Contents lists available at ScienceDirect

Remote Sensing of Environment

journal homepage: www.elsevier.com/locate/rse

Convergence of daily light use efficiency in irrigated and rainfed C3 and C4crops

Anatoly A. Gitelsona,b,⁎, Timothy J. Arkebauerc, Andrew E. Suykerb

aMapping and Geoinformation Engineering, Faculty of Civil & Environmental Engineering, Israel Institute of Technology (TECHNION), Haifa, Israelb School of Natural Resources, University of Nebraska-Lincoln, Lincoln, NE 68583-0973, USAc Department of Agronomy and Horticulture, University of Nebraska-Lincoln, Lincoln, NE 68583-0817, USA

A R T I C L E I N F O

Keywords:Light use efficiencyPrimary productionIncident radiation

A B S T R A C T

The goal of this study is to quantify variability of daily light use efficiency (LUE) based on radiation absorbed byphotosynthetically active vegetation (LUEgreen) in C3 and C4 crops. Data contained GPP, fAPAR, total and greenleaf area index taken over 16 site years of irrigated and rainfed maize and 8 site years of soybean in 2001–2008including years with quite strong drought events. LUEgreen in irrigated and rainfed sites were statistically in-distinguishable showing low sensitivity to water availability. Seasonal changes of LUEgreen remained remarkablysmall over a wide range of water supply, leaf area index and weather conditions. The magnitude and compo-sition of incident radiation affected the magnitude of the day-to-day LUEgreen change – increases in incident PARcaused statistically significant decreases of LUEgreen. Convergence of LUEgreen to a narrow range in irrigated andrainfed crops brought important implications for understanding mechanisms of plant response to stress andremote estimation of primary production in crops.

1. Introduction

The amount of photosynthesis or primary production of a plantcanopy is largely determined by the amount of photosynthetically ac-tive radiation (PAR) absorbed by the vegetation; i.e.,APAR= fAPAR×PARin where fAPAR is the fraction of absorbed PARand PARin is the photosynthetically active radiation incident on thecanopy. This is further modified by the efficiency with which this ab-sorbed light is converted to fixed carbon, light use efficiency (LUE).

There are two contrasting view points on the temporal behavior ofLUE; namely,

(a) LUE is a widely variable biophysical characteristic (e.g. Turneret al., 2003; Kergoat et al., 2008),

(b) there is a functional convergence of LUE in vegetation predicted bythe concept of an optimization of limited resource allocation (Field,1991; Field et al., 1995; Goetz and Prince, 1999; Ruimy et al., 1996,1999).

We argue that for accurate remote estimation of gross primaryproduction (GPP), it is necessary to quantify the effect of LUE on GPPand, thus, to understand how variability in LUE affects the accuracy of

remote GPP estimation.There are many definitions of LUE. Definitions differ in the form of

fixed carbon in the numerator (e.g., CO2, aboveground biomass, totalbiomass) and in the time frame over which LUE is determined (e.g.,seconds, hours, days, years). Definitions also vary in the form the de-nominator takes, specifically, the radiation range over which the ‘light’is measured. Moreover, the ‘light’ term may be based on (a) incidentradiation (e.g., Nichol et al., 2000; Suyker et al., 2005; Barton andNorth, 2001), (b) total absorbed radiation (Monteith, 1972; Normanand Arkebauer, 1991; Lindquist et al., 2005; Kergoat et al., 2008), or (c)radiation absorbed by photosynthetically active/green vegetation (Hallet al., 1992; Viña and Gitelson, 2005; Rossini et al., 2012). LUE basedon total APAR, which is a common expression of LUE in field studies,incorporates measurements of canopy light absorption that do notdistinguish the contribution of green from brown/yellow componentsto the overall absorption. LUE based on radiation absorbed by photo-synthetically active/green vegetation (LUEgreen) is not confounded bychanging pigmentation and canopy structure during plant growth andsenescence, its properties are clearly different from the other two de-finitions (Gitelson and Gamon, 2015; Gitelson et al., 2017); thus,LUEgreen provides the most mechanistically sound definition. The un-certainty estimates for GPP using different LUE definitions is a critical

https://doi.org/10.1016/j.rse.2018.08.007Received 28 March 2018; Received in revised form 17 July 2018; Accepted 7 August 2018

⁎ Corresponding author at: 812 Rabin Bld., Mapping and Geoinformation Engineering, Faculty of Civil & Environmental Engineering, Israel Institute of Technology(TECHNION), Haifa, Israel.

E-mail address: agitelson2@unl.edu (A.A. Gitelson).

Remote Sensing of Environment 217 (2018) 30–37

0034-4257/ © 2018 Elsevier Inc. All rights reserved.

T

component of the total error budget in the context of remotely sensedbased estimations of GPP.

We took advantage of the Carbon Sequestration Program at theUniversity of Nebraska-Lincoln (UNL), which presented a unique pos-sibility for comparison of multiyear LUE at closely located irrigated andrainfed maize and soybean sites. The data sets allowed us to answer apivotal question: is LUEgreen in irrigated and rainfed crops widelyvariable or conservative and how strongly does the change in LUEgreenmodulate GPP? Specifically, the goal is to quantify:

• long term seasonal constitutive and short term day-to-day faculta-tive changes in LUEgreen of irrigated and rainfed crops having dif-ferent photosynthetic pathways, physiology, phenology, leaf struc-ture and canopy architecture,

• LUEgreen response to drought.

2. Methods

Three AmeriFlux sites (Mead Irrigated/US - Ne1, Mead IrrigatedRotation/US - Ne2, and Mead Rainfed Rotation/US - Ne3), located atthe UNL Eastern Nebraska Research and Extension Center near Mead,Nebraska, USA, were studied during growing seasons from 2001 to2008 as part of the Carbon Sequestration Project at UNL (http://ameriflux.lbl.gov/). Each field is approximately 50 ha and all are within4 km of each other. Two sites, site 1 and site 2, were equipped withcenter pivot irrigation systems, while site 3 was a rainfed that did notreceive supplemental water. Site 1 was planted in continuous maize;site 2 and site 3 were both planted in a maize-soybean rotation withmaize in odd years (2001, 2003, 2005 and 2007) and soybean in evenyears (2002, 2004, 2006 and 2008). Further information regarding thestudy sites can be found in Suyker and Verma (2010).

Hourly incoming PAR (PARin) was measured by point quantumsensors (LI-190, LI-COR Inc., Lincoln, Nebraska) placed 5m above thesurface pointing toward the sky at each site. Daytime PARin values werecalculated by integrating the hourly measurements during a day fromsunrise to sunset (period when PARin exceeded 1 μmol m−2 s−1).Daytime PARin values are reported in MJm−2 d−1 (Suyker and Verma,2012).

Daytime potential PAR (PARpot) is the maximal value of daytimePARin that is possible at a site when the concentrations of atmosphericgases and aerosols are minimal. PARpot thus incorporates the seasonalchanges in hours of sunshine (i.e., day length) and it varies graduallythroughout the growing season (Gitelson et al., 2012). At each site,daytime PARpot was calculated as a maximal value of daytime PARin foreach day of year (DOY) for each of the eight years of observation. Thedifference, ΔPAR=PARpot− PARin, reveals the influence of dailyweather fluctuations. Low values of PARin (cloudy and/or hazy days)correspond to high ΔPAR, while high PARin values (sunny days) cor-respond to low differences. Thus, the ΔPAR differences reflect day-to-day weather variations that are not affected by the seasonal change ofday length.

There are two facets of LUE: the first reflects the rapidly changing,day-to-day and the second reflects the slowly changing, seasonal orontogenetic changes. These two have been termed the “facultative” and“constitutive” responses, respectively (Gamon and Berry, 2012). In thisstudy, following Gamon and Berry (2012), we define “facultative LUE”as those that change on a daily time scale, and “constitutive LUE” asthose that change over much longer seasonal time scale.

For the facultative component of LUE, irradiance is particularlycritical due to the asymptotic shape of the photosynthetic light responserelationship that results in a progressive lowering of LUE as a plant isexposed to higher irradiance (e.g., Gamon and Berry, 2012). An un-derstanding of the effects of incident irradiance on the GPP vs. APARrelationship and LUE is essential for remote estimation of GPP usingLUE models. We used a PARin constraint criterion to select days whensites were under “clear skies” conditions and PARin was greater than the

80% of PARpot required for optical remote sensing. It has been shownthat PARin above 80% of PARpot corresponds to clear sky conditions forTM, ETM and daily MODIS images collected over the US Corn Belt(Gitelson et al., 2012; Peng et al., 2013).

Within each of three study sites were six small plot areas(20m×20m) representing all major occurrences of soil and cropproduction zones at the site (Verma et al., 2005). For each of theseplots, leaf area index (LAI) was estimated from destructive samplestaken at 10–14 day intervals during the growing season from 2001through 2008. On each sampling date, plants from a one meter length ofeach of two rows within each plot were collected and the total numberof plants recorded. Plants were kept on ice and transported to the la-boratory where they were separated into green leaves, senesced (yellowor brown) leaves, stems, and reproductive components. Green and se-nesced leaves were run through an area meter (LI-3100, Li-Cor, Inc.,Lincoln, Nebraska, USA) and the total leaf area per plant was de-termined. For each plot, the total leaf area per plant was multiplied bythe plant population (determined by counting plants in each plot) toobtain a total LAI. Total LAI for the six plots were then averaged as asite-level value (additional details in Viña et al., 2011). Green leaveswere handled in the same way to obtain the green LAI (LAIgreen). SinceLAI values change gradually during the growing season, daily total LAIand LAIgreen values were interpolated based on measurements onsampling dates for each site in each year.

In three small plots (four at the rainfed Site 3), soil volumetric watercontents (VWC) in the root zone were monitored continuously at fourdepths (0.10, 0.25, 0.5, and 1.0m) using dielectric permittivity sensors(Theta probe ML2x, Delta-T Devices, Cambridge, UK).

Multiple quantum sensors were placed at each study site to collecthourly incoming PAR (PARin), PAR reflected by the canopy and soil(PARout), PAR transmitted through the canopy (PARtransm) and PARreflected by the soil (PARsoil). PARin was measured using point quantumsensors (LI-190, Li-Cor Inc., Lincoln, Nebraska) 5m above the surfacepointing toward the sky. PARout was measured with point quantumsensors aimed downward placed at 5m above the ground. PARtransm

was measured with 5 line quantum sensors (LI-191, Li-Cor Inc., Lincoln,Nebraska) placed diagonally across E-W planted rows, at about 2 cmabove the ground, pointing upward and PARsoil was measured with aline quantum sensor placed diagonally about 12 cm above the ground,pointing downward (further details in Hanan et al., 2002; Burba, 2005).All daily values of radiation were computed by integrating the hourlymeasurements during a day when hourly PARin exceeded1 μmol m−2 s−1. Daily values of the fraction of PAR absorbed by thewhole canopy (fAPARtotal) were then calculated as (Goward andHuemmerich, 1992; Viña and Gitelson, 2005):

= − − +fAPAR (PAR PAR PAR PAR )/PARtotal in out transm soil in

During the vegetative growth stages, when LAIgreen is equal to totalLAI, fAPARtotal represents fAPAR used for photosynthesis. However,during the reproductive and senescence stages fAPARtotal remainedinsensitive to decreases in crop greenness (Hatfield et al., 1984; Galloet al., 1985; Viña and Gitelson, 2005) since both photosynthetic andnon-photosynthetic components intercepted PARin, with progressivelyless used for photosynthesis – Fig. 1 (see also Hall et al., 1992; Viña andGitelson, 2005). To obtain a measure of the fAPAR absorbed solely bythe photosynthetic component of the vegetation, fAPARgreen was esti-mated as (Hall et al., 1992):

= ×fAPAR fAPAR (LAI /total LAI).green total green

Crop GPP was measured by the eddy covariance method. Each sitewas equipped with an eddy covariance tower and meteorological sen-sors, with which measurements of CO2 fluxes, water vapor, and energyfluxes were obtained continuously. Daytime net ecosystem exchange(NEE) values were computed by integrating hourly CO2 fluxes collectedduring a day when PARin exceeded 1 μmolm−2 s−1. Daytime estimatesof ecosystem respiration (Re) were obtained from the night CO2

A.A. Gitelson et al. Remote Sensing of Environment 217 (2018) 30–37

31

exchange-temperature relationship (e.g., Xu and Baldocchi, 2003;Verma et al., 2005). GPP was then obtained by subtracting Re from NEEas: GPP=NEE - Re (note: sign convention is such that fluxes toward thesurface are positive). GPP values are presented in units of g Cm−2 d−1

(details in Verma et al., 2005; Suyker et al., 2004).Daytime PAR absorbed by the whole canopy (APARtotal) was cal-

culated as the product of fAPARtotal and daytime incoming PAR. PARabsorbed only by the photosynthetic component of the vegetation wascalculated as: APARgreen= fAPARgreen× PARin. LUE of photo-synthetically active vegetation was calculated as LUEgreen=GPP /APARgreen which is a quantitative measure of the efficiency of conver-sion of APARgreen into fixed carbon (Gitelson and Gamon, 2015) at thecanopy scale.

3. Results

3.1. GPP vs. APARgreen relationships

Analysis of GPP vs. APARgreen relationships for the two crops wasthe first step in understanding and quantifying variability of LUE. Wecompared the relationships for 16 site years of maize data (12 irrigatedand 4 rainfed) and for 8 site years of soybean data (4 irrigated and 4rainfed) taken during eight years (2001 through 2008). Nine maizehybrids and three soybean cultivars were grown during the eight years.These data allowed the study of LUE in crops under the rather harshweather conditions of eastern Nebraska including years with very dif-ferent temperature regimes and water supplies. The GPP vs. APARgreen

relationships for very different irrigation regimes were linear and veryclose for both crops, although relationships for soybean have a widerscatter around the best-fit functions (Fig. 2). The relationship for irri-gated maize GPP= (2.24 ± 0.011)×APAR was almost identical tothat in rainfed maize GPP= (2.22 ± 0.018)×APAR. NRMSE of GPPestimation by these equations was below 13.3% for irrigated and 11.1%for rainfed maize. For irrigated soybean NRMSE of GPP estimation bythe relationship GPP= (1.46 ± 0.02)×APARgreen was 18.6% and forrainfed soybean NRMSE by the relationship GPP= (1.42 ±0.017)×APARgreen was 16%. A heteroscedastic t-test showed with95% probability that the relationships of GPP vs. APARgreen were notstatistically different for irrigated and rainfed sites for either crop.

3.2. Long term seasonal constitutive changes in LUEgreen

In both crops, LUEgreen increased in the beginning of the season andthen leveled off as APARgreen exceeded 5MJm−2 d−1 in maize and

4MJm−2 d−1 in soybean (Fig. 3); at that time green LAI in both cropswas> 2m2m−2. With the onset of senescence, LUEgreen decreased.This decrease was more pronounced and occurred more quickly insoybean than in maize. Importantly, the seasonal constitutive change ofLUEgreen was alike in both irrigated and rainfed sites. LUEgreen dis-tributions in irrigated and rainfed sites were almost identical for bothcrops (Fig. 4); a t-test showed with 95% probability that the LUE dis-tributions for irrigated and rainfed sites were not statistically different.

The standard deviation of LUEgreen in maize across the seasons(constitutive change) was below 14.7% in rainfed sites (during fouryears) and below18.4% in irrigated sites (during eight years) -Table 1A. In soybean, the standard deviation was higher than in maize -24.9% and 28.6% in rainfed and irrigated sites, respectively (Table 1B).Significantly, mean LUEgreen in irrigated and rainfed sites were veryclose; t-tests showed (95% probability) no significant difference be-tween mean LUEgreen in irrigated and rainfed sites for either crop.

The LUEgreen variability early in the season may be related to un-certainties of fAPAR measurements, as crop LAI is low and leaves areclumped into rows (e.g., Burba, 2005). In senescence stages, theLUEgreen decrease was more pronounced in soybean than in maize dueto sharp decreases in leaf chlorophyll content (Mock and Pearce, 1975;Setiyono et al., 2007). In both crops, the decrease of LUEgreen in se-nescence stages was likely due to overestimation of APARgreen as it wascalculated using a subjective measure of greenness, green LAI (Gitelsonet al., 2014). For the same destructively determined green LAI, leafchlorophyll content in reproductive and senescence stages may be sig-nificantly lower than that in vegetative stages (Ciganda et al., 2009;Peng et al., 2011). To make accurate quantification of constitutiveLUEgreen changes across sites and years, GPP and APARgreen data for theperiod June 1st to August 30 were used (e.g., Turner et al., 2003, Schullet al., 2015). These criteria eliminated days early in the growing seasonas green LAI < 2m2m−2 corresponding to APARgreen below5MJm−2 d−1 for maize and 4MJm−2 d−1 for soybean (Fig. 3) whenuncertainties of APARgreen measurement were greatest. The month ofSeptember was omitted from comparisons because in late senescencestages foliage was rapidly changing from green to yellow and brownand LUEgreen calculated using green LAI may therefore be biased. Inaddition, at that time green vegetation fraction drops and uncertaintiesin the measurement of fAPAR increased significantly.

Applying the temporal constraint criteria allowed a more accurateestimate of the variability of LUEgreen (Table 1, June 1–August 30;Fig. 5). A t-test showed that no significant differences between LUEgreenin rainfed and irrigated sites existed (risk level α=0.05). MeanLUEgreen values in irrigated and rainfed sites were very close: 2.23 vs.

Fig. 1. Fraction of total absorbed PAR (fAPAR total) measured by four sensors and fAPAR absorbed by photosynthetically active green vegetation (fAPAR green) in(A) irrigated maize in 2003 and (B) irrigated soybean in 2002. (For interpretation of the references to color in this figure legend, the reader is referred to the webversion of this article.)

A.A. Gitelson et al. Remote Sensing of Environment 217 (2018) 30–37

32

2.24 g CMJ−1 in maize and 1.47 vs. 1.43 g CMJ−1 in soybean (Table 1,Fig. 5). Standard deviation of LUEgreen was 12.5% and 10.8% in irri-gated and rainfed maize, respectively, and 12.8% and 15.2% in irri-gated and rainfed soybean, respectively.

3.3. Short term facultative changes in LUEgreen

Short term day-to day facultative LUEgreen oscillated around the

almost horizontal lines representing the constitutive LUEgreen vs. DOYrelationships with a STD=12.5% in 16 site-years of data in maize and11.5% in 8 site-years of data in soybean (Fig. 6 for rainfed maize andsoybean). Oscillations of LUEgreen at both irrigated and rainfed sitesfrequently coincided with ΔPAR=PARpot-PARin. Importantly, almostevery increase of PARin (i.e., decrease of ΔPAR) corresponded to a de-crease in LUEgreen, i.e., a decrease in photosynthetic efficiency. Therewas a consistent response of the magnitude of LUEgreen to changes in

Fig. 2. Gross Primary Production (GPP) vs. absorbed PAR (APAR) relationships in irrigated and rainfed crops based on 16 site years of maize data (A) and 8 site yearsof soybean data (B). Solid lines are best-fit functions for irrigated and rainfed sites; dashed lines represent two standard deviations of the relationships.

Fig. 3. LUEgreen plotted versus absorbed PAR (APAR) in vegetative (left panels) and reproductive (right panels) stages for irrigated and rainfed (A, B) maize and (C,D) soybean. Arrows show direction of increase (vegetative stage, left-hand panels) and decrease (reproductive and senescence stages, right-hand panels) of absorbedphotosynthetically active radiation during the early (A and C) and late (B and D) portions of the growing season.

A.A. Gitelson et al. Remote Sensing of Environment 217 (2018) 30–37

33

PARin resulting in linear LUEgreen vs. ΔPARin relationships withR2=0.44 for maize and 0.42 for soybean (p < 0.01) and quite smallstandard error of LUE estimation by PARin: 7.11% in maize and 8.15%in soybean. The relationships for irrigated and rainfed sites were not

statistically significant at the α=0.05 level. This strongly suggeststhat, in many cases, excessive PARin that cannot be efficiently utilizedby the plants was the reason for the decrease of photosynthetic activity.

An additional factor contributing to the close LUEgreen vs. ΔPARrelationship was a likely increase in the fraction of diffuse radiation athigh ΔPAR enhancing absorption of radiation (Norman and Arkebauer,1991; Gu et al., 2002; Turner et al., 2003). However, our analyses werelimited to PARin within 20% of PARpot with cloudiness coefficient(Turner et al., 2003) below 0.2. Thus, likely the effects of diffuse lightweren't as dramatic as shown in Turner et al. (2003) and Norman andArkebauer (1991).

3.4. Drought effects on LUEgreen

Availability of irrigated and rainfed sites of maize in odd years andsoybean sites in even years allowed for the study of the effect of droughton LUEgreen. Several drought events with significant decreases in soilmoisture occurred; the most severe droughts during the 2001–2008period occurred in 2002 and 2003 (Suyker and Verma, 2010). Wecompared the temporal behavior of LUEgreen at irrigated and rainfedsoybean sites in 2002 and maize sites in 2003. As indicators of theLUEgreen change we used the differences between actual LUEgreen andthe mean LUEgreen for each crop (16 site-years for maize and 8 site-years

Fig. 4. Frequency distributions of LUEgreen normalized to sample number in (A) maize irrigated (625 samples) and rainfed (189 samples) and (B) soybean irrigated(186 samples) and rainfed (184 samples). A t-test showed (95% probability) that the LUE distributions for irrigated and rainfed sites were not statistically different.

Table 1Minimal, maximal, mean, and standard deviation (STD) of LUEgreen of irrigatedand rainfed maize (A) and soybean (B) sites. CV, %= (STD/mean LUEgreen) iscovariance coefficient.

Whole season June 1–August 30

Rainfed Irrigated Rainfed Irrigated

(A)Min LUEgreen, g CMJ−1 0.53 0.49 1.20 0.72Max LUEgreen, g CMJ−1 4.02 9.54 2.75 3.23Mean LUEgreen, g CMJ−1 2.17 2.16 2.23 2.24STD, g CMJ−1 0.32 0.39 0.24 0.28CV, % 14.7 18.4 10.8 12.5

(B)Min LUEgreen, g CMJ−1 0.34 0.11 0.34 0.53Max LUEgreen, g CMJ−1 2.11 2.6 2.11 2.12Mean LUEgreen, g CMJ−1 1.35 1.43 1.43 1.47STD, g CMJ−1 0.34 0.37 0.23 0.19CV, % 24.9 28.6 15.2 12.8

Fig. 5. LUEgreen plotted versus PAR absorbed by photosynthetically active green vegetation (APARgreen) for irrigated and rainfed (A) maize and (B) soybean for theperiod June 1 to August 30 when APARgreen was below 5MJm−2 d−1 for maize and 4MJm−2 d−1 for soybean and green leaf area index in both crops exceeded2m2m−2. Descriptive statistics of LUEgreen for this period are given in the right-hand sections of Table 1. Solid lines are best-fit functions. (For interpretation of thereferences to color in this figure legend, the reader is referred to the web version of this article.)

A.A. Gitelson et al. Remote Sensing of Environment 217 (2018) 30–37

34

for soybean) during the season (Fig. 7).With a decrease of the 0.5 m soil volumetric water content (VWC) at

the rainfed maize site in 2003 (Fig. 7A), LUEgreen remained very close orhigher than the mean value for at least 20 days (DOY 180 to 200).During this period fAPARgreen decreased almost synchronously with thedecrease in GPP (Fig. 7A in Gitelson et al., 2015), thus, LUEgreen re-mained quite stable. At that time, leaf chlorophyll content kept in-variant and the decrease in fAPARgreen was likely caused by photo-protective mechanisms such as changes in leaf inclination/leaf rolling,chloroplast avoidance movement, etc. Later, when VWC dropped below0.25m3m−3, LUEgreen in the rainfed site decreased and was smallerthan in the irrigated site, although it did not drop below the standarddeviation of multiyear mean LUEgreen.

In soybean (Fig. 7B) when the 0.5 m VWC at the rainfed site wasbelow 0.25m3m−3, LUEgreen oscillated around its multiyear meanvalue not dropping below the STD of LUEgreen. Thus, during droughtevents in both crops the day-to-day LUEgreen oscillated with a magni-tude about half of the standard deviation of the multiyear meanLUEgreen.

4. Discussion and conclusions

For the data used in this paper it was shown that LUEgreen is affected

by many factors, specifically cloudiness coefficient (Suyker and Verma,2012) as well as daytime temperature, vapor pressure deficit, andphenology (in Nguy-Robertson et al., 2015). Our results showed thatthe effect of all these factors on LUEgreen resulted in normalized stan-dard deviations of LUEgreen for irrigated and rainfed maize of 11.9%and for irrigated and rainfed soybean of 13.3% thus demonstratingconvergence of LUEgreen to quite a narrow range.

In both maize and soybean, C3 and C4 crops, LUEgreen in irrigatedand rainfed sites are statistically indistinguishable showing low sensi-tivity to water availability. This conclusion is also supported by findingsthat (i) about 90% of GPP variation in crops and grasslands is explainedby total canopy chlorophyll content (Gitelson et al., 2003; Peng et al.,2011; Gitelson et al., 2015; Sakowska et al., 2016; Wu et al., 2009), (ii)assuming an invariant LUEgreen, GPP in crops and grasslands may beaccurately retrieved from close range and satellite data (e.g., Gitelsonet al., 2006, 2012; Harris and Dash, 2010, 2011; Peng et al., 2013;Rossini et al., 2012, 2014; Sakowska et al., 2014, 2016; Wu et al.,2009), and (iii) the maximum daily LUE based on PAR absorption bycanopy chlorophyll, unlike other expressions of LUE, tends to convergeacross biome types (Zhang et al., 2018).

Seasonal constitutive changes of LUEgreen remained remarkablysmall over a wide range of water supply in rainfed and irrigated maizeand soybean, crops with different photosynthetic pathways, leaf

Fig. 6. LUEgreen vs. day of year (DOY) and ΔPAR= (PARpot− PARin) vs. DOY relationships for (A) rainfed maize in 2005 and (B) soybean in 2006. Dashed linesindicating multiyear mean and standard deviation of LUEgreen for the period June 1 to August 30 when APARgreen was below 5MJm−2 d−1 for maize and4MJm−2 d−1 for soybean.

Fig. 7. The difference (ΔLUEgreen) between measured and multiyear mean (16 sites-years for maize and 8 site-years for soybean) LUEgreen values in irrigated andrainfed sites during years with drought events in (A) 2003 maize and (B) 2002 soybean. Black lines represent soil average volumetric water content (0.5m depth)measured at four locations in rainfed sites. Dashed lines are the standard deviations of the multiyear mean LUEgreen values for maize and soybean.

A.A. Gitelson et al. Remote Sensing of Environment 217 (2018) 30–37

35

structures and canopy architectures. The magnitude and composition ofincident radiation affect the magnitude of the day-to-day facultativeLUEgreen change. Increases in incident PAR caused decreases ofLUEgreen.

Is limited resource availability and high resource acquisition costs atrainfed sites a reason for efficient resource use and convergence ofLUEgreen? Such a scenario results in an optimization of resource allo-cation, which then results in a maximization of carbon gains and aconvergence on a narrow range of LUEgreen (Field, 1991; Goetz andPrince, 1999). In this case, the plant response to stress is a decrease inAPARgreen such that LUEgreen remains relatively invariant.

To make conclusions about facultative and constitutive LUEgreenchanges in other crops and vegetation types and to identify the causesunderlying the LUEgreen values, it is necessary to assess fAPARgreen andLUEgreen using consistent procedures (Gitelson and Gamon, 2015).

Our results bring important implications for remote estimating ofprimary production in crops. Convergence of LUEgreen allows the use ofsimple robust gross primary production models and also allows a betterunderstanding of the role and constraints of LUEgreen in process-basedmodels. Assuming invariant LUEgreen, the models based on either thecanopy/stand/community chlorophyll content or green LAI may facil-itate accurate assessments of primary production and plant optimiza-tion patterns at multiple scales, from leaves to canopies and entire re-gions.

Acknowledgements

This research was supported by NASA NACP grant No.NNX08AI75G and by the U.S. Department of Energy: (a) EPSCoR pro-gram, Grant No. DE-FG-02-00ER45827 and (b) Office of Science (BER),Grant No. DE-FG03-00ER62996. We sincerely appreciate the supportand the use of facilities and equipment provided by the Center forAdvanced Land Management Information Technologies (CALMIT) theSchool of Natural Resources and the Carbon Sequestration Program atthe University of Nebraska-Lincoln. AG greatly acknowledges supportby the European Union FP7 Marie Curie International IncomingFellowship at Technion. We are very thankful to RSE Associate EditorPablo J. J. Zarco-Tejada and three anonymous reviewers for their veryconstructive and helpful reviews.

References

Barton, C.V.M., North, P.R.J., 2001. Remote sensing of canopy light use efficiency usingthe photochemical reflectance index - model and sensitivity analysis. Remote Sens.Environ. 78, 264–273.

Burba, G., 2005. Water and Energy Fluxes in Native Tallgrass Prairie and CultivatedWheat Ecosystems. University of Nebraska-Lincoln thesis: Natural Resource Sciences(Ph.D.), Lincoln, NE (87 pp.).

Ciganda, V., Gitelson, A.A., Schepers, J., 2009. Non-destructive determination of maizeleaf and canopy chlorophyll content. J. Plant Physiol. 166, 157–167.

Field, C.B., 1991. Ecological scaling of carbon gain to stress and resources availability. In:Mooney, H.A., Winner, W.E., Pell, E.J. (Eds.), Response of Plants to Multiple Stresses.Academic Press, London, pp. 35–65.

Field, C.B., Randerson, J.T., Malmström, C.M., 1995. Global net primary production:combining ecology and remote sensing. Remote Sens. Environ. 51, 74–88. https://doi.org/10.1016/0034-4257(94)00066-V.

Gallo, K.P., Daughtry, C.S.T., Bauer, M.E., 1985. Spectral estimation on absorbed pho-tosynthetically active radiation in corn canopies. Remote Sensing of Environment 17,221–232.

Gamon, J.A., Berry, J.A., 2012. Facultative and constitutive pigment effects on thePhotochemical Reflectance Index (PRI) in sun and shade conifer needles. Isr. J. PlantSci. 60, 85–95.

Gitelson, A.A., Gamon, J.A., 2015. The need for a common basis for defining light-useefficiency: implications for productivity estimation. Remote Sens. Environ. 156,196–201.

Gitelson, A.A., Verma, S.B., Vina, A., Rundquist, D.C., Keydan, G., Leavitt, B., Arkebauer,T.J., Burba, G.G., Suyker, A.E., 2003. Novel technique for remote estimation of CO2

flux in maize. Geophys. Res. Lett. 30 (9), 1486. https://doi.org/10.1029/2002GL016543.

Gitelson, A.A., Viña, A., Verma, S.B., Rundquist, D.C., Arkebauer, T.J., Keydan, G.,Leavitt, B., Ciganda, V., Burba, G.G., Suyker, A.E., 2006. Relationship between grossprimary production and chlorophyll content in crops: implications for the synopticmonitoring of vegetation productivity. J. Geophys. Res. 111, D08S11. https://doi.

org/10.1029/2005JD006017.Gitelson, A.A., Peng, Y., Masek, J.G., Rundquist, D.C., Verma, S., Suyker, A., Baker, J.M.,

Hatfield, J.L., Meyers, T., 2012. Remote estimation of crop gross primary productionwith Landsat data. Remote Sens. Environ. 121, 404–414.

Gitelson, A.A., Peng, Y., Arkebauer, T.J., Schepers, J., 2014. Relationships between grossprimary production, green LAI, and canopy chlorophyll content in maize:Implications for remote sensing of primary production. Remote Sens. Environ. 144,65–72.

Gitelson, A.A., Peng, Y., Arkebauer, T.J., Suyker, A.E., 2015. Productivity, absorbedphotosynthetically active radiation, and light use efficiency in crops: implications forremote sensing of crop primary production. J. Plant Physiol. 177, 100–109. https://doi.org/10.1016/j.jplph.2014.12.015.

Gitelson, A.A., Gamon, J.A., Solovchenko, A., 2017. Multiple drivers of seasonal changein PRI: implications for photosynthesis 2. Stand level. Remote Sens. Environ. 190,198–206. https://doi.org/10.1016/j.rse.2016.12.015.

Goetz, S.J., Prince, S.D., 1999. Modelling terrestrial carbon exchange and storage: evi-dence and implications of functional convergence in light-use efficiency. In: Fitter,A.H., Raffaelli, D. (Eds.), Advances in Ecological Research. 28. Academic Press, SanDiego, pp. 57–92.

Goward, S.M., Huemmerich, K.E., 1992. Vegetation canopy PAR absorptance and theNormalized Difference Vegetation Index: an assessment using SAIL model. RemoteSens. Environ. 39, 119–140.

Gu, L., Baldocchi, D.D., Verma, S.B., Black, T.A., Vesala, T., Falge, E.M., Dowty, P.R.,2002. Advantages of diffuse radiation for terrestrial ecosystem productivity. J.Geophys. Res. 107, 4050.

Hall, F.G., Huemmrich, K.F., Goetz, S.J., Sellers, P.J., Nickeson, J.E., 1992. Satellite re-mote sensing of surface energy balance: success, failures and unresolved issues inFIFE. J. Geophys. Res. 97, 19061–19089.

Hanan, N.P., Burba, G., Verma, S.B., Berry, J.A., Suyker, A., Walter-Shea, E.A., 2002.Inversion of net ecosystem CO2 flux measurements for estimation of canopy PARabsorption. Glob. Chang. Biol. 8, 563–574.

Harris, A., Dash, J., 2010. The potential of the MERIS Terrestrial Chlorophyll Index forcarbon flux estimation. Remote Sens. Environ. 114, 1856–1862.

Harris, A., Dash, J., 2011. A new approach for estimating northern peatland gross primaryproductivity using a satellite-sensor-derived chlorophyll index. J. Geophys. Res. 116,G04002. https://doi.org/10.1029/2011JG001662.

Hatfield, J.L., Asrar, G., Kanemasu, E.T., 1984. Intercepted photosynthetically activeradiation estimated by spectral reflectance. Remote Sensing of Environment 14,65–75.

Kergoat, L., Lafont, S., Arneth, A., Le Dantec, V., Saugier, B., 2008. Nitrogen controlsplant canopy Light-Use-Efficiency in temperate and boreal ecosystems. J. Geophys.Res. 113 (G04017), 1–19.

Lindquist, J.L., Arkebauer, T.J., Walters, D.T., Cassman, K.G., Dobermann, A., 2005.Maize radiation use efficiency under optimal growth conditions. Agron. J. 97, 72–78.

Mock, J.J., Pearce, R.B., 1975. An ideotype of maize. Euphytica 24, 613.Monteith, J.L., 1972. Solar radiation and productivity in tropical ecosystems. J. Appl.

Ecol. 744–766.Nguy-Robertson, A., Suyker, A., Xiao, X., 2015. Modeling gross primary production of

maize and soybean croplands using light quality, temperature, water stress, andphenology. Agric. For. Meteorol. 213, 160–172. https://doi.org/10.1016/j.agrformet.2015.04.008.

Nichol, C.J., Huemmrich, K.F., Black, T.A., Jarvis, P.G., Walthall, C.L., Grace, J., Hall,F.G., 2000. Remote sensing of photosynthetic-light-use efficiency of boreal forest.Agric. For. Meteorol. 101, 131–142.

Norman, J.M., Arkebauer, T.J., 1991. Predicting canopy photosynthesis and light useefficiency from leaf characteristics. In: Modeling Crop Photosynthesis – FromBiochemistry to canopy. American Society of Agronomy and Crop Science Society ofAmerica, pp. 75–94 CSSA Special Publication. No. 19.

Peng, Y., Gitelson, A.A., Keydan, G.P., Rundquist, D.C., Moses, W.J., 2011. Remote esti-mation of gross primary production in maize and support for a new paradigm basedon total crop chlorophyll content. Remote Sens. Environ. 115, 978–989. https://doi.org/10.1016/j.rse.2010.12.001.

Peng, Y., Gitelson, A.A., Sakamoto, T., 2013. Remote estimation of gross primary pro-ductivity in crops using MODIS 250m data. Remote Sens. Environ. 128, 186–196.

Rossini, M., Cogliati, S., Meroni, M., Migliavacca, M., Galvagno, M., Busetto, L.,Cremonese, E., Julitta, T., Siniscalco, C., Morra di Cella, U., Colombo, R., 2012.Remote sensing-based estimation of gross primary production in a subalpine grass-land. Biogeosciences 9, 2565–2584. https://doi.org/10.5194/bg-9-2565-2012.

Rossini, M., Migliavacca, M., Galvagno, M., Meroni, M., Cogliati, S., Cremonese, E., Fava,F., Gitelson, A., Julitta, T., Morra di Cella, U., Siniscalco, C., Colombo, R., 2014.Remote estimation of grassland gross primary production during extreme meteor-ological seasons. Int. J. Appl. Earth Obs. Geoinf. 29, 1–10.

Ruimy, A., Dedieu, G., Saugier, B., 1996. TURC: a diagnostic model of continental grossprimary productivity and net primary productivity. Glob. Biogeochem. Cycles 10 (2),269–285. https://doi.org/10.1029/96GB00349.

Ruimy, A., Kergoat, L., Bondeau, A.T., the participants of the P. N. P. P. model inter-comparison, 1999. Comparing global models of terrestrial net primary productivity(NPP): analysis of differences in light absorption and light-use efficiency. Glob.Chang. Biol. 5 (S1), 56–64.

Sakowska, K., Vescovo, L., Marcolla, B., Juszczak, R., Olejnik, J., Gianelle, D., 2014.Monitoring of carbon dioxide fluxes in a subalpine grassland ecosystem of the ItalianAlps using a multispectral sensor. Biogeosciences 11, 4695–4712.

Sakowska, K., Juszczak, R., Gianelle, D., 2016. Remote sensing of grassland biophysicalparameters in the context of the Sentinel-2 satellite mission. J. Sens., 4612809.https://doi.org/10.1155/2016/4612809.

Schull, M.A., Anderson, M.A., Houborg, R., Gitelson, A.A., Kustas, W.P., 2015. Thermal-

A.A. Gitelson et al. Remote Sensing of Environment 217 (2018) 30–37

36

based modeling of coupled carbon, water and energy fluxes using nominal light useefficiencies constrained by leaf chlorophyll observations. Biogeosciences 12,1511–1523. www.biogeosciences.net/12/1511/2015/ https://doi.org/10.5194/bg-12-1511-2015.

Setiyono, T.D., Weiss, A., Specht, J., Bastidas, A.M., Cassman, K.G., Dobermann, A., 2007.Understanding and modeling the effect of temperature and daylength on soybeanphenology under high-yield conditions. Field Crop Res. 100, 257–271.

Suyker, A.E., Verma, S.B., 2010. Coupling of carbon dioxide and water vapor exchangesof irrigated and rainfed maize–soybean cropping systems and water productivity.Agric. For. Meteorol. 150 (4), 553–563.

Suyker, A.E., Verma, S.B., 2012. Gross primary production and ecosystem respiration ofirrigated and rainfed maize–soybean cropping systems over 8 years. Agric. For.Meteorol. 165, 12–24. https://doi.org/10.1016/j.agrformet.2012.05.021.

Suyker, A.E., Verma, S.B., Burba, G.G., Arkebauer, T.J., Walters, D.T., Hubbard, K.G.,2004. Growing season carbon dioxide exchange in irrigated and rainfed maize. Agric.For. Meteorol. 124, 1–13.

Suyker, A.E., Verma, S.B., Burba, G.G., Arkebauer, T.J., 2005. Gross primary productionand ecosystem respiration of irrigated maize and irrigated soybean during a growingseason. Agricultural and Forest Meteorology 131, 180–190.

Turner, D.P., Urbanski, S., Bremert, D., Wofsy, S.C., Meyers, T.P., Gower, S.T., Gregory,M., 2003. A cross-biome comparison of daily light use efficiency for gross primary

production. Glob. Chang. Biol. 9, 383–395.Verma, S.B., Dobermann, A., Cassman, K.G., Walters, D.T., Knops, J.M., Arkebauer, T.J.,

Suyker, A.E., Burba, G.G., Amos, B., Yang, H., Ginting, D., Hubbard, K.G., Gitelson,A.A., Walter-Shea, E.A., 2005. Annual carbon dioxide exchange in irrigated andrainfed maize-based agroecosystems. Agric. For. Meteorol. 131, 77–96.

Viña, A., Gitelson, A.A., 2005. New developments in the remote estimation of the fractionof absorbed photosynthetically active radiation in crops. Geophys. Res. Lett. 32,L17403.

Viña, A., Gitelson, A.A., Nguy-Robertson, A.L., Peng, Y., 2011. Comparison of differentvegetation indices for the remote assessment of green leaf area index of crops.Remote Sens. Environ. 115, 3468–3478. https://doi.org/10.1016/j.rse.2011.08.010.

Wu, C., Niu, Z., Tang, Q., Huang, W., Rivard, B., Feng, J., 2009. Remote estimation ofgross primary production in wheat using chlorophyll-related vegetation indices.Agric. For. Meteorol. 149, 1015–1021.

Xu, L., Baldocchi, D.D., 2003. Seasonal trends in photosynthetic parameters and stomatalconductance of blue oak (Quercus douglasii) under prolonged summer drought andhigh temperature. Tree Physiol. 23 (13), 865–877.

Zhang, Y., Xiao, X., Wolf, S., Wu, J., Wu, X., Gioli, B., et al., 2018. Spatio-temporalconvergence of maximum daily light use efficiency based on radiation absorption bycanopy chlorophyll. Geophys. Res. Lett. 45, 3508–3519. https://doi.org/10.1029/2017GL076354.

A.A. Gitelson et al. Remote Sensing of Environment 217 (2018) 30–37

37