars have been superseded by new,higher yielding cultivars.Picture 9. Seedling vigor of zinc-inefficient Gatcher wheat

growing in zinc-deficient soil, showing the importance of seedzinc content to establishment (Rengel and Graham, 1995a).


genetic control(Moussavi-Nik, 1997).This means that seed ofgenotypes collectedfrom the same site (soiltype) and season maydiffer notably innutrient content. In anextensive series ofexperiments involving11 wheat genotypes,seed for each genotypewas selected from eightlocalities in SouthAustralia and tested ateight locations over twocontrasting climates.

233

Since trace element efficiency is onlyone objective in an overall breedingprogram, it should be incorporated intothe breeding effort and fitted in withstrategies and methodologies used forselecting for disease resistance, yield,adaptation, and quality. Once potentialparents have been chosen andappropriate crosses made, an importantpart of the strategy is to set up nurserieswhich with plant or line selection willenhance the frequency of genes in thebreeding population for trace elementefficiency.

A critical consideration in such nurseriesis seed source, given that the traceelement content of seeds can greatlyaffect early plant vigor and even finalgrain yield (previous section). Seed lowin Mn content, harvested from Mndeficient sites, is best for screening forMn efficiency; however, such seed maynot be available in a normal breedingprogram. It is important therefore to useseed as close to the same Mn content aspractical, and as low as possible(probably not seed produced on aresearch station, where soil fertility tendsto be higher than on farms). Allcomparisons should be made using seedfrom the same original nursery. Toreduce the effects of seed Mn content onselection, the Mn selection nurseriesshould be grown in soils with an acutedeficiency. In South Australia selectionsites for Mn efficiency have a highlycalcareous soil in which Mn is tightlybound and poorly available to inefficientplants.

The flow chart in Figure 4 details howselecting for Mn efficiency isincorporated into the wheat breedingprogram in South Australia, as a modelfor any micronutrient efficiency trait.Note that it is no different fromincorporating selection for resistance to aroot disease such as cereal cystnematode, where much of the screeningis done in specially chosen field sites.

Early generation screeningIn early generations populationsexpected to segregate for Mn efficiencyare grown in spaced-plant nurseries insoil known to be acutely deficient in Mn.Bulk F2 and/or F3 populations areplanted in long rows using a precisionseeder with 10 cm between seeds in arow. Rows are 30 cm apart. Thisarrangement allows for easy observationof single plants.

In these soils even small changes inavailable manganese can cause largedifferences in plant growth response. Tobe able to get some indication of this andto take it into account when selecting inthe nursery, indicator rows are sown atregular intervals throughout the nursery,usually as a pair every 7th and 8th row.Cultivar Yarralinka, recognized as beingvery manganese efficient, and Millewa, avery inefficient variety, are planted aspaired rows. Contrasting growth betweenthese two control cultivars indicates verydeficient areas where meaningfulselection of single plants from adjacentrows can be carried out. Where thegrowth between the two rows is similar,the Mn deficiency is not such a limitingfactor or there are other more limitingfactors and selection for Mn efficiencywill not be effective.

Long rows are better than short plotsbecause they will traverse across thevariability that occurs and there willnearly always be sections of each rowwhere selection can be practiced. Singleplants or single heads are selected andthese are treated in a routine way in thebreeding program to select for otherattributes. Whether the selections fromeach cross are bulked or whether theyare kept as separate progeny, thepopulation resulting from this nursery isenhanced for Mn efficiency genes. Thisselection process can be repeated.

Later generation screeningIn later generations lines should beevaluated for yield in locations, soils,and farmers’ fields where Mn problemsare expected as part of the range of sitesused in assessing adaptation. Thus Mnefficiency will be measured as part ofthe region’s overall genotype xenvironment effects. In areas wherethere are strong abiotic factors limitingyield, the breeder should be testing andselecting in the presence of the stressand selecting for stress tolerance ratherthan for yield per se. Such sites are notusually found on research stations thathave been chosen and/or managed foryield potential and that are free ofabiotic stresses.

To get a measure of Mn deficiencydirectly (which the breeder needs toknow to do further selection andcrossing), lines are assessed in a knownMn-deficient site in replicated split-plotexperiments, with genotypes as themain plots and different levels of Mnfertilizer as the subplots. Fertilizers (Pand N) are applied at normalrecommendations, and there are nil andapplied Mn subplots. On Mn fertilizedsubplots an initial dose is applied to thesoil at seeding as manganeseoxysulphate; further applications aremade as foliar sprays of manganesesulphate at tillering and again duringstem elongation if needed.

Control cultivars include Yarralinka andMillewa, as well as cultivars beinggrown commercially in the area. Theperformance of this pair indicates theresponsiveness and success of theexperiment. Yield data are spatiallyanalyzed using ASREML software, andmean yields with and without appliedMn are plotted against each other(Figure 3).

MICRONUTRIENTS

234

Figure 4. The Roseworthy modified pedigree wheat breeding program, describing the integration of breeding for cereal cystnematode (CCN) resistance and manganese efficiency within the program. Heavy boxes mark specific activities for Mn efficiency.

F1

F2 bulks of routine crosses Special rust crosses Crosses for Mn

Space planted in appropriate disease nursery

Nat. Rust ControlProgram

Space planted inMn deficient site

Heads bulk threshed andrecycled through other nurseries

if necessary

Single head selections taken from desirable plants

F3 head hills in field rustnursery 60,000

Recycled if necessary forgreater homozygosity

F4 Unreplicated yield trial; single site representative ofthe whole breeding region. Approx 10,000 entries +

augmented checks

F5 Yield evaluation trials; 3 sites,single rep, including a site with high

soil boron. 3000 entries

F5 Yield at Mn deficient site 1site, 2 reps families from Mn

crosses

F5 CCN tolerance trial 1 or 2 CCNinfested sites families from CCN

crosses

NRCP for rust(2000) SARDI for

CCN (500)

F6 Replicated yield trials; 3-6 sitesrepresenting the whole wheat growing

region. 1000 entries

Entries separated into trials and sites onthe basis of boron, CCN, grain hardness,Mn efficiency and residual segregation

Septoria nursery2 sites

CCN pots test15,000 single seedlings

Short head rows of CCN resistantselections. Up to 4000 rows

Lab tests forcoleoptile lengthand boron tol.

F7F8 stage Yield trialsup to 8 sites for some entries

500 entries

Mn efficiency tests; Mn deficientsite, 4 reps, split plot with and

without Mn fertilization

CCN tolerance tests CCN site(s)4 reps, split plot with and

without nematicide

Special stripe rustnursery Horsham,Vic

SouthernCollaborative

Trials. NSW, Vic,SA, WA

Advanced crossbred trials; 12 sites,common trial, 4 reps; includes many

varieties 90 entries

SARDI Stage 3 trials; 5 sites, 4 reps30 entries plus advanced lines from

other breeders

Special trials for CCN; Mnefficiency and Zn efficiency

Seed purificationrows

Detailed disease,protein tests etc.for homogeneity

Advanced crossbred trials as before SARDI stage 4 trials; up to20 sites in SA

Special trials for CCN, Mnand Zn efficiency split

plot

Multiplicationrows, carefullyobserved forhomogeneity

DiseaseProgressNursery

Advanced and stage 4 trials; detailed tests repeated Multiplication plots; individualplant progenies kept separate

RELEASEBASIC SEED (and commercial evaluation

if needed)


Crossing Blockcrosses made - A/B,A/B/ /C, BC’s

235

Use of MolecularMarkers in Screeningfor MicronutrientEfficiency

Molecular markers simplifyselection for micronutrientefficient genotypesSelection of micronutrient-efficient linesfrom segregating populations is desirablebut difficult to do in either field trials orglasshouse and growth-chamber pot bio-assays. The primary reason for this is therequirement for specific growingconditions to maximize genotypedifferentiation. This is not alwaysattainable in the field and, consequently,screening in field trials is expensive andmay not be successful in some seasonssuch as droughts. Even in the glasshouseor growth chamber where day length,temperature and lighting can becontrolled, variation in expression of thetrait can be affected by other factors, suchas the nutrient status of the seed (Uren etal., 1988) and the length of time andconditions in which the soil used to growthe plants was stored (Webb et al.,1993a).

Another difficulty to overcome is theproblem of differentiating heterozygotes,even those which may be intermediate in

phenotype, from homozygotes. Ourstudies have shown that this is not easilydone, even where the trait is primarilycontrolled by a single gene that canexpress (depending on environment) asdominant, semi-dominant, or recessive,as appears to be the case for manganeseefficiency in barley (Pallotta et al.,1999). Separating genotypes insituations where several genes controlthe trait, as is likely to be the case forhexaploid wheat, is exceedingly difficultand requires extensive progeny testing.

Identification of micronutrient efficientgenotypes in early generations of crossesis virtually impossible since eachindividual plant represents a differentgenotype. In addition, genotypes shouldbe tested at both low and sufficientmicronutrient availability to compensatefor genetic variation in plant responsedue to segregation of genes independentof the nutrient effect.

Where traits are difficult to reliablyassay, as are micronutrient efficiencytraits, there is a strong case for the use ofmarker assisted selection (MAS). Ourgroup has identified several RFLPs(restriction fragment lengthpolymorphisms) closely linked to amajor gene (Mel 1) controllingmanganese efficiency in barley (Pallottaet al., 1999); these are currently beingused for early generation selection andto facilitate rapid backcrossing of thetrait into elite breeding lines.

A second major locus in barley, Mel 2,has just been mapped. Work is underway to genetically characterize and mapthe manganese efficiency trait in thedurum wheat cultivar ‘Stojocri.’ Resultsindicate the trait is semi-dominant, aswas found in our studies on barley andsegregation in an F2 population fitting atwo-gene model (Khabaz-Saberi et al.,1998). Genetic variation for manganeseand zinc efficiency has been reported in

hexaploid wheats (Graham, 1988b;Graham et al., 1992; Grewal andGraham, 1997). Efficient wheats have ayield advantage when grown inenvironments where zinc or manganeseis limiting. Mapping genes controllingthese traits in both species would assistthe efficient breeding for these traits.

Marker-assisted selection is especiallyuseful in mainstream breeding programsto accelerate backcrossing. The trait ofinterest can be followed in successivebackcross generations by MAS with noneed for phenotyping, given that theoverall performance of the recurrentparent is usually well known. Twobreeding schemes are shown in Figures 5and 6, for the backcross and intercrossprograms. Marker-assisted selection cansubstantially reduce the time required forvariety development.

MICRONUTRIENTS

Figure 3. The regression of yield ofgenotypes without manganese fertilizer onyield with Mn fertilizer, in a field trial onMn-deficient calcareous sand at Marion Bay.

Yield without Mn fertilizer1200

1000

800

600

400

200

00 200 400 600 800 1000 1200

Yield with Mn fertilizer

Donor parent X Recurrent parent

MAS for donor gene ateach backcross F1

BC3 F1

MAS for donor plant

Doubled haploids

Field trial/bioassay selection of individualscarrying donor trait

Summer multiplication

Advanced trials - S3+S4

MAS for recurrentparent background

Advanced trials - S4

Summer multiplication

Advanced trials + commercial scale quality evaluation

Commercial release

Figure 5. The flow-chart indicates whereMAS (marker-assisted selection) can be usedin a backcrossing program. The schemefacilitates the rapid incorporation of one ormore desired traits into an elite background.

236

However, precautions need to be taken toreduce the variations in kernel size, theposition of the kernel in the spike, andthe conditions under which differentspikes of the same plant mature. It isnecessary to carefully control theripening conditions of the primary DHplants and to select for uniformity ofkernel size.

Doubled haploid populations offer otheradvantages over recombinant inbredpopulations when molecular markers areavailable, particularly if the molecularmarker for a trait is a co-dominantmarker because the need for progenytesting is eliminated. Doubled haploids incombination with molecular markersprovide a powerful tool for earlygeneration screening of micronutrient-efficient genotypes.

Molecular marker-assisted selection canbe used to pre-screen the F2 plants usedfor DH production and increase theproportion of DH lines carrying thedesirable gene(s). This is illustrated inFigure 7 for a single gene formicronutrient uptake efficiency where F2putative donor plants are tested for thepresence of molecular markers closelylinked to the efficiency gene. Selection ofF2 donor plants may be restricted tohomozygotes only (homozygoteselection) or may also includeheterozygotes (allele enrichment). Acomparison between homozygote donor

plant selection, allele enrichment, and noMAS is made in Table 1, which tabulatesthe proportion of F2s and F2-derivedDHs expected when selecting forindependently segregating genes at 1, 2,5, or 10 gene loci.

It can be seen from Table 1 that withhomozygote selection, 100% of the DHsproduced will possess the desired allelesirrespective of the number of gene lociunder selection. However, the proportionof donor plant F2s which arehomozygous at each locus decreasesexponentially with increasing number ofloci so that the probability of selectingfive genes homozygous for the desiredallele would be 9.77 x 10-4, or less than1/1,000. This is actually less favorablethan using no donor plant selection at all,where 3.13 x 10-2, or around 3% of theDHs derived from unselected F2s, wouldbe expected to carry all five favorablealleles.

With allele enrichment, the proportion ofF2 donors carrying at least one favorableallele from each of five genes would be24%. It would be expected that 14% ofDHs produced from these selecteddonors would carry all five desirableable alleles. That is, approximately 1 in 4F2 plants can be selected as donors forDH production and among these 14%will carry the desirable allele at each ofthe five gene loci. Allele enrichment isclearly a more efficient procedure forcombining DH technology and MASthan homozygote selection, particularlywhen larger numbers of genes are beingscreened.

Doubled haploids for theelucidation of molecularmarkersDoubled haploids are also very usefulfor genetic and mapping studies ofmicronutrient efficiency traits, as alllines in a DH population represent theresult of a single meiotic event, whereas


Crossing

MAS for F1 Doubledgene Glasshouse haploids

enrichment

F2 bulk-field

F3 bulk-fieldSingle-plant selection

F4 field plots(1 site)

MAS formultiple

traits F5 yield trials(1 rep, 3 sites)

Figure 6. The flow chart indicates wheremarker-assisted selection (MAS) can be usedin an intercross breeding program. MASwould be used where either a backcross or atopcross is involved.

Figure 7. Selection of F2 donor plants with and without marker-assisted selection (MAS).

Parent 1ee X Parent 2EEinefficient efficient

F1 EE

+MAS +MAS -MAS

F2DH EE Ee eE ee EE Ee eE ee EE Ee eE eedonors

Homozygote Allele No donorselection enrichment selection

Doubled haploids Doubled haploids Doubled haploids

Doubled haploids aid inscreening for micronutrient-efficient cultivarsMarker-assisted selection is especiallyuseful when combined with doubledhaploid (DH) technology. If molecularmarker technology is unavailable, theuse of DH populations provides anoption for improved selection ofmicronutrient-efficient lines. Doubledhaploid populations make earlygeneration screening by traditionalmethods more accurate because allprogeny of each DH line are geneticallyidentical and true-breeding, enablingreplication of tests, which is desirable.

237

F2s are the summation of a male plus afemale meiosis. The DH lines thereforereflect the results of recombinationprecisely without the necessity forstatistical inference.

In both wheat and barley DHs are beingwidely used for mapping many differentgenetic traits. In addition to the barleymanganese efficiency genes describedabove, a DH population derived from across between two zinc-efficient breadwheat lines, cultivar ‘Trident’ andbreeding line 88ZWK043, exhibitedtransgressive segregation in a potbioassay for zinc efficiency; segregationdata suggests several genes control thetrait in this cross.

Conclusions

Genotypic variation exists for toleranceto practically every abiotic stress inevery crop investigated. The level oftolerance available in elite germplasm isagronomically valuable and justifiesbreeding efforts in most cases, not just

those for which there are no othersuccessful agronomic solutions.Inheritance varies from simple toquantitative, and both genotype and soiltype, as well as climate and season,affect the expression of a trait. Oftenthese traits are still manageable in abreeding program, and the G × Einteraction is not prohibitive of theeffort.

Screening for deficiency tolerance traitsis, however, much more difficult than fortoxicity tolerance traits, owing to quitesophisticated inducible systems thatrespond to deficiency in the plant bymobilizing nutrient bound in the soil thatis not otherwise available. Strongexpression of these systems is theobjective. However, in view of thedifficulty in developing fast methods forscreening for efficiency (such asseedling selection in pots) or theirlimited ability to reflect field screening,molecular marker assisted selection isconsidered important for success inpractical breeding programs, and anumber of major gene loci alreadyidentified make this possible.


Ascher, J.S., and Graham, R.D. 1993. Agronomicvalue of seed with high nutrient content. In:Wheat in Heat-Stressed Environments:Irrigated, Dry Areas and Rice-WheatFarming Systems. D.A. Saunders and G.P.Hettel (eds.). Dinajpur, Bangladesh: UNDP/ARC/BARI/CIMMYT. pp. 297-308.

Asher, C.J. 1981. Limiting external concentrationsof trace elements for plant growth: Use offlowing solution culture techniques. J. PlantNutr. 3:163-180.

Asher, C.J. 1987. Crop nutrition during theestablishment phase: Role of seed reserves.In: Crop Establishment Problems inQueensland: Recognition, Research andResolution. I.M Wood, W.H. Hazard, and F.From (eds.). Aust. Inst. Agric. Sci.Occasional Publication No. 20.

Bouis, H.E. 1994. The effect of income ondemand for food in poor countries: Are ourdatabases giving us reliable estimates? J.Dev. Economics 44:199-226.

Bowling, D.J.F., Graham, R.D., and Dunlop, J.1978. The relationship between the cellelectrical potential difference and salt uptakein the roots of Helianthus annuus. J. Exp.Bot. 29:35-140.

Brown, J.C. 1956. Iron chlorosis. Annu. Rev. PlantPhysiol. 7:171-190.

Brown, J.C. 1978. Mechanism of iron uptake byplants. Plant Cell Environ. 1:249-257.

Brown, J.C., and Ambler, J.E. 1973. “Reductants”released by roots of iron-deficient soybeans.Agron. J. 65:311-314.

Brown, J.C., and Wann, E.V. 1982. Breeding foriron efficiency: Use of indicator plants. J. ofPlant Nutr. 5:623-635.

Cakmak, I., Derici, R., Torun, B., Tolay, I., Braun,H.J., and Schlegel, R. 1997. Role of ryechromosomes in improvement of zincefficiency in wheat and triticale. In: PlantNutrition for Sustainable Food Productionand Environment. T. Tando, K. Fujita, T.Mae, H. Matsumoto, S. Mori, and J. Seikiya(eds.). Kluwer Academic Publishers,Dordrecht, The Netherlands.

Campbell, T.A., Rathjen, A.J., Paull, J.G., andIslam, A.K.M.R. 1998. Method for screeningbread wheat for tolerance to boron.Euphytica 100:131-135.

Chantachume, Y. 1995. Genetic studies on thetolerance of wheat to high concentration ofboron. Ph.D. thesis. The University ofAdelaide, Adelaide, South Australia.

Chantachume, Y., Rathjen, A.J., Paull, J.P., andShepherd, K.W. 1994. Genetic studies onboron tolerance of wheat (Triticum aestivumL.). In: Genetics and Molecular Biology ofPlant Nutrition, Abstracts of the FifthInternational Symposium, Davis, CA. p. 141.

Table 1. Selection strategies for combining doubled haploid (DH) technology with markerassisted selection. Probability of obtaining F2 donor plants homozygous or heterozygous forthe desired allele(s) at n loci and the proportion of DHs expected to be homozygous for thesealleles when only homozygotes are used as DH donor plants (homozygote selection), wheneither homozygotes or heterozygotes are used as DH donor plants (allele enrichment), orwhen selection is not practiced on donor plants.

No donorHomozygote selection Allele enrichment selection

F2s DHs F2s DHs DHshomozygous for homozygous for homo- or hetero- homozygous for homozygous for

No. of desired allele at desired allele at zygous for desired desired allele at desired allele atgene loci each locus each locus allele at each locus each locus each locus

(1 / 4)n (1)n (3 / 4)n (2 / 3)n (1 / 2)n

1 0.25 1 0.75 0.67 0.502 6.25 x 10-2 1 0.56 0.45 0.255 9.77 x 10-4 1 0.24 0.14 3.13 x 10-2

10 9.54 x 10-7 1 5.6 x 10-2 1.7 x 10-2 9.77 x 10-4

MICRONUTRIENTS

238

Clark, R.B. Yusuf, Y., Ross, W.M., andMaranville, J.W. 1982. Screening forsorghum genotypic differences to irondeficiency. J. Plant Nutr. 5:587-604.

Cooper, K.V., Graham, R.D., and Longnecker,N.E. 1988. Triticale: A cereal for manganesedeficient soils. In: International Symposiumon Manganese in Soils and Plants:Contributed Papers. M.J. Webb, R.O. Nable,R.D. Graham, and R.J. Hannam (eds.).Manganese Symposium, Adelaide.pp 113-116.

Epstein, E. 1972. Mineral Nutrition of Plants:Principles and Perspectives. Wiley & Sons,New York.

Fehr, W.R. 1982. Control of iron deficiencychlorosis in soybeans by plant breeding. J.Plant Nutr. 5:611-621.

Fox, R.H. 1978. Selection for phosphorusefficiency in corn. Commun. Soil Sci. PlantAnal. 9:13-37.

Foy, C.D., Chaney, R.L., and White, M.C. 1978.The physiology of metal toxicity in plants.Annu. Rev. Plant Physiol. 29:511-566.

Furlani, P.R., Clark, R.B. Ross, W.M., andMaranville, J.W. 1982. Variation and geneticcontrol of aluminum tolerance in sorghumgenotypes. In: Genetic Specificity inMineral Nutrition of Plants. M.R. Saric(ed.). Scientific Assemblies (Serbian Acad.Sci., Belgrade) 13:363-370.

Gilmour, R.M., Cullis, B.R., and Verbyla, A.P.1997. Accounting for natural and extraneousvariation in the analysis of fieldexperiments. J. Agric. Biol. Environ. Stat.2:269-293.

Graham, J.H., Leonard, R.T., and Menge, J.A.1981. Membrane-mediated decrease in rootexudation responsible for phosphorusinhibition of vascular-arbuscular mycorrhizaformation. Plant Physiol. 68:548-552.

Graham, R.D. 1983. Effects of nutritional stresson susceptibility to disease with particularreference to trace elements. Adv. Bot. Res.10:221-276.

Graham, R.D. 1984. Breeding for nutritionalcharacteristics in cereals. Adv. Plant Nutr.1:57-102.

Graham, R.D. 1987. Triticale, a cereal formicronutrient-deficient soils. InternationalTriticale Newsletter No. 1. University ofNew England, Armidale.

Graham, R.D. 1988a. Development of wheatswith enhanced nutrient efficiency: Progressand potential. In: Wheat ProductionConstraints in Tropical Environments. A.R.Klatt (ed.). Mexico, D.F.: CIMMYT. pp.305-320.

Graham, R.D. 1988b. Genotypic differences intolerance to manganese deficiency. Chapter17. In: Manganese in Soils and Plants. R.D.Graham, R.J. Hannam, and N.C. Uren (eds.).Kluwer Academic Publishers, Dordrecht,The Netherlands. pp. 261-276.

Graham, R.D. 1990. Breeding wheats fortolerance to micronutrient deficient soil:Present status and priorities. In: Wheat forthe Nontraditional Warm Areas. D.A.Saunders (ed.). Mexico, D.F.: CIMMYT. pp.315-332.

Graham, R.D., and Pearce, D.T. 1979. Thesensitivity of hexaploid and octoploidtriticales and their parent species to copperdeficiency. Aust. J. Agric. Res. 30:791-799.

Graham, R.D., Anderson, G.D., and Asher, J.S.1981. Absorption of copper by wheat, ryeand some hybrid genotypes. J. Plant. Nutr.3:679-686.

Graham, R.D., Davies, W.J., Sparrow, D.H.B., andAscher, J.S. 1983. Tolerance of barley andother cereals to manganese-deficientcalcareous soils of South Australia. In:Genetic Aspects of Plant Nutrition. M.R.Saric and B.C. Loughman (eds.). MartinusNijhoff/Dr W Junk, The Hague.pp. 339-345.

Graham, R.D., and Rovira, A.D. 1984. A role formanganese in the resistance of wheat plantsto take-all. Plant Soil 78:441-444.

Graham, R.D., Ascher, J.S., Ellis, P.A.E., andShepherd, K.W. 1987. Transfer to wheat ofthe copper efficiency factor carried on ryechromosome arm 5RL. Plant Soil99:107-114.

Graham, R.D., and Webb, M.J. 1991.Micronutrients and resistance and toleranceto disease. Chapter 10. In: Micronutrients inAgriculture. 2nd ed. J.J. Mortvedt et al.(eds.). Soil Science Society of America,Madison, WI. pp. 329-370.

Graham R.D., Ascher J.S., and Hynes S.C. 1992.Selecting zinc-efficient cereal genotypes forsoils of low zinc status. Plant Soil146:241-250.

Graham, R.D., and Ascher, J.S. 1993. Nutritionallimitations of subsoils. In: Plant Nutritionfrom Genetic Engineering to Field Practice.N.J. Barrow (ed.). Kluwer Acad. Publ.,Dordrecht, The Netherlands. pp. 739-742.

Graham, R.D., and Welch, R.M. 1996. Breedingfor staple-food crops with highmicronutrient density. In: InternationalWorkshop on Food Policy and AgriculturalTechnology to Improve Diet Quality andNutrition. Agricultural Strategies forMicronutrients, Working Paper No. 3.Washington, D.C.: International Food PolicyResearch Institute. 82 pp.

Graham, R.D., Senadhira, D., and Ortiz-Monasterio, I. 1997. A strategy for breedingstaple-food crops with high micronutrientdensity. In: Plant Nutrition - For SustainableFood Production and Environment. T. Andoet al. (eds.). Kluwer Academic Publishers,Japan. pp. 933-937.

Grewal, H.S., and Graham, R.D. 1997. Seed zinccontent influences early vegetative growthand zinc uptake in oilseed rape (Brassicanapus and Brassica juncea) genotypes onzinc-deficient soil. Plant Soil 192:191-197.

Grewal, H.S., Graham, R.D., and Rengel, Z. 1996.Genotypic variation in zinc efficiency andresistance to crown rot disease (Fusariumgraminearum Schw. Group 1) in wheat.Plant Soil 186:219-226.

Grundon, N.J. 1980. Effectiveness of soildressings and foliar sprays of coppersulphate in correcting copper deficiency ofwheat (Triticum aestivum L.) in Queensland.Aust. J. Exp. Agric. Anim. Husb.20:717-723.

Grundon, N.J., and Best, E.K. 1981. Tolerance ofsome winter and summer crops to copperdeficiency. In: Copper in Soils and Plants.J.F. Loneragan, A.D. Robson, and R.D.Graham (eds.). Academic Press, Sydney.p 360.

Harry, S.P. 1982. Tolerance of wheat, rye andtriticale to copper and zinc deficiency insoils of low and high pH. M.Ag. Sc. thesis,University of Adelaide, Adelaide, SouthAustralia.

Hartwig, E.E., Jones, W.F., and Kilen, T.C. 1991.Identification and inheritance of inefficientzinc absorption in soybean. Crop Sci.31:61-63.

Haynes, J.L., and Robbins, W.R. 1948. Calciumand boron as essential factors in the rootenvironment. J. Amer. Soc. Agron.40:795-803.

Holloway, R.E. 1991. Factors affecting the growthof wheat roots in the subsoil of Upper EyrePeninsula. M.Ag.Sc. thesis. University ofAdelaide, Adelaide, South Australia.

Holloway, R.E. 1996. Zn as a subsoil nutrient forcereal. Ph.D. thesis, University of Adelaide,Adelaide, South Australia,

Huang, C., Webb, M.J., and Graham, R.D. 1994a.Effect of pH on Mn absorption amongbarley genotypes in a chelate-bufferednutrient solution. Plant Soil155/156:437-440.

Huang, C., Webb, M.J., and Graham, R.D. 1994b.Mn efficiency is expressed in barleygrowing in soil system but not in solutionculture. J. Plant Nutr. 17:83-95.

Huang, C., Webb, M.J., and Graham, R.D. 1996.Pot size affects expression of Mn efficiencyin barley. Plant Soil 178:205-208.

Huber, D.M., and Wilhelm, N.S. 1988. The role ofmanganese in resistance to plant diseases.Chapter 11. In: Manganese in Soils andPlants. Graham, R.D., Hannam, R.J., andUren, N.C. (eds.). Kluwer AcademicPublishers, Dordrecht, The Netherlands.pp. 155-173.

IRRI. 1979. Annual Report, International RiceResearch Institute, Los Banos, ThePhilippines.


239

Jones, G.B., and Belling, G.B. 1967. Themovement of copper, molybdenum andselenium in soils as indicated by radioactivetracers. Aust. J. Agric. Res. 18:733-740.

Khabaz-Saberi, H., Graham, R.D., and Rathjen,A.J. 1998. Inheritance of Mn efficiency indurum wheat. J. Plant Nutr. 22:11-21.

Khabaz-Saberi, H., Graham, R.D., Ascher, J.S.,and Rathjen, A.J. 2000. Quantification of theconfounding effect of seed Mn content inscreening for Mn efficiency in durum wheat.J. Plant Nutr. 23 (7):855-866.

Loneragan, J.F., Kirk, G.J., and Webb, M.J. 1987.Translocation and function of zinc in roots.J. Plant Nutr. 10:1247-1254.

Longnecker, N.E., and Uren, N.C. 1990. Factorsinfluencing variability in manganese contentof seeds, with emphasis on barley (Hordeumvulgare) and white lupins (Lupinus albus).Aust. J. Agric. Res. 41:29-37.

Longnecker, N.E., Marcar, N.E., and Graham,R.D. 1991. Increased manganese content ofbarley seeds can increase grain yield inmanganese-deficient conditions. Aust. J.Agric. Res. 42:1065-1074.

Majumder, N.D., Rakshit, S.C., and Borthakur,D.N. 1990. Genetic effects on uptake ofselected nutrients in some rice (Oryza sativaL.) varieties in phosphorus-deficient soil.Plant Soil 123:117-120.

Marcar, N.E., and Graham, R.D. 1986. Effect ofseed manganese content on the growth ofwheat (Triticum aestivum) under manganesedeficiency. Plant Soil 96:165-173.

Marcar, N.E., and Graham, R.D. 1987. Toleranceof wheat, barley, triticale and rye tomanganese deficiency during seedlinggrowth. Aust. J. Agric. Res. 38:501-511.

Marschner, H., Romheld, V., and Kissel, M. 1986.Different strategies in higher plants inmobilization and uptake of iron. J. PlantNutr. 9:695-713.

McCarthy, K.W., Longnecker, N.E., Sparrow,D.H.B., and Graham, R.D. 1988. Inheritanceof manganese efficiency in barley (Hordeumvulgare L.). In: International Symposium onManganese in Soils and Plants: ContributedPapers. M.J. Webb, R.O. Nable, R.D.Graham, and R.J. Hannam (eds.).Manganese Symposium, Adelaide.pp. 121-122.

McCay-Buis, T.S., Huber, D.M., Graham, R.D.,Phillips, J.D., and Miskin, K.E. 1995.Manganese seed content and take-all ofcereals. J. Plant Nutr. 18:1711-1721.

Moody, D.B., Rathjen, A.J., Cartwright, B., Paull,J.G., and Lewis, J. 1988. Genetic diversityand geographical distribution of tolerance tohigh levels of soil boron. In: Proc. 7th Int.Wheat Genetics Symp. 13-19 July 1988. T.E.Miller and R.M.D. Koebner (eds.).Cambridge Laboratory, IPSR. Cambridge.pp. 859-865.

Moussavi-Nik, M. 1997. Seed quality and cropestablishment in wheat. Ph.D. Thesis.University of Adelaide, Adelaide, SouthAustralia.

Nable, R.O., and Loneragan, J.F. 1984.Translocation of manganese in subterraneanclover (Trifolium subterraneum L. cv. SeatonPark). II. Effects of leaf senescence and ofrestricting supply of manganese to part of asplit root system. Aust. J. Plant Physiol.11:113-118.

Pallotta, M.A., Khabaz-Saberi, H., Lewis, J.,Graham, R.D., and Barker, S.J. 1999.Breeding for tolerance to nutritional stress:Molecular mapping of loci for manganeseefficiency in barley and durum wheat. In:Proceedings of the 11th Australian PlantBreeding Conference, Adelaide. P.Langridge, A. Barr, G. Auricht, G. Collins, A.Granger, D. Handford, and J. Paull (eds.).2:144-145.

Paull, J.G. 1990. Genetic studies on the toleranceof wheat to high concentrations of boron.Ph.D. thesis, University of Adelaide,Adelaide, South Australia.

Pedler, J.F. 1994. Resistance to take-all disease byMn-efficient wheat cultivars. Ph.D. Thesis,University of Adelaide, Adelaide, SouthAustralia. 210 p.

Pollard, A.S., Parr, A.J., and Loughman, B.C.1977. Boron in relation to membranefunction in higher plants. J. Exp. Bot. 28:831-839.

Ponnamperuma, F.N. 1976. Screening rice fortolerance to mineral stresses. In: PlantAdaptation to Mineral Stress in ProblemSoils. M.J. Wright (ed.). Cornell Univ. Agric.Exp. Stn., Ithaca, New York.pp. 341-353.

Pope, D.T., and Munger, H.M. 1953a. Heredity andnutrition in relation to magnesium deficiencychlorosis in celery. Proc. Am. Soc. Hort. Sci.61:472-480.

Pope, D.T., and Munger, H.M. 1953b. Theinheritance of susceptibility to borondeficiency in celery. Proc. Am. Soc. Hort.Sci. 61:481-486.

Rao, V.S., Gangwar, M.S., and Rathore, V.S. 1977.Genotypic variation in distribution of totaland labelled zinc and availability of zinc (Aand L values) to soybeans grown in mollisol.J. Agric. Sci. 8:417-420.

Reid, D.A. 1976. Screening barley for aluminumtolerance. In: Plant Adaptation to MineralStress in Problem Soils. M.J. Wright (ed.).Ithaca, Cornell Univ. Agric. Exp. Sta.pp. 269-275.

Rengel, Z. 1992. Role of calcium in aluminiumtoxicity. New Phytol. 121:499-513.

Rengel, Z., and Graham, R.D. 1995a. Importanceof seed Zn content for wheat growth on Zn-deficient soil. I. Vegetative growth. Plant Soil173:259-266.

Rengel, Z., and Graham, R.D. 1995b. Importanceof seed Zn content for wheat growth on Zn-deficient soil. II. Grain yield. Plant Soil173:267-274.

Rengel, Z., and Graham, R.D. 1995c. Wheatgenotypes differ in Zn efficiency whengrown in chelate-buffered nutrient solution.I. Growth. Plant Soil 176:307-316.

Rengel, Z., and Graham, R.D. 1995d. Wheatgenotypes differ in Zn efficiency whengrown in chelate-buffered nutrient solution.II. Nutrient uptake. Plant Soil 176:317-324.

Rengel, Z., and Graham, R.D. 1996. Uptake ofzinc from chelate-buffered nutrient solutionsby wheat genotypes differing in Znefficiency. J. Exp. Bot. 47:217-226.

Rengel, Z., and Jurkic, V. 1992. Genotypicdifferences in wheat Al tolerance. Euphytica62:111-117.

Rengel, Z., Graham, R.D., and Pedler, J.F. 1993.Manganese nutrition and accumulation ofphenolics and lignin as related to differentialtolerance of wheat genotypes to the take-allfungus. Plant Soil 151:255-263.

Rerkasem, B., Bell, R.W., and Loneragan, J.F.1990. Effects of seed and soil boron on earlyseedling growth of black and green gram(Vigna mungo and V. radiata). In: PlantNutrition-Physiology and Applications. M.L.van Beusichem (ed.). Kluwer AcademicPublishers, Dordrecht, The Netherlands.pp. 281-285.

Rerkasem, B., and Jamjod, S. 1997. Geneticvariation in plant response to low boron andimplications for plant breeding. In: Boron inSoils and Plants: Reviews. B. Dell, P.H.Brown, and R.W. Bell (eds.) KluwerAcademic Publisher, Dordrecht, TheNetherlands. pp.169-180.

Ripperger, H., and Schreiber, K. 1982.Nicotianamine and analogous amino acids,endogenous iron carriers in higher plants.Heterocycles 17:47-461.

Romheld, V., and Marschner, H. 1981. Irondeficiency stress induced morphological andphysiological changes in root tips ofsunflower phytosiderophore in roots ofgrasses. Physiol. Plant. 53:354-360.

Rose, I.A., Felton, W.L., and Banke, L.W. 1981.Response of four soybean varieties to foliarzinc fertilizer. Aust. J. Exp. Agric. Anim.Husb. 21:236-240.

Saxena, S.C., and Chandel, A.S. 1992. Effect ofzinc fertilization on different varieties ofsoybean (Glycine max). Indian J. Agric. Sci.62:695-697.

Sparrow, D.H., and Graham, R.D. 1988.Susceptibility of zinc-deficient wheat plantsto colonization by Fusarium graminearumSchw. Group 1. Plant Soil 112:261-266.

Streeter, T.L. 1998. Role of Zn nutritional statuson infection of Medicago species byRhizoctania solani. Ph.D. thesis. TheUniversity of Adelaide, Adelaide, SouthAustralia.

MICRONUTRIENTS

240 J.S. ASCHER-ELLIS ET AL.

Thongbai, P., Graham R.D., Neate, S.M., andWebb, M.J. 1993a. Interaction between zincnutritional status of cereals and Rhizoctoniaroot rot. II. Effect of zinc on disease severityof wheat under controlled conditions. PlantSoil 153:215-222.

Thongbai, P., Hannam, R.J., Graham, R.D., andWebb, M.J. 1993b. Interaction between zincnutritional status of cereals and Rhizoctoniaroot rot severity. I. Field observations. PlantSoil 153:207-214.

Ulrich, A., and Ohki, K. 1966. Potassium. In:Diagnostic Criteria for Plants and Soils.H.D. Chapman (ed.). Riverside, Univ.California Press. pp. 362-393.

Uren, N.C. 1982. Chemical reduction at the rootsurface. J. Plant Nutr. 5:515-520.

Uren, N.C., Asher, C.J., and Longnecker, N.E.1988. Techniques for research on manganesein soil-plant systems. In: Manganese in Soilsand Plants. R.D. Graham, R.J. Hannam, andN.C. Uren (eds.). Martinus NijhoffPublishers, Dordrecht, The Netherlands. pp.309-328.

Wall, J.R., and Andrus, C.F. 1962. The inheritanceand physiology of boron response in thetomato. Am. J. Bot. 49:758-762.

Webb, M.J., Dinkelaker, B.E., and Graham, R.D.1993a. The dynamic nature of Mnavailability during storage of a calcareoussoil: Its importance for plant growthexperiments. Soil Biol. Fert. 15:9-15.

Webb, M.J., Norvell, W.A., Welch, R.M., andGraham, R.D. 1993b. Using a chelate-buffered nutrient solution to establishcritical tissue levels and the solution activityof Mn2+ required by barley (Hordeumvulgare L.). Plant Soil 153:195-205.

Weiss, M.G. 1943. Inheritance and physiology ofefficiency in iron utilization in soybeans.Genetics 28:253-268.

Welch, R.M., Webb, M.J., and Loneragan, J.F.1982. Zinc in membrane function and itsrole in phosphorus toxicity. In: PlantNutrition. Proceedings of the Ninth Int.Plant Nutrition Colloquium. A. Scaife (ed.).Commonwealth Agricultural Bureaux,Slough. pp. 710-715.

Welch, R.M., and House, W.A. 1983. Factorsaffecting the bioavailability of mineralnutrients in plant foods. In: Crops as Sourcesof Nutrients for Humans. R.M. Welch andW.H. Gabelman (eds.). Am. Soc. Agron.,Madison, WI. pp 37-54.

Welch, R.M., House, W.A., Beebe, S., Senadhira,D., Gregorio, G., and Cheng, Z. 1999.Testing iron and zinc bioavailability ingenetically enriched bean (Phaseolusvulgaris L.) and rice (Oryza sativa L.) usinga rat model. In: Improving Human Nutritionthrough Agriculture: The Role ofInternational Agricultural Research. IRRI/IFPRI Workshop, Los Banos, Philippines.pp. 2-16.

Wilhelm, N.S., Graham, R.D., and Rovira, A.D.1990. Control of Mn status and infectionrate by genotype of both host and pathogenin the wheat-take all interaction. Plant Soil123:267-75.

Yeo, A.R., and Flowers T.J. 1986. Salinityresistance in rice (Oryza sativa L.) and apyramiding approach to breeding varietiesfor saline soils. Aust. J. Plant Physiol.13:161-173.

Zarcinas, B.A., Cartwright, B., and Spouncer,L.R. 1987. Nitric acid digestion and multielement analysis of plant material byinductively coupled plasma spectrometry.Commun. Soil Sci. Plant Anal. 18:131-136.

Appl i cati o n of Phy s iol ogy i n Wheat Br eedi n g

M. P. Re y nold s , J. I . Or tiz -M

o n aste r io, a nd A. McNa b, E d itor s

M.P. Reynolds,J.I. Ortiz-Monasterio,and A. McNab,Editors

ISBN: 970-648-077-3

International Maize and Wheat Improvement CenterApdo. Postal 6-641, 06600 Mexico, D.F., Mexico

www.cimmyt.org

application of physiology in wheat breeding

Documents

tilletia indica

preharvest

suppressing

vapor pressure

carbon isotope

recombinant

kluwer academic

polymerase