Modelling the effects of post-heading heat stress on biomass growth of winter wheat

Agricultural and Forest Meteorology 247 (2017) 476–490

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology journal homepage: www.elsevier.com/locate/agrformet

Research Paper

Modelling the effects of post-heading heat stress on biomass growth of winter wheat

MARK

Bing Liua,b, Senthold Assengb, Anning Wanga, Shenghao Wanga, Liang Tanga, Weixing Caoa, ⁎ ⁎ Yan Zhua, , Leilei Liua, a National Engineering and Technology Center for Information Agriculture, Key Laboratory for Crop System Analysis and Decision Making, Ministry of Agriculture, Jiangsu Key Laboratory for Information Agriculture, Jiangsu Collaborative Innovation Center for Modern Crop Production, Nanjing Agricultural University, Nanjing, Jiangsu 210095, PR China b Department of Agricultural and Biological Engineering, University of Florida, Gainesville, FL 32601, USA

A R T I C L E I N F O

A B S T R A C T

Keywords: Heat stress WheatGrow Biomass production Leaf area index Leaf photosynthesis Model improvement

Climate change scenarios project an increase in the frequency of heat stress events, making it critical to quantify adverse heat stress effects on wheat production. Biomass growth determines much of grain yield in winter wheat, but it is substantially reduced under heat stress during the reproductive phase. In this study, leaf photosynthesis, biomass production, and leaf area index (LAI) dynamics were measured under various heat stress treatments in a 4-year phytotron experiment with two winter wheat cultivars. Heat stress at anthesis and during grain filling accelerated the measured degradation of leaf chlorophyll (SPAD) and resulted in a lower leaf photosynthesis rate with decreased final biomass growth. The observed relationships between leaf photosynthesis, LAI, and high temperatures were integrated into the WheatGrow model. In this study, we introduced a new cultivar parameter into the model to simulate cultivar difference in the sensitivity of biomass growth to heat stress. The new heat stress routines in the WheatGrow model significantly improved the simulated growth dynamics and the root mean square error (RMSE) with an independent validation data set for LAI and final aboveground biomass by 40% and 57% under heat stress treatments, respectively. This improvement in the crop model WheatGrow enables more reliable studies on climate change impacts and reduces uncertainties in simulations, particularly the impacts of extreme temperature events on crop growth and yields.

1. Introduction

Prasad et al., 2011). During the reproductive period, heat stress usually accelerates the degradation of leaf chlorophyll and results in lower leaf photosynthesis rate with smaller green leaf area (Prasad et al., 2011; Zhao et al., 2007). A reduction in both photosynthesis rate and green leaf area index will reduce biomass growth and consequently grain yield. Also, higher temperatures during heat stress could increase plant respiration (Atkin and Tjoelker, 2003; Kaše and Čatský, 1984), which could decrease biomass growth. Diverse genetic differences of heat tolerance in the response of photosynthesis to heat stress has been found in wheat (Feng et al., 2014; Wang et al., 2015) and could be a key trait to adapt wheat to climate change (Semenov et al., 2014; Stratonovitch and Semenov, 2015). Process-based crop models have been widely used to assess the impact of climate on crop production (Challinor et al., 2014; Rosenzweig et al., 2014). However, only few studies have dealt with heat stress effects in crop models. When testing 30 wheat models, Asseng et al. (2015) found a wide range of model responses to increasing temperature, especially under higher temperature conditions.

Increased temperature variability under climate change will lead to more heat stress events during crop production, which poses additional risks on global food security (IPCC, 2012). Extreme heat stress events will become more frequent in many main wheat-producing regions (Asseng et al., 2011; Gouache et al., 2012; Gourdji et al., 2013; Lobell et al., 2015; Lobell et al., 2012; Semenov and Shewry, 2011; Teixeira et al., 2013). For instance, in China, the largest wheat producer in the world, heat stress already had significant negative impacts on wheat yields during the last 50 years (Liu et al., 2014; Tao et al., 2015). Several studies have reported the severe negative effects of heat stress on wheat growth and yield; these negative effects occur especially during the reproductive period (Farooq et al., 2011; Pradhan et al., 2012; Prasad and Djanaguiraman, 2014; Tashiro and Wardlaw, 1990; Wardlaw et al., 1989). One of the most sensitive crop growth processes affected by heat stress is biomass accumulation (including photosynthesis and respiration), affecting grain yield (Feng et al., 2014; ⁎

Corresponding authors. E-mail addresses: [email protected] (Y. Zhu), [email protected] (L. Liu).

http://dx.doi.org/10.1016/j.agrformet.2017.08.018 Received 27 March 2017; Received in revised form 23 July 2017; Accepted 15 August 2017 0168-1923/ © 2017 Elsevier B.V. All rights reserved.

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

In our previous study (Liu et al., 2016a), four wheat models (APSIMWheat, CERES-Wheat, Nwheat, and WheatGrow) were systematically evaluated with observed heat stress dataset from environment-controlled phytotron experiments. That study indicated that most crop models only represent parts of heat stress effects on biomass production. Grain yield simulations in most crop models depend on the simulation of biomass growth; therefore, it is critical for crop models to simulate biomass growth accurately. Some studies have implied that improving the algorithms and functions under heat stress could be helpful to enhance the performances of crop models in wheat (Liu et al., 2016a; Stratonovitch and Semenov, 2015). For example, a leaf senescence acceleration function recently incorporated into the Sirius model improved the model predictions under high post-anthesis temperature (Stratonovitch and Semenov, 2015). However, few studies have focused on improving crop models when simulating leaf photosynthesis and biomass growth under heat stress. In this study, with detailed observed data of leaf photosynthesis and biomass growth, we examined and improved the response of biomass growth to post-heading heat stress in a wheat crop model, which will enable better simulations of wheat production under climate change. The objectives of this study were: (1) to determine the post-heading heat stress effects on leaf photosynthesis, leaf area index and biomass growth in wheat and (2) to use this knowledge to improve the predictions of biomass growth under heat stress with the WheatGrow model.

27 °C) with a maximum temperature of 27 °C, was used as a check or control treatment. The average temperature in T1 was 22 °C, which has been considered as the optimal temperature for post-heading period in wheat (Porter and Gawith, 1999). T2, T3, T4, and T5 were used as heat stress treatments. Temperature and humidity in the phytotrons were controlled precisely to simulate daily temperature and humidity fluctuations in the ambient environment to capture the actual response of wheat to heat stress in the field as realistically as possible. Day-night temperature fluctuations shown in the supplementary material (Fig. S1) were similar to the ambient temperatures. After a heat stress period, plants were moved out of the phytotrons and maintained at normal ambient environmental conditions until harvest. The fertilization applied was 18.3 g N m−2, 10.2 g P2O5 m−2, and 18.3 g K2O m−2 prior to sowing, and another 18.3 g N m−2 during jointing stage of wheat. All other cultivation practices, such as irrigation and pesticide application, were performed according to local standards of wheat cultivation to ensure no water or nitrogen stress in the experiments. Meteorological records, including daily temperature, rainfall, and radiation during wheat growing season, were measured by Dynamet–1 K (Dynamet Inc., USA) near the experiment sites. More details about our experiment can be found in Liu et al. (2016a).

2. Materials and methods

2.2.1. Biomass and leaf area index Plant samples were taken before and after heat stress treatments. On the day prior to heat stress treatments, plant samples were taken once to obtain initial growth variables before heat stress. After heat stress treatments, plant samples were taken every seven days during the growing seasons 2010–2011, 2011–2012, and 2012–2013, and every five days during the growing season 2013–2014 until maturity. Samples of ten plants in one pot were analyzed with three replications. Sample plants were separated into different plant tissues including stem and sheath, green leaves, senescence leaves, grain, peduncle, and chaff. At maturity, twelve pots were harvested for each treatment to obtain grain yields, total aboveground biomass, yield components, and grain protein concentration. Green leaf area was measured with LI-3000 leaf area meter (LI-COR, Lincoln, NE, USA).

2.2. Sampling and measurements

2.1. Environment-controlled phytotron experiments Environment-controlled phytotron experiments were conducted for four years at Nanjing (118.78°E, 32.04°N) and Rugao (120.33°E, 32.23°N) in Jiangsu Province of China. Winter wheat (Yangmai16 and Xumai30) was planted in plastic pots, with a plant density of 10 plants per pot. The height and inside diameter of pots were 30 cm and 25 cm. Sowing dates in the four growing seasons were November 1, November 6, November 4, and November 5. Wheat was grown in pots installed outside, in an ambient environment, with no environmental control, before and after the heat stress treatments. Once wheat developed into the appropriate growth stages (anthesis or grain filling), wheat was transferred into phytotrons to be exposed to different heat stress conditions. Table 1 summarizes the heat stress treatments, including two cultivars (Yangmai16 and Xumai30); five temperature levels (Tmin/ Tmax): 17/27 °C (T1), 21/31 °C (T2), 25/35 °C (T3), 29/39 °C (T4), and 33/43 °C (T5); three heat stress durations (3 days, 6 days, and 9 days); and two heat stress stages (anthesis and grain filling). There were 27 pots for sampling after heat stress and measuring grain yield at harvest for each treatment. Heat stress treatments at anthesis and the grain filling stage started when wheat began flowering and 10 days after anthesis (DAA10), respectively. According to previous studies (Farooq et al., 2011; Liu et al., 2014), 30 °C was selected as the temperature threshold of heat stress for winter wheat cultivars. Therefore, T1 (17/

2.2.2. Leaf chlorophyll content The significant positive relationship between SPAD and chlorophyll content has been shown in several previous studies especially during the grain filling period (Uddling et al., 2007), and leaf senescence is the main reason for the lose of leaf greenness during grain filling period. Therefore, leaf chlorophyll status were reflected with SPAD values in this study. Leaf chlorophyll SPAD content was determined using a chlorophyll meter SPAD502 (Soil Plant Analysis Development; Minolta, Japan) at the same time with plant samples were taken in the four growing seasons. Top three leaves on each stem were sampled separately, and there were five replications for each leaf position from

Table 1 Summary of heat stress treatments in environment-controlled phytotron experiments. Cultivar

Growing season

Site

Starting time of treatment

Duration

Yangmai16

2010–2011 2011–2012 2012–2013 2013–2014

Nanjing Nanjing Nanjing Rugao

Anthesis, Anthesis, Anthesis, Anthesis,

D3 D3 D3 D3

Xumai30

2011–2012 2012–2013 2013–2014

Nanjing Nanjing Rugao

Anthesis, DAA10 Anthesis, DAA10 Anthesis, DAA10

DAA10 DAA10 DAA10 DAA10

(3d), (3d), (3d), (3d),

Temperature regime (Tmin/Tmax) D6 D6 D6 D6

(6d) (6d) (6d) (6d), D9 (9d)*

D3 (3d), D6 (6d) D3 (3d), D6 (6d) D3 (3d), D6 (6d), D9 (9d)*

*DAA10: 10 days after anthesis. *D9 (9d): only for treatments during anthesis, not for treatments starting from DAA10. *T1: the control or check treatment.

477

T1 T1 T1 T1

(17 °C/27 °C), (17 °C/27 °C), (17 °C/27 °C), (17 °C/27 °C),

T2 T2 T2 T3

(21 °C/31 °C), (21 °C/31 °C), (21 °C/31 °C), (25 °C/35 °C),

T3 T3 T3 T4

(25 °C/35 °C), (25 °C/35 °C), (25 °C/35 °C), (29 °C/39 °C),

T4 (29 °C/39 °C) T4 (29 °C/39 °C) T4 (29 °C/39 °C) T5(33 °C/43 °C)

T1 (17 °C/27 °C), T2 (21 °C/31 °C), T3 (25 °C/35 °C), T4 (29 °C/39 °C) T1 (17 °C/27 °C), T2 (21 °C/31 °C), T3 (25 °C/35 °C), T4 (29 °C/39 °C) T1 (17 °C/27 °C), T3 (25 °C/35 °C), T4 (29 °C/39 °C), T5 (33 °C/43 °C)

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

under heat stress, and led to the improved version of WheatGrow model. Both the original and improved versions of WheatGrow model in this study include the improvement on modelling heat stress effects on wheat post-heading phenology from Liu et al. (2016b). The variables of the improved model are listed in the Appendix A.

different plants in different pots. 2.2.3. Leaf photosynthesis The response of leaf initial light use efficiency (ε) to heat stress was determined by measuring light response curve of leaf photosynthesis under various temperature levels in growing season 2013–2014. The LICOR “light curve” program in LI-6400 system (LI-COR Biosciences, Lincoln, NE, USA) was used to measure light response curve of leaf photosynthesis rates. The light intensities in the light curve were as follows: 1500, 1200, 1000, 800, 600, 400, 200, 100, 50, 20, 10, 0 μmol m−2 s−1, respectively. Unheated wheat flag leaves were measured at a wide range of temperatures (16–44 °C, with an interval of 2 °C). To get the response of leaf initial light use efficiency to high temperature during all growth stages, we measured light response curves of fully unfolded flag leaves under various temperatures at heading and flowering stage, because wheat flag leaves in these stages usually did not begin the senescence process. At each temperature level, three replicates from different plants and different pots were measured. The value of leaf apparent initial quantum yield of light as the linear regression slope of light response curve at low light level (< 200 μmol m−2 s−1) can be determined directly from the “light curve” after measurements. Leaf net photosynthesis rate after heat stress treatments were also measured on the same day that plant samples were taken. All photosynthesis measurements were conducted from 8:30 a.m. to 12:00 a.m. and from 2:30 p.m. to 4:30 p.m., with the light level set at 1500 μmol m−2 s−1 and CO2 concentration maintained at a constant 380 ppm using a CO2 injecting system. Leaf temperature was kept at 25 °C during the measurements. The top two and three leaves were measured for each treatment with three replicates in the growing season 2010–2013 and growing season 2013–2014, respectively.

2.4. Model calibration We compared the simulations from the improved WheatGrow model and the original version to test the performances of the newly developed algorithms in simulating post-heading heat stress effects on wheat biomass growth. Independent observed data were used for model calibration and validation to obtain a reasonable model evaluation, respectively. Experiments in growing season 2010–2011 and 2011–2012 were used for calibration of genetic parameters for the two winter wheat cultivars, and the reminder of the experimental data were used for model validation (Table 1). The genetic parameters in the original and improved models were calibrated to best match the estimated wheat growth variables with observed data, respectively. 2.5. Sensitivity analysis Next, we performed tests on the sensitivity of the newly improved WheatGrow model to various post-heading simulated heat stress events. For this analysis, first we tested the responses of LAI to heat stress at different days after anthesis. Six days of simulated heat stress events with two temperature levels (25/35 °C and 29/39 °C) were applied to wheat post-anthesis period every four days. Then the responses of final aboveground biomass were investigated under heat stress at different days after anthesis. Simulated heat stress events with four temperature levels (25/35 °C and 29/39 °C) and three durations (3d, 6d, and 9d) were applied at post-anthesis period starting from the first day to the 30th day after anthesis during sensitivity analysis. In addition, responses of aboveground biomass were tested under different HTS within a range of 0–20. Six days of heat stress with two temperature levels (25/35 °C and 29/39 °C) were applied at anthesis and grain filling (10 days after anthesis) in the simulated heat stress during sensitivity analysis of HTS. Other genetic parameters in the sensitivity simulations came from Yangmai16. The daily temperature during the days without heat stress was fixed at 27/17 °C in the sensitivity analysis.

2.3. WheatGrow model In this study, based on the observed heat stress effects on the biomass growth related processes from environment-controlled phytotron experiments, we developed new algorithms to simulate heat stress effects wheat biomass growth, and integrated the algorithms into the WheatGrow model to improve the performance of WheatGrow model under heat stress. The WheatGrow model is a process-based wheat model, which can predict wheat phenology, photosynthesis and biomass production, biomass partitioning and organ establishment, and grain yield and quality formation under various environmental factors and management practices. The WheatGrow model has six submodels, including apical and phenological development (21-35Yan et al., 2000), photosynthesis and biomass accumulation (Liu et al., 2000b), biomass partitioning and organ establishment (Liu et al., 2000a), grain yield and quality formation (Cao et al. 2002), and soil water balance (Hu et al., 2004) and nitrogen dynamics (Zhuang et al., 2004). Evaluations of the WheatGrow model have been carried out in many winter wheat ecological regions across China (Lv et al., 2013; Zhao et al., 2010). Post-heading biomass production in wheat is defined by net growth between anthesis and maturity, which is determined by leaf photosynthesis duration, leaf area index, leaf photosynthesis rate, and plant respiration. In the recent version of WheatGrow (v3.1), a newly developed function for the effect of heat stress was incorporated into the phenology submodel to improve the predictions of post-heading duration under heat stress (Liu et al., 2016b). Here, we focus on the effects of heat stress on leaf area index, leaf photosynthesis rate, and plant respiration when simulating heat stress effects on wheat biomass production. According to observed impacts of heat stress on wheat biomass growth, we developed new algorithms to simulate heat stress effects on leaf area index (LAI), initial light use efficiency (ε), and leaf chlorophyll content. The new heat stress routines were incorporated into the original WheatGrow model to improve the simulation of biomass growth

2.6. Model validation To achieve a balanced picture of model performance, we employed four statistical indices for assessing and comparing model performances. They were: (1) root mean square error (RMSE); (2) the coefficient of determination (R2); (3) MSEs/MSE: Overall systematic error (MSEs) relative to total mean squared error (MSE); and (4) Index of agreement (IA). MSEs/MSE was taken to identify how much or what proportion of simulation errors was systematic in nature. It can have values within the range [0,1], and values close to 1 indicate the high proportion of systematic error in the total error. IA was used as a more general indicator of modelling efficiency and varies between 0 and 1. The values of IA closer to 1 indicate the better simulation quality. The calculation of above four statistical indices was the same as Rötter et al. (2012). 3. Observed heat stress effects on biomass growth of wheat Negative heat stress effects on several main biomass growth processes have been found, including leaf area index, leaf photosynthesis rate, leaf SPAD, aboveground biomass. With same experimental data set, we have found that heat stress at anthesis and during grain filling accelerated leaf senescence; the rate of leaf senescence acceleration increased with increasing heat stress (Liu et al., 2016a). 478

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 1. Dynamic of leaf net photosynthesis rate (Pn) of flag leaf and second leaf from the top after heat stress (a and b) at anthesis and (c and d) grain filling for cv.Yangmai16 in growing season 2013–2014. Duration of heat stress treatments at anthesis and grain filling were 6 days (D6). These treatments are indicated by the horizontal bar at the lower left-hand corner of each graph.

Fig. 2. Relationship between leaf net photosynthesis rate (Pn) and leaf chlorophyll concentration (SPAD) after heat stress treatments in (a) the flag and (b) the second leaves from the top in growing season 2013–2014.

Both flag and second leaves from the top have shown decreased photosynthesis rate under heat stress treatments (T3, T4, and T5) at anthesis and grain filling stages, compared with control treatments (T1) (Fig. 1). Generally, heat stress has more negative impacts on the second leaf from the top than the flag leaf, which means that lower leaves are more sensitive than upper leaves. Fig. 2 shows the significant positive

relationship between leaf net photosynthesis rate and leaf chlorophyll concentration (SPAD value) after heat stress treatments with the flag and the second leaves from the top. When investigating the relationship between heat stress and SPAD, we found that leaf SPAD decreased significantly with increased heat stress after heat stress treatments at both anthesis and grain filling stages (Fig. 3). Therefore, we concluded 479

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 3. Dynamics of leaf chlorophyll concentration (SPAD) after heat stress treatments at anthesis (a and b) and grain filling (c and d) in the flag and second leaves from the top in cv.Yangmai16 during growing season 2013–2014.

that the acceleration in the degradation of leaf chlorophyll due to heat stress contributed to lower leaf photosynthesis after heat stress. Finally, observed final aboveground biomass at maturity decreased significantly with heat stress at anthesis and grain filling (p < 0.01). And with same temperature level and duration, heat stress at anthesis led to more reductions in final aboveground biomass at maturity than heat stress at grain filling in both cultivars (Liu et al., 2016a).

in the calculation of leaf area index (LAI) to quantify the heat stress effects on LAI (Eq. (1)):

LAIsene, i = Wsene leaf , i × SLA × SHT , i

(1)

In Eq. (1), SLA is the specific leaf area during post-heading period, which is a genetic parameter. Wsene_leaf, i is the senesced leaf dry weight on day i under normal conditions, which is calculated with total aboveground biomass weight and partitioning index of leaf, as shown in Eq. (2).

4. Model description: simulating heat stress effects on biomass growth Based on observed impacts of heat stress on leaf area index (LAI) from pot experiments, leaf photosynthesis, new algorithms for simulating heat stress effects on LAI, initial light use efficiency (ε), and leaf chlorophyll content were developed. Also, the simulation results of leaf respiration under heat stress were tested against the measured data.

Wsene leaf , i = Wi × PILVGi − Wi − 1 × PILVGi − 1

(2)

PILVGi = 0.809/(1 + 51.4 × exp(−0.112 × (56 − PDTi )))

(3)

Where Wi is the total aboveground biomass on day i, and PILVGi is the partitioning index for leaf. PDTi is the physiological development time on day i, which is calculated in phenology submodel. According to Asseng et al. (2011), a linear relationship was used to quantify the heat stress effects on SHT, as shown in Eq. (4):

4.1. Heat stress effects on leaf area index (LAI)

SHT , i = 1 + HTS ×

Many previous studies have reported that heat stress can accelerate the senescence of green leaf area (Shah and Paulsen, 2003; Tewolde et al., 2006; van Herwaarden et al., 1998). According to the approach of previous simulation studies (Asseng et al., 2011; Stratonovitch and Semenov, 2015), we employed an accelerated leaf senescence factor SHT

AHDDi GDDRGP

(4)

In Eq. (4), GDDRGP represents the growing degree days (GDD) requirements for wheat reproductive growth phase (RGP) in wheat, which was an ecological parameter and set as 520 °Cd in winter wheat 480

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al. 3

according to previous analysis of post-heading duration with many winter wheat cultivars across different agro-ecological zones (Cao et al., 2008). AHDDi ( °Cd) is the accumulated daily heat degree days (HDD) on day i. A genetic parameter, high temperature sensitivity (HTS), was used to quantify the genetic differences in heat tolerance in Eq. (4), which has been suggested by Liu et al. (2016a). Heat stress was quantified with AHDD in Eq. (5):

DTAR =

i=1

∑

HDDj (5)

j=1

As we focus on post-heading heat stress, HDD (heat degree days) are calculated starting from the date of heading. Therefore, AHDDi (°Cd) is the accumulated daily HDD from the heading date to the ith date after heading. The HDDj (°C) is the accumulation of hourly temperature above the high temperature threshold (Th) on the jth date after heading, as in Eqs. (6)-(7):

HDDj =

1 24

HDt

In Eq. (10), ε is the initial light use efficiency (mol mol ), and varies with temperature (Eq. (12)). Amax is the maximum photosynthesis rate when photosynthetic active radiation (PAR) is saturation intensity (μmolCO2m−2s−1), and was affected by genetic and environmental factors (Eq. (13)). Il,a[i][j] is the intercepted PAR by the jth canopy layer at time point i (μmol m−2s−1), which was calculated based on the leaf area index and the instantaneous solar radiation at time point i. The exact time for each time point in WheatGrow model has daily variation during the growing season, depending on the day length, and was calculated as Eq. (11).

Tt < Th Tt ≥ Th

(6)

(7)

In Eq. (7), Th represents the high temperature threshold. As slightly different temperature threshold was used for different eco-type cultivars in previous studies (Farooq et al., 2011; Luo, 2011; Porter and Gawith, 1999), we introduced Th as an ecological parameter in the improved WheatGrow model, which means different values of Th for cultivars from different eco-regions. 30 °C has been chosen as the threshold temperature when heat stress starts during the reproductive phase of winter wheat in China (Deng et al., 2009). In addition, differences in measured leaf photosynthesis rate and biomass between T1 (17/27 °C) and T2 (21/31 °C) under heat stress at anthesis and grain filling were observed between our observations from pot experiments. Therefore, Th is set as 30 °C when post-heading heat stress starts in winter wheat. Tt is the hourly temperature, which is determined by daily maximum and minimum temperatures with a cosine function as in Eq. (8) (Matthews and Hunt, 1994). The cosine function, as well as other hourly temperature estimating methods, have been shown to be able to capture the temperature dynamic reliably as used in several crop models (Reicosky et al., 1989).

Tt =

(10)

= 1, 2, 3, 4, 5) −1

t=1

0 HDt = ⎧ Tt − Th ⎨ ⎩

(9)

P [i][j] = Amax × [1 − exp(−ε × Il, a [i][j]/ Amax )] (i=1, 2, 3. j

24

∑

P [i][j] × WGUSS [i] × WGUSS [j]

j=1

Where P[i][j] is the instantaneous photosynthesis rate of the jth canopy layer at time point i (μmolCO2m−2s−1). WGUSS[i] is the Gaussian integration weight for different time points, which was set as 0.28, 0.44 and 0.28 for time point 1, 2 and 3, respectively. WGUSS[j] is the Gaussian integration weight for different canopy layers, which was set as 0.12, 0.24, 0.28, 0.24, and 0.12 for the five canopy layers. Based on the maximum leaf photosynthesis rate and initial light use efficiency (ε), the asymptotic negative exponential model is used to calculate the leaf instantaneous photosynthesis rate at each time point for each canopy layer (Eq. (10)):

i

AHDDi =

5

∑∑

Tmax + Tmin Tmax − Tmin t −14 + × cos(π × ) 2 2 12

t[i] = 12 + 0.5 × DL × DIS [i] ( i= 1, 2, 3)

(11)

Where, DL is the day length (h), DIS is the Gaussian distance coefficient for each time point, which was set as 0.11, 0.50, and 0.89 for time point 1, 2, and 3, respectively. t[i] is the exact time for each time point i (h). 4.2.1. Heat stress effects on initial light use efficiency (ε) In the original model of WheatGrow, initial light use efficiency (ε) was a constant parameter, with a value of 0.05. However, a linear negative relationship between measured ε and temperature (from 16 to 44 °C) was found with our phytotron experiments in the growing season 2013–2014. In the light response curves (Fig. 4a), ε, known as the slopes of linear regression between leaf net photosynthesis rate (Pn) and photosynthetically active radiation (PAR) under low light intensity, decreased from 32 to 44 °C. From the perspective of photosynthetic physiology, ε mainly indicates the enzyme activity in photosynthetic processes, which can be affected by temperature (Ehleringer and Björkman, 1977). In previous studies (Ehleringer and Pearcy, 1983; Shi et al., 2004), decreased ε has been observed with increasing temperature after temperature exceeded optimal value in several C3 plants. Therefore, similar to previous studies (Ehleringer and Pearcy, 1983; Shi et al., 2004), a linear response function of ε to temperature was fitted from our observed data (Fig. 4b). No significant differences between two cultivars were found in the relationship between ε and temperature, so measured ε and temperature from two cultivars were pooled together. The new version of the WheatGrow model incorporates this function (Eq. (12)) into the photosynthesis and biomass accumulation submodel.

(8)

In this equation, t is the number of hours of a day (t = 1, 2… 24). Tmax and Tmin indicate the daily maximum and minimum temperatures (°C). 4.2. Heat stress effects on leaf photosynthesis Process-based leaf level photosynthesis or radiation use efficiency (RUE) models have been the two main approaches in simulating daily biomass production in crop models. Huge differences can be found in model complexity between these different approaches. For example, a RUE model can be simple, while a leaf-level photosynthesis model, such as the FvCB model, could be relatively complex due to numerous sub processes and parameters (Yin and Struik, 2009). In the photosynthesis and biomass accumulation submodel of WheatGrow, daily total dry matter was calculated from daily total gross assimilation of canopy and respiration consumption. The daily total canopy biomass assimilation rate (DTAR) is the weighted sum of leaf photosynthesis from five canopy layers at three time points (from noon to sunset) over the day. Gaussian integration weight was introduced to calculate daily canopy photosynthesis, by integrating the instantaneous photosynthetic rate at three time points over the day and integrating the five canopy layers into the whole canopy, as shown in Eq. (9).

ε (T ) = −0.001 × T + 0.0708 15oC ≤ T ≤ 45oC

(12)

4.2.2. Heat stress effects on maximum photosynthesis rate (Amax) In the original WheatGrow model, maximum photosynthesis rate (Amax) was determined by genetic and environmental factors, as shown in Eq. (13).

Amax = AMX × FCO2 × FA × FT × min (FN , WDF )

(13)

The AMX is the maximum CO2 assimilation rate, which is a genetic parameter. FCO2, FA, FT, FN, and WDF are the impacts of ambient CO2 481

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 4. (a) Measured light response curves of flag leaves with different temperatures and (b) Relationship between measured flag leaf initial light use efficiency and temperature for two winter wheat cultivars in growing season 2013–2014 in a phytotron.

quantify the effects of decreased chlorophyll concentration on posttreatment leaf photosynthesis. As shown in Eq. (15), the accumulated heat degree days (AHDD) was used to quantify the FHT:

concentration, leaf physiological age, temperature, nitrogen, and water conditions. The temperature factor FT is calculated as Eq. (14):

⎧ ⎪ sin ⎪ FT = ⎨ ⎪ ⎪ cos ⎩

0

(

T24H − Tb Tol − Tb

T24H < Tb or T24H > Tm ×

π 2

)

Tol ≤ T24H < Tou

1

(

T24H − Tou Tm − Tou

FHT = max(0,1 − HTS ×

Tb ≤ T24H ≤ Tol

×

π 2

)

Tou ≤ T24H ≤ Tm

AHDD ) GDDRGP

(15)

As similar with effects of heat stress on LAI, we used a genetic parameter HTS to quantify the cultivar differences in relative heat tolerance. AHDD was calculated with Eqs. (5)–(8). The maximum photosynthesis rate (Amax) in the new improved model was now determined as in Eq. (16):

(14)

In Eq. (14), T24H is the daily 24-h average temperature. Tb, Tol, Tou, and Tm are the base temperatures, the upper optimal temperatures, lower optimal temperatures, and maximum temperatures; these are set as 0 °C, 15 °C, 25 °C, and 45 °C in WheatGrow. As shown in Fig. 5, FT decreases to 0 with increasing temperature when daily temperatures exceed the upper optimal temperature. However, FT could only be used to quantify heat stress effects on leaf photosynthesis during heat stress treatments, with the combination of high temperature effects on initial light use efficiency (Eq. (12)). Another obvious negative impact of heat stress is the decline in leaf photosynthesis after heat stress treatments, which can be another factor contributing to the reduction in biomass growth. Several previous studies have suggested that photosynthesis rate after heat stress treatments decreased due to an acceleration of leaf chlorophyll degradation (Farooq et al., 2011; Zhao et al., 2007). To simulate the heat stress effects on the degradation of leaf chlorophyll, we incorporate another factor (FHT) in Eq. (13), which can

Amax = AMX × FCO2 × FA × FT × min (FN , WDF , FHT )

(16)

4.3. Respiration simulation under heat stress in WheatGrow model The biomass growth depends not only on leaf photosynthesis, but also on plant respiration. In the original WheatGrow model, the relationship between dark respiration rate of plant organs (R) and temperature (T) was quantified as Eq. (17): (T − To)/10 R = R (To) × Q10

(17)

R(To) is the dark respiration rate under standard reference temperature (To), and varies with different organs (Penning de Vries et al., 1989). To is set as 25 °C for wheat. Q10 is the respiration coefficient, set as 2.0. Hourly temperatures, estimated with Eq. (8), were used to calculate dark respiration. To test simulation of the dark respiration under heat stress with Eq. (17), we compared the simulated and observed dark respiration rate at leaf scale under different temperatures. Observed leaf respiration rate can be obtained as the photosynthesis rate when PAR was set to 0 in the light response curve. Relative good agreement can be found between simulated and observed dark respiration rate (Fig. 6). Therefore, no further algorithms were developed to improve the simulation of dark respiration under heat stress for the WheatGrow model. The newly developed algorithms (Eqs. (1)–(16)), including heat stress effects on LAI, initial light use efficiency (ε), and leaf chlorophyll content, were incorporated into WheatGrow model to improve the simulations of biomass growth under heat stress. 5. Results 5.1. Model validation The first two years of experimental data were used to calibrate the genetic parameters including high temperature sensitivity (HTS) for the

Fig. 5. Effects of daily 24-h average temperature on maximum leaf photosynthesis rate (FT).

482

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 8. For control treatment (T1), both the original and improved WheatGrow model can simulate final aboveground biomass pretty well. For the heat stress treatments (T3, T4, and T5), heat stress at anthesis and grain filling apparently reduced the growth in aboveground biomass. As temperature increased during heat stress, negative impact of heat stress on biomass growth increased. For example, aboveground biomass growth almost stopped immediately after 6 days of heat stress events at 33/43 °C (T5D6) at anthesis. Similar to the simulation of LAI, the original WheatGrow model obviously overestimated final aboveground biomass under heat stress in model calibration and validation (Figs. 8 and 9). Incorporating heat stress routines (Eqs. (1)–(16)) significantly improved the predictions of biomass growth. The biomass differences between different temperature treatments can be reproduced very well with the improved WheatGrow model (Fig. 9). The statistical evaluation in Table 3 also resulted into similar conclusions. Simulation errors from both the original and improved model were very small for treatments without heat stress (T1). The RMSE of aboveground biomass predictions under heat stress treatments with improved model for cv.Yangmai16 and cv.Xumai330 were 668 kg ha−1 and 552 kg ha−1, both decreased about 57%, compared with these from the original model. The R2 and IA for simulation of final aboveground biomass increased about 0.24 and 0.44 for cv.Yangmai16, 0.26 for cv.Xumai30, with the improved WheatGrow model. The overall systematic error (MSEs) relative to total mean squared error (MSE) decreased about 90% for both cultivars under heat stress treatments. In summary, the heat stress module significantly improved the prediction of the final aboveground biomass in WheatGrow under various heat stress treatments in phytotron experiments.

Fig. 6. Comparison of observed (symbols) and simulated (solid line) leaf dark respiration rate under different temperatures for two winter wheat cultivars in growing season 2013–2014 in a phytotron.

two winter wheat cultivars. The parameter HTS for cv.Yangmai16 and cv.Xumai30 were 2.6 and 3.5 in the improved WheatGrow model with observed data during growing seasons of 2010–2011 and 2011–2012, which suggested that cv.Xumai30 was slightly more heat tolerant than cv.Yangmai16. Similar results can be found in our previous regression analysis (Liu et al., 2016a). After model calibration, both the original and improved WheatGrow model were evaluated with observed datasets from the growing seasons of 2012–2013 and 2013–2014. Predictions of leaf area index, leaf biomass, and aboveground total biomass from the two versions of WheatGrow model were compared with observed data.

5.2. Sensitivity analysis After evaluating the improved WheatGrow model with datasets from our phytotron experiments, we conducted a sensitivity analysis to test the improved model under various simulated heat stress treatments. Fig. 10 shows the response of LAI to heat stress occurred at different days after anthesis. As heat stress occurs closer to maturity, fewer impacts of heat stress on LAI are observed. For example, duration of LAI degradation (from maximum LAI at anthesis to minimum LAI at maturity) was shortened by 8.0 days for a six days heat stress event with temperature of 29/39 °C occurred at the 5th day after anthesis, while it was only shortened by 3.0 days for heat stress occurring at the 25th day after anthesis. Also, heat stress with higher temperatures (29/ 39 °C) resulted in the faster decline of LAI, compared to the heat stress with 25/35 °C. For example, a heat stress of six days with temperatures of 25/35 °C and 29/39 °C occurred at the 5th day after anthesis decreased durations of LAI degradation by about 4.0 and 8.0 days, respectively. Fig. 11 shows the response of final aboveground biomass to heat stress occurred at different days after anthesis. Similar with the response of LAI, heat stress occurring at an earlier growth stage resulted in more reduction in aboveground biomass, which agrees with our observation that heat stress at anthesis reduced more aboveground biomass than heat stress during grain filling (Liu et al., 2016a). As an example, a six days heat stress event with temperature of 29/39 °C occurred at the first day after anthesis and the 10th day after anthesis lead to 19% and 9% reduction of final aboveground biomass, respectively. Higher temperature and longer duration of heat stress events resulted in more reductions in final aboveground biomass. Compared with temperatures of 25/35 °C, a six days heat stress event with temperature of 29/39 °C at anthesis decreased about 10% final aboveground biomass. Fig. 12 shows the response of final aboveground biomass to newly introduced cultivar parameter heat stress sensitivity (HTS). The improved WheatGrow model predicted a smaller final aboveground

5.1.1. Simulation of leaf biomass and leaf area index (LAI) Fig. 7 shows the comparison of simulated LAI dynamics and observed values with cv.Yangmai16 in growing season 2013–2014, as an example. The original WheatGrow model obviously overestimated LAI under heat stress at anthesis and grain filling. Compared with T1, the accelerated senescence under T3, T4, and T5 resulted in smaller LAI, while the original WheatGrow model simulated almost no obvious differences among the four temperature treatments. Fig. S2 and Fig. S3 provide a 1:1 comparison that helps show the overestimated LAI and leaf biomass with original model. After integrating the heat stress routines (Eqs. (1)–(16)), the improved model reasonably simulated LAI dynamics under different heat stress treatments. The statistical indices in Table 2 confirmed the significant improvement in the predictions of LAI and leaf biomass under heat stress treatments (T2, T3, T4, and T5). For example, the RMSE of LAI for cv.Yangmai16 and cv.Xumai30 with improved model decreased about 40% and 60% under heat stress treatments, compared to these statistics from the original model. The IA of LAI with improved model increased about 0.13 and 0.17 for cv.Yangmai16 and for cv.Xumai30, respectively. The overall systematic error (MSEs) relative to total mean squared error (MSE) under heat stress treatments decreased about 88% and 76% for cv.Yangmai16 and cv.Xumai30, respectively. Similar statistical results can be found with leaf biomass. The RMSE of leaf biomass for cv.Yangmai16 and cv.Xumai30 were 123.12 and 100.81 kg ha−1, decreased about 37% and 42%, respectively, compared to these measurements from the original model. Generally, the improved model gives more accurate predictions of LAI and leaf biomass under heat stress. 5.1.2. Simulation of aboveground total biomass The comparison of observed and simulated final aboveground biomass with the original and the improved WheatGrow was shown in

483

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 7. Comparison of simulated and observed leaf area index dynamics of cv.Yangmai16 with the original and the improved WheatGrow model under heat stress at anthesis (a and b) and grain filling (c and d) in growing season 2013–2014.

6. Discussion

biomass with increasing HTS, which agreed with the assumption in the heat stress equations that a larger HTS means a more sensitive cultivar to heat stress. When comparing the sensitivity analysis results of heat stress at anthesis and grain filling, the final aboveground biomass was more affected by heat stress at anthesis than at grain filling (Fig. 12). For example, a heat stress of six days with temperatures of 29/39 °C at anthesis and grain filling reduced final aboveground biomass by about 9.0% and 5.0% when HTS was set as 3.0.

Increasing heat stress events are projected to affect crop production (Lesk et al., 2016; Teixeira et al., 2013). Therefore, it is critical to take heat stress events into account when assessing climate change impacts on crop production (Barlow et al., 2015). Recent studies have indicated the need to improve crop models under heat stress (Barlow et al., 2015; Lobell et al., 2012; Rezaei et al., 2015). While other studies have investigated the effects of heat stress on biomass growth, photosynthesis

Table 2 Statistical indices in the validation of the original and the improved WheatGrow model in estimating leaf area index (LAI) and leaf biomass in environment-controlled phytotron experiments under different heat stress levels and durations. Variable

LAI

Cultivar

Yangmai16 Xumai30

Leaf biomass

Yangmai16 Xumai30

Treatment

Original model

Improved model

RMSE

MSES/MSE

IA

RMSE

MSES/MSE

IA

T1 T2,T3,T4,T5 T1 T2,T3, T4,T5

0.57 0.51 0.39 0.67

0.47 0.40 0.19 0.52

0.92 0.83 0.97 0.82

0.59 0.30 0.34 0.26

0.21 0.05 0.22 0.12

0.92 0.96 0.98 0.99

T1 T2,T3,T4,T5 T1 T2,T3,T4,T5

131.38 196.02 134.82 174.48

0.20 0.57 0.47 0.66

0.98 0.88 0.98 0.91

139.90 123.12 110.90 100.81

0.28 0.10 0.30 0.35

0.97 0.97 0.99 0.99

484

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 8. Comparison of simulated and observed final aboveground biomass with the original and the improved WheatGrow model in model calibration (a and b) and validation (c and d).

in this study. The relationship between ε and high temperature has been reported for other species in several studies (Ehleringer and Pearcy, 1983). By fitting the negative relationship with measured dataset into the WheatGrow model, the leaf photosynthesis simulation during the heat stress treatments was improved. The improved relationship between ε and high temperature stress here, provides a surrogate of the biochemical relationship between CO2 fixation rates and high temperatures used in more process based models (e.g., FvCB model) (Yin and Struik, 2009). Similar to the temperature factor (FT) in the simulation of leaf photosynthesis in the WheatGrow model, other crop models, such as APSIM-Wheat, simulate a reduced biomass growth during heat stress events (APSRU, 2014; Bernacchi et al., 2013). In our environmentcontrolled phytotron experiments, heat stress treatments have accelerated the degradation of chlorophyll in wheat leaves, and the lower chlorophyll concentration could result directly in a lower photosynthesis rate after the heat stress event (Fig. 4), consistent with other studies (Feng et al., 2014; Prasad et al., 2011; Zhao et al., 2007). As leaf chlorophyll is the main pigment for leaf photosynthesis in chloroplast, leaf chlorophyll concentration is critical for biomass production in plant leaves (Gitelson et al., 2003). Therefore, incorporating heat stress effects on leaf chlorophyll concentration improved the simulations of the photosynthesis rate during and after heat stress treatments. The decline in leaf area index (LAI) is another important response of a crop to heat stress (Shah and Paulsen, 2003; van Herwaarden et al., 1998). An acceleration of leaf area senescence under heat stress has

or LAI (Farooq et al., 2011; Feng et al., 2014; Salvucci and CraftsBrandner, 2004), the main focus of this study was to incorporate heat stress effects on wheat biomass growth into a wheat crop model, an aspect mostly ignored so far (Barlow et al., 2015; Rezaei et al., 2015). To improve WheatGrow model’s ability to predict wheat production in China and elsewhere in the world with increasing heat stress events under future climate change, we conducted four years environmentcontrol phytotron experiments and collected detailed observed wheat data under heat stress (Table 1). Instead of improving crop yields predictions by directly reducing final grain yields or harvest index under heat stress (Moriondo et al., 2011), we used a step-by-step approach to improve our WheatGrow model under heat stress by including heat stress effects on key crop growth processes, such as phenology development, LAI, biomass growth, and grain development. A previous study improved the simulation of phenology development under heat stress with the WheatGrow model (Liu et al., 2016b). Here, we improved the simulation of biomass growth, by adding and integrating a new reduction function for photosynthesis rate and leaf area index under heat stress (Eqs. (1)–(16)). In WheatGrow, a negative exponential routine was used to quantify leaf-level photosynthesis and then aggregated to whole canopy level biomass growth, which is less empirical than a RUE model but still less complex than the FvCB model. The responses of the two main variables in the negative exponential routine especially under heat stress, including the initial light use efficiency (ε) and the maximum photosynthesis rate, were re-calibrated with observed data under heat stress

485

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 9. Comparison of simulated and observed aboveground biomass dynamics of cv.Yangmai16 with the original and the improved WheatGrow model under heat stress at anthesis (a and b) and grain filling (c and d) in growing season 2013–2014.

Table 3 Statistical indices in the validation of the original and improved WheatGrow model in estimating final aboveground biomass in environment-controlled phytotron experiments with different heat stress treatments. Cultivar

Treatment

Original model

Improved model

RMSE

R2

MSES/MSE

IA

RMSE

R2

MSES/MSE

IA

Yangmai16

T1 T2, T3, T4, T5

329.06 1607.20

0.69 0.49

0.59 0.93

0.90 0.48

320.34 668.32

0.71 0.73

0.52 0.01

0.90 0.92

Xumai30

T1 T2, T3, T4, T5

374.69 1276.24

0.81 0.62

0.01 0.94

0.96 0.71

353.58 552.93

0.82 0.88

0.05 0.09

0.96 0.97

one of the most important strategies to study in adapting to heat stress under climate change (Gouache et al., 2012; Semenov et al., 2014; Stratonovitch and Semenov, 2015). Genetic diversity of heat tolerance has been found in previous studies, not only in the response of final grain yields, but also in key crop growth processes. For biomass growth processes, cultivars with increased heat stress tolerance can maintain higher photosynthesis rate than cultivars with more heat stress susceptibility through a better antioxidant defense system and more heat shock proteins (HSPs), which both could help to stabilize macro-molecules against heat-induced damage. Also a stay-green trait can help to maintain higher green leaf area and longer green leaf duration (Farooq et al., 2011). The canopy temperature, which directly determines a heat stress impact, can be another source for heat stress sensitivity. As suggested by Cossani and Reynolds (2012), cultivars with deeper roots can result in more water uptake and cooler canopy temperature and hence more tolerant to high air temperatures. For example, Feng et al.

been incorporated into some wheat crop models (Asseng et al., 2011; Stratonovitch and Semenov, 2015). Similarly, a heat stress factor based on a linear relationship between heat stress (heat degree days, HDD) and leaf senescence was integrated into WheatGrow. The model validation demonstrated that the improved WheatGrow model could reproduce several heat stress treatments and captured the higher sensitivity to heat stress at anthesis compared to the period of grain filling (Liu et al., 2016a). All three drivers (shortened grain filling duration, reduced LAI, reduced photosynthesis) to reduced biomass growth under heat stress, have been suggested in the literature to be potentially critical when describing heat stress impacts (Farooq et al., 2011), but their impacts on biomass growth vary with the size and the timing of the heat stress. Therefore, all three kinds of effects on biomass growth should be quantified to improve model performance under heat stress. Exploring genetic differences to heat stress has been suggested as

486

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 10. Response of simulated leaf area index (LAI) to simulated heat stress events starting at different days after anthesis with the improved WheatGrow model in a sensitivity analysis. Simulated heat stress events included two temperature levels (25/35 °C and 29/39 °C) and one duration (6d).

compared several kinds of approaches (e.g. empirical, and energy balance) in simulating canopy-air temperature differences and heat stress impacts on wheat. They found that the use of canopy temperature to drive heat stress effects did improve simulations compared to using only air temperature to a relatively minor extent. Models that simulated yield well under heat stress had varying skill in simulating canopy temperature. Some models had very poor simulations of canopy temperature but performed well in simulating grain yield. Their results highlighted the need to more systematically understand and model heat stress events in wheat. Also, the differences between canopy temperature and air temperature are mainly associated with plant water conditions, especially under drought conditions (Siebert et al., 2014). Water stress can increase canopy temperature and may exacerbate heat stress impacts on crops. In our improved model, the interaction between water stress and heat stress were not quantified (Eq. (16)). But full irrigation was applied in our experiments, which means that water stress had marginal effects on the estimation of heat stress impacts in our experiments. Therefore we used air temperature to quantify heat stress impacts for this study, and model evaluation has indicated the

(2014) have shown differences in photosynthetic characteristics with two different heat-resistant winter wheat cultivars, including photosynthesis rates, chlorophyll content, PSII function, and carboxylation activity. Similarly, in our study, cultivar Xumai30 has shown to be more heat tolerant than cultivar Yangmai16, as final aboveground biomass was less impacted in Xumai30 than Yangmai16 under heat stress (Liu et al., 2016a). Therefore, an empirical heat stress sensitivity parameter (HTS) has been used to quantify the sensitivity differences to heat stress (Shi et al., 2015; Stratonovitch and Semenov, 2015). The validation with observed data together with the sensitivity analysis demonstrated that the improved WheatGrow model can now simulate the response of two different cultivars to various heat stress events during post-heading periods, a critical prerequisite for applying crop models to assessing climate change impacts under increasing heat event scenarios in the future. Canopy temperature has been found to be directly related to heat stress impacts on crop biomass growth (Siebert et al., 2014). However, the estimation of canopy temperature depends on canopy energy balance, which is a relatively complex process. Webber et al. (2015)

Fig. 11. Response of simulated final aboveground biomass to simulated heat stress events starting at different days after anthesis with the improved WheatGrow model in sensitivity analysis. The simulated heat stress event included two temperature levels (25/35 °C and 29/39 °C) and three durations (3d, 6d, and 9d).

487

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Fig. 12. Response of simulated final aboveground biomass to different values of heat stress sensitivity parameter (HTS) with the improved WheatGrow model in sensitivity analysis. Simulated heat stress events included two temperature level (25/35 °C and 29/39 °C) and one duration (6d) and started at (a) anthesis and (b) 10 days after anthesis.

grain yield (RMSE for grain yield was reduced by 0.5 t ha−1 for two cultivars), but some overestimation of grain yields under heat stress still can occur even with the improved WheatGrow model (Fig. S4). The overestimation in some cases could be due to effects of heat stress on other key processes in yield formation, which were not considered in the model improvement, like the effect of heat stress during anthesis on grain number (Liu et al., 2016a; Rezaei et al., 2015). Additional studies should be conducted to assist in improving crop models in simulating heat stress effects during anthesis on grain number.

reasonable responses of improved WheatGrow model to heat stress under well-watered conditions. Future studies need to consider the interaction between heat stress and water stress for better quantifying heat stress and water stress impacts. In this study, we integrated a heat stress-response algorithms into the WheatGrow model and improved the performance of this model under post-heading heat stress. The new algorithms can be used in other wheat models to improve their accuracy under heat stress. For example, as suggested in our previous model evaluation study, the simulation of heat stress effects on wheat leaf senescence may substantially improve the biomass growth prediction for DSSAT-CERESWheat model (Liu et al., 2016a). For other crops, the same algorithms could be implemented to improve the simulation of heat stress responses, but this will likely require different, crop-specific parameters. The data we used to improve the WheatGrow model were from phytotron experiments. Although heat stress treatments were short episodes, a response of wheat growth to heat stress treatments in pot experiments in a phytotron can be different to field conditions (van Herwaarden et al., 1998). In addition, the impact of heat stress on leaf photosynthesis and biomass production can be altered by other environmental conditions, like water stress, which were not considered in our experiment. Pradhan et al. (2012) showed that the combined effect of drought and heat stress was more detrimental on leaf photosynthesis than individual stress. Studies suggested that drought and heat stress often occur simultaneously in the field (Pradhan et al., 2012; Semenov and Shewry, 2011). In this study, wheat crops which are exposed to very high temperature (such as 39 °C) might not be able to take up enough water even under well-watered conditions and experience additional water stress. Therefore, the yield reduction under these extreme high temperature treatments considers a combined heat and possible water stress in such a situation. Additional field observations are needed to test the improved model under field conditions. As biomass growth determines the final grain yield, improvements in biomass growth simulation have reduced the prediction error of

7. Conclusions Using measured data of leaf photosynthesis, biomass production, and leaf area index (LAI) dynamics from environment-controlled phytotron experiments, we found that heat stress at anthesis and during grain filling accelerated the measured degradation of leaf chlorophyll (SPAD) and resulted in a lower leaf photosynthesis rate and decreased final biomass growth. Observed impacts of heat stress on leaf photosynthesis and LAI were quantified and new algorithms were developed to simulate heat stress effects on LAI, initial light use efficiency (ε), and leaf chlorophyll content. The new heat stress routines incorporated into the WheatGrow model significantly improved the simulated growth dynamics under a range of heat stress treatments. The new WheatGrow model enables improved assessments of climate change impacts with increasing heat event scenarios. Acknowledgements This work was supported by the National High-Tech Research and Development Program of China (2013AA102404), the National Natural Science Foundation of China (31271616), the 111 project (B16026), the Priority Academic Program Development of Jiangsu Higher Education Institutions(PAPD), and the China Scholarship Council.

Appendix A See Table A1.

488

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

Table A1 Description of variables used in the improved WheatGrow model. Id

Abbreviation

Unit

Description

1 2 3 4 5 6 7

ε AHDD AMX Amax DIS DL DTAR

mol mol−1 °Cd μmolCO2m−2s−1 μmolCO2m−2s−1

8

FA

9 10

FHT FT

11 12

FCO2 FN

13

GDDRGP

°Cd

14 15

HDD HTS

°Cd

16

Il,a

μmol m−2s−1

17 18 19 20 21 22 23 24 25 26

LAI LAIsene P PDT PILVG Q10 R SLA SHT To

27 28 29

Th W Wsene_leaf

30

WDF

31

WGUSS

Leaf initial light use efficiency Accumulated daily HDD after heading maximum CO2 assimilation rate Actual CO2 assimilation rate Gaussian distance coefficient Day length Daily total canopy biomass assimilation rate Leaf physiological age factor on leaf photosynthesis Heat stress factor on leaf photosynthesis Temperature factor on leaf photosynthesis CO2 factor on leaf photosynthesis Leaf nitrogen factor on leaf photosynthesis Growing degree days requirements for wheat reproductive growth phase Heat degree days High temperature sensitivity, genetic parameter Intercepted photosynthetic active radiation Leaf area index Daily senesced leaf area index Instantaneous photosynthesis rate Physiological development time Partitioning index for leaf Respiration temperature coefficient Dark respiration rate of plant organs Specific leaf area Leaf senescence factor due to heat stress Standard reference temperature for dark respiration Threshold temperature for heat stress Total aboveground biomass Daily senesced leaf dry weight under normal conditions Water stress factor on leaf photosynthesis Gaussian integration weight

h KgCH2O ha−1d−1

μmolCO2m−2s−1

μmolCO2m−2s−1 10−4 m2 g−1 °C °C Kg ha−1 Kg ha−1

Appendix B. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.agrformet.2017.08.018.

characteristics of the wheat dry hot wind in northern China. J. Anhui Agric. Sci. 20, 107. Ehleringer, J., Björkman, O., 1977. Quantum yields for CO2 uptake in C3 and C4 plants dependence on temperature, CO2, and O2 concentration. Plant Physiol. 59 (1), 86–90. Ehleringer, J., Pearcy, R.W., 1983. Variation in quantum yield for CO2 uptake among C3 and C4 plants. Plant Physiol. 73 (3), 555–559. Farooq, M., Bramley, H., Palta, J.A., Siddique, K.H.M., 2011. Heat stress in wheat during reproductive and grain-filling phases. Crit. Rev. Plant Sci. 30 (6), 491–507. Feng, B., et al., 2014. Effect of heat stress on the photosynthetic characteristics in flag leaves at the grain-Filling stage of different heat-Resistant winter wheat varieties. J. Agron. Crop Sci. 200 (2), 143–155. Gitelson, A.A., Gritz, Y., Merzlyak, M.N., 2003. Relationships between leaf chlorophyll content and spectral reflectance and algorithms for non-destructive chlorophyll assessment in higher plant leaves. J. Plant Physiol. 160 (3), 271–282. Gouache, D., et al., 2012. Evaluating agronomic adaptation options to increasing heat stress under climate change during wheat grain filling in France. Eur. J. Agron. 39 (0), 62–70. Gourdji, S.M., Sibley, A.M., Lobell, D.B., 2013. Global crop exposure to critical high temperatures in the reproductive period: historical trends and future projections. Environ. Res. Lett. 8 (2), 024041. Hu, J., Cao, W., Luo, W., Zhu, Y., 2004. Quantification of water stress factor for crop growth simulation II. Model establishment and validation. Acta Agronomica Sinica 30 (5), 460–464 (in Chinese with English Abstract). IPCC, 2012. Managing the Risks of Extreme Events and Disasters to Advance Climate

References APSRU, 2014. The APSIM Wheat Module. https://www.apsim.info/Documentation/ Model,CropandSoil/CropModuleDocumentation/Wheat.aspx. Asseng, S., Foster, I., Turner, N.C., 2011. The impact of temperature variability on wheat yields. Global Change Biol. 17 (2), 997–1012. Asseng, S., et al., 2015. Rising temperatures reduce global wheat production. Nat. Clim. Change 5 (2), 143–147. Atkin, O.K., Tjoelker, M.G., 2003. Thermal acclimation and the dynamic response of plant respiration to temperature. Trends Plant Sci. 8 (7), 343–351. Barlow, K.M., Christy, B.P., O'Leary, G.J., Riffkin, P.A., Nuttall, J.G., 2015. Simulating the impact of extreme heat and frost events on wheat crop production: a review. Field Crop Res. 171, 109–119. Bernacchi, C.J., et al., 2013. Modelling C3 photosynthesis from the chloroplast to the ecosystem. Plant Cell Environ. 36 (9), 1641–1657. Cao, W., 2002. Simulating organ growth in wheat based on the organ-weight fraction concept. Plant Prod. Sci. 5 (3), 248–256. Cao, W., Zhu, Y., Tian, Y., Yao, X., Liu, X., 2008. Digital Farming Technology, 1. Science Press, Beijing (572 pp). Challinor, A.J., et al., 2014. A meta-analysis of crop yield under climate change and adaptation. Nat. Clim. Change 4 (4), 287–291. Cossani, C.M., Reynolds, M.P., 2012. Physiological traits for improving heat tolerance in wheat. Plant Physiol. 160 (4), 1710–1718. Deng, Z., Xu, J., Huang, L., Zhang, S., 2009. Research summary of the damage

489

Agricultural and Forest Meteorology 247 (2017) 476–490

B. Liu et al.

activation state of Rubisco as a limiting factor in photosynthesis. Physiol. Plant. 120 (2), 179–186. Semenov, M.A., Shewry, P.R., 2011. Modelling predicts that heat stress, not drought, will increase vulnerability of wheat in Europe. Sci. Rep. 1, 66. Semenov, M.A., Stratonovitch, P., Alghabari, F., Gooding, M.J., 2014. Adapting wheat in Europe for climate change. J. Cereal. Sci. 59 (3), 245–256. Shah, N.H., Paulsen, G.M., 2003. Interaction of drought and high temperature on photosynthesis and grain-filling of wheat. Plant Soil 257 (1), 219–226. Shi, P., Zhang, X., Zhong, Z., 2004. Response of apparent quantum yield of winter wheat to temperature and intracellular CO2 concentration under low atmospheric pressure in Tibetan plateau. Sci. China: Earth Sci. 48 (34), 182–188. Shi, P.H., et al., 2015. Modeling the effects of post-anthesis heat stress on rice phenology. Field Crop Res. 177, 26–36. Siebert, S., Ewert, F., Rezaei, E.E., Kage, H., Graß, R., 2014. Impact of heat stress on crop yield—on the importance of considering canopy temperature. Environ. Res. Lett. 9 (4), 044012. Stratonovitch, P., Semenov, M.A., 2015. Heat tolerance around flowering in wheat identified as a key trait for increased yield potential in Europe under climate change. J. Exp. Bot. 66 (12), 3599–3609. Tao, F.L., Zhang, Z., Zhang, S., Rotter, R.P., 2015. Heat stress impacts on wheat growth and yield were reduced in the Huang-Huai-Hai Plain of China in the past three decades. Eur. J. Agron. 71, 44–52. Tashiro, T., Wardlaw, I.F., 1990. The effect of high-temperature at different stages of ripening on grain set, grain weight and grain dimensions in the semi dwarf wheat banks. Ann. Bot. 65 (1), 51–61. Teixeira, E.I., Fischer, G., van Velthuizen, H., Walter, C., Ewert, F., 2013. Global hot-spots of heat stress on agricultural crops due to climate change. Agr. Forest Meteorol. 170 (0), 206–215. Tewolde, H., Fernandez, C.J., Erickson, C.A., 2006. Wheat cultivars adapted to postheading high temperature stress. J. Agron. Crop Sci. 192 (2), 111–120. Uddling, J., Gelang-Alfredsson, J., Piikki, K., Pleijel, H., 2007. Evaluating the relationship between leaf chlorophyll concentration and SPAD-502 chlorophyll meter readings. Photosynthesis Res. 91 (1), 37–46. van Herwaarden, A.F., Angus, J.F., Richards, R.A., Farquhar, G.D., 1998. ‘Haying-off', the negative grain yield response of dryland wheat to nitrogen fertiliser III. The influence of water deficit and heat shock. Aust. J. Agric. Res. 49 (7), 1095. Wang, X., Dinler, B.S., Vignjevic, M., Jacobsen, S., Wollenweber, B., 2015. Physiological and proteome studies of responses to heat stress during grain filling in contrasting wheat cultivars. Plant Sci.: Int. J. Exp. Plant Biol. 230, 33–50. Wardlaw, I., Dawson, I., Munibi, P., Fewster, R., 1989. The tolerance of wheat to high temperatures during reproductive growth. I: Survey procedures and general response patterns. Crop Pasture Sci. 40 (1), 1–13. Webber, H., et al., 2015. Canopy temperature for simulation of heat stress in irrigated wheat in a semi-arid environment: A multi-model comparison. Field Crop Res. 202, 21–35. Yan, M., Cao, W., Luo, W., Jiang, H., 2000. A mechanistic model of phasic and phenological development of wheat. I: Assumption and description of the model. J. Appl. Ecol. 11 (3), 355–359. Yin, X., Struik, P.C., 2009. C3 and C4 photosynthesis models: an overview from the perspective of crop modelling. NJAS − Wageningen J. Life Sci. 57 (1), 27–38. Zhao, H., Dai, T.B., Jing, Q., Jiang, D., Cao, W.X., 2007. Leaf senescence and grain filling affected by post-anthesis high temperatures in two different wheat cultivars. Plant Growth Regul. 51 (2), 149–158. Zhao, Y., Tang, L., Cao, W., Zhu, Y., 2010. Adaptability evaluation of a wheat growth model (WheatGrow). J. Triticeae Crops 30 (3), 443–448 (in Chinese with English abstract). Zhuang, H., Cao, W., Jiang, S., Wang, Z., 2004. Simulation on nitrogen uptake and partitioning in crops. Syst. Sci. Comp. Stud. Agric. 20 (1), 5–11.

Change Adaptation. Cambridge University Press, Cambridge, UK. Kaše, M., Čatský, J., 1984. Maintenance and growth components of dark respiration rate in leaves of C3 and C4 plants as affected by leaf temperature. Biol. Plant. 26 (6), 461–470. Lesk, C., Rowhani, P., Ramankutty, N., 2016. Influence of extreme weather disasters on global crop production. Nature 529 (7584), 84–87. Liu, T., et al., 2000a. Quantitative simulation on dry matter partitioning dynamic in wheat organs. J. Triticeae Crops 21 (1), 25–31 (in Chinese with English Abstract). Liu, T., Cao, W., Luo, W., Wang, S., Yin, J., 2000b. A simulation model of photosynthetic production and dry matter accumulation in wheat. J. Triticeae Crops 21 (3), 26–30 (in Chinese with English Abstract). Liu, B., et al., 2014. Post-heading heat stress and yield impact in winter wheat of China. Global Change Biol. 20 (2), 372–381. Liu, B., et al., 2016a. Testing the responses of four wheat crop models to heat stress at anthesis and grain filling. Global Change Biol. 22 (5), 1890–1903. Liu, B., et al., 2016b. Modelling the effects of heat stress on post-heading durations in wheat: a comparison of temperature response routines. Agr. Forest Meteorol. 222, 45–58. Lobell, D.B., Sibley, A., Ortiz-Monasterio, J.I., 2012. Extreme heat effects on wheat senescence in India. Nat. Clim. Change 2 (3), 186–189. Lobell, D.B., et al., 2015. The shifting influence of drought and heat stress for crops in northeast Australia. Global Change Biol. 21 (11), 4115–4127. Luo, Q., 2011. Temperature thresholds and crop production: a review. Clim. Change 109 (3), 583–598. Lv, Z.F., Liu, X.J., Cao, W.X., Zhu, Y., 2013. Climate change impacts on regional winter wheat production in main wheat production regions of China. Agr. Forest Meteorol. 171, 234–248. Matthews, R.B., Hunt, L.A., 1994. Gumcas − a model describing the growth of cassava (Manihot-Esculenta l crantz). Field Crop Res. 36 (1), 69–84. Moriondo, M., Giannakopoulos, C., Bindi, M., 2011. Climate change impact assessment: the role of climate extremes in crop yield simulation. Clim. Change 104 (3-4), 679–701. Penning de Vries, F.W.T., Jansen, D.M., Ten Berge, H.F.M., Bakema, A., 1989. Simulation of Ecophysiological Processes of Growth in Several Annual Crops Vol. 29. Simulation Monographs, Pudoc/IRRI, Wageningen/Los Baños, pp. 29. Porter, J.R., Gawith, M., 1999. Temperatures and the growth and development of wheat: a review. Eur. J. Agron. 10 (1), 23–36. Pradhan, G.P., Prasad, P.V.V., Fritz, A.K., Kirkham, M.B., Gill, B.S., 2012. Effects of drought and high temperature stress on synthetic hexaploid wheat. Funct. Plant Biol. 39 (3), 190–198. Prasad, P.V.V., Djanaguiraman, M., 2014. Response of floret fertility and individual grain weight of wheat to high temperature stress: sensitive stages and thresholds for temperature and duration. Funct. Plant Biol. 41 (12), 1261–1269. Prasad, P.V.V., Pisipati, S.R., Momcilovic, I., Ristic, Z., 2011. Independent and combined effects of high temperature and drought stress during grain filling on plant yield and chloroplast EF-Tu expression in spring wheat. J. Agron. Crop Sci. 197 (6), 430–441. Rötter, R.P., et al., 2012. Simulation of spring barley yield in different climatic zones of Northern and Central Europe: a comparison of nine crop models. Field Crop Res. 133, 23–36. Reicosky, D.C., Winkelman, L.J., Baker, J.M., Baker, D.G., 1989. Accuracy of hourly air temperatures calculated from daily minima and maxima. Agr. Forest Meteorol. 46 (3), 193–209. Rezaei, E.E., Webber, H., Gaiser, T., Naab, J., Ewert, F., 2015. Heat stress in cereals: mechanisms and modelling. Eur. J. Agron. 64 (0), 98–113. Rosenzweig, C., et al., 2014. Assessing agricultural risks of climate change in the 21 st century in a global gridded crop model intercomparison. Proc. Natl. Acad. Sci. U. S. A. 111 (9), 3268–3273. Salvucci, M.E., Crafts-Brandner, S.J., 2004. Inhibition of photosynthesis by heat stress: the

490