Ia

SKC

a

AA

KRVV

1

fmcpttigcd

eefttk

0d

Ecological Modelling 222 (2011) 2530– 2541

Contents lists available at ScienceDirect

Ecological Modelling

journa l h o me pa g e: www.elsev ier .com/ locate /eco lmodel

nterpreting spatial heterogeneity of crop yield with process model and remote sensing

hi Hu, Xingguo Mo ∗

ey Laboratory of Water Cycle and Its Related Land Processes, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, PRhina

r t i c l e i n f o

rticle history:vailable online 12 June 2011

eywords:emote sensing datacmax

IP model

a b s t r a c t

A process-based crop growth model (Vegetation Interface Processes (VIP) model) is used to estimate cropyield with remote sensing over the North China Plain. Spatial pattern of the key parameter—maximumcatalytic capacity of Rubisco (Vcmax) for assimilation is retrieved from Normalized Difference of VegetationIndex (NDVI) from Terra-MODIS and statistical yield records. The regional simulation shows that theagreements between the simulated winter wheat yields and census data at county-level are quite wellwith R2 being 0.41–0.50 during 2001–2005. Spatial variability of photosynthetic capacity and yield inirrigated regions depend greatly on nitrogen input. Due to the heavy soil salinity, the photosyntheticcapacity and yield in coastal region is less than 50 �mol C m−2 s−1 and 3000 kg ha−1, respectively, whichare much lower than that in non-salinized region, 84.5 �mol C m−2 s−1 and 5700 kg ha−1. The predicted

−1

yield for irrigated wheat ranges from 4000 to 7800 kg ha , which is significantly larger than that ofrainfed, 1500–3000 kg ha−1. According to the path coefficient analysis, nitrogen significantly affects yield,by which water exerts noticeably indirect influences on yield. The effect of water on yield is regulated, to acertain extent, by crop photosynthetic capacity and nitrogen application. It is believed that photosyntheticparameters retrieved from remote sensing are reliable for regional production prediction with a process-based model.

. Introduction

Due to the growing population and diminishing hectare ofarmland, grain production increasing by per unit of area in the

ain agricultural area, such as North China Plain (NCP), is criti-al for the nation food security. A basic and emergent issue forroduction enhancement is to investigate what factors limitinghe current production and how to ameliorate the situation. Evenhough process-based crop growth model can simulate crop yieldn different environmental and management situations, it still facesreat challenges to capture the spatial and temporal variability ofrop yield due to the uncertainty of spatial environmental drivingata (e.g., weather, soil, management, irrigation, fertilization).

Remote sensing has been widely used for retrieving this kind ofnvironmental information in recent years, due to its high inher-nt spatial and temporal density. Through some forms of empiricalunction or physical scattering models, remote sensing data can be

ranslated into data which are corresponding to forcing parame-ers or state variables in ecological modeling (Mo et al., 2005). Twoinds of data are most frequently been derived, one is the surface

∗ Corresponding author. Tel.: +86 10 64889307.E-mail address: moxg@igsnrr.ac.cn (M. Xingguo).

304-3800/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.ecolmodel.2010.11.011

© 2010 Elsevier B.V. All rights reserved.

climatic data and the other is some vegetation-related information.The surface climatic data mainly include air temperature (Hamdiet al., 2009), humidity (Lohnert et al., 2009), precipitation and radi-ation (Michaelides et al., 2009; Troy and Fwood, 2009). The secondkind of data includes individual environmental factors (e.g., sur-face temperature, soil moisture content) and vegetation growthindexes (e.g., LAI, Vcmax), which are usually expressed as parame-ters in the model and their spatial pattern are hard to obtain. Withthe support of remote sensing, vegetation-related information canbe readily obtained and the modeling capability of process-basedmodel is intensified. This is confirmed by Ottle and Vidal-Madjar(1994) and Gillies et al. (1997) who used the surface tempera-ture to improve the simulation of energy balance components,Burke et al. (1998) who utilized the surface soil moisture con-tent derived from microwave data to update the process modelingof bare soil, and Matsushita and Tamura (2002) who successfullysimulated net primary productivity in East Asia with the aid ofLAI retrieved from remote sensing. However, there is no reporton photosynthetic capacity pattern retrieved from remote sensingdata yet. Previously, the regional photosynthetic capacity is defined

based on the data at some typical sites corresponding to yieldlevels (Mo et al., 2009), because there is no empirically or theoret-ically derived transfer function between Vcmax and remote sensingdata.

l Mode

tesatcvtnatcaise

trtyp

2

2

Npgwts1s57mfpwsl

2

ewoTe2

cWidi

ses

H. Shi, M. Xingguo / Ecologica

Several literatures have attributed spatial variation of crop yieldo regional climate variables (Twine and Kucharik, 2009; Lobellt al., 2009), irrigation facility (Mo et al., 2009; Reidsma et al., 2009),oil nutrient status (Shahandeh et al., 2005; David et al., 2005)nd a combination of all the above-mentioned factors. Accordingo Kaufmann and Snell (1997), roughly 19% of the variability inorn yield in USA is due to climate variables and about 74% of theariability can be explained by soil variables (e.g., capital, labor, fer-ilizers, pesticides, etc.). In the North China Plain, the effect of water,itrogen and soil alkalization on crop yield have been studied sep-rately. It is reported that the available nitrogen is closely relatedo wheat yield and thousand-grain weight with partial correlationoefficients of 0.57 and 0.50, respectively (Chen et al., 2008). It islso shown that water is the limiting factor to winter wheat yieldn the northern part of NCP when nitrogen is sufficient, and soilalinity is the key factor limiting crop yield in costal regions (Mot al., 2005; Jin et al., 1999).

In this paper, we design an approach to retrieve photosyn-hetic capacity pattern from remote sensing data for simulation ofegional crop yield. Based on the pattern of crop yield, some ques-ions are focused: (1) what factors show significant effects on cropield and how they impact wheat yield; (2) how to increase currentroduction level.

. Materials and methods

.1. Region description

The Beijing–Tianjin–Hebei region is located at north part of theorth China Plain, which covers Beijing, Tianjin and part of Hebeirovince and extends from latitude 36◦05′N to 40◦24′N and lon-itude 113◦11′E to 119◦45′E. The prevailing cropping system is aheat–maize rotation, which represents traditional cropping sys-

em in the North China Plain and accounts for 90% of the cerealowing area. The mean annual temperature is between 3.8 ◦C and3.1 ◦C and varies gradually from subhumid in the southern toemi-arid in the northern part. The annual precipitation is about00–800 mm, which distributes irregularly among seasons. About5% of the rainfall occurs from late June to September during theaize growth stage, and only 25% occurs in wheat growing period,

rom October to May. Due to insufficient rainfall in wheat growingeriod as well as insufficient surface water for irrigation, ground-ater is the major irrigation water source in the study area. The

oil texture in this region is categorized as silt, sandy silt and sandyoam (Fig. 1).

.2. Model introduction

The ecological system model used in this study is a dynamiccosystem model (Vegetation Interface Processes (VIP) model)hich simulates radiation, water, heat and CO2 transfer processes

ver the crop growing seasons (Mo et al., 2004; Mo and Liu, 2001).he main biochemical processes in the model include land surfacenergy balance, water cycle as well as carbon cycle (Fig. 2, Mo et al.,009).

Energy fluxes are described with a two-source scheme that dis-erns the canopy and soil surface separately (Shuttleworth and

allace, 1985). Transfer of solar radiation within the crop canopys simulated with 20 sub-layers that distinguishes the direct andiffuse components of both the visible (300–700 nm) and the near

nfrared (700–1300 nm) fractions (Mo and Liu, 2001).

Water cycle deals with the total above-canopy evapotran-

oiration and soil moisture transfer. The total above-canopyvapotransoiration which consists of soil evaporation, canopy tran-piration and its intercepted water evaporation can be simulated,

lling 222 (2011) 2530– 2541 2531

respectively, in the form similar to the Penman–Monteith equa-tion (Monteith and Unsworth, 1990). The soil moisture budget isestimated using a six-layer scheme and described with Darcy’s law.

Carbon cycle includes assimilation via photosynthesis, cropgrowth and soil organic matter decomposition schemes. Thecrop growth model includes photosynthetic production, the dailygrowth of crop dry mass is expressed as the balance of gross pho-tosynthesis, respiration and senescence, more details can be seenin Mo et al. (2005).

The model focuses on biochemical mechanism on photosynthe-sis, crop dry mass formation and water consumption, because theenergy transfer, water cycle and carbon cycle are interacted viaevapotranspiration, stomata conductance and photosynthesis. Inthis paper, we mainly pay attention to the photosynthesis of C3(wheat) crop. Considering the highly nonlinear response of pho-tosynthesis to incident light, photosynthetic rates for sunlit andshaded leaves are estimated separately, then summed and up-scaled to the canopy using a stomatal conductance/photosynthesisrelationship (Leuning et al., 1995). The detailed biochemicalapproaches of photosynthesis of C3 (wheat) crop, according toFarquhar et al. (1980), Collatz et al. (1991) and Sellers et al. (1996),are employed. The net assimilation rate of CO2 (An, � mol C m−2 s−1)is expressed as:

An = min(AR, AE, AS) − Rd (1)

where AR is the Rubisco limiting rate, AE is the light-limited rate,AS is the capacity for the export or utilization of the products ofphotosynthesis, Rd is the leaf respiration rate (� mol C m−2 s−1). TheRubisco limiting rate AR is expressed as:

AR = Vc

[ci − � ∗

ci + Kc(1 + O2/Ko)

](2)

where Vc is the maximum catalytic capacity of Rubisco(� mol C m−2 s−1), ci, O2 are intercellular concentration of CO2and O2 (Pa), respectively. � * is the CO2 compensation point (Pa,=0.5O2/S, S is the Rubisco specificity for CO2 relative to O2); Kc,Ko are Michaelis constant for CO2 and O2, respectively (Pa). In VIPmodel, Vc decline exponentially with an extinction coefficient kv

and expressed as:

Vc = Vc max exp(−kvL) (3)

where Vcmax is the values of Vc at top of canopy, kv is taken as 0.6and L is the leaf area index. For more details about AE and AS, pleaserefer to Mo and Liu (2001).

A key parameter in this photosynthetic scheme is the maxi-mum catalytic capacity of Rubisco (Vcmax) which is determined bythe amount of activated enzyme (Rubisco) present. The value ofVcmax is affected by any factor which may affect the amount of acti-vated enzyme (Rubisco) present. When there are no environmentalstresses, the concentration of Rubsico in the leaf has highest leveland crop maintains its maximum Vcmax value. During leaf senes-cence or under environmental stresses, Rubsico decomposes andits concentration in leaf falls, resulting in Vcmax value declining.Although Vcmax is the mean value in growing seasons in the VIPmodel (so we would not pay more attention to temporal variationof Vcmax in this study), it is still affected by the degree of environ-mental stresses, leading to spatial variation of Vcmax. The VIP modelhas not yet dealt with nitrogen cycle mechanically in this study. Asnitrogen is a key factor to determine the photosynthetic capabilityand Vcmax value, the influence of soil nitrogen, soil water and their

interactions is assumed to be accounted for by Vcmax values here.Thus, to some extent Vcmax is considered to be an empirical param-eter reflecting the background of crop photosynthetic capacity inthis study.

2532 H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541

Fig. 1. Geographical and soil texture map of Beijing–Tianjin–Hebei region (gray line is the provincial boundary).

Fig. 2. The framework of the VIP model (from Mo et al., 2009).

H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541 2533

F erage

M

2

ilua

2

mssgiullStmslr1iAsam

tttciNsi

ig. 3. Spatial and interannual variations of NDVI [(a) is the spatial variation of avarch to June)].

.3. Data sources

Three types of data are used, namely geographical, meteorolog-cal and agricultural data. Geographical data are used as the basicand surface information for crop and soil. Meteorological data aresed as atmospheric forces to drive VIP model. Agricultural datare used for model and crop parameter validation.

.3.1. GIS and RS dataGIS data consist of land surface characterization data and Nor-

alized Difference of Vegetation Index (NDVI) data. The landurface characterization includes topography, land use map andoil maps. The digital elevation model is obtained from the topo-raphic map at 1:250,000 scale. Land-use data from TM imagesn 2000, at the scale of 1:100,000 (http://www.resdc.cn/), aresed. The land-use is classified into six types, which are farm-

and, mixed forest, grassland, urban land, water area and unutilizedand. In this study, only farmland is taken into consideration.oil maps involve saline–alkaline land map, soil physical tex-ure map and soil total nitrogen map. In saline–alkaline land

ap, the soil is divided into three types: alkali–saline soil, sandyoil and sajong black soil. The alkali–saline soil is categorized asight, medium and heavy alkali–saline types, which is virtuallyetrieved from the resources satellite images in the spring from978 to 1981 (Mo et al., 2009). Soil physical texture data are dig-

tized from a 1:14,000,000 scale map (Institute of soil science,cademia Sinica, 1986). Soil total nitrogen content is divided intoix degrees: <0.05%, 0.05–0.07%, 0.07–0.1%, 0.1–0.15%, 0.15–0.2%nd 0.2–0.4%. These data are retrieved from a 1:4,000,000 scaleap (www.geodata.cn/Portal/index.jsp).The Terra-Modis 16-days maximum composite NDVI from 2001

o 2008, at 1 km resolution, are used. All the geographic informa-ion and remote sensing images have been processed for projectionransformation (Lambert projection), geometrical and atmosphericorrection which aim at removing of cloud effect. The average NDVI

n June and interannual variability of accumulative (March to June)DVI are shown in Fig. 3a and b, respectively. In most part of the

tudy area, the coefficient of variation of annual accumulative NDVIs between 0.43% and 31.5%.

NDVI in Juneand (b) is coefficient of variation of annual accumulative NDVI (from

2.3.2. Meteorological dataWith the inverse distance square method, the daily mete-

orological data are interpolated to the whole region with themeteorological data (2001–2008) from 88 stations in and aroundNCP. These meteorological data include daily average, maximumand minimum air temperature, humidity, wind speed, precipitationand sunshine duration, which are used to drive the model. Exceptsunshine duration, all variables above are corrected with elevationabove sea level.

2.3.3. Agricultural dataThe agricultural statistical data include irrigation amount, fer-

tilization amount and grain yield of winter wheat at county scale ofBeijing, Tianjin and Hebei province from 2001 to 2005. These statis-tical data come from the China Agricultural Yearbook 2002–2006and provincial agricultural annals.

2.4. Approach to establish photosynthetic capacity pattern

The model is run with 1 km grid and 1 h time step from 2001to 2008. The spatial pattern of crop production is affected byagricultural management factors such as, irrigation facilities anddrainage tunnels which may influence irrigation efficiency; chem-ical fertilizers which may affect soil fertility. All these agriculturalmanagement factors should be taken into account during crop yieldsimulation, but the spatial pattern data of these factors are notavailable. In this study, we assume that the function of all theseagricultural management can be eventually expressed by crop pho-tosynthetic capacity (Vcmax).

As Vcmax varies with environmental factors, it is time-consumingto obtain its raster data of the study area through measurement.Feasible solution is to establish relationship between Vcmax andother available regional factors to get photosynthetic capacity pat-tern of crop. The NDVI is based on different properties of greenvegetation in reflecting the incident solar radiation. Healthy green

vegetation usually has a high value while stressed vegetation orvegetation with small leaf area has reduced NDVI value. In recentyears, many studies have been conducted to establish the relation-ships between NDVI and crop yield (Ren et al., 2008; Moriondo et al.,

2534 H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541

Table 1Irrigation treatments in wheat system.

Treatments Before winter Green return stage Elongation stage Booting stage Grain filling stage

1 – 80 – – –2 – 80 – 70 –3 70 80 – 70 –4 70 80 70 70 –

T

2atcyi

2

ccdcmacrcipcslaiy

3

3

mLwOsCabrr

3

mtrti(uc

5 70 75

he values indicate irrigation amount (mm).

007). It may be possible to build the relationship between NDVInd Vcmax through crop yield. Following steps are taken to deducehis relationship: (1) seeking the relationship between NDVI androp yield, (2) seeking the relationship between Vcmax and cropield predicted by VIP model, (3) with the aid of crop yield, obtain-ng the relationship between Vcmax and NDVI.

.5. Data analysis

To find the principal factors influencing final grain yield, simpleorrelation and path coefficient analysis are applied. Since yieldomponents occur successively and may interact in compensationuring plant development, simple correlation may not provide aomplete picture of the importance of each component in deter-ining grain yield. Thus, path coefficient analysis is conducted

fter simple correlation analysis to help dividing the correlationoefficient into direct and indirect effects caused by the mutualelationships among yield components. Expressed as direct pathoefficient and indirect path coefficient, respectively, the direct andndirect effect values indicate the influence degree of each inde-endent variable on dependent variable. The more details of pathoefficient analysis can be seen in del Moral et al. (2003). In thistudy, correlation coefficients and path coefficients among the fol-owing variables are computed from the mean values, over yearsnd counties, namely, (1) nitrogen application rate; (2) rainfall; (3)rrigation amount; (4) simulated grain yield; (5) statistical grainield.

. Results

.1. Model verification

To validate the growth process predicted by VIP model, fieldeasurements were carried out in a winter wheat field at

uancheng Agricultural Experimental Station (37◦53′N, 114◦41′E,ith elevation at 50.1 m) during the whole growing season (fromctober 2006 to June 2007). The five irrigation schedules are

hown in Table 1 and each treatment was replicated three times.omparisons between the simulated and the measured biomassesbove ground are presented in Fig. 4. The simulated abovegroundiomasses are in fair agreement with the observations. The cor-elation coefficient is 0.98 (R2 = 0.97), which demonstrates theationality and reliability of the model.

.2. Relationship establishment and validation

Since environmental stress and NDVI change with years, nor-ally, the Vcmax map would also change with years. Hence, using

he Vcmax value calculated by NDVI of a particular year will giveise to errors in yield simulation. However, the coefficient of varia-ion of accumulative NDVI in most part of NCP is lower than 8.74%,

ndicating no significant difference with annual accumulative NDVIFig. 2b) in NCP from 2001 to 2008. This make it is possible to sim-late crop yield of some other years with certain Vcmax which isalculated by NDVI of a particular year (taken as 2001 in this study).

70 70 70

Based on the climatic characteristics of NCP and phenological char-acters of winter wheat, the accumulative NDVI from March to Junein 2001 is used to establish the linear relationship between yieldand NDVI (Fig. 5a), the regressive equation is as follows:

yyield = 747.71xNDVI + 2371.40 (R2 = 0.52) (4)

where yyield is the county-level yield of winter wheat, xNDVI is theaverage county-level value of accumulative NDVI.

Since the relationship between Vcmax and yield is controlled bythe mechanism of crop growth, it can be used to scale up the model.In this study, the Vcmax and yield data of Shijiazhuang are used toestablish the relationship between Vcmax and yield (Fig. 5b), theregressive equation is as follows:

yVc max = 17.46e0.0003xyield (R2 = 0.99) (5)

where yVc max is the value of Vcmax, xyield is the winter wheat yield.The relationship between Vcmax and NDVI can be derived from

the two above equations and is expressed as follows:

yVc max = 17.46e(0.23xNDVI+0.71) (6)

where yVc max is the value of Vcmax at a grid, xNDVI is the value ofNDVI at corresponding grid.

Using the accumulative NDVI (March to June) of 2001, the valueof Vcmax at each grid can be calculated by Eq. (6). Based on these val-ues of Vcmax, the wheat yields from 2001 to 2005 in the study areaare simulated by VIP model. The comparisons of simulated and sta-tistical yields of each county are presented in Fig. 6. The simulatedand statistical yields are in reasonable agreement with correla-tion coefficients varying from 0.64 to 0.71 (R2 ranges between 0.41and 0.50). The average yield of statistical and simulated shows abetter correlation coefficient (0.72, R2 = 0.51) than those of eachyear, demonstrating the rationality of Eq. (6). The correlation coef-ficients between simulated and statistical yields from 2002 to 2005are not lower than that in 2001, indicating the reasonability ofthe method. However, the Vcmax map depends on near real-timeremote sensing data, the precision of yield prediction by a certainVcmax map lie on the variation degree of environmental factors. Thismethod therefore cannot be used for longer-term projection. Sincethe environmental factors did not change greatly in recent years(the coefficient of variation of accumulative NDVI from 2001 to2008 in most part of NCP is lower than 8.74%), from the above val-idation, using this method to predict near few years yield in NCP isacceptable.

3.3. Spatial variation of Vcmax

With the relationship between NDVI and Vcmax, it is able to getthe spatial pattern of Vcmax over whole region via NDVI (Fig. 7a).The value of Vcmax ranges from 40 to 120 �mol C m−2 s−1. In thenorthern, western and southern areas where soil are free fromalkali–saline, Vcmax value is over 60 �mol C m−2 s−1 and its spatial

variation is highly consistent with nitrogen application (R2 = 0.64).Since the synthesis of Rubisco is closely related to the nitrogenconcentration throughout the lifespan of a leaf (Imai et al., 2008),Vcmax value is linearly related to total leaf nitrogen or available

H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541 2535

F vegroi 2, m3,t abov

nnsnV

Fc

ig. 4. Comparisons of simulated and measured aboveground biomass [(a) the abon treatment 1, treatment 2, treatment 3, treatment 4, treatment 5 of Table 1; m1, mreatment 4, treatment 5 of Table 1and (b) comparisons of simulated and measured

itrogen (Bruck and Guo, 2006; Makino, 2003). In the leaves ofitrogen-starved plants, Rubisco loses earlier than that in nitrogen-

ufficient plants (Crafts-Brandner et al., 1998). This illustrates thatitrogen limitation will reduce the average Rubisco level andcmax value in the lifespan of a leaf. When leaf nitrogen con-

ig. 5. Relationship between NDVI and Vcmax [(a) is the relationship between yield and Nurve].

und biomass changing with time; s1, s2, s3, s4, s5: simulated aboveground biomass m4, m5: measured aboveground biomass in treatment 1, treatment 2, treatment 3,eground biomass].

tent varies from 0.74 to 3.56 g m−2, Vcmax value varies from 32.3to 173 �mol C m−2 s−1 correspondingly (Muller et al., 2005). This

variation tendency is similar with that in this study, Vcmax valuecan reach 100–120 �mol C m−2 s−1 in the western area where Napplication is 240–260 kg ha−1 and irrigation demand is guaran-

DVIand (b) is the relationship between yield and Vcmax, the solid line is the fitting

2536 H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541

s with

tsafttT

Fi

Fig. 6. Comparisons of the simulated yield

eed; as N application decreases to 100–140 kg ha−1, Vcmax valueteps down to 60–80 �mol C m−2 s−1 in the middle area. The aver-ge Vcmax value is 84.5 �mol C m−2 s−1 in non-alkali–saline soil

−2 −1

or which Mo et al. (2009) use 90 �mol C m s . Additionally,he retrieved Vcmax values are compared with those reported inhe literature for NCP and measured values in other sites, seeable 2.

ig. 7. Spatial pattern of Vcmax and wheat yield [(a) is the spatial variation of Vcmax and (bs not farmland without crop yield data].

the measured yields from 2001 to 2005.

Due to poor soil and insufficient irrigation, Vcmax value is lowerthan 50 �mol C m−2 s−1 in the middle and eastern portions ofthe study area. Since accumulation of Na+ or Cl− in the leaf and

poor compartmentation in the vacuoles have been suggested asthe main cause of reduction in photosynthesis of salt-stressedplant (Tester and Davenport, 2003; James et al., 2006), heavysalinized soil and groundwater make Vcmax value decrease to

) is the spatial variation of average wheat yield from 2001 to 2008. The blank area

H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541 2537

Table 2Examples of Vcmax value that have been reported in the literature for NCP and some international sites.

References Value (�mol C m−2 s−1) Comments on growing environment of wheat

Yin et al. (2009) 58.5 Wheat was growing in contains (66.6 × 88.8 × 35 cm), the soil each contain had 3.47 gnitrogen and no extra nitrogen was applied, temperature is about 16 ◦C

65.8 The temperature and soil condition is the same as above, but nitrogen was addedtwice: 5.13 g before sowing and 2.16 g at stem elongation per contain

Alonso et al. (2009) 93.2 The wheat was watered twice a week and being fed once a week with a nutrientsolution

Kothavala et al. (2005) 93 Daily average temperature ranged from −3.7 in winter to 33.9 in summer,precipitation is about 835 mm/year. The soil is silt–clay loam

Wang et al. (2006) 55 The model is used in North China Plain, the environment is similar with that in thisstudy

Mo et al. (2009) 90 The environment is similar with that in this study, 30, 60, 90 was applied in heavyalkali–saline soil, medium alkali–saline soil and all other soil types, respectively60

30Vcmax

125, 1respec

4hM

3

tvaswidcJoa

Ft

Muller et al. (2005) 32.3–173

0 �mol C m−2 s−1 in the eastern coastal zone. This value is a bitigher than 30 �mol C m−2 s−1 which is used in the same area byo et al. (2009).

.4. Spatial variation of crop yield

Crop almost exclusively depends on the ability of photosyn-hetic organisms to sequester CO2 into organic carbon. The spatialariation of Vcmax is in high consistency with that of crop yield,s shown in Fig. 7. In the northeast and southeast portions of thetudy area, since the precipitation is less than 150 mm during theheat growing seasons and no irrigation is guaranteed, this result

n low yield under 3000 kg ha−1. In the west region where irrigationemand is guaranteed, the yield is more than 5000 kg ha−1 which is

onsistent with that of plot-level experiments (Zhang et al., 1999;in et al., 1999) and regional simulation by Mo et al. (2005). Obvi-usly, water stress is one of the main factors that limit crop yieldnd result in spatial variation.

ig. 8. Soil total nitrogen map and nitrogen application rate map [(a) is the soil total nitrohese data are collected from statistical yearbook].

was correlated linearly with leaf nitrogen, the value of Vcmax is 32.3, 58.2, 103,73 �mol C m−2 s−1 when leaf nitrogen is 0.74, 1.19, 2.24, 2.61, 3.56 g m−2,tively

Apart from water availability, there are still other factorsimpacting crop yield. Soil fertility and soil salinity are two ofthem. Farmland soil fertility is determined by natural fertility andapplied fertility. The major nutrient limiting crop grow is nitro-gen in this area, so we mainly discuss nitrogen fertility following.At first glance, the spatial variation of natural fertility (Fig. 8a)is not consistent with that of yield (Fig. 7b). For further analysis,the average yields in different natural fertility levels are shown inFig. 9a. It is found that the highest yield (5981.3 kg ha−1) occurresin grids with 0.15–0.2% natural nitrogen, whereas the yield is only4069.4 kg ha−1 in grids with the highest natural nitrogen (0.2–0.4%),showing crop yield is not only limited by natural nitrogen fertility.As shown in Fig. 8b, nitrogen application in Beijing–Tianjin–Hebeiregion varies from 37 kg ha−1 to 260 kg ha−1. The highest nitro-gen application is in the western and northern parts of the study

area while the nitrogen application is lower in the mid and eastof the study area. This variation tendency is similar to that of cropyield (correlation coefficient between nitrogen applying amount

gen map and (b) is the spatial variation of nitrogen application rate in county level,

2538 H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541

F diffea

ad2ib

eprsisa(ig2w1msNsta2sc

4

4

stpr

ig. 9. Average wheat yield in different soil conditions [(a) are the average yields inlkali–saline soil].

nd crop yield is 0.71). The optimal nitrogen fertility is 96 kg ha−1

uring growth seasons and 155 kg ha−1 before sowing (Cui et al.,006), but the total nitrogen of natural soil is 0.05–0.4% which is

nsufficient for plant. Therefore, soil fertility is mainly determinedy input chemical fertilizers.

Soil alkalization is the main limiting factor of crop yield in theastern coastal zone and some inland regions which have a largeortion of alkali–saline soil. According to Eynard et al. (2005), theeduction in yield normally ranges from 10% to 90% for wheat due tooil salinity. It is shown that the yield above 5000 kg ha−1 occurresn non-alkali–saline soil, whereas the average yield in alkali–salineoil is less than 3000 kg ha−1. It is found that the yield in heavylkali–saline soil is lower than that in medium alkali–saline soilFig. 9b), indicating crop yield decreases with increasing soil salin-ty. Similar results are also reported in naked oat that thousandrain weights are reduced as salinity stress increase (Zhao et al.,009), in cotton that dry mass is reduced by 22%, 52%, and 84%hen the soil salinity level increases from 2.4 dS m−1 to 7.7 dS m−1,

2.5 dS m−1 and 17.1 dS m−1, respectively (Chen et al., 2010). Someechanical investigations point out nitrogen use efficiency in non-

aline soil exceeds that in saline soil by 15% (Irshad et al., 2008), uptake is independent of N rates and mainly influenced by soil

alinity at high salinities (Chen et al., 2010). These demonstratehat soil alkalization reduces nitrogen availability (Li et al., 2008)nd exaggerates the water stress due to osmosis of salt (James et al.,006). As a consequence, crop yield is usually at low level even ifufficient fertilization and irrigation are available in this kind ofonditions, such as the eastern coastal zone.

. Discussion

.1. Correlation and path coefficient analysis

Correlation analyses for simulated and statistical yield are

hown in Table 3. Three yield affecting factors (nitrogen applica-ion rate, irrigation amount and rainfall) are all significantly andositively correlated with simulated and statistical yield. The cor-elation coefficient of modeled grain yield with irrigation amount

rent soil total nitrogen contents and (b) are the average yields in various degree of

(0.583) and rainfall (0.579) are similar with each other, which arelower than that with nitrogen application rate (0.719). In order toobtain further information of the relationships among traits andtheir effects on grain yield, path coefficient analysis is performed(Table 4). Direct effects obtain in the path coefficient analysisshow that only nitrogen application rate has exerted noticeablyinfluence on grain yield (path coefficient is 0.501), demonstrat-ing that nitrogen has most significant effect on yield. Generally,water limitation affects processes such as crop growth and pho-tosynthesis. Nitrogen is the main nutrient limiting yield and themajor factor controlling the rate of biomass accumulation (Jensenet al., 1990).

Although irrigation amount and rainfall do not significantlydirect impact grain yield (their path coefficients are only 0.152and 0.165, respectively), they keep obvious indirect effects ongrain yield via nitrogen application rate (their indirect path coef-ficients are 0.301 and 0.339, respectively), indicating that thepositive influence of water on yield in nitrogen-sufficient condi-tion is more significant than that in nitrogen-deficient condition.In another word, increasing N investment improves WUE (Brueckand Senbayram, 2009), which is similar with that in this study.Additionally, some other similar results are also obtained ondrum wheat (Cabrera-Bosquet et al., 2007), wheat (Jensen andChristensen, 2004) and tomato (Kutuk et al., 2004) with plot exper-iments. Further mechanical studies about interaction of nitrogenand water point out that increased N investment brings positiveeffect on photosynthetic apparatus (Ranjith et al., 1995), but nocounterbalancing effect on stomatal conductance (Harvey and vanden Driessche, 1999; Welander and Ottoson, 2000), these resultin crop assimilation rate increasing, water use efficiency improv-ing and crop yield enhancement. Therefore, the impact of water onyield is regulated, to a certain extent, by nitrogen application, justas the result of path coefficient analysis presenting.

4.2. Yield limited factors

With respect to the associations among characters and grainyield, path coefficient analysis gives a clearer picture than

H. Shi, M. Xingguo / Ecological Modelling 222 (2011) 2530– 2541 2539

Table 3Correlation coefficients among the characters studied for simulated and statistical yield.

Irrigation amount (mm) Rainfall (mm) Simulated yield (kg ha−1) Statistical yield (kg ha−1)

N application rate (kg ha−1) 0.601** 0.477** 0.719** 0.651**

Irrigation amount (mm) – 0.492** 0.583** 0.565**

Rainfall (mm) 0.492** – 0.579** 0.606**

** Significant at 0.01 probability level.

Table 4Path coefficient analysis of simulated and statistical yield.

Pathway Simulated yield Statistical yield

Nitrogen application rate Direct effect 0.501** 0.392**

Indirect effect via irrigation amount 0.092 0.098Indirect effect via rainfall 0.126 0.161Correlation 0.719** 0.651**

Irrigation amount Direct effect 0.152 0.163Indirect effect via nitrogen application rate 0.301** 0.236**

Indirect effect via rainfall 0.130 0.166Correlation 0.583** 0.565**

Rainfall Direct effect 0.165 0.187Indirect effect via nitrogen application rate 0.339** 0.339**

nt

stnaoggGc1

Fw

Indirect effect via irrigation amouCorrelation

** Significant at 0.01 probability level.

imple correlations analysis. Simple correlation analysis revealshat nitrogen application rate, irrigation amount and rainfall all sig-ificantly affect grain yield (correlation coefficient is 0.719, 0.583,nd 0.579, respectively), however, path coefficient analysis showsnly nitrogen application rate exerts significant direct influence onrain yield, other factors put indirect impacts on yield via it. Nitro-

en is therefore the most important factor impacting grain yield.enerally, water limitation and nitrogen availability are the mainonstraints on yield (Araus et al., 2002; Passioura, 2002). Prior to980s these two factors bring imponderable influence to crop pro-

ig. 10. The effect of photosynthetic capacity on yield [(a) is the yield variation tendencith same irrigation amount in different Vcmax levels].

0.075 0.0800.579** 0.606**

duction in NCP. However, the soil conditions are greatly improvedafter year’s efforts by: (1) popularizing crop straw returning toincrease soil organic matter; (2) increasing chemical fertilizer inputand popularizing balanced fertilization technique. These effortsassist the regional soil condition in two ways: firstly improving thesoil structure, water-holding and fertilizer-conserving capability

and secondly ameliorating soil nutrition condition and enhanc-ing complementary effect among nutrient elements. In these soilremediation areas, water is the principal yield limiting factor, theyield gap between potential and water-limited yield varies from

y with Vcmax in five randomly selected sitesand (b) is the yield variation tendency

2 l Mod

5e

cstccic(ti1(rastiAmmde

5

oiVois

tivd0ahrmadicAocpt

A

(Mtn

540 H. Shi, M. Xingguo / Ecologica

00 kg ha−1 in the south to 5400 kg ha−1 in the north of NCP (Wut al., 2006).

In the coastal region, crop photosynthetic capacity is heavilyonstricted by salinization, so it can be enhanced by improvingoil condition. Five sites are chosen randomly to show variationendency of yield with photosynthetic capacity in different soilonditions, the yield in good soil condition (2000 kg ha−1) whererop photosynthetic capacity is high (Vcmax = 120 �mol C m−2 s−1)s twice of that in heavily alkali–saline soil (1000 kg ha−1) whererop photosynthetic capacity is low (Vcmax = 40 �mol C m−2 s−1)Fig. 10a). The influence of irrigation on yield rises as crop pho-osynthetic capacity increases. With 240 mm irrigation, the yieldncreases 3500 kg ha−1 in high crop photosynthetic capacity and000 kg ha−1 in low crop photosynthetic capacity, respectivelyFig. 10b). These demonstrate that the effect of water on yield isegulated, to a certain extent, by crop photosynthetic capacity. Soillkalization reduces nitrogen availability (Li et al., 2008) and photo-ynthetic capacity (Tester and Davenport, 2003; James et al., 2006),his limits the effect of water on yield in alkali–saline soil. Therefore,mproving soil condition is the main priority in alkali–saline soil.s suggested by Eynard et al. (2005), crop yield in alkali–saline soilay be enhanced by nutrient management (especially N), wateranagement (irrigation with good quality water and appropriate

rainage) and the use of soil amendments (manures and gypsum,tc.).

. Conclusion

The process-based ecosystem model (VIP model), with the aidf remote sensing, has been used to estimate regional wheat yieldn the northern portion of NCP. Through established relationship ofcmax with NDVI, the photosynthetic capacity pattern is obtainedver the region. The yield predictions of winter wheat are well val-dated with statistical yields, illustrating the ability of the model inimulating regional grain yield.

The spatial patterns of average wheat yield and photosyn-hetic capacity are strongly influenced by soil salinity degree,rrigated/rainfed condition and nitrogen supplying capacity. Spatialariability of photosynthetic capacity and yield in irrigated regionsepend greatly on applied nitrogen with correlation coefficients of.8 and 0.71, respectively. With increased soil salinity, crop yieldnd photosynthetic capacity decreased correspondently. Due to theeavy soil salinity, the photosynthetic capacity and yield in coastalegion are less than 50 �mol C m−2 s−1 and 3000 kg ha−1, which areuch lower than that in no salinity region, 84.5 �mol C m−2 s−1

nd 5700 kg ha−1. The crop yield will be doubled if the soil con-ition is improved in rainfed condition. The modeled yield for

rrigated wheat ranges from 4000 to 7500 kg ha−1, which is signifi-antly larger than that of rainfed ranging from 1500 to 3000 kg ha−1.lthough water is of importance for crop production, its influencen yield is determined, to a certain extent, by crop photosyntheticapacity. The yield is expected to increase 3500 kg ha−1 in high crophotosynthetic capacity and 1000 kg ha−1 in low crop photosyn-hetic capacity with 240 mm irrigation, respectively.

cknowledgements

This study was jointly supported by 973 Basic Research Project

2010CB428404), International Cooperation Project (0911) of China

OST, and the Innovation project of CAS (KZCX2-YW-Q06-1). Manyhanks to the two anonymous reviews and the Editor whose perti-ent comments have greatly improved the quality of this paper.

elling 222 (2011) 2530– 2541

References

Araus, J.L., Slafer, G.A., Reynolds, M.P., Royo, C., 2002. Plant breeding and drought inC3 cereals: what should we breed for? Ann. Bot. 89, 925–940.

Alonso, A., Perez, P.A., Martinez-Carrasco, R., 2009. Growth in elevated CO2

enhances temperature response of photosynthesis in wheat. Physiol. Plant. 135,109–120.

Burke, E.P., Gurney, R.J., Simmonds, L.P., O’Neill, P.E., 1998. Using a modellingapproach to predict soil hydraulic properties from passive microwave measure-ments. IEEE Trans. Geosci. Remote Sens. 36, 454–462.

Bruck, H., Guo, S.W., 2006. Influence of N form on growth and photosynthesis ofPhaseolus vulgaris L. plants. J. Plant Nutr. Soil Sci. 169, 849–856.

Brueck, H., Senbayram, M., 2009. Low nitrogen supply decreases water-use efficiencyof oriental tobacco. J. Plant Nutr. Soil Sci. 172, 216–223.

Collatz, G.J., Ball, J.T., Grivet, C., Berry, J.A., 1991. Physiological and environmentalregulation of stomatal conductance, photosynthesis and transpiration: a modelthat includes a laminar boundary layers. Agric. For. Meteorol. 54, 107–136.

Crafts-Brandner, S.J., Holzer, R., Feller, U., 1998. Influence of nitrogen deficiency onsenescence and the amounts of RNA and proteins in wheat leaves. Physiol. Plant.102, 192–200.

Cui, Z.L., Chen, X.P., Li, J.L., Xu, J.F., Shi, L.W., Zhang, F.S., 2006. Effect of N fertil-ization on grain yield of winter wheat and apparent N losses. Pedosphere 16,806–812.

Cabrera-Bosquet, L., Molero, G., Bort, J., Nogues, S., Araus, J.L., 2007. The combinedeffect of constant water deficit and nitrogen supply on WUE, NUE and Delta C-13in durum wheat potted plants. Ann. Appl. Biol. 151, 277–289.

Chen, S.R., Xiao, W.Z., Zhu, Y.K., Wang, L.G., 2008. Spatial variability and correla-tion of soil nutrient and wheat yield. Trans. Chin. Soc. Agric. Mach. 39, 140–143(Chinese).

Chen, W.P., Hou, Z.A., Wu, L.S., Liang, Y.C., Wei, C.Z., 2010. Effects of salinity andnitrogen on cotton growth in arid environment. Plant Soil 326, 61–73.

del Moral, L.F.G., Rharrabti, Y., Villegas, D., Royo, C., 2003. Evaluation of grain yieldand its components in durum wheat under Mediterranean conditions: an onto-genic approach. Agron. J. 95, 266–274.

David, C., Jeuffroy, M.H., Henning, J., Meynard, J.M., 2005. Yield variation in organicwinter wheat: a diagnostic study in the Southeast of France. Agron. Sustain. Dev.25, 213–223.

Eynard, A., Lal, R., Wiebe, K., 2005. Crop response in salt-affected soils. J. Sustain.Agr. 27, 5–50.

Farquhar, G.D., Von, C.C., Berry, J.A., 1980. A biochemical model of photosyntheticCO2 assimilation in leaves of C3 species. Planta 149, 78–99.

Gillies, R.R., Carlson, T.N., Cui, J., Kustas, W.P., Humes, K.S., 1997. A verification of the‘triangle’ method for obtaining surface soil water content and energy fluxes fromremote measurements of the normalized difference vegetation index (NDVI) andsurface radiant temperature. Int. J. Remote Sens. 18, 3145–3166.

Harvey, H.P., van den Driessche, R., 1999. Nitrogen and potassium effects on xylemcavitation and water-use efficiency in poplars. Tree Physiol. 19, 943–950.

Hamdi, R., Deckmyn, A., Termonia, P., Demaree, G.R., Baguis, P., Vanhuysse, S., et al.,2009. Effects of historical urbanization in the Brussels capital region on sur-face air temperature time series: a model study. J. Appl. Meteorol. Climatol. 48,2181–2196.

Institute of Soil Science, Academia Sinica, Cartographers, 1989. The Soil Atlas ofChina [Map]. Cartographic Publishing House, Beijing.

Imai, K., Suzuki, Y., Mae, T., Makino, A., 2008. Changes in the synthesis of Rubisco inrice leaves in relation to senescence and N influx. Ann. Bot. 101, 135–144.

Irshad, M., Eneji, A.E., Yasuda, H., 2008. Comparative effect of nitrogen sources onmaize under saline and non-saline conditions. J. Agron. Crop Sci. 194, 256–261.

Jensen, A., Lorenzen, B., Spelling-Ostergaard, H., Kloster-Hvelplund, E., 1990. Radio-metric estimation of biomass and N content of barley grown at different N levels.Int. J. Remote Sens. 11, 1809–1820.

Jin, M.G., Zhang, R.Q., Sun, L.F., Gao, U.F., 1999. Temporal and spatial soil water man-agement: a case study in the Heilonggang region, PR China. Agr. Water Manage.42, 173–187.

Jensen, B., Christensen, B.T., 2004. Interactions between elevated CO2 and added N:effects on water use, biomass, and soil 15-N uptake in wheat. Acta Agr. Scand.Sect. B, Soil Plant Sci. 54, 175–184.

James, R.A., Munns, R., Von Caemmerer, S., Trejo, C., Miller, C., Condon, T., 2006.Photosynthetic capacity is related to the cellular and subcellular partitioning ofNa+, K+ and Cl− in salt-affected barley and durum wheat. Plant Cell Environ. 29,2185–2197.

Kaufmann, R.K., Snell, S.E., 1997. A biophysical model of corn yield: integratingclimatic and social determinants. Am. J. Agr. Econ. 79, 178–190.

Kutuk, C., Cayci, G., Heng, L.K., 2004. Effects of increasing salinity and N-15 labelledurea levels on growth, N uptake, and water use efficiency of young tomato plants.Aust. J. Soil Res. 42, 345–351.

Kothavala, Z., Arain, M.A., Black, T.A., Verseghy, D., 2005. The simulation of energy,water vapor and carbon dioxide fluxes over common crops by the Canadian LandSurface Scheme (CLASS). Agric. For. Meteorol. 133, 89–108.

Leuning, R., Kelliher, F.M., De Pury, D.G., Schulze, E.D., 1995. Leaf nitrogen, photo-synthesis, conductance and transpiration: scaling from leaves to canopies. PlantCell Environ. 18, 1183–1200.

Li, W.Q., Xiao-Jing, L., Khan, M.A., Gul, B., 2008. Relationship between soil character-istics and halophytic vegetation in coastal region of North China. Pakistan J. Bot.40, 1081–1090.

Lohnert, U., Turner, D.D., Crewell, S., 2009. Ground-based temperature and humidityprofiling using spectral infrared and microwave observations. Part I: simulated

l Mode

L

M

M

M

M

M

M

M

M

M

M

O

P

R

H. Shi, M. Xingguo / Ecologica

retrieval performance in clear-sky conditions. J. Appl. Meteorol. Climatol. 48,1017–1032.

obell, D.B., Cassman, K.G., Field, C.B., 2009. Crop yield gaps: their importance, mag-nitudes, and causes. Ann. Rev. Environ. Resour. 34, 179–204.

onteith, J.H., Unsworth, M.H., 1990. Principle of Environmental Physics. EdwardArmold, London.

o, X.G., Liu, S.X., 2001. Simulating evapotranspiration and photosynthesis of winterwheat over the growing season. Agric. For. Meteorol. 109, 20322.

atsushita, B., Tamura, M., 2002. Integrating remotely sensed data with an ecosys-tem model to estimate net primary productivity in East Asia. Remote Sens.Environ. 81, 56–66.

akino, A., 2003. Rubisco and nitrogen relationships in rice: leaf photosynthesis andplant growth. Soil Sci. Plant Nutr. 49, 319–327.

o, X.G., Liu, S.X., Lin, Z.H., 2004. Simulating temporal and spatial variation of evap-otranspiration over the Lushi basin. J. Hydrol. 285, 125–142.

o, X., Liu, S., Lin, Z., Xu, Y., Xiang, Y., McVicar, T.R., 2005. Prediction of crop yield,water consumption and water use efficiency with a SVAT-crop growth modelusing remotely sensed data on the North China Plain. Ecol. Model. 183, 301–322.

uller, J., Wernecke, P., Diepenbrock, W., 2005. LEAFC3-N: a nitrogen-sensitiveextension of the CO2 and H2O gas exchange model LEAFC3 parameterised andtested for winter wheat (Triticum aestivum L.). Ecol. Model. 183, 183–210.

oriondo, M., Maselli, F., Bindi, M., 2007. A simple model of regional wheat yieldbased on NDVI data. Eur. J. Agron. 26, 266–274.

ichaelides, S., Levizzani, V., Anagnostou, E., Bauer, P., Kasparis, T., Lane, J.E., 2009.Precipitation: measurement, remote sensing, climatology and modeling. Atmos.Res. 94, 512–533.

o, X.G., Liu, S.X., Lin, Z.H., Guo, R.P., 2009. Regional crop yield, water consumptionand water use efficiency and their responses to climate change in the NorthChina Plain. Agric. Ecosyst. Environ. 134, 67–78.

ttle, C., Vidal-Madjar, D., 1994. Assimilation of soil moisture inferred from infraredremote sensing in a hydrological model over the HAPEX MOBILHY region. J.Hydrol. 158, 241–264.

assioura, J.B., 2002. Environmental biology and crop improvement. Funct. Plant

Biol. 29, 537–546.

anjith, S.A., Meinzer, F.S., Perry, M.H., Thom, M., 1995. Partitioning of carboxylaseactivity in nitrogen stressed sugarcane and its relationship to bundle sheathleakiness to CO2, photosynthesis and carbon isotope discrimination. Aust. J. PlantPhysiol. 22, 903–911.

lling 222 (2011) 2530– 2541 2541

Ren, J.Q., Chen, Z.X., Zhou, Q.B., Tang, H.J., 2008. Regional yield estimation for winterwheat with MODIS-NDVI data in Shandong, China. Int. J. Appl. Earth Obs. Geoinf.10, 403–413.

Reidsma, P., Ewert, F., Boogaard, H., van Diepen, K., 2009. Regional crop modellingin Europe: the impact of climatic conditions and farm characteristics on maizeyields. Agric. Syst. 100, 51–60.

Shuttleworth, W.J., Wallace, J.S., 1985. Evaporation from sparse crops—an energycombination theory. Quart. J. Roy. Meteorol. Soc. 111, 839–855.

Sellers, P.J., Randall, D.A., Collatz, G.J., Berry, T.A., Field, C.B., 1996. A revised land sur-face parameterization (SIB2) for atmospheric GCMs. Part I: model formulation.J. Climate 9, 676–705.

Shahandeh, H., Wright, A.L., Hons, F.M., Lascano, R.J., 2005. Spatial and temporalvariation of soil nitrogen parameters related to soil texture and corn yield. Agron.J. 97, 772–782.

Tester, M., Davenport, R., 2003. Na+ tolerance and Na+ transport in higher plants.Ann. Bot. 91, 503–527.

Troy, T.J., Fwood, E., 2009. Comparison and evaluation of gridded radiation productsacross northern Eurasia. Environ. Res. Lett. 4, 1–7.

Twine, T.E., Kucharik, C.J., 2009. Climate impacts on net primary productivity trendsin natural and managed ecosystems of the central and eastern United States.Agric. For. Meteorol. 149, 2143–2161.

Welander, N.T., Ottoson, B., 2000. The influence of low light, drought and fertilisationon transpiration and growth in young seedlings of Quercus robur L. For. Ecol.Manage. 127, 139–151.

Wu, D.R., Yu, Q., Lu, C.H., Hengsdijk, H., 2006. Quantifying production potentials ofwinter wheat in the North China Plain. Eur. J. Agron. 24, 226–235.

Wang, J., Yu, Q., Li, J., Li, L.H., Li, X.G., Yu, G., et al., 2006. Simulation of diurnal vari-ations of CO2, water and heat fluxes over winter wheat with a model coupledphotosynthesis and transpiration. Agric. For. Meteorol. 137, 194–219.

Yin, X.Y., Struik, P.C., Romero, P., Harbinson, J., Evers, J.B., Van Der Putten, P.E.L.,et al., 2009. Using combined measurements of gas exchange and chlorophyllfluorescence to estimate parameters of a biochemical C-3 photosynthesis model:a critical appraisal and a new integrated approach applied to leaves in a wheat

(Triticum aestivum) canopy. Plant Cell Environ. 32, 448–464.

Zhang, H., Wang, X., You, M., Liu, C., 1999. Water-yield relations and water-useefficiency of winter wheat in the North China Plain. Irrigation Sci. 19, 37–45.

Zhao, G.Q., Ma, B.L., Ren, C.Z., 2009. Salinity effects on yield and yield componentsof contrasting naked oat genotypes. J. Plant Nutr. 32, 1619–1632.