root biomass, root - inpaagroeco.inpa.gov.br/reinaldo/ribarbosa_prodcient_usu_visitantes... · root...

Root biomass root shoot ratio and belowgroundcarbon stocks in the open savannahs of RoraimaBrazilian Amazonia

Reinaldo Imbrozio BarbosaAC Jhonson Reginaldo Silva dos SantosBMariana Souza da CunhaB Tania Pena PimentelA and Philip Martin FearnsideAC

ANational Institute for Research in the Amazon Department of Environmental DynamicAv Andreacute Arauacutejo 2936 CEP 69060-000 ManausAmazonas Brazil

BFederal University of Roraima Post-graduate Program in Natural Resources Campus ParicaranaAv Cap Ene Garcez 2413Bairro Aeroporto 69304-000 Boa Vista ndash Roraima Brazil

CCorresponding author Email philipfearnsidegmailcom

Abstract Biomass of roots the root shoot ratio (ratio of below- to abovegroundbiomass) and carbon stocks belowground(to 100-cm depth) were estimated in different open savannah environments in the extreme north of the Brazilian AmazonSampling was conducted in permanent plots established in two open savannah areas in the state of Roraima We identifiedfour phytopedounits in the27plots sampled in twoareas four indrygrasslandsonArgisolUltisol soils (DG-Arg) eight indrygrasslands on LatosolOxisol soils (DG-Lts) five in a mosaic of grasslands with savannah-parkland on LatosolOxisol soils(GP-Lts) and 10 in seasonally flooded (wet) grasslands on HydromorphicEntisol soils (WG-Hyd) Fine roots (lt2mmdiameter) dominated the 0ndash100-cm vertical profile in the four phytopedounits (gt925) Biomass of the roots in WG-Hyd(2952 715Mg handash1) was significantly higher as compared with the other phytopedounits studied although the carbonstocks did not differ among the phytopedounits (620ndash721MgC handash1) The largest concentration of roots was found in theupper three 10-cm sections of the soil profile ranging from 563 to 829 in the four environments The root shoot ratiobased only on living biomass of roots with diameter 2mm (standard Intergovernmental Panel on Climate Changemethodology) ranged from 0 for seasonally flooded grasslands to 007ndash020 for unflooded grasslands on clay soils Theresults indicate that the root shoot ratio (expansion factor) for belowground biomass in open savannah ecosystems in thenorthern Amazon are low and differ from the default values used in Brazilrsquos reference report to the Climate Convention

Received 10 December 2011 accepted 30 April 2012 published online 11 July 2012

Introduction

Savannah is a common type of vegetation in the tropics includingtheNeotropics (Solbrig et al 1996 Furley 1999) Their terrestrialcoverage has been estimated to be 16ndash19 106 km2 dependingon the ecogeographical definitions used (Scholes and Hall 1996Asner et al 2004) Most studies on the vertical and horizontalstructures of these ecosystems are aimed at identifying structuralpatterns associated with biological diversity and abovegroundbiomass or carbon stocks Root biomass and carbon are often notreported on a small scale because they demand so much timeand effort to sample However even with the small number ofstudies some reviews suggest that roots in tropical savannahand grassland ecosystems represent a major compartment forcarbon accumulation (Jackson et al 1996 Mokany et al 2006)Estimates of these stocks are important for national inventoriesof greenhouse gases under the UN Framework Convention onClimate Change

Generally distribution production and accumulation of rootsare related to water availability in the soil (climatic seasonality)

which is a variable with strong temporal fluctuations in themore superficial soil layers in savannahs and grasslands (SanJoseacute et al 1982 Baruch 1994 Delitti et al 2001) Other factorshave been investigated in order to understand which processesregulate subterranean biomass such as physionomic structure(Sarmiento and Vera 1979 Castro and Kauffman 1998) nutrientavailability (Kellman and Sanmugadas 1985 February andHiggins 2010) human alterations (Fiala and Herrera 1988)fire (Menaut and Cesar 1979 Castro-Neves 2007) and grazing(Pandey and Singh 1992 Milchunas and Lauenroth 1993McNaughton et al 1998) These studies make it possible toassess parameters such as root shoot ratio (the lsquoexpansionfactorrsquo used for inferring belowground biomass fromaboveground biomass measurements) and rates of growth ofroot biomass under different successional paths which arecrucial to the understanding of belowground carbon allocationHowever no study has been conducted inAmazonian savannahs

In savannahs the thicker roots represent an important reservoirof carbon at greater depths especially in physiognomies that are

CSIRO PUBLISHING

Australian Journal of Botany 2012 60 405ndash416httpdxdoiorg101071BT11312

Journal compilation CSIRO 2012 wwwpublishcsiroaujournalsajb

more densely populated by trees (Abdala et al 1998 Schenk andJackson2002a)However theseopen-vegetationenvironmentsarecharacterised by having large grassy expanses and a low-densitytree component Most roots in these ecosystems are located in themost superficial layers of the soil and are characterised by havingsmall diameters (lt2mm) (Knoop andWalker 1985) It is estimatedthat the different forms of savannahs (open and wooded) contain~20 of all of the fine-root biomass on Earth to a depth of 30 cm(Jackson et al 1997) This category of roots has a high rate ofreplenishment in tropical grasslands and savannahs making it acritical component in sequesteringatmospheric carbonbyallowingconstant accumulation of organic matter in the soil (Stanton1988 Gill and Jackson 2000 Chen et al 2004) However theIntergovernmentalPanelonClimateChange(IPCC2006) suggeststhat this category of roots (lt2mm in diameter) should not beincluded as part of the lsquobelowground biomassrsquo because it isdifficult to distinguish it empirically from soil organic matter

Savannahs grasslands and other natural non-forestecosystems occupy ~200 000 km2 (~5) of the BrazilianAmazonia (Santos et al 2007) Although the area of theseecosystems is substantial all existing studies on the biomassof roots in natural vegetation in the Amazonian biome in Brazilare from forest ecosystems due to the much larger area offorests (Klinge 1973 Luizatildeo et al 1992 Thompson et al1992 Nepstad et al 1994 Cattanio et al 2004) Estimatesof the temporal dynamics and spatial distribution of rootsin different phytophysionomies of Amazon savannahs arenonexistent representing an important lacuna in ourknowledge about this potential belowground carbon reservoirin the Amazon These data are important for understandingthe role of these environments in mitigating global warming(IPCC 2007)

The first Brazilian inventory of greenhouse gases whichwas submitted to the Climate Convention (UN FrameworkConvention on Climate Change) in 2004 did not consider theroots explaining that lsquothe consideration of carbon below ground(roots) is complex andwas not included in this inventoryrsquo (BrazilMCT 2004 p 146) Of course considering the carbon stock inroots to be zero and consequently considering the emission fromthis source to be zero after clearing the vegetation represents asubstantial underestimation especially in savannahs where thegreat majority of the biomass is below ground The secondBrazilian inventory used the default root shoot ratios valuespresented by IPCC (2003 p 3109 table 343) to estimatetotal biomass for all grassland and savannah environmentslisted in Brazil MCT (2010 pp 236ndash237) This was done bothfor the cerrados of central Brazil and for Amazon savannahsThe root shoot ratio is an expansion factor used to estimatebelowground biomass from aboveground biomass (IPCC 2006p 67) However use of the IPCC default values for calculatingbelowgroundbiomass in open-vegetation systemsunder differentenvironmental conditions can cause undesired distortion in thefinal values for total biomass

Within this context our goal was to estimate the biomass ofroots indifferent open savannah environments inRoraima inviewof the combination of two effects (phytophysionomic structureand soil class forming a lsquophytopedounitrsquo) The phytopedounitspresented here are similar to the lsquolandscape system unitsrsquo definedby Sombroek et al (2000) A classification including both

vegetation and soil effects is important in order to preventphytophysionomies with the same structure on different soilclasses from being analysed as the same ecological unit Thisstrategy includes spatial variations that occur along differentedaphic gradients that may affect the modelling of rootbiomass (Espeleta and Clark 2007)

The current study included the following questions (i) Doesthe allocation of total carbon to roots differ among opensavannahs on different phytopedounits (ii) Does the verticaldistribution of root biomass differ among these environmentsand (iii) Does the ratio between the biomass of roots (totaland 2mm) and the aerial biomass (root shoot) differ amongthe phytopedounits investigated Our results represent anopportunity to reformulate the estimates of belowgroundbiomass and carbon stocks in Amazonian savannahsproviding appropriate regional values for open-vegetationsystems with low densities of trees and shrubs

Materials and methodsSavannahs of RoraimaSavannahs of Roraima are part of the Rio Branco-Rupununicomplex located in the triple frontier between BrazilVenezuela and Guyana (Beard 1953 Eden 1970) Altogetherthese continuous savannahs cover 68 103 km2 the Brazilianpart being ~43 103 km2 (~63) (Barbosa et al 2007 Barbosaand Campos 2011) In general these savannahs are located onpoor soils with high frequency of fire and strong climaticseasonality that directly influences the fluctuation of thewatertable and the phytophysionomic structure (Miranda et al2003 Barbosa and Fearnside 2005a) The climate type of thiswhole region is Awi according to the Koumlppen classification withaverage rainfall of ~1650mmyearndash1 the peak of the dry periodis between December and March and the rainy period betweenMay and August (Barbosa 1997) These savannahs have a widevariety of phytophysionomies ranging from grasslands that aretotally devoid of trees to densely populated types on differentsoil classes (Brazil Projeto RADAMBRASIL 1975 Barbosaand Fearnside 2005b) The Venezuelan llanos have structureand species composition that are similar to those of the savannahsof Roraima (San Joseacute and Farintildeas 1983 Medina and Silva 1990)and neither of these should be confused with the savannahs(cerrados) of central Brazil (Eiten 1978)

Study areasThe studywas carried out in two savannah areas that have samplegrids for a Research Program on Biodiversity (i) Aacutegua BoaExperimental Station (AB) and (ii) Cauameacute or lsquoMonte CristorsquoCampus (MC) (Fig 1) The grids are composed of walking trailsin the NorthndashSouth (N-S) and EastndashWest (E-W) directions thatcross the area at intervals of 500m All sampling was performedbased on permanent plots (10m 250m) that are systematicallydistributed at points equidistant from the intersections of E-Wand N-S trails Each plot is an independent sampling unit thatfollows the contour line established beginning from the initialpicket This configuration was adopted to minimise the effects oftopographic variability in each plot (Magnusson et al 2005) Allplots are individually classified by soil class and vegetation

406 Australian Journal of Botany R I Barbosa et al

physionomy The general descriptions of the sample sites aregiven below

Aacutegua Boa Experimental Station (AB)This experimental station of the Brazilian Enterprise forAgriculture and Ranching Research (EMBRAPA-Roraima) islocated ~35 km south of the city of Boa Vista on the BR-174Highway (25104900N 2530600N and 604401400W 426002700W)The grid area is 616 ha and relief is typically flat with an averagealtitude of 777 13m Seventeen of the 22 terrestrial plots inthis grid were sampled The soil classes determined by BrazilrsquosNational Soil Survey andConservation Service indicate thatmostof the area has low fertility and high aluminium toxicity (BrazilSNLCS 1996)

Most of the grid is seasonally flooded grasslands with variousspecies of Poaceae and Cyperaceae (Arauacutejo and Barbosa2007) In this area the soils are typically hydromorphic and ofsandy texture due to an association of Gleysols with quartzo-arenitic Neosols (Entsols) A smaller part of the grid has twotypes of savannah on clay soils that are not exposed to periodicflooding (dry grasslands) and are characterised by the highdensity of the tree-bush component (i) low density (lt5canopy cover) represented by grassland savannahs mixedwith scrubby savannah (shrublands) and (ii) medium density(5ndash20) characterised by shrublands mixed with grassland and

savannah-parkland In this sector of the grid the soil is welldrained and problems of flooding are not present

Cauameacute Campus (MC)The Cauameacute Campus known as lsquoMonte Cristorsquo belongs to theFederal University of Roraima and is situated ~15 km north ofthe city of Boa Vista on the BR-174 Highway segment thatleads to the border with Venezuela (2380700N 24001100N and604902500W 605202800W) The grid has an area of 498 ha withan average altitude of 773 49m The relief -is flat to gentlyrolling and is derived from the Apoteri Geological FormationThis area is the most densely wooded type on clay soils Ten ofthe 12 terrestrial plots in this grid were sampled The soil classeswere determined by Benedetti et al (2011) indicating thatthis grid has soils with better drainage as compared with theAacutegua Boa grid

Sampling design and proceduresThe sampling period for collections and field assessments wasbetween 3 June 2009 and 27 February 2010 Sampleswere pairedin all plots between the rainy and the dry seasonWe adopted thiscriterion in order to avoid distortions that would either under- oroverestimate biomass depending on the collection period Thiswas necessary because there is strong seasonal variation in rootproduction in grassland and savannah areas (Neill 1992)

Fig 1 Location of the two sample areas established in savannahs (lavrado) in Roraima Brazil

Savannah root biomass and carbon in Brazilian Amazonia Australian Journal of Botany 407

Our first goal was to quantify total aboveground biomassthrough direct methods (for herbaceous vegetation) andindirect methods (for trees and bushes) We sampled rootsusing two methods (i) direct (destructive) to understand thevertical distribution of small-diameter roots which aregenerally associated with grasses and herbs and (ii) indirect(regression) to calculate the total biomass of the root crown inthe tree-bush component Although the term lsquoroot crownrsquo isusually used to refer to roots located immediately below thesurface of the soil under the main stems of the plants (Snowdonet al 2000) we use this term to specify coarse roots at thetransition point between stem and soil including all roots10mm in diameter up to 1-m depth The term lsquoroot crownrsquohas been used by Abdala et al (1998) to refer to roots in thisdiameter and depth range that are located directly beneath theaerial portion of the tree thereby distinguishing these rootsfrom roots of the same diameter located in the open spacesbetween the trees However in the case of open savannahs inRoraima where trees are widely spaced and root diameterdistributions are dominated by small- and medium-diameterroots the biomass of roots 10mm in diameter is negligiblein the open spaces and a separate category for these roots wouldhave minimal effect on the overall total

Total aerial biomass was estimated from the sum of itstwo components (i) herbaceous and (ii) tree-bush Herbaceousbiomass was defined as lsquograssesrsquo (Poaceae Cyperaceaeseedlings small dicots and litter) and woody individuals withdiameter at the base (Db) lt2 cm measured at 2 cm abovethe ground We sampled this group by establishing foursubsampling points in each permanent plot The firstsubsample was established just to the right (R) of the 50-mpicket perpendicularly at a distance of 5m from the referenceline for the central trail in the permanent plot This procedure wasperformedalternately using thepicket at 100m(L-left) 150m(R)and 200m (L)

After marking the four points we used a 1-m2 metal frame todelimit the area for destructive sampling All individuals in thisgroup within the metal frame were cut close to the ground usingmetal blades They were then weighed to obtain the wet weightcorresponding to the subsample point A composite sample ofherbaceous biomass (80ndash150 g) was brought to the laboratory fordetermination of its dryweight after drying in an oven at 70ndash75Cuntil constant weight The total herbaceous biomass in each plotwas estimated by discounting the water content from the totalfresh weight of each subsample and then calculating a simplemean of the four subsamples

To estimate the total carbon corresponding to the herbaceousbiomass we used the carbon content (C) in the form of aweighted average of the different components of this group asdescribed by Barbosa (2001) The weighted average of C wascalculated separately for each experiment station 344 (MC)and 362 (AB)

Live tree-bush biomass was defined as the group of woodyindividuals composed of two vertical strata (tree and bushor shrub) as set in Miranda et al (2002) and Barbosa et al(2005) The area used for sampling the arboreal stratum was10m 250m while the shrubs were sampled in a subplot(2m 250m) The central trail of the plot was always used asthe baseline for the sampling All individuals in the tree-bush

group were identified taxonomically and inventoried bymeasuring biometric parameters Db = diameter of the base ofthe stemmeasuredat 2 cmabove thegroundD30 = diameter of thestem measured at 30 cm height Dc = diameter of the canopycalculated as the average of the largest and smallest individualcrown diameter Ht = total height defined as the distance from theinsertion of the stem in the ground to the top of the canopy Theseparameters were used to indirectly estimate the biomass of eachtree-bush individual based on the regression model developedby Barbosa and Fearnside (2005b) for savannahs in RoraimaTree-bush biomass of each plot was derived from the sum of allindividual biomasses

The carbon corresponding to the tree-bush biomass of speciesinventoried in the two gridswas estimated fromdata derived fromBarbosa (2001) for biomass of savannah species in Roraimaaccording to the weighting given in Supplementary materialpart A

Total biomass of rootsDirect method (destructive)

The sampling of root biomass using the direct (destructive)method came from the same four subsampling points establishedfor herbaceous biomass estimates The goal of this methodwas toobtain mean data for each plot at five depths in the soil column0ndash10 cm 10ndash20 cm 20ndash30 cm 30ndash40 cm and 40ndash50 cm Thismethod checks the inventory and the vertical distribution patternof small-diameter roots present along the altitudinal gradient inthe plots In general these roots are associatedwith grasses herbsand the lateral roots of the small-diameter individuals of the tree-bush component

Each subsample was obtained at the exact geometric centre ofthe metal frame used to delimit the area for destructive samplingof the herbaceous biomass We used a soil collector measuring08m in length by 01m in diameter adapted for collecting soil atdepths up to 05m Each sample was placed directly in a plasticbag identified individually by depth and was then weighed toobtain the net weight in the field The samples were thenforwarded to the laboratory for separation of the roots and fordetermination of air-dried weight These weights allowedcalculation of soil density (dry weight of the soil divided by itssaturated volume) for each 10-cm section of the soil column

Weseparated live rootsmanually packing them inplastic bagsidentified by the plot the subsample point and the depth in theprofileAfter this initial screening and separation the residual soilwas subjected to the floatation process as suggested by McKellet al (1961) This method consists of addingwater to the residualsoil so that the lighter plantmaterial thatwas not visible in thefirstseparation would float and could be collected and added to theroots separated in the previous stage After this process the rootswere placed in a drying oven at 70ndash75C until they attainedconstant weight

Throughout the process all of the collected material wasseparated by diameter category (d) using the classes suggestedby Abdala et al (1998) dlt 2mm (very fine and fine roots) 2 dlt10mm (medium) and d10mm (coarse) After sortingwashing and drying all categories were weighed individuallyto obtain the mean biomass of roots by diameter category depthsection plot and phytopedounit


Indirect method (root crown)This method was to estimate the biomass of the lsquoroot crownrsquo

(as defined above) of the individual trees or bushes withD302 cm In open savannah ecosystems of Roraima the tree-bush component is present at low density and our destructivemethod therefore cannot providevalues for thebiomassof the rootcrown of trees and shrubsWe therefore used an indirect methodapplying the linear regression of Abdala et al (1998) to estimatethis category We assumed that the values derived from theregression correspond to the lsquoroot crownrsquo (roots10mm indiameter) connected to the trees and bushes with D302 cmup to 1-m depth

Laboratory analysesAll of the biomass of roots collected by the direct methodwas separated by plot and vertical section of depth and thenground using a knife mill The samples were then sent to theSoil and Plant Thematic Laboratory at the National Institutefor Research in the Amazon Manaus Amazonas Brazil fordetermination of carbon content (C) The C was determinedusingaCHNAuto-Analyzer (VarioMAXElementarGermany)This equipment performs the analysis by combustion at hightemperatures followed by reduction (Nelson and Sommers1996) In the case of C for the root crowns of trees andshrubs we used the same values for aboveground tree-bushbiomass

Data analysisThe differences between all values obtained for the herbaceousand tree-bush components and for the total biomass in eachenvironment were verified using the KruskalndashWallis non-parametric test (H005) (Zar 1999) In cases where the nullhypothesis (equal means) was rejected the StudentndashNewmanndashKeuls test was applied for multiple comparisons (P lt 005)

All values for root biomass (total biomass and biomass bydiameter category) were transformed into units of weight per unitarea for each 10-cm section of the soil profile Using the resultsfor the biomass of roots for each section of the 0ndash50-cm soilprofile we derived an estimate for the 50ndash100-cm section Thisestimate was to combine the information from this methodwith the same profile (0ndash100 cm) adopted for calculating theroot crown For both each result obtained from destructivesubsamples (0ndash50 cm) was applied using an exponential model

(Y = a b endashX) with a unique value for the 50ndash100-cm sectionfor each subsample (see Jackson et al 1996)

Different sections up to 1-m depth were summed to determinethe biomass in the vertical soil column We also used theKruskalndashWallis test (H005) to assess biomass differences (bydiameter category and total) in the soil column and the verticaldistribution patterns of biomass in all environments The totalbiomass per unit area for roots to 1-m depth was added to thevalues estimated by regression for root crowns in the tree-bushstratum

Carbon allocation in aerial biomass and roots was calculatedfrom the multiplication of each of these groups by thecorresponding carbon fraction For calculating the root shootratio for each phytopedounit we used the values of liveaboveground and belowground biomass and carbon (direct andindirectmethods)Wealso carriedout a separate analysis for rootswith diameter2mm (direct and indirect methods) The purposeof this second analysis was to provide values for Amazon opensavannahs that could be used in the national inventory asrecommended by the IPCC (2006 p 8)

Results

Aboveground biomass and carbon

Four phytopedounits were observed in the 27 plots sampled inthe two experimental grids in open savannahs in RoraimaThe main tree-bush species present in dry grasslands both onArgisols (DG-Arg) and on Latosol (DG-Lts) as well as in themore densely wooded landscapes (GP-Lts) were Curatellaamericana L (Dilleniaceae) Byrsonima crassifolia (L) Kunth(Malpighiaceae) and B coccolobifolia Kunth (Malpighiaceae)In wet grasslands (WG-Hyd) woody individuals were not foundwith D302 cm

Herbaceous and tree-bush biomasses differ significantlyamong environments (Table 1) The total herbaceous biomassof WG-Hyd (901 286Mg handash1) was the largest valueamong all of the phytopedounits although it only differedfrom the DG-Arg environment No significant difference wasdetected between the largest total biomass (WG-Hyd) and theother phytopedounits due to a greater presence of the tree-bushcomponent in the dry grasslands and in the mosaic of grasslandswith parkland savannahs

The WG-Hyd phytopedounit had the largest carbon stock inthe herbaceous component (326 104MgChandash1) but GP-Lts

Table 1 Aerial biomass distribution by group in different phytopedounits in two grids of open savannahs in Roraima Brazil(mean sd)

Values in parentheses represent the plotrsquos live component (Mg handash1) with the litter (dead biomass) already discounted Different lowercaseletters indicate distinct significance amongvalues in each column(StudentndashNewmanndashKeuls testPlt005)AB=AacuteguaBoaDG-Arg = drygrasslands on Argisols DG-Lts = dry grasslands on Latosols GP-Lts =mosaic of grasslands with savannah-parkland on Latosols

MC=CauameacuteMonte Cristo WG-Hyd=wet grasslands on Hydromorphic soils

Phytopedounit Number of plots (n) Herbaceous Tree-bush TotalAB MC

DG-Arg 0 4 525 plusmn 036a (459) 106 plusmn 068bc 631 plusmn 088a (565)DG-Lts 5 3 674 plusmn 191ab (589) 060 plusmn 108b 734 plusmn 196a (650)GP-Lts 2 3 610 plusmn 278ab (534) 276 plusmn 159c 887 plusmn 243a (810)WG-Hyd 10 0 901 plusmn 286b (765) 000a 901 plusmn 286a (765)


(347 089MgC handash1) had the largest total aboveground carbonstock (Fig 2) In this environment the tree-bush componentrepresents a greater proportion of the biomass and has greatercarbon content per unit of weight

Belowground total biomass and carbon

The total biomass of roots determined for the 0ndash100-cm profile(direct + indirect method) of WG-Hyd (2952 715Mg handash1)was greater as compared with the other phytopedounits studied(Table 2) The phytopedounits with tree-bush biomass allhave the smallest values for root biomass Fine roots (lt2mm)in diameter dominated the 0ndash100-cm vertical profile for thefour open savannah phytopedounits evaluated in Roraima Theconcentration of this category reached 100 inWG-Hyd andwasbetween 925 and 979 in the other environments Themedium-diameter roots (2ndash10mm) were found in three landscapes that

contained tree-bush biomass with no significant differencesbeing detected between these phytopedounits for this diametercategory Coarse roots (10mm) were determined only by theindirect method with no concentration of this diameter classbeing detected by the direct (destructive) method

The carbon content (C) in the root biomass measured by thedirect method varied from 248 inWG-Hyd where herbaceousbiomass predominated on sandy soil to 317 in GP-Lts whichwas the environment with the greatest presence of tree-bushbiomass on clay soil (Table 3)

Vertical distribution

The vertical distribution of root biomass for the fourphytopedounits evaluated in open savannahs of Roraimameasured by the direct method followed a pattern ofexponential decrease with the greatest values in the 0ndash10-cmsection and smaller values in the subsequent sections (Fig 3)The largest concentration of roots (fine and medium) wasfound in the first three sections of the vertical profile of thesoil (0ndash30 cm) ranging from 560 to 646 in the fourenvironments Taking into consideration only these threesections of depth the biomass of the roots of WG-Hyd wassignificantly different from the other environments that had atree-bush component

Root shoot ratio (biomass and carbon)

WG-Hyd was the environment with the highest absoluteroot shoot ratio taking into consideration the totalaboveground and belowground live biomass (Table 4) Thevalues of the root shoot ratios calculated on the basis ofcarbon were smaller than those calculated on the basis ofbiomass in all phytopedounits Using only the carbonvalues for roots 2mm in diameter the root shoot ratio havethe highest values in GP-Lts and DG-Arg (both with high tree-bush biomass)

00

05

10

15

20

25

30

35

40

45

DG-Arg DG-Lts GP-Lts WG-Hyd

Car

bon

stoc

ks (

MgC

handash

1 )

Phytopedounits

GrassTrees and shrubs

Fig 2 Distribution of aboveground biomass carbon stock in twocomponents (herbaceous and tree-bush) for the four phytotopedounitssampled in open savannahs of Roraima Brazil

Table 3 Carbon content (C) and total carbon stock (mean sd) in roots of open savanna phytopedounits in Roraima Brazil(0ndash100-cm depth)

MeanCwascalculatedbyweightingbetweendirect and indirectmethodsThere is nosignificantdifferencebetweenvalueswith the sameletter on each line (StudentndashNewmanndashKeuls test P lt 005) See Table 1 for phytopedounit definitions

Phytopedounit Sub-total (direct method) Sub-total (indirect method) TotalMgC handash1 C MgC handash1 C MgC handash1 C

DG-Arg 628 plusmn 029a 3110 plusmn 290 041 plusmn 034bc 4680 669 plusmn 029a 3206DG-Lts 604 plusmn 115a 2703 plusmn 263 021 plusmn 014b 4673 625 plusmn 112a 2769GP-Lts 662 plusmn 200a 3173 plusmn 410 058 plusmn 010c 4609 721 plusmn 185a 3289WG-Hyd 710 plusmn 165a 2480 plusmn 616 000a ndash 710 plusmn 165a 2480

Table 2 Distribution of root biomass (mean sd) by diameter category (0ndash100 cm)Different lowercase letters in each column indicate a distinct difference among values (StudentndashNewmanndashKeuls test P lt 005)

See Table 1 for phytopedounit definitions

Phytopedounit Fine roots Medium roots Coarse roots Total(lt2mm) Mg handash1 (2ndash10mm) Mg handash1 ( 10mm) Mghandash1 Mg handash1

DG-Arg 2027 plusmn 139a 026 plusmn 015b 087 plusmn 072bc 2140 plusmn 247aDG-Lts 2214 plusmn 147a 014 plusmn 010b 033 plusmn 033b 2262 plusmn 221aGP-Lts 2049 plusmn 169a 039 plusmn 015b 126 plusmn 022c 2214 plusmn 490aWG-Hyd 2952plusmn 240b 000a 000a 2952 plusmn 715b


Discussion


All of the environments evaluated had tree-bush biomass valueswithin the expected range for open savannahs in Roraima(005ndash364Mg handash1) (Barbosa and Fearnside 2005b) Howeverour values for total herbaceous biomass are closer to the valuesfound by Castro and Kauffman (1998) (60ndash75Mg handash1) andCastro-Neves (2007) (62ndash104Mg handash1) for cerrado areas nearBrasiacutelia (in central Brazil) than to those found by Barbosaand Fearnside (2005b) for open savannahs in Roraima(255ndash418Mg handash1) Both studies in cerrado carried out theirsampling at the peak of the dry period because they wereinterested in estimating the emission of greenhouse gases byburning Our research tookmeasures over thewet and dry periodsto avoid biasing the mean for each environment In the short termthis method entails higher values for herbaceous biomass due tomore favourable edaphic conditions in thewet season and even inthe early months of the dry season as in the case of WG-HydThis environment is characterised by the dominance of a grassystratum that makes use of the soil moisture in order to expand itsaboveground biomass In the long term savannah ecosystemswhere water availability is not limiting or that are protected fromfire tend to increase their belowground biomass (Sarmiento 1984San Joseacute et al 1998)

Our results imply that while the herbaceous and tree-bushcomponents are heterogeneous amongst themselves the resultsfor aboveground total biomass and carbon of each phytopedounitmay be considered homogeneous representing the same set ofopen savannah environments inRoraimaThe phytophysionomicgroups with low or average density of trees and bushes havetotal biomass and carbon that are similar to exclusively grassyenvironments under periodic flooding regardless of the soil type

Belowground biomass and carbon

Our values for total biomass of live roots (direct + indirectmethods) for WG-Hyd are higher than the 114ndash189Mg handash1

presented by Sarmiento and Vera (1979) for savannah gradientsbetweengrasslands andwoodlands in theVenezuelan llanosup to2-m depth However despite differences in the sampling depthmore than 90 of the roots found in the study by Sarmiento andVera are located in the top 60 cm which is a value very close to76ndash83 of our study to 50 cm In contrast the biomass of rootsderived from studies in the cerrado of central Brazil is largerAbdala et al (1998) estimated a total value of 411Mg handash1

(live + dead) in a 62-m profile for a cerrado lsquosensu strictorsquo ondark red latosol near Brasiacutelia of which ~233Mg handash1 were in thefirst 50 cm (excluding the root crowns) Similarly Castro andKauffman (1998) foundvalues ranging from163 to 529Mg handash1

(2m) for live roots in different savannah types ranging fromgrassland (campo limpo or lsquoclean fieldrsquo) to woodland (cerradodenso or lsquodense cerradorsquo) also located close to Brasiacutelia with~80 concentrated in the first 50 cm of depth

Differences between our results for total belowground livebiomass inRoraima and those for the cerrado of central Brazil areclearly due to sampling being performed at sites with differentphytophysionomies depths and burning schemes Despite thiscontrast it is possible to infer that regardless of the depth orsavannah type the total biomass of live roots in areas of opensavannah in Roraima with a low or medium presence of the tree-bush component are closer to those in the Venezuelan llanos thanto those of the cerrado of central Brazil This should be expected

00

15

30

45

60

75

90

00ndash10 10ndash20 20ndash30 30ndash40 40ndash50 50ndash100

Roo

t bio

mas

s (M

g ha

ndash1)

Depth (cm)

DG-Arg DG-Lts GP-Lts WG-Hydb

b

bb

a

c

a aa a

b

aa

aaa

a

aa

a

a

aa

a

Fig 3 Vertical distribution of root biomass (Mghandash1) by depth interval as estimatedby the directmethod (50ndash100 cm calculatedby exponential regression) in four open savannah phytotopedounits evaluated in Roraima Values with the same letter in eachdepth interval have no significant difference between means as determined the by StudentndashNewmanndashKeuls test (Plt 005)

Table 4 Root shoot ratio in different phytopedounits sampled in opensavannas of Roraima Brazil for total (live) biomass and carbon and

separately for roots with $2mm diameterSee Table 1 for phytopedounit definitions

Phytopedounit Total 2mmBiomass Carbon Biomass Carbon

DG-Arg 379 333 020 024DG-Lts 348 264 007 008GP-Lts 273 241 020 024WG-Hyd 386 256 0 0


since both the Venezuelan llanos and the open savannahs ofRoraima have similar species composition physionomicstructure soil type and rainfall regime (San Joseacute and Farintildeas1983 Medina and Silva 1990 Miranda et al 2002)

Another important inference is that the WG-Hydphytopedounit which is grassy and seasonally flooded canhave a large absolute increment in the biomass of live rootseven in hydromorphic soils This observation was also made byMenaut and Cesar (1979) when they investigated seven types ofsavannah in Lamto (IvoryCoast) also indicating that the biomassin wooded environments is almost always constant regardlessof the density of trees This contrastswith the general conclusionsof Castro and Kauffman (1998) in the cerrado of centralBrazil indicating that dominance of aboveground woodybiomass is reflected in increased belowground biomass Inour study total carbon allocated to roots did not differbetween the phytopedounits evaluated in open savannahs ofRoraima supporting the idea of uniformity among the openenvironments studied with low or no tree density

The concentration of fine roots in the first layers of the soil intropical savannah and grasslands is a pattern detected globallyOliveira et al (2005) observed that up to 1-m depth fine rootsrepresented ~90of the total determined for two types of cerrado(campo limpo lsquograsslandrsquo and campo sujo lsquoscrubby savannahrsquo) incentralBrazil In ageneral review Jackson et al (1996) calculated57 (990Mg handash1) as the average proportion of fine roots in theupper 30 cm of soil in tropical savannahs and grasslands Ourstudy indicates that in open savannahs of Roraima these figuresare higher in absolute terms and can reach values almost doublethe general average found by Jackson and collaborators forfine roots to 30-cm depth (115ndash191Mg handash1 or 55ndash65 forthe 0ndash100-cm profile)

The most significant example is the WG-Hyd savannah typewhich is seasonally flooded and has 100 fine roots (lt2mm)throughout the sampled soil column The plants in this type ofenvironment are fully adapted to soilswith sandy texture periodicflooding and aluminium toxicity but this savannah type has thelargest biomass of roots even under these unfavourable edaphicconditions In part this expansion is explained by the prolongedmaintenance of moisture in the soil in these phytopedounits evenduring the dry season WG-Hyd has the largest concentration ofroots between 0 and 10-cmdepth (264) and the lowest between50 and 100-cm depth (171) suggesting that the exploitationof nutrients in this soil is very superficial Environments withgreater presence of grasses are more efficient in absorbingwater and nutrients in the upper soil layers because of the highconcentrations of fine roots (Knoop and Walker 1985) Inaddition sandy soils can also have a positive effect on rootbiomass increase as compared with soils with more clayeysoil texture (Silver et al 2000) Roots with smaller diameterhave higher surface area relative to their size or weight andare more effective in capturing water and nutrients (Newman1966 Vitousek and Sanford 1986) Nutrient-poor tropicalenvironments therefore tend to have larger quantities of fineroots in the upper layers of the soil with high rates ofreplacement (turnover rate) and better capacity to absorbnutrients (Jordan and Escalante 1980 Priess et al 1999)

The larger-diameter roots (2mm) are essential for thecalculation of the total belowground biomass and carbon

stock even in environments with low tree-bush density as inthe case of open savannahs in Roraima The direct methodallowed us to sample medium-diameter roots (2ndash10mm) underconditions of lateral rooting Adding this medium-root biomassto the biomass determined by the indirect method for coarseroots (10mm) indicates that 0ndash75 (0ndash165Mg handash1) of thetotal belowground biomass in savannahs with low tree-bushdensity in the far northern part of Amazonia is live roots with2mm diameter In the more-wooded environments of thecerrado of central Brazil the biomass of this component canreach values gt20Mg handash1 (Abdala et al 1998 Castro-Neves2007) depending on the tree-bush structure and density

The smaller carbon content (C) in roots found in the0ndash50-cm soil column suggests a direct relationship with thelarge quantity of fine roots found in all of the savannah typesstudied For example Manlay et al (2002) also found lowvalues for carbon content (298ndash351 C) for fine roots underagricultural crops established in savannah areas in West AfricaCarbon content values lower than 40 are not common in theliterature but can be expected where the material analysed doesnot have lignified parenchyma Fine roots are characterised asnon-ligneous almost all being without bark and with a short lifecycle (McClaugherty et al 1982) These smaller-diameter rootsdie steadily throughout the year and quickly disappear fromthe system (Yavitt and Wright 2001) providing an importantsource of organic matter and mineral nutrients for maintenanceand functioning of ecosystems (Luizatildeo et al 1992)

Gill and Jackson (2000) presented a range of 064ndash088 for theturnover rate in open environments in tropical zones (grasslandsshrublands and wetlands) Taking the midpoint of this range(076) and applying the results derived for the total carbonstock of roots in the phytopedounits evaluated in Roraima(Table 3) we estimate an annual carbon cycling on the orderof 47ndash55MgChandash1 (0ndash100 cm) In temperate forest ecosystemsit is estimated that ~13 of this carbon is used in the productionof new roots (Nadelhoffer and Raich 1992) but there are noestimates of this for tropical Amazonian savannahs andgrasslands


The distribution pattern of root biomass within the vertical soilprofile observed in fourphytopedounitswas typically exponentialwith most of the roots concentrated in the surface layersThis pattern is the same as that observed in other studies ofNeotropical savannahs and grasslands in northern SouthAmerica(Sarmiento and Vera 1979 San Joseacute et al 1982) in CentralAmerica (Kellman and Sanmugadas 1985 Fiala and Herrera1988) and in the Brazilian cerrados (Abdala et al 1998Castro and Kauffman 1998 Delitti et al 2001 Rodin 2004Oliveira et al 2005 Castro-Neves 2007 Paiva and Faria 2007)This is also an overall global pattern observed in other ecosystemswhere the great majority of roots is concentrated in the top 30 cmof the soil (Jackson et al 1996 Schenk and Jackson 2002a)

In our case the extinction or decay coefficient (b) cannot becalculated using the formula of Gale and Grigal (1987) becauseour datawere divided into 10-cm sections only up to 50-cmdepthwhereas at least 100 cm would need to be so divided for acalculation of this type (see Jackson et al 1996) However


based on the decay pattern in the current data up to 30-cm depthwe suggest that all environments investigated have a superficialroot distribution with WG-Hyd (grassy environment on sandysoil) beingmost prominent (b30 cm = 095) as comparedwith tree-bush environments on clay soils (b30 cm = 096)

These values are lower than those given in the general reviewof Jackson et al (1996) for the tropical savannah and grasslandbiome (0972) but this couldbe a reflectionof the small numberofstudies available at the time of the review (five in Africa one inIndia and one in Cuba) Recent investigations in the cerrado ofcentral Brazil found values of 097 (cerrado stricto sensu) and099 (campo sujo lsquoshrublandsrsquo) (Rodin 2004) and ranging from088 to 092 for cerrado stricto sensu under different burningregimes (Castro-Neves 2007) This variation in values indicatesthat b is very variable and is highly dependent on the time ofsampling (dry or wet season) soil type (clay or sand) drainage ofthe environment (hydromorphy) and phytophysionomy (grassyor different forms of wooded savannah)

Root shoot ratio

Use of the root shoot ratio as an indicator of the relationshipbetween the belowground and aerial biomass (total live) isvery important because it can serve as an estimator ofbelowground carbon based on a simple biometric survey ofaboveground biomass with lower costs (Schenk and Jackson2002b) Realistic root shoot ratios are necessary to improve theaccuracy of estimates of root biomass and to estimate the effectsof management and land-use change in national inventories ofgreenhouse gas emissions (Mokany et al 2006) In our study wecalculate the root shoot ratio based on biomass and carbonThis latter form provides ecological values closer to reality forcalculation of stock production and ecosystemproductivity Thisis because the carbon content (C) of the different aboveground

components is not the same as that applied to belowgroundbiomass In ecosystems where the biomass of fine roots isoverwhelmingly superior to the other categories as in the caseon the open savannahs of Roraima the carbon content can belower causing the root shoot ratio based on biomass to notrepresent the ecosystem faithfully

The values of the root shoot ratio based on total live biomassvaried between 27 and 38 reflecting discrepancies between thetotal values above and below ground for all phytopedounitsHigher ratios (3ndash5) were determined in the savannahs in Lamto(Ivory Coast) indicating greater total belowground biomass toa depth of 1m as compared with aerial biomass (Menaut andCesar 1979) However these values are extremely variable anddependent on the depth of sampling In the cerrado of centralBrazil Castro and Kauffman (1998) found high values forsavannahs with low tree density (56ndash77) and smaller valuesfor more wooded phytophysionomies (26ndash29) to 2-m deptheven without including any estimate for the biomass of the rootcrowns Thus although the phytopedounits investigated inRoraima are limited by the low density of tree individuals ourvalues are closer to those of the wooded cerrado environmentsof Castro and Kauffman (1998) than to grassland environmentsOur results indicate that the total biomass of roots (0ndash100-cmdepth) is a component of great importance in the open savannahenvironments of Roraima representing 24ndash33 times the totalcarbon allocated to aboveground biomass

The IPCC (2006 p 472) suggests that fine roots (lt2mm) arean integral part of the soil and therefore should be consideredin the calculations of soil carbon To have a valid correction forthis it is necessary to disaggregate the results and use onlythe categories of roots 2mm in diameter This is required toprevent double counting of inventory values derived for soilcarbon stocks Thus using our results for roots with diameter2mm for open savannah phytopedounits studied in Roraima

Table 5 Root shoot ratio in the formused by the IPCC (ratio the biomass of roots$2mm in diameter to live aboveground biomass) from the currentstudy and recalculated from published studies on other Brazilian savannahs

Sa = open woodland Sd = dense woodland Sg = grasslands Sp = savannah parkland

Phytophysionomy IBGElegendA

Depth (m) Root Mg handash1

(2mm)ShootMg handash1

RS(2mm)

Reference

Cerrado Sensu Stricto Sa 62 2140 3458 062 BCampo limpo (grassland) Sg 20 1157 290 399 CCampo sujo (shrublands) Sg 20 2137 390 548Cerrado Sensu Stricto (open cerrado) Sa 20 3302 1760 188Cerrado Sensu Stricto (dense cerrado) Sd 20 3756 1840 204Cerrado Sensu Stricto (biennial precocious) Sa 05ndash10 3815 2100 182 DCerrado Sensu Stricto (biennial modal) Sa 05ndash10 4361 2900 150Cerrado Sensu Stricto (biennial late) Sa 05ndash10 3918 2290 171Cerrado Sensu Stricto (quadrennial) Sa 05ndash10 3986 2630 152Dry grassland (Argisol) Sg 10 114 565 020 EDry grassland (Latosol) Sg 10 047 650 007Mosaic grasslandsshrublands (Latosol) SgSp 10 165 810 020Campo uacutemido (Hydromorphic) Sg 10 000 765 000

ABrazilian vegetation classification code determined by IBGE (1992)BAbdala et al (1998) includes live and dead rootsCCastro andKauffman (1998) doesnot include tap rootsTheseauthors consideredfine roots tobelt6mmdiameterThebiomassof rootslt2mmwasestimatedas 29 of the total of roots in a 2-m profile according to Abdala et al (1998)

DCastro-Neves (2007) uses Abdala et al (1998) for calculation of coarse roots (0ndash100 cm) and a direct method for estimating fine roots up to 50-cm depthEThis study


the root shoot ratio based on biomass is between 0 (seasonallyflooded grasslands) and 007ndash020 (grasslands with low treedensity) or between 0 and 008ndash024 based on carbon (seeTable 4) These values are lower than those indicated as thedefault values by the IPCC (2003 p 3109 table 343 IPCC2006 p 68 table 61) for sub-tropicaltropical grassland (16)woodlandsavannah (05) and shrubland (28) However theIPCC strongly suggests that default values only be used whenthe country does not have regional values that better reflect theecosystem (IPCC 2006 p 68)

The second Brazilian inventory used the default values for allgrassland and savannah environments as listed in Brazil MCT(2010 pp 236ndash237) This was done both for the cerrados ofcentral Brazil (for which published estimates of belowgroundbiomass existed) and for Amazon savannahs (for which thepresent study provides the first estimates) Although few innumber it is possible to make inferences about the root shootratio for Brazilian savannahs including cerrados (Table 5) Forexample the Brazilian estimates for root shoot ratio (roots2mm) vary tremendously depending on the vegetation typefire regime seasonality of the watertable soil class and samplingdepth Environments in central Brazil with greater abovegroundbiomass and that are not influenced by the watertable haveroot shoot ratios from 7 to 27 times higher than those inAmazonian grasslands and savannahswith low arboreal biomass

We therefore propose a reformulation of the calculationsfor the next Brazilian national inventory We suggest aminimum standardisation of 1-m depth for the estimates ofbelowground biomass and carbon in addition to region-specific values for root shoot ratios with different ratios forAmazonian grasslandssavannahs and for central Braziliancerrados This calculation strategy would bring advantagesin avoiding the use of empirical default values from the IPCC(2003 2006) thereby providing more realistic values for totalbiomass and carbon for ecosystems with open vegetation inBrazil

Conclusions

The total biomass of roots of seasonally flooded grasslands ishigher than the root biomass of grasslands with low tree-bushdensity although the total belowground carbon stock does notdiffer among phytopedounits

The vertical distribution pattern of root biomass follows anexponential model with the largest concentration of roots beingin the more superficial layers of the soil This pattern does notdiffer among phytopedounits

The total biomass of roots (direct + indirect methods) in opensavannah environments of Roraima represents a pool 24ndash33times the total carbon stocked in aboveground biomass

The expansion factor (root shoot ratio) used by IPCC forbelowground biomass in roots2mm diameter starting fromlive aboveground biomass is zero for seasonally floodedgrasslands of Roraima (in northern Amazonia) For unfloodedgrasslands with low densities of trees the values of this factorrange from007 to020onabiomass basis or from008 to024ona carbon basis

The standardisation of the minimum sampling depth andthe use of region-specific values for root shoot ratios to

calculate belowground biomass in grasslands and savannahs isadvantageous because it provides more realistic values of totalbiomass and carbon

Supplementary material

Supplementary material with details regarding the carbonconcentration (C) of the main tree and shrub species anddensity (number ha1) and basal area (cm2 ha1) of the tree-bush component present in the phytopedounits is available fromthe Journalrsquos website

AcknowledgementsWe thank the Federal University of Roraima and the Brazilian Enterprisefor Agriculture and Ranching Research for permission to use the ResearchProgram on Biodiversity grids installed in their experimental stations Theproject was financed by the Institutional Research Program of the NationalInstitute for Research in the Amazon (PPIINPA-PRJ 01218) linked tothe CNPq research group lsquoEcology and Management of NaturalResources of Savannahs of Roraimarsquo and INCT-Servamb (MCT) CAPESgranted postgraduate scholarships (Masters degree) to JRS Santos andMS Cunha CNPq granted productivity fellowships to RI Barbosa andPM Fearnside

References

Abdala GC Caldas LS Haridasan M Eiten G (1998) Above andbelowground organic matter and root shoot ratio in a cerrado inCentral Brazil Brazilian Journal of Ecology 2 11ndash23

Arauacutejo ACO Barbosa RI (2007) Riqueza e diversidade do estrato arboacutereo-arbustivo de duas aacutereas de savanas em Roraima Amazocircnia BrasileiraMens Agitat 2(1) 11ndash18

Asner GP Elmore AJ Olander LP Martin RE Harris AT (2004) Grazingsystems ecosystem responses and global change Annual Review ofEnvironment and Resources 29 261ndash299doi101146annurevenergy29062403102142

Barbosa RI (1997) Distribuicatildeo das chuvas em Roraima In lsquoHomemambiente e ecologia no estado de Roraimarsquo (Eds RI Barbosa EJGFerreira EGCastellon) pp 325ndash335 (INPAManaus Amazonas Brazil)

Barbosa RI (2001) Savanas da Amazocircnia emissatildeo de gases do efeito estufa ematerial particulado pela queima e decomposicatildeo da biomassa acimado solo sem a troca do uso da terra em Roraima Brasil PhD ThesisEcology Instituto Nacional de Pesquisas da Amazocircnia Universidade doAmazonas Manaus Amazonas Brazil

Barbosa RI Campos C (2011) Detection and geographical distribution ofclearing areas in the savannas (lsquolavradorsquo) of Roraima using Google Earthweb tool Journal of Geography and Regional Planning 4(3) 122ndash136

Barbosa RI Fearnside PM (2005a) Fire frequency and area burned in theRoraima savannas of Brazilian Amazonia Forest Ecology andManagement 204 371ndash384 doi101016jforeco200409011

Barbosa RI Fearnside PM (2005b) Above-ground biomass and the fate ofcarbon after burning in the savannas of Roraima Brazilian AmazoniaForest Ecology and Management 216 295ndash316doi101016jforeco200505042

Barbosa RI Nascimento SP Amorim PAF Silva RF (2005) Notas sobre acomposicatildeo arboacutereo-arbustiva deumafisionomiadas savanasdeRoraimaAmazocircnia Brasileira Acta Botanica Brasilica 19 323ndash329doi101590S0102-33062005000200015

Barbosa RI Campos C Pinto F Fearnside PM (2007) The lsquoLavradosrsquo ofRoraima Biodiversity and Conservation of Brazilrsquos AmazonianSavannas Functional Ecosystems and Communities 1(1) 29ndash41

Baruch Z (1994) Responses to drought and flooding in tropical foragegrasses I Biomass allocation leaf growth and mineral nutrients Plantand Soil 164 87ndash96 doi101007BF00010114


Beard JS (1953) The savanna vegetation of northern tropical AmericaEcological Monographs 23 149ndash215 doi1023071948518

Benedetti UG Vale JF Jr Schaefer CEGR Melo VF Uchocirca SCP (2011)Gecircnese quiacutemica e mineralogia de solos derivados de sedimentosPliopleistocecircnicos e de rochas vulcacircnicas baacutesicas em Roraima norteamazocircnico Revista Brasileira de Ciencia do Solo 35 299ndash312doi101590S0100-06832011000200002

Brazil MCT (2004) lsquoBrazilrsquos initial national communication to the UnitedNations Framework Convention on Climate Changersquo (Ministeacuterio deCiecircncia e Tecnologia Brazil)

Brazil MCT (2010) lsquoSegunda comunicacatildeo nacional do Brasil agrave Convencatildeo-Quadro das Nacotildees Unidas sobre mudanca do clima Vol 1rsquo (Ministeacuteriode Ciecircncia e Tecnologia Brazil) Available at httpwwwmctgovbrupd_blob0213213909pdf [Verified 11 April 2010]

Brazil Projeto RADAMBRASIL (1975) lsquoLevantamento dos recursosnaturais Vol 8rsquo (Ministeacuterio das Minas e Energia Rio de Janeiro)

Brazil SNLCS (1996) lsquoLevantamento semidetalhado dos solos e aptidatildeoagriacutecola das terras do campo experimental aacutegua boa do CPAF-RR estadode Roraimarsquo (Empresa Brasileira de Pesquisa Agropecuaacuteria ServicoNacional de Levantamento e Conservacatildeo do Solo Rio de Janeiro)

Castro EA Kauffman JB (1998) Ecosystem structure in the BrazilianCerrado a vegetation gradient of aboveground biomass root mass andconsumption by fire Journal of Tropical Ecology 14 263ndash283doi101017S0266467498000212

Castro-Neves BM (2007) Efeito de queimadas em aacutereas de cerrado strictusensu e na biomassa de raiacutezesfinas PhDThesis Universidade deBrasiacuteliaBrazil

Cattanio JH Anderson AB Rombold JS Nepstad DC (2004) Phenologylitterfall growth and root biomass in a tidal floodplain forest in theAmazon estuary Revista Brasileira de Botacircnica 27 703ndash712

Chen X Eamus D Hutley LB (2004) Seasonal patterns of fine-rootproductivity and turnover in a tropical savanna of northern AustraliaJournal of Tropical Ecology 20 221ndash224doi101017S0266467403001135

Delitti WBC Pausas JG Burger DM (2001) Belowground biomass seasonalvariation in twoNeotropical savannahs (BrazilianCerrados)withdifferentfire histories Annals of Forest Science 58 713ndash721doi101051forest2001158

Eden M (1970) Savanna vegetation in the northern Rupununi Guyana TheJournal of Tropical Geography 30 17ndash28

Eiten G (1978) Delimitation of the Cerrado concept Plant Ecology 36169ndash178 doi101007BF02342599

Espeleta JF Clark DA (2007) Multi-scale variation in fine-root biomass in atropical rain forest a seven-year study Ecological Monographs 77377ndash404 doi10189006-12571

February EC Higgins SI (2010) The distribution of tree and grass roots insavannas in relation to soil nitrogen and water South African Journal ofBotany 76 517ndash523 doi101016jsajb201004001

Fiala K Herrera R (1988) Living and dead belowground biomass and itsdistribution in some savanna communities in Cuba Folia Geobotanica23 225ndash237

Furley P (1999) The nature and diversity of neotropical savanna vegetationwith particular reference to the Brazilian cerrados Global Ecology andBiogeography 8 223ndash241

Gale MR Grigal DF (1987) Vertical distributions of northern tree species inrelation to successional status Canadian Journal of Forest Research17 829ndash834 doi101139x87-131

Gill RA Jackson RB (2000) Global patterns of root turnover for terrestrialecosystems New Phytologist 147 13ndash31doi101046j1469-8137200000681x

IBGE (1992) lsquoManual teacutecnico da vegetacatildeo brasileirarsquo (Instituto Brasileirode Geografia e Estatiacutestica Diretoria de Geociecircncias Departamento deRecursos Naturais e Estudos Ambientais Rio de Janeiro)

IPCC (2003) lsquo2003 IPCC good practice guidance for land use land-usechange and forestryrsquo (Eds J Penman M Gytarsky T Hiraishi T KrugD Kruger R Pipatti L Buendia KMiwa T Ngara K Tanabe F Wagner)(Institute for Global Environmental Strategies for the IntergovernmentalPanel on Climate Change Kanagawa)

IPCC (2006) lsquo2006 IPCC guidelines for national greenhouse gas inventoriesPrepared by the National Greenhouse Gas Inventories Programmersquo(Eds HS Eggleston L Buendia K Miwa T Ngara K Tanabe)(Intergovernmental Panel on Climate Change and Institute for GlobalEnvironmental Strategies Kanagawa)

IPCC (2007) Climate change 2007 synthesis report IntergovernmentalPanel on Climate Change Plenary XXVII Valencia Spain 12ndash17November 2007rsquo Working Group contributions to the FourthAssessment Report IPCC Geneva

Jackson RB Canadell J Ehleringer JR Mooney HA Sala OE Schulze ED(1996) A global analysis of root distributions for terrestrial biomesOecologia 108 389ndash411 doi101007BF00333714

Jackson RB Mooney HA Schulze ED (1997) A global budget for fine rootbiomass surface area and nutrient contents Ecology 94 7362ndash7366

JordanCF EscalanteG (1980)Root productivity in anAmazonian rain forestEcology 61 14ndash18 doi1023071937148

Kellman M Sanmugadas K (1985) Nutrient retention by savannaecosystems I retention in the absence of fire Journal of Ecology 73935ndash951 doi1023072260159

Klinge H (1973) Root mass estimation in lowland tropical rain forests ofcentral Amazonia Brazil Tropical Ecology 14 29ndash38

Knoop WT Walker BH (1985) Interactions of woody and herbaceousvegetation in a southern African savanna Journal of Ecology 73235ndash253 doi1023072259780

LuizatildeoFLuizatildeoRChauvelA (1992)Premiers reacutesultats sur la dynamiquedesbiomasses racinaires etmicrobiennes dans un latosol drsquoAmazonie centrale(Breacutesil) sous forecirct et sous pacircturage Cahiers Orstom serie Peacutedologie(Gent) 27 69ndash79

Magnusson WE Lima AP Luizatildeo R Luizatildeo F Costa FRC Castilho CVKinupp VF (2005) RAPELD A modification of the Gentry method forbiodiversity surveys in long-term ecological research sites BiotaNeotropica 5(2) httpwwwbiotaneotropicaorgbrv5n2ptabstractpoint-of-view+bn01005022005

ManlayRJChotte JLMasseD Laurent JY FellerC (2002)Carbon nitrogenand phosphorus allocation in agro-ecosystems of aWest African savannaIII Plant and soil components under continuous cultivation AgricultureEcosystems amp Environment 88 249ndash269doi101016S0167-8809(01)00220-1

McClaugherty CA Aber JD Melillo JM (1982) The role of fine roots in theorganic matter and nitrogen budgets of two forested ecosystems Ecology63 1481ndash1490 doi1023071938874

McKell CM Wilson AM Jones MB (1961) A flotation method for easyseparation of roots from soil samples Agronomy Journal 53 56ndash57doi102134agronj196100021962005300010022x

McNaughton SJ Banyikwa FF McNaughton MM (1998) Root biomass andproductivity in a grazing ecosystem the Serengeti Ecology 79 587ndash592doi1018900012-9658(1998)079[0587RBAPIA]20CO2

MedinaE Silva JF (1990) Savannas of northernSouthAmerica a steady stateregulated by waterndashfire interactions on a background of low nutrientavailabilityJournalofBiogeography17 403ndash413 doi1023072845370

Menaut JC Cesar J (1979) Structure and primary productivity of LamtoSavannas Ivory Coast Ecology 60 1197ndash1210 doi1023071936967

Milchunas DG Lauenroth WK (1993) Quantitative effects of grazing onvegetation and soils over a global range of environments EcologicalMonographs 63 327ndash366 doi1023072937150

Miranda IS Absy ML Rebecirclo GH (2002) Community structure of woodyplants of Roraima savannahs Brazil Plant Ecology 164 109ndash123doi101023A1021298328048


Mokany K Raison RJ Prokushkin AS (2006) Critical analysis of root shootratios in terrestrial biomes Global Change Biology 12 84ndash96doi101111j1365-24862005001043x

Nadelhoffer KJ Raich JW (1992) Fine root production estimates andbelowground carbon allocation in forest ecosystems Ecology 731139ndash1147 doi1023071940664

NeillC (1992)Comparisonof soil coringand ingrowthmethods formeasuringbelowground production Ecology 73 1918ndash1921 doi1023071940044

Nelson DW Sommers LE (1996) Total carbon organic carbon and organicmatter In lsquoMethods of soil analysis part 3 ndash chemical methodsrsquo SSSABook Series No 5 pp 9611010 (SSSA amp ASA Madison WI)

Nepstad D Carvalho CR Davidson EA Jipp PH Lefebvre PA NegreirosGH Silva ED Stone TA Trumbore SA Vieira S (1994) The role of deeproots in the hydrological and carbon cycles of Amazonian forests andpastures Nature 372 666ndash669 doi101038372666a0

NewmanEI (1966)Amethodof estimating the total length of root in a sampleJournal of Applied Ecology 3 139ndash145 doi1023072401670

OliveiraRSBezerra LDavidsonED Pinto FKlinkCANepstadDMoreiraA (2005)Deep root function in soil water dynamics in cerrado savannas ofcentral Brazil Functional Ecology 19 574ndash581doi101111j1365-2435200501003x

Paiva OA Faria GE (2007) Estoque de carbono do solo sob cerrado sensustricto no Distrito Federal Brasil Revista Troacutepica ndash Ciecircncias Agraacuterias eBioloacutegicas 1(1) 59ndash65

Pandey CB Singh JS (1992) Rainfall and grazing effects on net primaryproductivity in a tropical savanna India Ecology 73 2007ndash2021doi1023071941451

Priess J ThenCFoumllsterH (1999)Litter andfine-root production in three typesof tropical premontane rain forest in SE Venezuela Plant Ecology 143171ndash187 doi101023A1009844226078

Rodin P (2004) Distribuicatildeo da biomassa subterracircnea e dinacircmica de raiacutezesfinas em ecossistemas nativos e em uma pastagemplantada noCerrado doBrasilCentralMScThesisUniversidadedeBrasiacutelia InstitutodeCiecircnciasBioloacutegicas Brazil

San Joseacute JJ Farintildeas MR (1983) Change in tree density and speciescomposition in a protected Trachypogon savanna Venezuela Ecology64 447ndash453 doi1023071939963

San Joseacute JJBerradeFRamirez J (1982)Seasonal changesofgrowthmortalityand disappearance of belowground root biomass in the Trachypogonsavanna grass Acta Oecologica ndash Oecologia Plantarum 3 347ndash352

San Joseacute JJ Montes RA Farintildeas MR (1998) Carbon stocks and fluxes in atemporal scaling from a savanna to a semi-deciduous forest ForestEcology and Management 105 251ndash262doi101016S0378-1127(97)00288-0

Santos CPF Valles GF Sestini MF Hoffman P Dousseau SL Homemde Mello AJ (2007) Mapeamento dos Remanescentes e OcupacatildeoAntroacutepica no Bioma Amazocircnia In lsquoAnais do XIII Simpoacutesio Brasileirode Sensoriamento Remoto (Florianoacutepolis Brasil 21ndash26 abril 2007)rsquopp 6941ndash6948 (Instituto Nacional de Pesquisas Espaciais Satildeo Joseacute dosCampos Brazil) Available at httpmartedpiinpebrrepdpiinpebrsbsr80200611180125mirror=dpiinpebrbanon20031210193054ampmetadatarepository=dpiinpebrsbsr8020061118012531[Verified 8 April 2009]

Sarmiento G (1984) lsquoThe ecology of neotropical savannasrsquo (HarvardUniversity Press Cambridge MA)

Sarmiento G VeraM (1979) Composicioacuten estructura biomasa y produccioacutenprimaria de diferentes sabanas en los Llanos occidentales de VenezuelaBoletin de la Sociedad Venezolana de Ciencia Natural 34(136) 5ndash41

SchenkHJ JacksonRB(2002a)Theglobalbiogeographyof rootsEcologicalMonographs 72 311ndash328doi1018900012-9615(2002)072[0311TGBOR]20CO2

Schenk HJ Jackson RB (2002b) Rooting depths lateral root spreads andbelow-groundabove-ground allometries of plants in water-limitedecosystems Journal of Ecology 90 480ndash494doi101046j1365-2745200200682x

Scholes RJ Hall DO (1996) The carbon budget of tropical savannaswoodlands and grasslands In lsquoGlobal change effects on coniferousforests and grasslandsrsquo SCOPE Scientific Committee on Problemsof the Environment (v 56) (Eds AI Breymeyer DO Hall JM MelilloGI Agren) pp 69ndash100 (Wiley amp Sons Chichester UK)

Silver WL Neff J McGroddy M Veldkamp E Keller M Cosme R (2000)Effects of soil texture on belowground carbon and nutrient storage in alowland Amazonian forest ecosystem Ecosystems 3 193ndash209doi101007s100210000019

Snowdon P Eamus D Gibbons P Khanna P Keith H Raison JKirschbaum M (2000) Synthesis of allometrics review of root biomassanddesignof futurewoodybiomass sampling strategiesTechnicalReportno 17 National Carbon Accounting System Canberra)

Solbrig OT Medina E Silva JF (1996) Biodiversity and tropical savannaproperties a global view In lsquoFunctional roles of biodiversity A globalperspectiversquo (Eds HA Mooney JH Cushman E Medina OE SalaED Schulze) pp 185ndash211 SCOPE Scientific Committee onProblems of the Environment (Wiley amp Sons Chichester)

Sombroek WG Fearnside PM Cravo M (2000) Geographic assessmentof carbon stored in Amazonian terrestrial ecosystems and their soils inparticular In lsquoGlobal climate change and tropical ecosystems Advancesin soil sciencersquo pp 375ndash389 (CRC Press Boca Raton FL)

StantonNL (1988)The underground in grasslandsAnnual Reviewof Ecologyand Systematics 19 573ndash589doi101146annureves19110188003041

Thompson J Proctor J Viana V Milliken W Ratter JA Scott DA (1992)Ecological studies on a lowland evergreen rain forest on Maraca IslandRoraima Brazil I Physical environment forest structure and leafchemistry Journal of Ecology 80 689ndash703 doi1023072260860

Vitousek PM Sanford RL (1986) Nutrient cycling in moist tropical forestAnnual Review of Ecology and Systematics 17 137ndash167doi101146annureves17110186001033

Yavitt JB Wright SJ (2001) Drought and irrigation effects on fine rootdynamics in a Tropical Moist Forest Panama Biotropica 33 421ndash434

Zar JH (1999) lsquoBiostatistical analysisrsquo 4th edn (Prentice-Hall EnglewoodCliffs NJ)


wwwpublishcsiroaujournalsajb

more densely populated by trees (Abdala et al 1998 Schenk andJackson2002a)However theseopen-vegetationenvironmentsarecharacterised by having large grassy expanses and a low-densitytree component Most roots in these ecosystems are located in themost superficial layers of the soil and are characterised by havingsmall diameters (lt2mm) (Knoop andWalker 1985) It is estimatedthat the different forms of savannahs (open and wooded) contain~20 of all of the fine-root biomass on Earth to a depth of 30 cm(Jackson et al 1997) This category of roots has a high rate ofreplenishment in tropical grasslands and savannahs making it acritical component in sequesteringatmospheric carbonbyallowingconstant accumulation of organic matter in the soil (Stanton1988 Gill and Jackson 2000 Chen et al 2004) However theIntergovernmentalPanelonClimateChange(IPCC2006) suggeststhat this category of roots (lt2mm in diameter) should not beincluded as part of the lsquobelowground biomassrsquo because it isdifficult to distinguish it empirically from soil organic matter

Savannahs grasslands and other natural non-forestecosystems occupy ~200 000 km2 (~5) of the BrazilianAmazonia (Santos et al 2007) Although the area of theseecosystems is substantial all existing studies on the biomassof roots in natural vegetation in the Amazonian biome in Brazilare from forest ecosystems due to the much larger area offorests (Klinge 1973 Luizatildeo et al 1992 Thompson et al1992 Nepstad et al 1994 Cattanio et al 2004) Estimatesof the temporal dynamics and spatial distribution of rootsin different phytophysionomies of Amazon savannahs arenonexistent representing an important lacuna in ourknowledge about this potential belowground carbon reservoirin the Amazon These data are important for understandingthe role of these environments in mitigating global warming(IPCC 2007)

The first Brazilian inventory of greenhouse gases whichwas submitted to the Climate Convention (UN FrameworkConvention on Climate Change) in 2004 did not consider theroots explaining that lsquothe consideration of carbon below ground(roots) is complex andwas not included in this inventoryrsquo (BrazilMCT 2004 p 146) Of course considering the carbon stock inroots to be zero and consequently considering the emission fromthis source to be zero after clearing the vegetation represents asubstantial underestimation especially in savannahs where thegreat majority of the biomass is below ground The secondBrazilian inventory used the default root shoot ratios valuespresented by IPCC (2003 p 3109 table 343) to estimatetotal biomass for all grassland and savannah environmentslisted in Brazil MCT (2010 pp 236ndash237) This was done bothfor the cerrados of central Brazil and for Amazon savannahsThe root shoot ratio is an expansion factor used to estimatebelowground biomass from aboveground biomass (IPCC 2006p 67) However use of the IPCC default values for calculatingbelowgroundbiomass in open-vegetation systemsunder differentenvironmental conditions can cause undesired distortion in thefinal values for total biomass

Within this context our goal was to estimate the biomass ofroots indifferent open savannah environments inRoraima inviewof the combination of two effects (phytophysionomic structureand soil class forming a lsquophytopedounitrsquo) The phytopedounitspresented here are similar to the lsquolandscape system unitsrsquo definedby Sombroek et al (2000) A classification including both

vegetation and soil effects is important in order to preventphytophysionomies with the same structure on different soilclasses from being analysed as the same ecological unit Thisstrategy includes spatial variations that occur along differentedaphic gradients that may affect the modelling of rootbiomass (Espeleta and Clark 2007)

The current study included the following questions (i) Doesthe allocation of total carbon to roots differ among opensavannahs on different phytopedounits (ii) Does the verticaldistribution of root biomass differ among these environmentsand (iii) Does the ratio between the biomass of roots (totaland 2mm) and the aerial biomass (root shoot) differ amongthe phytopedounits investigated Our results represent anopportunity to reformulate the estimates of belowgroundbiomass and carbon stocks in Amazonian savannahsproviding appropriate regional values for open-vegetationsystems with low densities of trees and shrubs

Materials and methodsSavannahs of RoraimaSavannahs of Roraima are part of the Rio Branco-Rupununicomplex located in the triple frontier between BrazilVenezuela and Guyana (Beard 1953 Eden 1970) Altogetherthese continuous savannahs cover 68 103 km2 the Brazilianpart being ~43 103 km2 (~63) (Barbosa et al 2007 Barbosaand Campos 2011) In general these savannahs are located onpoor soils with high frequency of fire and strong climaticseasonality that directly influences the fluctuation of thewatertable and the phytophysionomic structure (Miranda et al2003 Barbosa and Fearnside 2005a) The climate type of thiswhole region is Awi according to the Koumlppen classification withaverage rainfall of ~1650mmyearndash1 the peak of the dry periodis between December and March and the rainy period betweenMay and August (Barbosa 1997) These savannahs have a widevariety of phytophysionomies ranging from grasslands that aretotally devoid of trees to densely populated types on differentsoil classes (Brazil Projeto RADAMBRASIL 1975 Barbosaand Fearnside 2005b) The Venezuelan llanos have structureand species composition that are similar to those of the savannahsof Roraima (San Joseacute and Farintildeas 1983 Medina and Silva 1990)and neither of these should be confused with the savannahs(cerrados) of central Brazil (Eiten 1978)

Study areasThe studywas carried out in two savannah areas that have samplegrids for a Research Program on Biodiversity (i) Aacutegua BoaExperimental Station (AB) and (ii) Cauameacute or lsquoMonte CristorsquoCampus (MC) (Fig 1) The grids are composed of walking trailsin the NorthndashSouth (N-S) and EastndashWest (E-W) directions thatcross the area at intervals of 500m All sampling was performedbased on permanent plots (10m 250m) that are systematicallydistributed at points equidistant from the intersections of E-Wand N-S trails Each plot is an independent sampling unit thatfollows the contour line established beginning from the initialpicket This configuration was adopted to minimise the effects oftopographic variability in each plot (Magnusson et al 2005) Allplots are individually classified by soil class and vegetation































Results




















00

05

10

15

20

25

30

35

40

45


Car

bon

stoc

ks (

MgC

handash

1 )

Phytopedounits












Discussion










00

15

30

45

60

75

90


Roo

t bio

mas

s (M

g ha

ndash1)

Depth (cm)


b

bb

a

c

a aa a

b

aa

aaa

a

aa

a

a

aa

a





















Root shoot ratio










RS(2mm)

Reference








Conclusions










References

















































































































Results




















00

05

10

15

20

25

30

35

40

45


Car

bon

stoc

ks (

MgC

handash

1 )

Phytopedounits












Discussion










00

15

30

45

60

75

90


Roo

t bio

mas

s (M

g ha

ndash1)

Depth (cm)


b

bb

a

c

a aa a

b

aa

aaa

a

aa

a

a

aa

a





















Root shoot ratio










RS(2mm)

Reference








Conclusions










References





























































































00

05

10

15

20

25

30

35

40

45


Car

bon

stoc

ks (

MgC

handash

1 )

Phytopedounits












Discussion










00

15

30

45

60

75

90


Roo

t bio

mas

s (M

g ha

ndash1)

Depth (cm)


b

bb

a

c

a aa a

b

aa

aaa

a

aa

a

a

aa

a





















Root shoot ratio










RS(2mm)

Reference








Conclusions










References




















































































Discussion










00

15

30

45

60

75

90


Roo

t bio

mas

s (M

g ha

ndash1)

Depth (cm)


b

bb

a

c

a aa a

b

aa

aaa

a

aa

a

a

aa

a





















Root shoot ratio










RS(2mm)

Reference








Conclusions










References


































































































Root shoot ratio










RS(2mm)

Reference








Conclusions










References























































































Conclusions










References




















































































root biomass, root - inpaagroeco.inpa.gov.br/reinaldo/ribarbosa_prodcient_usu_visitantes... · root...

Documents