Global warming is an indisputable reality that will continue into the foreseeable future (Qin et al., 2007; Zhou et al., 2016; Sun et al., 2017). The global surface temperature increased linearly by 0.85°C from 1880 to 2012 and is expected to increase by 1.0–3.7°C by the end of the 21st century (IPCC, 2014), with high-temperature events increasing globally (Betts et al., 2018). In the future, global warming will lead to the multiple and frequent occurrences of extreme meteorological disasters (Battisti and Naylor, 2009; Tramblay et al., 2012). Extremely high temperatures may also lead to large fluctuations in global and Chinese agricultural production (Lesk et al., 2016), and climate change puts global agricultural production, and even food security, at risk (Granderson, 2014; Hiwasaki et al., 2014; Jiao et al., 2017). China, which is a major global agricultural producer, is affected by a complex and variable climate, including its monsoon climate. Thus, China is one of the countries, which are most greatly affected by meteorological disasters (Zuo et al., 2016; He et al., 2017). The climate change that experienced in China is basically consistent with the trends in the global climate change, but its warming rate is greater than the global average (Wang et al., 2015; Li et al., 2017). In the future, China’s warming rate is predicted to reach 0.09–0.61°C (10 yr)^–1 (Taylor et al., 2012; Liu et al., 2014), and extreme high-temperature events, such as heat waves, may also increase (Dong et al., 2014; Liu J. et al., 2018; Zhan et al., 2018).

Rice is an important food crop in China, which plays an important role in the world rice production. The rice cultivation area in China accounts for ~18.5% of the global rice-cropping area, and the total output ranks first in the world, accounting for ~27.7% (Zhang et al., 2005; Zhu et al., 2015). According to the rice cultivation system, Chinese rice can be divided into single- and double-cropping rice, the cultivation areas of which are determined by agroclimatic resources (Duan and Zhou, 2012). The middle and lower reaches of the Yangtze River represent the largest rice-growing belt in China, with a rice planting area of ~20 million hm², which accounts for ~67% of the national yield (Meng et al., 2016). Statistics have shown that the planting area of single-cropping rice has continuously increased since the mid-1970s, with an adjustment to the planting structure, reaching more than 40% of the total rice-planted area (Wang, 2011). Under subtropical high-temperature conditions, the middle and lower reaches of the Yangtze River are prone to the persistent hot weather from July to August and have experienced strong, wide-ranging extreme high-temperatures lasting more than 10 days every year since 1999 (Liu X. C. et al., 2018). The duration of the high-temperature periods has increased, especially in plain areas (Yang et al., 2016; Meng et al., 2016). Usually, single-cropping rice in the middle and lower reaches of the Yangtze River is in its earliest flowering period in July and August, and it is most sensitive to high temperatures during this time. Continuous maximum daily temperatures exceeding 35°C result in decreased rice production and reduced quality (Hou et al., 2015). Previous studies mostly assessed the effects of heat-damage to single-cropping rice in the past, and there are currently few reports of future projections based on coupled climate models. Therefore, the projection of heat injury to single-cropping rice production in the middle and lower reaches of the Yangtze River, China, under climate warming scenarios, is of great importance for the early adoption of heat injury defense measures to protect rice yield and quality.

The development of numerical models provides a means for studying the future climate change and resulting extreme climate events (Betts et al., 2018; Tokarska et al., 2018). The latest Coupled Mode Intercomparison Phase 5 (CMIP5) earth system model has a high resolution and provides a global climate change database of nearly 30 climate models for current simulations and future projections. It has been widely applied to evaluate and estimate temperature, precipitation, sea surface temperature, wind speed, and other climatic factors (Han et al., 2015; Palerme et al., 2017; Lin et al., 2018). This study will project and reveal the occurrence probability of heat injury to single-cropping rice in China and its spatial distribution in the early, middle, and late 21st century by using CMIP5-coupled climate models under different Representative Concentration Pathway (RCP) emission scenarios (RCP2.6, RCP4.5, and RCP8.5) as well as observational data from agrometeorological stations. These findings will provide a scientific basis to enhance the disaster prevention and reduction and manage the single-cropping rice cultivation system.

Sixteen global climate models were used in this study, for which the maximum daily temperature data from CMIP5 for 1861–2099 were simultaneously provided in historical and RCP2.6, RCP4.5, and RCP8.5 scenario simulations. Owing to the different spatial resolutions of the output variables of the global climate models (Table 1), the model data were interpolated to a resolution of 0.1° × 0.1° by using a bilinear interpolation method. To project the heat injury to single-cropping rice, we used 1986–2005 (P0) as a reference period for the current climate, which is consistent with the definition of IPCC AR5, and divided the 21st century into three periods: P1, the early period (2016–35); P2, the middle period (2046–65); and P3, the late period (2080–99).

Heat injury to single-cropping rice in China typically occurs in the middle and lower reaches of the Yangtze River. The study region was thus located in the middle and lower reaches of the Yangtze River, including the entire territories of Jiangsu, Hubei, and Zhejiang, central and southern Anhui, northwestern Hunan, and northeastern Jiangxi provinces (Fig. 1). Based on site information from the national agrometeorological observation stations, we extracted the phenological data on single-cropping rice from the past five years and then calculated the average daily sequence of phenological phases at each station. Spatial interpolation was performed by inverse-distance weighting, and latitudinal and longitudinal information from future climate scenario data were used to extract the daily sequence of crop developmental phases.

Figure 1 Planting areas of single-cropping rice in the middle and lower reaches of the Yangtze River, China.

Figure options

Heat injury of single-cropping rice in the middle and lower reaches of the Yangtze River mainly occurred at the heading–flowering stage. High temperature at the heading–flowering stage affected the cracking rate and pollen fertility of anthers, which in turn reduced the seed-setting rate. Based on previous research and the actual conditions in the middle and lower reaches of the Yangtze River, a maximum daily temperature of ≥ 35.0°C was used as the index of heat injury at the heading–flowering stage of single-cropping rice, and its duration (day) was used to determine the level of heat injury (Table 2) (General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China and Standardization Administration of China, 2008; Zhang et al., 2011). When performing heat injury grading, we exclude mild from moderate and exclude moderate from severe.

When projecting the heat injury of single-cropping rice, the occurrence probability of mild heat injury was based on the number of occurrences $ F_m^n$ of mild heat injury under the n-th global climate model in the m-th year at a certain grid. The total number of occurences of mild heat injury that occurred in a certain period (M years) of the grid was defined as below:

The total number of occurrences of mild heat injury that occurred in the N climate models during a certain period of the grid was calculated by

The occurrence probability of mild heat injury during a certain period at a grid (P_mil) was expressed as

where M is the research period [with P0, P1, P2, and P3 referring to the reference (1986–2005), early (2016–35), middle (2046–65) and late (2080–99) periods respectively, and M = 20]; N is the number of climate models used.

The occurrence probability of moderate (P_mod) and severe (P_sev) heat injury during a certain period at a grid-based location was calculated as P_mil. Thus, the occurrence probability of total heat injury in single-cropping rice during a certain period at a grid-based location was calculated as following:

where P_total, P_mil, P_mod, and P_sev indicate the occurrence probability of the total, mild, moderate and severe heat-damage levels in single-cropping rice during a certain period at a grid-based location.

Single-cropping rice in the middle and lower reaches of the Yangtze River was often affected by heat injury at the heading and flowering stages, which resulted in a significant decline in yield. The changes in the occurrence probability of heat injury to single-cropping rice are shown in Fig. 2. In the reference period P0 (1986–2005), the occurrence probability of different levels of heat injury to single-cropping rice was ranked as mild > moderate > severe, with the average occurrence probabilities of 11.3%, 6.1%, and 5.2%, respectively. Under RCP2.6, RCP4.5, and RCP8.5 scenarios, the occurrence probabilities of heat injury to single-cropping rice showed increasing trends in the early, middle, and late periods of the 21st century, which were greater than that in P0. The occurrence probability of heat injury increased as emission scenarios increased during the same period. In particular, the occurrence probability of heat injury reached its highest levels, with an average of ~48% in the late 21st century under the RCP8.5 scenario.

Figure 2 Occurrence probability of heat injury to single-cropping rice under different RCP climate scenarios (RCP2.6, RCP4.5, and RCP8.5). The inset plot in the upper left corner shows the occurrence probability for the reference period (P0, 1986–2005). Mil, Mod, Sev, and Total indicate the mild, moderate, severe, and total occurrence probability of heat injury, respectively; P1: early period (2016–35); P2: middle period (2046–65); P3: late period (2080–99). Black lines from top to bottom of box-whisker plot indicate the occurrence probabilities of 95%, 75%, 50%, 25%, and 5% quantiles; black dots indicate the average values.

Figure options

The spatial distribution of the occurrence probability of total heat injury to single-cropping rice under different climate scenarios is shown in Fig. 3. In P0, the occurrence probability of heat injury was high in the middle planting area and low in the east (Fig. 3). The average occurrence probabilities of heat injury in northern Jiangsu, central Anhui, and southeastern Hubei provinces were 30%–50%, while the probabilities in the other regions were less than 30%. Compared with P0, the occurrence probabilities of heat injury all increased, but the distribution patterns did not change under the RCP2.6, RCP4.5, and RCP8.5 scenarios. In addition, with the increasing emission levels and time, the distribution range of the occurrence probability of heat injury to single-cropping rice exceeded 50%. In particular, in the late 21st century under the RCP8.5 scenario, the occurrence probability of heat injury to single-cropping rice in the planting area (in addition to Zhejiang and northern Jiangxi provinces) was greater than 50%.

Figure 3 Spatial distributions of the occurrence probability of total heat injury to single-cropping rice under RCP climate scenarios (RCP2.6, RCP4.5, and RCP8.5). P0: the reference period (1986–2005); P1: early period (2016–35); P2: middle period (2046–65); P3: late period (2080–99).

Figure options

The spatial distributions of the occurrence probabilities of mild (Fig. 4), moderate (Fig. 5), and severe (Fig. 6) heat injury to single-cropping rice show that the mild level of heat injury to single-cropping rice was the most prevalent, followed by the moderate and severe levels during P0. The probabilities of mild and moderate heat injury were both less than 20%, whereas that of severe heat injury in central Anhui and southern Hubei provinces was greater than 20%. In the early, middle, and late 21st century, under the RCP scenarios, the distribution of the occurrence probability of heat injury was mainly at the mild level, with low occurrences of moderate and severe heat injury levels. The overall distribution pattern was high in the middle planting area and low in the east. In addition, the spatial distribution range of the high occurrence probability of heat injury at each level increased with time and increasing emissions in the late 21st century.

Figure 4 As in Fig. 3, but for the mild heat injury .

Figure options

Figure 5 As in Fig. 3, but for the moderate heat injury.

Figure options

Figure 6 As in Fig. 3, but for the severe heat injury.

Figure options

According to the projections of A_heat to single-cropping rice in the 21st century under the RCP scenarios by using a linear regression method (Fig. 7), the occurrence areas of mild (A_mil), moderate (A_mod), severe (A_sev), and total (A_tol) heat injury showed significant linear increases over time during the 21st century, and A_heat values at each level were the greatest under the RCP8.5 scenario, followed by the RCP4.5 and RCP2.6 scenarios. In P0, the A_heat values at different levels were ranked as: mild > moderate > severe, accounting for 6.6%, 4.3%, and 3.6%, respectively, of the total planting areas in the middle and lower reaches of the Yangtze River ( Table 3). In the early, middle, and late 21st century under the RCP scenarios, A_heat was still ranked as mild > moderate > severe ( Table 3), but there were differences in the linear increase rates among the different levels of heat injury over time (Fig. 7). Under the RCP2.6 scenario, the increase rates of A_mil, A_mod, and A_sev to single-cropping rice were 2.93 × 10³ (R² = 0.82), 2.44 × 10³ (R² = 0.89), and 2.04 × 10³ (R² = 0.59) ha yr^–1, respectively. Under the RCP4.5 scenario, there was no significant difference in the increase rates of A_mil and A_sev (~6.9 × 10³ ha yr^–1), and the rates were slightly lower when a moderate level of heat injury occurred. Under the RCP8.5 scenario, the increase rate of A_sev was the greatest, at 15.02 × 10³ ha yr^–1 (R² = 0.87), and those of A_mil and A_mod were ~10 × 10³ ha yr^–1.

Figure 7 Occurrence area of different levels of heat injury to single-cropping rice under RCP climate scenarios (RCP2.6, RCP4.5, and RCP8.5) for (a) mild, (b) moderate, (c) severe, and (d) total heat injury.

Figure options

The variable A_tol to single-cropping rice was significantly larger than that during the reference period under the RCP2.6, RCP4.5, and RCP8.5 scenarios, and ranked as RCP8.5 > RCP4.5 > RCP 2.6 in all the three periods of the 21st century ( Fig. 7d). A_tol showed a significant linear increase over time during the 21st century, and the increase rates of A_tol were 35.33 × 10³ (R² = 0.90), 19.89 × 10³ (R² = 0.94), and 7.41 × 10³ (R² = 0.78) ha yr^–1 under the RCP8.5, RCP4.5, and RCP 2.6 scenarios, respectively. Thus, A_tol reached its maximum value in the late 21st century, accounting for about 24.5% (RCP2.6), 39.9% (RCP4.5), and 58.6% (RCP8.5) of the total planting areas (Table 3).

The average occurrence probability of each level (mild, moderate, and severe) of heat injury to single-cropping rice in the 21st century was less than 50%, but the average occurrence probability of total heat injury was approximately 19%–53% under the RCP scenarios. Here, incidents with P_total ≥ 50% were defined as high-probability events, fewer of which occurred in single-cropping rice during the early 21st century, but increased significantly in the middle and late 21st century (Fig. 8). The maximum A_tol value occurred in the late 21st century under the RCP8.5 scenario and involved ~ 4.3 × 10⁶ ha, which constituted ~ 55% of the planting area (Fig. 8).

Figure 8 Areas with heat injury occurrence probability ≥ 50% and their proportion of the main planting area in single-cropping rice under different RCP climate scenarios (RCP2.6, RCP4.5, and RCP8.5). Notes see Fig. 3.

Figure options

Heat injury of single-cropping rice mainly occurs in the middle and lower reaches of the Yangtze River, China, from July to August when single-cropping rice is in the reproductive growth stage. Heat injury at this stage leads to a reduced seed setting rate, which seriously affects the production and quality of the rice crop. This study projected and analyzed the future occurrence probability, spatial distribution pattern, and changes in the occurrence area of different levels of heat injury to single-cropping rice in China based on CMIP5-coupled climate models under different emission scenarios (RCP2.6, RCP4.5, and RCP8.5).

In P0 (1986–2005), the average occurrence probability of heat injury to single-cropping rice in the middle and lower reaches of the Yangtze River was 20%, with the highest probability of mild, followed by moderate and then severe heat injury. This was consistent with previous findings of heat injury to single-cropping rice based on the observational data of meteorological and agrometeorological stations from 1961 to 2012 in the middle and lower reaches of the Yangtze River (Yang et al., 2016). Thus, the results of multi-model collection produced highly accurate predictions of heat injury to single-cropping rice in China (Xu et al., 2010). Throughout the 21st century, the occurrence probability of heat injury to single-cropping rice was significantly greater than that during P0 under the RCP scenarios, and it increased as the emissions increased over time. In particular, during the late 21st century under the RCP8.5 scenario, the average and maximum occurrence probabilities were ~ 48% and 80%, respectively. Based on the grades of heat injury and its corresponding reference (yield reduction rate), the severe heat injury was calculated to correspond to a yield reduction rate of > 15%. For single-cropping rice in China, the increase rates of the occurrence area affected by severe heat injury, which may result in the yield reduction rate of over 15%, were about 2.04 × 10 ³, 6.9 × 10³, and 15.02 × 10³ ha yr^–1 in 2006–99 under the RCP2.6, RCP4.5, and RCP8.5 scenarios, respectively (Fig. 7). Consequently, there is a high future risk of heat injury to single-cropping rice in the middle and lower reaches of Yangtze River, which may seriously impact the rice yield. The reduction and prevention of heat injury to single-cropping rice in this region are therefore still important for the future rice production (Xiong et al., 2016). In this study, heat injury of the single-cropping rice mainly occurred from late July to mid-August (heading–flowering stage). Moreover, an advanced planting date prolongs the vegetative growth duration and delays the reproductive growth phrase, which may lead to a shift towards higher temperatures for the heading–maturity phase and effectively avoid heat injury (Xie et al., 2013; Hu et al., 2017). In addition, screening and cultivation of high-temperature rice cultivars are also effective methods to deal with the changes of heat injury (Xie et al., 2013). Previous studies have shown that when the planting date remained the same, a switch to new cultivars could also prolong the growth duration for single-cropping rice. The prolongation of the growth duration associated with cultivar renewal not only compensates for the shortening of the growth duration induced by climate warming, but may also make full use of the improvement of local heat resources in single-cropping rice production (Tao et al., 2013; Hu et al., 2017). Hence, advances in the planting date of single-cropping rice and high-temperature-tolerant cultivars should be developed and popularized to deal with the increasing heat injury risk in the middle and lower reaches of the Yangtze River, China (Tao et al., 2013; Tao and Zhang, 2013; Guo et al., 2018).

The study of the distribution of heat injury to rice in China based on the observational data for 1961–2000 showed that the occurrence probability of heat injury was the greatest in the central part of in the Yangtze River basin, which is a major planting area of single-cropping rice (Xiong et al., 2016). In addition, the increased extreme high-temperature stress under the climate warming would increase the probability of single-cropping rice yield decrease in the Yangtze River basin (Tao and Zhang, 2013), which is consistent with our results. Compared with P0 (1986–2005), the spatial distribution patterns of the occurrence probabilities of mild, moderate, and severe heat injury in the middle and lower reaches of the Yangtze River were also high in the central planting area and low in the east. The high A_heat values expanded in the northwest and northeast directions over time under the RCP scenarios. Moreover, the areas with P_total ≥ 50% increased significantly in the middle and late 21st century, with the maximum A_heat value occurring in the late 21st century under the RCP8.5 scenario, accounting for ~ 55% of the planting area (Fig. 8). Under all the RCP scenarios, A_mil, A_mod, and A_sev of single-cropping rice significantly increased with the increase in emissions over time. The variable A_tol linearly increased at rates of ~ 7 × 10³, ~ 20 × 10³, and 35 × 10³ ha yr^–1 and reached the maximum values in the late 21st century, accounting for 24.5%, 39.9%, and 58.6% of the total planting area under the RCP2.6, RCP4.5, and RCP8.5 scenarios, respectively.

This study combined the simulation results of 16 global climate models to predict the heat injury to single-cropping rice in China in the future and, to a certain extent, improved the accuracy compared with single-model predictions (Xu et al., 2010; Tian et al., 2017). However, although the global climate models have improved considerably in recent years, there are still large uncertainties in the simulation of climate factors (Guo et al., 2013; Jiang and Wu, 2013), which may lead to uncertainty in projecting the heat-damage hazards to single-cropping rice. It should be pointed out that the results of this study were based on the results of CMIP5 model simulations, which had a large pattern dependence. The previous study pointed out that the uncertainty of the increasing temperature of the model prediction was mainly due to the difference in the climate sensitivity. Different climate models had significant differences in the warming response forced by the same CO₂ concentration because of their structural framework, especially the different model-driven data sets and different expressions of physical processes of the climate system (Anav et al., 2013; Chen and Zhou, 2016). Therefore, future climate change research needs to strengthen the coupling study of climate and CO₂, while accumulating observation data, and improve the coupling global climate model, to provide more accurate support for projecting agrometeorological disasters.

In terms of assessments of the changes in plant phenology under the future global warming scenarios, most previous studies focused on the relationship between plant phenological changes and environmental factors. However, less research has been conducted on mechanisms of the phenological influence factors regulating plant growth and reproduction (Wang et al., 2015; Li et al., 2018), which reduces the accuracy of phenological prediction. Moreover, little research has been conducted on the changes in phenology of crops, especially in rice, because the changes in crop phenology are not only affected by climatic factors, but are also strongly influenced by human factors. Moreover, variations in planting patterns and varieties of rice may increase the uncertainty of rice phenology prediction, which makes it difficult to accurately predict changes in the future crop phenology. In the present study, we predicted and analyzed the occurrence probabilities and spatial distributions of heat injury to single-cropping rice for 1961–2000 based on the assumption that the varieties, sowing dates, and growth periods in this region were constant, which may lead to some deviation in the research results. As a result, further study should focus on the development of more objective and reasonable technical methods (Taylor et al., 2012; Dong et al., 2014), examination of the mechanism of changes associated with the heat injury to single-cropping rice, and refining evaluation indicators to improve the accuracy of future predictions.

Acknowledgments. We thank Lesley Benyon and Alex Boon from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.