Impact of increased temperature on spring wheat yield in northern China

Global warming has been reported to cause reductions in crop yields. However, it was suggested that warming temperature might benefit crop productivity in some cool areas at high latitude. In this study, we conducted a 17- year field experiment (2002– 2018) on spring wheat in Inner Mongolia. Temperature changes during each growth stage of spring wheat were investigated. Responses of spring wheat yield to temperature changes during the specific growing stages were evaluated. Average annual maximum temperature ( T max ) and minimum temperature ( T min ) significantly increased over the past 17 years. However, T max did not show obvious increase trend during spring wheat growing seasons ( p = 0.0672). Furthermore, T max also had no distinct change before or after anthesis. T min significantly increased during the whole growing season, as well as in pre- and post- anthesis stages. Correlation analysis indicated that T max in the entire growing season and post- anthesis did not affect spring wheat yield, but high T max during pre- anthesis can improve grain yield. The T min during the life cycle and pre- anthesis both had positive relationship with grain yield. Moreover, elevated temperature from seedling to stem elongation can benefit tiller formation and thus increasing spike number, which contributed to the significant yield increase ( p = 0.0093). Overall, climate warming affect spring wheat yield in cool area, and increasing temperature that was below the optimum temperature can benefit wheat productivity.


| INTRODUCTION
With climate change, global air temperature is predicted to increase by ~1.0-1.7°C by 2050 (IPCC, 2013). Increased heat stress generally reduces crops yield as temperature is a critical factor affecting crops growth and development (Abdelrahman et al., 2020;Edreira et al., 2014;Telfer et al., 2018;Wang et al., 2018). Many studies indicated that climatic warming resulted in the decline of wheat productivity (Liu et al., 2014;Lobell et al., 2011;Zhang et al., 2013;Zheng et al., 2017). A study reported that each 1.0°C increase in mean temperature resulted in a 3%-10% reduction in wheat yield in China (You et al., 2009), whereas another study predicted that a 1.0°C increase in global average temperature might reduce wheat yield by 4%-6% (Asseng et al., 2015). However, global warming occurring in the northern high latitude (>45°N) was argued to be beneficial to crop production (Meng et al., 2014;Xiao et al., 2008). As the northern regions experience more obvious warming temperature trend in recent decades compared to global temperature change, many crops are reaching the optimal level for maximum photosynthesis rates, thus increasing temperature may enhance crop productivity in these regions (Meng et al., 2014). Moreover, rising temperature could expand crop cultivation into regions currently constrained by cold temperature (Olesen et al., 2007). Therefore, global warming might benefit crop production in Inner Mongolia (average altitude 1000 m) as a cool farming area located in northern China.
Inner Mongolia is one of the most important regions for spring wheat production. Its diurnal temperature variation makes it appropriate for spring wheat production (Cao et al., 2009) and grain yield from the area is about one-third of total spring wheat production in China (Zhao et al., 2017). Moreover, spring wheat is the main food source, and fluctuations in its yield directly threaten food security of the region's population (Dong et al., 2018).
Simulation models have been employed to demonstrate the negative effects of high temperature on spring wheat productivity in the cool regions (Xiao et al., 2017;Zhao et al., 2017). However, simulating sometimes is prone to errors and results always have a level of uncertainty (Vitantonio-Mazzini et al., 2020). Inversely, other studies have shown that increased temperature in these regions improved wheat yield (He et al., 2020;Xiao et al., 2008). Although investigations on the impacts of warming temperature on wheat yield were conducted (Xiao et al., ,2008(Xiao et al., , , 2016, wheat yields were usually evaluated against average meteorological changes over the entire growing season rather than specific growth stages that were most sensitive to environmental limitations (Xiao et al., 2008(Xiao et al., , 2016. Therefore, it is important to investigate the effects of temperature change, especially in critical growing stages, on spring wheat productivity through a long-term field experiment in the target region. To better understand the effects of temperature change on spring wheat in the cool region, this study analyzed results of a 17-year field experiment without water deficit in Ordos, Inner Mongolia. This 17-year study aimed to (i) investigate significant trends in temperature variability during specific growing phases of spring wheat, (ii) evaluate the relationship between grain yield of spring wheat and yield components, and (iii) identify the responses of grain yield to temperature change at varied growing stages of spring wheat.

| Experimental site conditions
This 17-year field experiment was conducted from 2002 to 2018 at the Ordos Academy of Agriculture and Animal Husbandry Sciences (110.0°E, 40.5°N, altitude 1010 m). The soil in the experimental field is classified as light loam texture. The upper 20 cm soil profile contained 22.7 g kg −1 organic matter, 1.02 g kg −1 total N, 117.5 mg kg −1 available K, and 16.2 mg kg −1 available P. Averaged air maximum temperature (T max ) and minimum temperature (T min ) were 14.8°C and 1.1°C in the past 45 years, respectively (recording at 1.5 m height). The average annual solar radiation and rainfall were 5791 MJ m −2 and 310 mm during the past 45 years, respectively.

| Experimental design
Yongliang 4, a widely sown spring wheat cultivar in China, was planted at a seeding rate of 375 kg ha −1 around March 29 each year for 17 growing seasons (Table 1). During the whole growth period, irrigation with surface flooding was timely supplemented to avoid drought stress. This experiment plot area was 25 m 2 (5 m × 5 m) with three | 3 of 11 YE Et al.
replications. Before seed sowing each year, basal fertilizer was broadcasted at a per hectare rate of 180 kg nitrogen fertilizer (urea), 175 kg phosphate fertilizer (diammonium phosphate) and 125 kg potash fertilizer (potassium sulfate). There is no other fertilizer application during the wheat growing season.
Before harvest, spike number was counted from 3 onemeter wheat rows, and grain number per spike was evaluated by counting from 50 plants in each plot. In mid-July, spring wheat was harvested from a 2 m 2 area to determine grain yield. Final yield was calculated under 13% moisture content. Thousand kernel weight (TKW, dry weight) was measured from the harvested wheat plants.
Meteorological data (air temperature, rainfall, sunshine hours) in 2002-2018 were downloaded from the China Meteorological Data Sharing Service System (http://www. cdc.nmic.cn). The daily solar radiation (SR, MJ m −2 ) was calculated from sunshine hours following the methods in Gao et al., (2018). The climatic data during the 17-year spring wheat growing periods were calculated based on the sowing and harvest dates of field experiment from 2002 to 2018.

| Statistical analyses
Regression analysis of grain yield with climate factors, and yield components were conducted using SPSS 17.0 (SPSS Inc). All of the figures in the manuscript were constructed using SigmaPlot 12.5 (Systat Software Inc).

| Trend of temperature change during the spring wheat growing seasons
From 2002 to 2018, trends of annual maximum temperature (T max ), minimum temperature (T min ), and mean temperature (T mean ) levels were significantly rising ( Figure S1). Temperature data were monitored and recorded from the sowing date to the harvest date of spring wheat for 17 years of production seasons (Table 1). Trends for both T min and T mean were found increasing in this experiment, while T max did not apparently increase (p = 0.0672; Figure 1a). Analyses of temperature changes at varied growing stages of spring wheat revealed no obvious increase in T max either before or after anthesis (Figure 1b,c). Inversely, T min significantly increased at both pre-anthesis and post-anthesis. T mean significantly rose during pre-anthesis, but no distinct change at post-anthesis (p = 0.2408; Figure 1). Consequently, we mainly analyzed the temperature changes before anthesis. Results showed no obvious changes from seed sowing (Z00) to Z30. From Z30 to Z60, T min and T mean exhibited significant increasing trend (Table 2).

| Variability of spring wheat grain yield from 2002 to 2018
During the present field experiment, grain yield of spring wheat ranged from 4.4 t ha −1 to 7.6 t ha −1 . The mean grain yield was 5.8 t ha −1 in this experiment (Figure 2). The mean spike number per unit area was 594 per m 2 with a coefficient of variation (CV) of 14.0%. The grains per spike and thousand kernel weight (TKW) were 27.9 and 44.6 g with  a CV of 20.0% and 6.3%, respectively. Additionally, grain yield and grain number per spike significantly increased in this experiment, but spike number and grain weight did not show obvious change trends. Results of correlation analysis suggested spike number per square meter determined grain yield (p = 0.0093), where more spikes resulted in the higher yield levels. However, both TKW and grain number per spike had no significant correlations with grain yield (Figure 3). Figure 4 shows the relationship between grain yield and temperature data of the whole growing period. A significant relationship between T max and grain yield was not discovered. Interestingly, T min and T mean showed obvious increasing trends as grain yield increased, with a rate of 829.8 and 712.1 kg ha −1°C−1 , respectively. The relationships Year 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 M e a n Year 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 M e a n Thousand kernels weight (g) between grain yield and pre-anthesis/post-anthesis temperature were further conducted. The results indicated that post-anthesis high temperature did not reduce spring wheat yield, but pre-anthesis temperature had positive correlation with grain yield, that is, spring wheat yield increased 427.1-545.2 kg ha −1 with each 1°C warming ( Figure 5). Table 3 summarizes the effects of temperature change during each growth stage before anthesis on grain yield. From Z00 to Z09, temperature fluctuation did not affect grain yield of spring wheat. However, higher temperature and high diurnal temperature range benefited grain yield from seedling to tillering. During tillering to jointing, high T min and T mean also presented positive relationship with spring wheat yield. The temperature changes over tillering to anthesis had no obvious effect on grain yield. Additionally, effective tiller number showed positive correlation with grain yield (Figure S2). High temperatures from Z20 to Z30 had beneficial effects on tiller number (Figure 6).

| Temperature changes during different growing phases are diverse
At this experiment site, average annual temperature (T max , T min , T mean ) significantly increased over past years, which were consistent with previous research results (Gao et al., 2020;Meng et al., 2014;Zhao et al., 2017;Zhang et al., 2015). However, the annual T mean in this study was obviously lower than that of low latitudes (<10°C vs <15°C, Zhang et al., 2015). Global warming has a close connection with crop phenology (Chmielewski et al., 2004). Increases in temperature generally shorten growth periods of spring wheat in northern China (Xiao et al., 2016). However, mean growing season temperature usually showed a relatively weak correspondence with that of critical stage, emphasizing the importance of separate analyses for temperature changes (Gourdji et al., 2013). Xiao et al., (2016) showed that different growth stages (vegetative, reproductive, and whole growth period) presented similar temperature rise, but field-observed data for the three growth stages had great ranges of variation due to phenological changes from 1981 to 2009 in northern China. But our results indicated that temperature change trends in different growing phases were not coincident. Moreover, the previous research usually did not evaluate the temperature change in specific growing stages, especially for the critical stages that determined final yield (Grassini et al., 2009). Therefore, climate change (including temperature) analysis should pay attention to changes in different growing stages. The detailed analyses of each growing phase could accurately assess the impact of climate change and in turn lead to adjusted sowing date to avoid abiotic stresses (Gao et al., 2018;Gourdji et al., 2013). F I G U R E 4 Relationships between grain yield with temperatures of the whole growing period. T max , T min , and T mean indicate maximum, minimum, and mean temperature, respectively | 7 of 11 YE Et al.
F I G U R E 5 Regression analyses between grain yield and temperatures preanthesis (a) and post-anthesis (b) 8 of 11 | YE Et al.

| Temperature changes in different growth stages have diverse effects on spring wheat
Climate change, especially the global warming, can affect all stages of crop growth from seedling emergence to maturity, suppressing the yield of major cereal crops (Hussain et al., 2019). In recent study about spring wheat in western and middle Inner Mongolia (including this study's experiment site), spring wheat yield will likely reduce with the increase in temperature (Zhao et al., 2017). Additionally, He et al., (2020) also indicated that temperature rise shortened the whole growth duration and key growth phases, thus reducing grain yield of spring wheat in China. However, our long-term field experiment results were inconsistent with these studies. No significant relationship between spring wheat yield and T max during the whole growing season was detected; inversely, increased T min can increase wheat yield. Although the whole growth duration was shortened, the specific growth phases were not obviously reduced. Moreover, other studies found that warming could benefit wheat production in some places in China, where the mean growing season temperature is low and water supply is not limiting (Fang et al. 2015;Zhao et al. 2016). For the whole wheat plant growth, the optimum temperature ranges from 15°C to 30°C (Porter and Gawith, 1999). In this experiment, the average T max during the entire life cycle and before anthesis was below 25°C. The average T mean during growing season ranged from 15°C to 20°C, which was suitable for wheat. T min and T mean pre-anthesis was 4-8°C and 10-16°C, respectively, which was below the optimum growth temperature. Accordingly, temperature rise contributed to increase spring wheat yield. He et al., (2020) also showed that average temperature increase from 8°C to 13°C can improve wheat yield, but the temperature increase from 14°C to 17°C resulted in yield loss. Unexpectedly, high temperature stress after anthesis did not reduce grain yield, even though the average T max during grain filling stage was 29.8°C over the 17 seasons. In some years, the average maximum temperature was over 31°C. It exceeded optimal temperatures for wheat growth (Porter & Gawith, 1999). During grain filling stage, high temperature was reported to lower activities of antioxidant enzymes and photosynthesis and significantly resulted in wheat yield decline (Fan et al., 2018), which is against our results.
The main reason might be that the combined drought and heat usually resulted in greater yield losses than either stress alone (Cairns et al., 2013). Drought stress was avoided by supplemental irrigation in this experiment. Moreover, irrigation not only fulfilled crops water demand but also mitigated crop heat stress (Li et al., 2020;Tack et al., 2017). We therefore suggest that the major limiting factor of yield in spring wheat in Inner Mongolia area may be drought stress instead of heat. In addition, the temperature during past 17 seasons showed that minimum temperature had higher increase rate with a slope of 0.1147, the increase slope of maximum temperature was only 0.0711, that is, T min had a faster increase rate than T max . Previous research also presented same change trends (Peng et al., 2004). Furthermore, numerous researches suggested that T min reduced crop yield (García et al., 2015;Mohammed & Tarpley, 2009a;Mohammed & Tarpley, 2009b;Peng et al., 2004). Our results showed that rising T min increased spring wheat yield (Figure 4), and both pre-anthesis T max and T min showed positive correlation with grain yield. This suggested increase of temperature (below the optimum temperature) in cool condition benefited spring wheat production.
In Inner Mongolia, the long seedling period and short panicle differentiation stage restricted the productivity of spring wheat (Dong et al., 2019). High temperature usually accelerated wheat development resulted in a shorter key stage duration (He et al., 2020), thus decreasing photosynthetically active radiation capture with negative results for biomass accumulation and final yield (García et al., 2015). However, this experiment indicated that increased temperature during seedling to jointing had beneficial effects on effective tiller number and final yield owing to improvement of temperature below optimized level. Additionally, our results showed that spike number variation contributed largely to yield fluctuation (p = 0.0093). We concluded that increasing temperature (below the optimum temperature) had positive effects on tillering differentiation in spring wheat, which contributed to grain yield increase of spring wheat.

| CONCLUSIONS
In this long-term field experiment, annual average T max and T min significantly increased, but during spring wheat growing T A B L E 3 Correlation analysis between grain yield and temperature in each growing stage | 9 of 11 YE Et al.
season T max did not show obvious increasing trend. Before anthesis, T max and T min significantly increased but T max did not increase after anthesis. The T max during the whole growing season did not affect spring wheat yield, but increased T min benefited wheat yield. High temperature before anthesis showed positive relationship with spring wheat yield. Moreover, the increasing temperature during seedling to stem elongation, contributed much to final yield owing to improved spike formation.