Wheat Yield Responses to Rising Temperature: Insights from North Indian Plains of India

Climate change and consequent variations in temperature pose a significant challenge for sustaining wheat 3 production systems globally. In this study, the potential impact of rising temperature on wheat yield in the north 4 Indian plains, India's major wheat growing region, was analyzed using panel data from the year 1981 to 2009. 5 This study deviates from the majority of the previous studies by including non-climatic factors in estimating the 6 impact of climate change. Two temperature measures were used for fitting the function, viz., Growing Season 7 Temperature (GST) and Terminal Stage Temperature (TST), to find out the differential impact of increased 8 temperature at various growth stages. Analysis revealed that there was a significant rise in both GST as well as 9 TST during the study period. The magnitude of the annual increment in TST was twice that of GST. Wheat 10 yield growth in the region was driven primarily by increased input resources such as fertilizer application and 11 technological development like improved varieties and management practices. Most importantly, the study 12 found that the extent of yield reduction was more significant for an increase in temperature at terminal crop 13 growth stages. The yield reduction due to unit increase in TST was estimated to be 2.26 % while rise in GST by 14 1 ◦ C resulted in yield reduction of 2.03%.

concern for future food security.

16
The potential impact of global warming on wheat yields has been investigated extensively using crop simulation

19
Further, missing values were imputed using a spline function for the respective states.

20
The data on climate variables were extracted from 1 x 1 degree daily gridded data obtained from the Indian

30
Wheat yield is adversely affected by terminal heat stress, especially under delayed sowing conditions 31 (Nagarajan 2005 Mann-Kendall statistic and Sen's slope estimator were used for examining the trends in average GST and 1 average TST. Non-parametric tests are preferred to conventional parametric methods because the former is more 2 powerful when probability distribution is skewed (Onoz and Bayazit 2003). Mann-Kendall statistic is a non-3 parametric way of identifying trends in time series data wherein the significance of trend is tested based on 4 normalized test statistic (Z-values). This test is commonly used to determine the direction of the trend and 5 whether it is increasing or decreasing based on signs of the Z values. A positive Z value indicates an increasing 6 trend, while a negative Z-value denotes a decreasing trend. The Mann-Kendall test statistic is given by equation Where Tj and Ti are climate variables in year 'j' and 'i' respectively given j>i.

11
If the number of observations is greater than 10, Mann-Kendall statistic assumed to follow a normal distribution 12 with variance equal to

14
We can use Z test to test the significance of the trend. The standard Z statistic, Zs is

16
If Zs>Ztablevalue, we reject the null hypothesis that no significant trend is present in the variable under 17 consideration.

18
The magnitude of underlying trend is quantified using Sen's slope estimator, wherein the slope is computed as

21
For a time series X within observations, there are possible N = n (n -1)/2 values of Q that can be calculated.

22
According to Sen's method, the overall estimator of slope is the median of Q's N values. The overall slope 23 estimator, Q* is thus: , N is even ………(6)

25
The Q * sign indicates the trend in data, while its value indicates the steepness of the trend.
Several methods have been employed in literature to estimate the economic impact of climate change on 4 agriculture, which can be broadly classified into following approaches viz., production function approach, data and provides an excellent structural form. But it fails to incorporate farmers' adaptation to minimize 7 harmful effects due to climate change, thereby results in biased estimates of climatic parameters. In Ricardian  We assumed wheat yield as a function of climate factors, inputs, production technology and land quality.

34
proportion of wheat in the total cultivated area (referred to as regional specialization hereon) is used as a proxy to capture the effect of variables like land quality, government support programs, etc., which facilitate wheat 36 cultivation in the region but tend to vary over time.
The wheat yield function is specified as follows, where represents wheat yield for state 'i' at time 't' (where time period t vary from 1980 to 2009),

3
represents the vector of inputs like human labor, power, seed, irrigation and other inputs, alludes to regional 4 specialization, is vector of climate variables, is the vector of regional dummies. Here, , , , , are 5 parameters to be estimated and is the error term. Two models, Model 1 and Model 2 were estimated 6 separately to quantify the impact of crop growing season temperature (GST) and terminal stage temperature 7 (TST) on wheat yield, respectively. By estimating two models, we tried to discern the marginal impact of GST 8 and TST on the productivity of wheat. Yield and input variables were transformed into logarithmic form to 9 reduce excess variations in values, whereas the regional specialization temperature variables were retained as   in GST was found to be highest for Rajasthan (0.053 0 C) followed by Haryana (0.051 0 C) and Punjab (0.048 0 C),

5
indicating that the North-Western regions of IGP are likely to be more vulnerable to heat stress. A significant 6 increasing trend was observed in TST in all selected states, while an annual increment in TST was found nearly 7 twice as that of GST.

19
The irrigation variable's coefficient was positive but not statistically significant, which is not in line with the 20 findings of previous studies. Zaveri and Lobell (2019), reported that wheat yield in India during 2000s was 13 21 percent higher than it would have been without irrigation, and it is to be noted that irrigation development in wheat cultivation has contributed significantly to reduce impact of heat sensitivity on yield, albeit the fact that 23 this effect is reducing in recent years. In our study, the data used for the analysis is of more recent years. In

32
The time trend was positive and significant, indicating that technological change positively contributed to yield 33 growth during the study period. Contrary to our expectation, regional specialization was found to be negatively

11
The values of the natural logarithm of Yield climate were plotted against the GST ( Figure 2) and TST (Figure 3).

12
The downward sloping trend lines indicate the negative impact of temperature on wheat yields.

13
( Insert Fig 3 here) 14 The optimum temperatures for wheat growth are ~22°C for vegetative development and 21°C for reproductive 15 development, while ~35.4°C is the maximum limit for grain filling (Porter and Gawith 1999

28
South Asia have been tabulated in Table 5.

29
Simulation models have been widely used to predict the potential impact of climate change on wheat yield in

32
2018). Therefore the predicted impact in the form of reduction in wheat yield by 2050 using our estimates would 33 be in a range of 3-6 percent, which concurs with simulation model results for the same period. However, some studies predicted that wheat yields are likely to increase due to a surge in CO2 concentration through the increase 1 in temperature above 3 0 C would cancel out potential gains (Lal et al. 1998).
2    al. 2011). We calculated the trend growth in wheat production in the five states following an exponential 1 growth model and about 2.43 percent per year. This turns out that the contribution of TFP in the output growth 2 is about 1.43 percent. This shows that the TFP growth in India's wheat production has occurred against the loss 3 caused by temperature rise, either as GST or TST. We compare the impact of temperature rise against the TFP 4 growth in wheat production in India. The annual production increment of wheat in the five states due to TFP growth turns out to be to the tune of 0.98 million tonnes at mean, whereas that of a one-unit increase in GST and

6
TST could turn out to be to the tune of 1.04 and 1.12 million tonnes, respectively.

19
This study tried to segregate the impact due to climatic and non-climatic factors on wheat yield. We also 20 analyzed two separate models, including growing season temperature (GST) and terminal stage temperature

21
(TST) to understand the extent of impact on yield due to rise in temperature at various stages of crop growth.
During the study period, there was a significant increase in GST and TST in major wheat-producing regions of 23 the IGP, whereas the latter's growth rate was almost twice the former. Annual increment in GST ranges from 24 0.021 0 C to 0.053 0 C while it ranges from 0.048 0 C to 0.087 0 C in case of TST. The relationship between wheat 25 yield and temperature (both GST and TST) was found to be negative and non-linear. The adverse effect of rising 26 temperature was more prominent in the terminal stages of wheat growth. As per our study, an increase in GST

29
The TFP growth contributes to about an increment of 0.98 million tonnes of wheat annually in the five IGP 30 states considered. But one unit increase in GST and TST could impact a production loss to the tune of 1.04 and 31 1.12 million tonnes, respectively. Therefore, the temperature rise depresses the TFP growth and, therefore, the 32 output growth in wheat in the selected IGP states. Thus, the undesirable effects of climate change on wheat efforts need to be made for developing adaptation strategies to moderate the impact of rising temperature.

Conflicts of Interest / Competing Interests :
3 The authors have no relevant financial or non-financial interests to disclose.

12
Availability of data and material:.

13
The datasets generated during and/or analysed during the current study are available from the corresponding 14 author on reasonable request.     Correlation between GST and Yield climate Figure 3 Correlation between TST and Yield climate Figure 4 State wise lowess plot of yield and seed application