Responses of Wheat Protein Content and Protein Yield to Future Climate Change in China during 2041–2060

Abstract

The nutritional value of wheat is measured by its grain protein content (PC) and is sensitive to climate change. The potential variations of future wheat PC under the influence of global warming varied among studies. Wheat PC data from China since 1980 were collected to explore the relationship between wheat PC and climatic variables, and Coupled Model Intercomparison Project 6 (CMIP6) models were used to project wheat PC and protein yield (PY) in China from 2041–2060. The results show that climatic variables during wheat heading to the maturation period have critical effects on wheat PC. The mean maximum air temperature and mean diurnal temperature range exhibited the greatest positive effects on wheat PC. The mean PC will increase under all shared socioeconomic pathway (SSP) scenarios, with significant rises in North China and the Guanzhong Plain, but a decrease in the Yangtze River Basin. Wheat PY with adaptations will increase, while that without adaptations will decrease. Global warming will increase wheat PC but decrease PY and protein production. These impacts could be mitigated by applying adaptation management. Our results enhance our understanding of wheat PC variation patterns and the possible response of wheat to future climate changes, and highlight the importance of applying suitable adaptations.

1. Introduction

It has been widely observed that climate change is happening around the world. The Intergovernmental Panel on Climate Change (IPCC) made it clear in their sixth report that the mean air temperature over the past decade was approximately 1.09 °C higher than in 1850–1900, that there had been greater warming on land than in the ocean, and that there had been greater warming in winter [1]. The daily minimum air temperature rise rate is approximately twice that of the daily maximum air temperature [2]. The diurnal temperature range (DTR) continues to decrease, with little change in the daily maximum air temperature. Human activities have played a dominant role in climate change over the past century, and the global warming trend continues [3].

Climate change affects grain crop production significantly [4,5]. Xiong et al. [6] used a spatially calibrated crop model to simulate idealized national cereal production from 1961–2010, and the results showed that climate factors led to a decrease in wheat yield by 9.7%; however, this reduction is partially offset by the fertilization effect of CO₂, resulting in an overall change of 0.9%. Wheat is one of the top three grain crops in China, with the largest cultivation area, and provides a large amount of protein for human beings. The nutritional quality of cereal crops is defined as the content of nutrients (protein, starch, and other nutrients) in the crop grains [7]. Wheat grain nutritional value is measured in terms of protein content (PC) [2], and the total nutritional value of grain crops is the real indicator of food security. The wheat PC of wheat is determined by both variety and environment [8]. Climate change effects on wheat production have already received considerable focus [9], while the effects of climate change on wheat PC have not been equally emphasized. How future climate change will affect wheat grain PC and its potential variations remains uncertain, and further research would enhance our understanding [10].

There is clear evidence that wheat PC is significantly affected by climate variables [11]. Climate factors such as air temperature [6,12], precipitation [13], DTR [14,15,16], and atmospheric CO₂ concentration [17,18] have an important influence on wheat grain PC. Grain filling to maturity is an important period for wheat grain protein accumulation. The increase in daily air temperature during the filling stage promotes an increase in wheat PC, and higher average temperatures result in faster filling rates. A reasonable combination of day and night air temperatures during grain filling is also important to ensure the grain filling rate and improve the wheat PC [19]. Water is indispensable in grain crops’ growth and development and is an important factor affecting crop quality [20,21]. Excessive precipitation will also reduce the total sunlight incidence on crops, thereby affecting the normal crop photosynthetic rate, which is not conducive to the formation of wheat grain protein. Increasing atmospheric CO₂ concentration has been driving global climate change, while also affecting plant growth and ecosystem productivity [11]. Elevated CO₂ concentration contributes to wheat protein production but decreases wheat grain PC [18].

It remains uncertain how wheat PC will change under the influence of global warming. Several studies have been performed to project wheat grain PC and protein yield for the future. Asseng et al. [9] projected global wheat PC under climate change scenarios by applying a multi-model combination of 32 wheat crop models. The results revealed an increase in wheat grain protein yield (2%), while wheat grain PC (−1.1%) decreased. However, Long et al. [10] projected the impact of climate change on wheat yield and nitrogen content in the Yellow River Basin from 2020 to 2050 under the Representative Concentration Pathway 8.5 (RCP8.5) scenario by combining the corrected refined meteorological grid data with the DSSAT-CERES-Wheat model, and indicated that the average nitrogen content (2.3%) of wheat grain in the region will increase in the future. By the mid-21st century, it is projected that the air temperature will increase by 0.7 °C [22], the DTR will decrease by 0.151 °C to 0.207 °C [23], and there will also be an increase in precipitation. Considering the influence of global warming, the projected change direction and amplitude of wheat grain PC need to be further studied. The changes in wheat grain PC under future climate change scenarios have become a relevant issue for Chinese agricultural scientists.

Wheat grain PC and protein yield play a critical part in food security for humans. The drivers of change in crop grain PC can be further explored [8]. Additionally, existing studies on crop quality focus on individual variety certification techniques or future projections for specific regions, and attention has only gradually been given to analyses and projections of crop yield at large scales. However, the ways that wheat grain PC will change in China under the influence of climate change should be given more attention. The specific objectives of this study were (1) to explore the relationship between climate variables and wheat grain PC with statistical analysis, (2) to determine the potential effects and changes of global warming on wheat grain PC and protein yield per unit area under three shared socioeconomic pathway (SSP) scenarios in China in the future period (2041–2060) relative to the historical period (1991–2010), and (3) to reveal the variations in future wheat grain protein yield per unit area under different adaptation scenarios. The results will contribute to deepening our knowledge about the relationship between global warming and food crop nutritional quality in China, which is crucial to the rational use of climate resources and to guaranteeing national food security.

2. Materials and Methods

2.1. Climate Data

2.1.1. Observation Data

The historical climate observation data used were derived from the “The dataset of fundamental daily surface meteorological elements in China (V3.0)” provided by the National Meteorological Information Center. This dataset contains the daily observation data and station information from 1961 to the present, gathered from 2474 weather stations in China. It has a high temporal resolution and is relatively continuous. In this study, the daily climate variables (including air temperature, maximum temperature, minimum temperature, precipitation, and relative humidity) of relevant meteorological stations in wheat cultivation areas of China since 1980 were selected to analyze the relationship between climate variables and wheat grain PC.

2.1.2. CMIP6 Model Output

Global climate data simulated by global climate models (GCM) from Coupled Model Intercomparison Project (CMIP) phase 6 were obtained from the CMIP6 official website (https://esgf.nci.org.au/projects/cmip6-nci, accessed on 24 August 2023). The daily output data of the mean near-surface air temperature and maximum and minimum near-surface air temperature from 1991 to 2010 were provided by EC-Earth3 and EC-Earth3-Veg-LR, respectively. To ensure consistency, this study focused on the first ensemble simulation (CMIP6: ‘r1i1p1f1′). EC-Earth3 [24] and EC-Earth3-Veg-LR [25] models were developed by the EC-Earth Consortium (Sweden), and have spatial resolutions of 512 × 256 (longitude × latitude) and 320 × 160 (longitude × latitude), respectively. These models offer continuous and long-time series data stretching back to 1850. It is reported that EC-Earth3 and EC-Earth3-Veg-LR models demonstrated excellent simulation capabilities for air temperature and DTR. The evaluation results indicated that these models not only perform well in simulating air temperature and DTR over China [26], but also demonstrate the best performance in simulating these variables in the main wheat cultivation areas [27]. Moreover, these models not only excel in capturing interannual variations of these variables [28], but also show the best simulation performance during the wheat-growing season [27].

The Shared Socioeconomic Pathway (SSP) offered in CMIP6 combined RCPs and other pathways, including CMIP5 updates and several new scenarios. SSP1-2.6, SSP2-4.5, and SSP5-8.5 represent scenarios in which the radiative sunpower stabilized at 2.6, 4.5, and 8.5 W/m² by the year 2100. The SSP1-2.6 scenario represents the low-emissions scenario, the SSP2-4.5 scenario is suitable for most countries pursuing sustainable development, and the SSP5-8.5 scenario represents high emissions, which reflects the impact of uncontrolled development [29,30]. Therefore, we used model output data for the historical period (1991–2010) and the future period (2041–2060) under the SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios in this study. The bilinear interpolation method was utilized to perform uniform interpolation of the model data onto a 5′ × 5′ grid, which corresponds to the grid position and resolution of the wheat yield dataset. Additionally, a simple bias correction was applied to the model data to ensure accuracy and consistency.

2.2. Wheat Yield and Grain PC

2.2.1. Wheat PC

Wheat PC is defined as the proportion of protein contained in a unit mass of wheat grain [31]. The nutritional value of wheat is measured by its grain PC [2], and the total nutritional value of grain crops is the real indicator of food security. In this study, the wheat grain PC data in China since 1980 were obtained from the Chinese National Knowledge Infrastructure (CNKI) by searching for relevant publications with wheat grain PC data. In this study, a total of two hundred sets of available PC data (Table S1) were collected, and the information regarding the year, location, grain PC, and relevant weather stations was recorded. The experimental conditions of different studies were controlled when collecting data. The experimental conditions were restricted to conventional tillage, control check (CK), or only applying nitrogen fertilizer and other similar conditions. Restricting experimental conditions can help reduce the influence of other experimental variables (such as potassium fertilizer, phosphate fertilizer, irrigation, soil conditions, planting date, and other field management measures) to a certain extent. The data covered the wheat planting areas of China (Figure 1) and does represent the wheat PC regional variations during the given period in the part of China considered by the reports.

2.2.2. Wheat Yield

The mean wheat yield and harvested area data in China for 1997–2003 of the Harvested Area and Yield for 175 Crops [32] dataset were obtained from the EarthStat official website (http://www.earthstat.org/harvested-area-yield-175-crops, accessed on 24 August 2023). This dataset was produced using a combination of statistical data collected from national, state, and county level censuses and a global cultivable land dataset generated with the MODIS land cover product and the GLC2000 land cover dataset on a 5′ × 5′ resolution grid, describing data such as the cultivation area and production of various crops in more than 70 countries and 6 continents around the world. The dataset has been widely used [33,34,35]. This dataset was verified using the wheat yield data obtained from the same publications with the wheat PC. The Root Mean Square Error (RMSE) between them is 0.036 t/hm² (N = 100), which indicates that this dataset can well represent regional variations in the real wheat yield. Considering the data availability and the accuracy of the dataset, the wheat yield and harvested area data were used in this study. All projections and analyses were conducted on grids of the wheat cultivation areas (Figure 1).

Future yield change data (

Table 1. Percentage of wheat yield change (%) in each grid under different SSP scenarios relative to the historical period (1991–2010).

Scenario	Management Adaptation	Temperate Regions	Tropical Regions
SSP1-2.6	With adaptation	0.12	15.82
SSP1-2.6	No adaptation	−5.02	6.02
SSP2-4.5	With adaptation	1.44	12.04
SSP2-4.5	No adaptation	−5.49	2.74
SSP5-8.5	With adaptation	4.33	1.94
SSP5-8.5	No adaptation	−6.19	−6.48

) for wheat under different scenarios were derived from the percentage of simulated yield change for food crops versus local temperature change for temperate and tropical regions relative to the 2000s, aggregated by the IPCC [36]. Zheng et al. [37] projected the winter wheat yield in the Guanzhong Plain by using SimCLIM with the CSM-CERES-Wheat model, and found that winter wheat yield showed a slight decrease in the south of the Guanzhong Plain in the 2050s based on RCP4.5 without considering adaptation scenarios. Geng et al. [38], using the Cobb–Douglas production function, analyzed the impacts of climate change on the winter wheat yield in Northern China and found that under adaptation scenarios, the winter wheat yield showed an increase both under RCP4.5 and RCP8.5 in the 2050s in Northern China. These projected changes share the same trends as that from IPCC. Therefore, the future yield change data were used for future projections and analyses in this study.

Crop management adaptations also make an essential contribution to wheat growth and production. Different management adaptations (cultivation adjustment, irrigation optimization, fertilizer optimization, and other management adaptations) were considered as “With adaptation (WA)” and “No adaptation (NA)”, according to IPCC AR5 [36].

2.3. Methods

In this study, a statistical model between climate and wheat grain PC was constructed by correlation analysis and stepwise multiple linear regression analysis. The wheat grain PC and yield were projected for the future period (2041–2060) under different SSP scenarios, and the changes relative to the historical period (1991–2010) were analyzed (Figure 2).

Wheat grain PC is determined by both genetic factors and environmental conditions [8]. Since significant differences in nitrogen fertilizer application rates existed in different experiments, the effect of nitrogen fertilizer application rates on PC was eliminated in this study by unifying nitrogen fertilizer application rates to the unfertilized level through a simple linear regression method.

When projecting wheat grain PC and protein yield under different future scenarios, it is assumed that (1) sufficient irrigation will be carried out through an artificial water supply during wheat growth to ensure sufficient water supply, (2) the planting area and spatial distribution of wheat in China will remain unchanged in the future, and (3) the field management and agronomic skills will be kept at the same level as nowadays under the scenario, without adaptation. These assumptions help eliminate the interference of other factors and enhance the comparability and interpretability of the results.

2.3.1. Meta-Analysis

Meta-analysis is a quantitative research approach that synthesizes the results of several independent studies on the same scientific problem for statistical analysis [39]. In this study, wheat PC data were obtained by collating the results of published studies on the subject. The research data used in this study were obtained from papers provided by CNKI which focus on climate change and agriculture. Keywords used for the literature search included wheat, climate change, and wheat PC. The information from weather stations in China was matched according to the origin of wheat to extract the corresponding climate observations.

2.3.2. Correlation Analysis

In this study, correlation analysis was applied to explore the relationship between wheat PC and climate factors. The Pearson correlation coefficient ($R$) is a statistic that describes the linear relationship between two variables. $R$ is:

$$R=\frac{{\sum }_{i=1}^n\left(x_i-\overline{x}\right)\left(y_i-\overline{y}\right)}{\sqrt{{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2\sum _{i=1}^n{\left(y_i-\overline{y}\right)}^2}}$$

(1)

where $x_i$ and $y_i$ are the ith values of the two random variables, $\overline{x}$ and $\overline{y}$ denote the means of each variable separately, and $n$ is the sample size. A two-tailed Student’s t-test was performed to estimate the significance level.

2.3.3. Regression Analysis

Multiple linear regression is a regression analysis that has only one dependent variable but contains multiple independent variables. It is widely used in scientific research [40,41,42]. Stepwise multiple linear regression was chosen to construct a statistical model for the relationship between climatic variables and wheat PC. When a multiple linear regression is performed, a covariance test between the independent variables is needed. If there is a high covariance between the independent variables entering the equation, the equation should be discarded. Multivariate regression analysis of climate anomalies and wheat PC was performed by a linear model as follows:

$${Pro}_w=\sum _{i=1}^N\left(a_i\times {Cli}_i\right)+C$$

(2)

where ${Pro}_w$ is the observed wheat protein content (%), $N$ is the number of weather factors included in the equation, ${Cli}_i$ is the ith climatic variable during the wheat growing period, $a_i$ is the ith coefficient of the ${Cli}_i$, and $C$ is a constant.

The RMSE is often applied to assess the goodness of fit of multiple linear regression [43,44]. The RMSE is calculated as follows:

$$RMSE=\sqrt{\frac{1}{N}\sum _{i=1}^N{\left(X_i-Y_i\right)}^2}$$

(3)

where $X_i$ and $Y_i$ are the simulation and observation values of wheat PC in the training set, respectively. $N$ is the number of samples in the training set. A smaller RMSE indicates that simulation is more consistent with observation.

2.3.4. Climate-Wheat Protein Content Relationship

The statistical model of climate–wheat PC was constructed as follows (Figure 2):

(1) Correlation analysis was performed between the wheat PC and historical observations of the corresponding weather stations to screen out the climatic variables with significant correlations. (2) The wheat grain PC was stratified and randomly sampled, then divided into a training set (70%) and a validation set (30%). (3) The training set was used to conduct a forward selection stepwise multiple regression analysis to derive a valid climate–wheat PC regression model. (4) The climate–wheat PC regression model was validated using the validation set.

2.4. Other Indicators

2.4.1. Climatic Indicators

The DTR was calculated as the difference between the maximum and minimum 2 m air temperature over 24 h, according to Xie et al. [23,27].

2.4.2. Developmental Periods

In this study, the following wheat key developmental periods were selected to investigate the contribution of climatic conditions to wheat grain PC: the whole growing period (greening to maturity), jointing to maturity (JM), and heading to maturity (HM).

The length of developmental periods is strongly dependent on the accumulated temperature and is usually expressed in degree-days because the number of days needed to reach a developmental stage is a function of temperature. Therefore, the accumulated temperature thresholds were used to divide the developmental stages of wheat at each grid point. It is assumed that wheat has reached a certain developmental period when the accumulated temperature reaches a particular value. In this way, the different stages of wheat fertility can be defined more precisely.

2.4.3. Protein Yield and Protein Production

The protein yield and protein production of each grid in the cultivation areas were calculated as follows:

$$Y_i={Pro}_i\times {WY}_i$$

(4)

$$P=\sum _{i=0}^N\left({Area}_i\times Y_i\right)$$

(5)

where $Y_i$ is the wheat protein yield (t/hm²) of the ith grid, ${Pro}_i$ is the wheat protein content (%), and ${Area}_i$ is the cultivation area (hm²) of the ith grid. ${WY}_i$ is the wheat yield (t/hm²) of the ith grid, $P$ is the protein production (t), and $N$ is the number of grids.

3. Results

3.1. The Relationships between Climate Variables and Wheat Protein Content

3.1.1. Correlation Analysis

In this study, climatic variables (

Table 2. Information on climatic variables.

Period	Indicator	Variable	Unit
The whole growing period Jointing to maturity (JM) Heading to maturity (HM)	Tacc0	>0 °C effective accumulated temperature	°C·d
	Tacc5	>5 °C effective accumulated temperature	°C·d
	Tacc10	>10 °C effective accumulated temperature	°C·d
	Tacc15	>15 °C effective accumulated temperature	°C·d
	Tmean	Mean air temperature	°C
	Tmax	Mean maximum air temperature	°C
	Tmin	Mean minimum air temperature	°C
	DTR	Mean diurnal temperature range	°C
	H	Mean relative humidity	%
	Pr	Total precipitation	mm

) among different growing periods were selected to determine the climate–wheat PC relationship. Although total sunshine hours have an important effect on wheat PC, the effects of total sunshine hours and CO₂ were not considered due to the lack of available GCM simulations under SSP scenarios.

Climatic variables with significant correlations were screened in

Table 3. Information on climatic variables with significant correlations with wheat protein content (PC).

Period	Indicator	Correlation Coefficient	Indicator	Correlation Coefficient
The whole growing period	Tacc0	0.212 **	Tmean	0.352 **
	Tacc5	0.229 **	DTR	0.297 **
	Tacc10	0.369 **	H	−0.177 *
	Tacc15	0.406 **	Pr	−0.275 **
Heading to maturity (HM)	Tacc0_hm	0.407 **	Tmax_hm	0.448 **
	Tacc5_hm	0.409 **	Tmin_hm	0.293 **
	Tacc10_hm	0.415 **	DTR_hm	0.343 **
	Tacc15_hm	0.410 **	H_hm	−0.315 **
	Tmean_hm	0.407 **	--	--
Jointing to maturity (JM)	Tacc0_jm	0.407 **	Tmax_jm	0.443 **
	Tacc5_jm	0.414 **	Tmin_jm	0.278 **
	Tacc10_jm	0.421 **	DTR_jm	0.355 **
	Tacc15_jm	0.413 **	H_jm	−0.352 **
	Tmean_jm	0.406 **	Pr_jm	−0.169 *

** indicates that the correlation relationship was significant at p < 0.01. * indicates that the relationship was significant at p < 0.05.

. Air temperature (0.448) and DTR (0.343) during the period from heading to maturity (HM) and the period from jointing to maturity (JM) exhibited significant positive correlations with wheat PC, while precipitation showed a negative correlation. The formation of wheat grain protein occurs mainly during the period from heading to maturity, and a suitably high air temperature helps to accelerate the filling rate of wheat grain. The higher the air temperature during the filling period, the faster the filling rate of wheat grains [45], and the grain PC will increase accordingly. A lower minimum air temperature will have an inhibitory effect on the respiration intensity of wheat [46], while a suitably low air temperature at night helps to reduce the consumption of organic materials such as protein by respiration, which can promote the accumulation of grain protein. Therefore, a greater DTR contributes to the accumulation of wheat grain protein and the formation of nutritional quality [14].

In addition, precipitation was negatively correlated (−0.169) with wheat PC under the assumption of sufficient irrigation. Excessive precipitation is detrimental to the formation of protein during wheat filling to maturity [47]. This is because too much precipitation tends to wash away nitrate from wheat roots, making N supply insufficient [48], and on the other hand, it affects photosynthesis and delays nutrient transit time [49].

3.1.2. Climate–Wheat PC Relationship

The training set was used to conduct a stepwise multiple linear regression of wheat PC (${Pro}_W$) and the corresponding climatic variables. A valid statistical model was established as follows:

$${Pro}_W=0.205\times Tmax\_hm+0.255\times DTR\_hm+4.314$$

(6)

where ${Pro}_W$ is the wheat grain protein content (%), $Tmax\_hm$ is the daily mean maximum air temperature (°C) during the heading to maturity period of wheat, and $DTR\_hm$ is the mean diurnal temperature range (°C) during the heading to maturity period of wheat.

The result of simple correlation coefficient decomposition between wheat PC and key climate variables are shown in

Table 4. Simple correlation coefficient decomposition between wheat protein content (PC) and key climate variables.

Variable	Path Coefficient	Path Coefficient (Direct)	Path Coefficient (Indirect)
			Tmax_hm	DTR_hm	Total
Tmax_hm	0.435	0.379	-	0.103	0.103
DTR_hm	0.332	0.232	0.147	-	0.147

.

Tmax_hm has the largest direct path coefficient, followed by DTR_hm. By analyzing various indirect path coefficients, it was found that DTR_hm has a relatively significant indirect effect on PC through Tmax_hm, with an indirect path coefficient of 0.147. Therefore, both Tmax_hm and DTR_hm play essential roles in increasing PC, with Tmax_hm having a greater impact on PC compared to DTR_hm. Zhou et al. [47] also indicated that temperature is the climate factor that has the greatest effect on wheat PC.

Although precipitation and other climatic variables also had significant effects on wheat PC, the contribution of precipitation was attenuated under the conditions of irrigation [47]. Therefore, they were not included in the model.

The statistical model can generally reflect the relationship between key climatic variables and wheat grain PC. Higher air temperature and greater DTR correlated with increased PC in wheat grains. The model was significant (R = 0.48, p < 0.001), and the simulation was close to the observation (RMSE = 1.52%, MAE = 1.21%). It indicates that the statistical model can approximate wheat PC. According to the validation results, this statistical model of climate and grain PC can be applied to the projection of wheat PC under future climate change scenarios.

3.2. Wheat Grain PC and Yield Simulation

Wheat PC and protein yields during historical (1991–2010) and future periods (2041–2060) were simulated by combining the statistical model of climate and PC with the historical and future climate data under different SSP scenarios.

3.2.1. Historical Simulation

The historical wheat PC and protein yield are shown in Figure 3 and

Table 5. Projected wheat protein content (PC), protein yield (PY), and protein production (P) under different SSP scenarios.

Scenario	Adaptations	PC (%)	ΔPC (%)	PY (t/hm2)	ΔPY (t/hm2)	P (×104 t)	ΔP (×104 t)
-	1991–2010	12.45	-	0.440	-	1221.46	-
SSP1-2.6	With adaptation	12.90	0.45	0.457	0.017	1257.98	36.52
SSP1-2.6	No adaptation	12.90	0.45	0.433	−0.007	1193.40	−28.06
SSP2-4.5	With adaptation	12.82	0.37	0.460	0.020	1260.12	38.66
SSP2-4.5	No adaptation	12.82	0.37	0.406	−0.034	1113.75	−107.71
SSP5-8.5	With adaptation	12.88	0.42	0.482	0.042	1316.40	94.94
SSP5-8.5	No adaptation	12.88	0.42	0.383	−0.057	1046.18	−175.28

. The average wheat PC in China was 12.45% (11–15%). The wheat PC showed a gradual increase from low latitudes to high latitudes and a gradual increase from coastal areas to inland, which is related to the climatic conditions and geomorphic features of the main wheat cultivation areas. The wheat protein yield shared similar geographical distribution characteristics as wheat yield, and the highest wheat protein yield was found in Northern China (NC).

3.2.2. Future Projection

Projected wheat PC and protein yield predictions for the future period under SSP1-2.6 are shown in Figure 4 and Table 5. The average wheat PC in China forecast was 12.90%, with an increase of 0.45% relative to the 1991 to 2010 period. Wheat protein yield (with adaptation) increased by 0.017 t/hm² and protein yield increased by 365,200 t relative to the historical period, while that with no adaptation decreased. Wheat protein yield increased by 0.024 t/hm² and protein yield increased by approximately 645,800 t under the scenario with adaptation compared to that with no adaptation.

The increase in Tmax_hm and DTR_hm during wheat growing periods favored the formation of grain protein. In

Table 6. Projected wheat protein content (PC), protein yield (PY), and protein production (P) under different SSP scenarios.

Scenario	Adaptations	ΔPC (%)		ΔPY (%)		ΔTmax_hm (%)		ΔDTR_hm (%)
		+	−	+	−	+	−	+	−
SSP1-2.6	With adaptation	83.88	16.12	85.43	14.57	67.66	32.34	86.56	13.44
	No adaptation			21.96	78.04
SSP2-4.5	With adaptation	75.01	24.99	98.22	1.78	67.50	32.50	65.05	34.95
	No adaptation			4.09	95.91
SSP5-8.5	With adaptation	77.46	22.54	99.76	0.24	60.20	39.80	72.15	27.85
	No adaptation			1.26	98.74

, wheat PC broadly increased (83.88%) in NC due to the wide increase in Tmax_hm (67.66%) and DTR_hm (86.56%). The Tmax_hm decreased significantly, while the DTR_hm varied slightly (13.44%) in the Yangtze River Basin (YRB), resulting in a decrease (16.12%) in wheat grain PC. In addition, wheat grain protein yield (with adaptation) showed a significant increase in NC under the SSP1-2.6 scenario because of increased wheat yield (with adaptation) and the increase in wheat PC. The increase in wheat yield in YRB was smaller than the decrease in grain PC, leading to a slightly lower protein yield. Despite a slight increase in wheat PC, there was a widespread decrease in wheat yield with no adaptation, which ultimately led to a decrease in protein yield.

Projected wheat PC and protein yield in the future period under SSP2-4.5 are shown in Figure 5 and Table 5. The average wheat PC in China was 12.81%, which was slightly lower than that under the SSP1-2.6 scenario and increased by 0.37% relative to the historical period. Protein production (with adaptation) also increased by 386,600 t, and protein yield increased by 0.02 t/hm², while that with no adaptation showed a greater decline. Compared to the scenario with no adaptation, wheat protein production with adaptation increased by approximately 1,463,700 t, and the protein yield increased by 0.054 t/hm².

Under the SSP2-4.5 scenario, the projection shows a similarly widespread increase in wheat grain PC in the future period (2041–2060) (Table 6). Wheat PC increased widely (75.01%) in NC and decreased in YRB (24.99%). There was a slight predicted decrease in the area of increased PC relative to SSP1-2.6. The majority of grids (67.50%) had increased Tmax_hm, but the percentage of grids with increased Tmax_hm decreased slightly. These grids were mainly distributed in NC, leading to an increase in grain PC. There was a greater decrease in Tmax_hm (32.49%) and a decreasing trend in DTR_hm (34.95%) in YRB, resulting in a greater decrease in wheat grain PC (24.99%). The protein yield (with adaptation) increased extensively (98.22%) under the SSP2-4.5 scenario, while the protein yield (no adaptation) showed an extensive decrease (95.91%). As a result of the increase in wheat yield (with adaptation) and wheat PC, there was a significant rise in wheat protein yield in NC. In YRB, the predicted increase in wheat yield was greater than the decrease in PC, resulting in a slight increase in wheat protein yield. Relative to the SSP1-2.6 scenario, there are more grids showing wheat protein yield increase under SSP2-4.5.

The projection in wheat PC and protein yield under SSP5-8.5 in the future period is shown in Figure 6 and Table 5. The average wheat PC in China was 12.88% and increased by 0.42% relative to the historical period. The distribution of wheat PC in China under the SSP5-8.5 scenario shared the same characteristics as that under the SSP1-2.6 and SSP2-4.5 scenarios, with the same significant increase in wheat PC in Northeast China (NEC) and NC. Relative to the historical period, protein yield (with adaptation) increased by 949,400 t and protein yield increased by 0.042 t/hm², while the scenario with no adaptation showed a greater decrease. The application of management adaptations contributed to the increase in the wheat protein yield.

The wheat PC showed a similar extensive increase in the future period (2041–2060) under the SSP5-8.5 scenario (Table 6). A greater increase in Tmax_hm and a slight decrease in DTR_hm was detected in NC, resulting in a slight increase in grain PC (77.46%). The wheat grain PC decreased considerably (22.54%) due to a greater and broader decrease in Tmax_hm (30.98%) with a decreasing trend in DTR_hm (27.85%) in YRB (22.54%). Under the SSP5-8.5 scenario, wheat protein yield (with adaptation) showed a further increase (99.76%), while protein yield (no adaptation) further decreased (98.74%). The wheat protein yield (with adaptation) increased greater because of the increased wheat yield (with adaptation) and the increasing wheat PC in NC. In YRB, a slight increase in wheat protein production was detected as the increase in wheat yield outweighed the decrease in wheat grain PC.

Under the scenario with adaptation, wheat protein production gradually increased under all SSP scenarios (

Table 7. Wheat grain protein production (×10⁴ t) under different SSP scenarios and different management adaptations.

Adaptation	1991–2010	SSP1-2.6	SSP2-4.5	SSP5-8.5
With Adaptation	1221.46	1257.98	1260.12	1316.40
No Adaptation	1221.46	1193.40	1113.75	1046.18
Difference	-	64.58	146.37	270.22

). Compared with the historical period, changes in protein production under the SSP1-2.6 scenario were closer to those under the SSP2-4.5 scenario, with an average increase of 370,000 t, while the change under the SSP5-8.5 scenario (949,400 t) was much greater. However, wheat grain protein production with no adaptation decreased significantly under all SSP scenarios, with the smallest change under the SSP1-2.6 scenario (280,600 t) and a much greater change under the SSP5-8.5 scenario (1,752,800 t). The application of management adaptations can contribute to the improvement of wheat protein production by improving wheat production.

Protein production was affected by the combination of climatic conditions and wheat production. The variability of key climatic variables during wheat growing periods resulted in PC differential variations. Both Tmax_hm and DTR_hm were extensively increased under the SSP1-2.6 scenario relative to the historical period, resulting in a great increase in wheat PC (83.88%) along with a slight increase in wheat production, thereby increasing wheat protein production. Compared to the SSP1-2.6 scenario, Tmax_hm changed slightly, and DTR_hm increased in a smaller area (65.05%) under the SSP2-4.5 scenario. There was a slight increase in protein production due to a greater increase in wheat yield than that under the SSP1-2.6 scenario. The proportion of increased grids of Tmax_hm and DTR_hm (72.15%) also decreased under the SSP5-8.5 scenario relative to the SSP1-2.6 scenario, with a smaller increase in PC. However, the change in wheat production under the SSP5-8.5 scenario was the highest among others, so protein production increased further relative to that under the SSP1-2.6 and SSP2-4.5 scenarios.

4. Discussion

4.1. Effects of Climate Factors on Wheat Protein Content and Protein Yield

Crop grain PC was significantly influenced by climate factors. Climatic variables during the period from heading to maturity exhibited stronger correlations with wheat PC relative to the periods from greening to maturity and jointing to maturity. It is worth noting that wheat needs suitable climatic conditions during key growing periods. Climate variables during these periods had significant effects on the formation of grain protein, including effective accumulated temperature (0.415), air temperature (0.407), DTR (0.343), relative humidity (−0.315), and total precipitation (−0.275). The higher the air temperature during the filling period, the faster the wheat grains filling rate [50], and a suitably low air temperature at night helps to reduce the consumption of organic materials such as protein by respiration, which can promote the accumulation of grain protein. A greater DTR contributes to the accumulation of wheat grain protein and the formation of nutritional quality [14,51]. Wang et al. [14] indicated that there is a suitable range for meteorological variables such as air temperature and total precipitation during the wheat growth periods. When these variables fall outside of those limits, variations will have negative effects. When the air temperature is in the optimum range, the wheat PC will increase as the air temperature rises gradually.

The precipitation showed a negative correlation (−0.275) with wheat PC. Zhou et al. [47] indicated that air temperature is the dominant factor in the irrigation area, but precipitation is more important in the rainfed area. In the absence of irrigation, an appropriate increase in precipitation helps wheat development, but under conditions of adequate irrigation, an increase in precipitation is detrimental to the accumulation of protein in wheat and thus has a negative effect [47,52]. Water supply for modern wheat cultivation in China is mainly through irrigation, which weakens the effect of precipitation on wheat grain protein. Since precipitation and irrigation will directly change the soil water content, it was assumed that the main wheat cultivation areas had been fully irrigated, so the effect of soil water content was not discussed.

The effects of other climate variables could not be ignored. Pan et al. [47] used stepwise regression analysis to explore the effect of sunshine hours and precipitation on wheat PC and indicated that the post-anthesis total sunshine hours contributed to the increase of wheat PC. Brankovic et al. [53] indicated that sunshine hours in March affect wheat PC. Relative humidity had a significant negative correlation (−0.315) with wheat PC, and the combined effect of high temperature and high humidity has had a significant impact on wheat production [54].

Under all SSP scenarios, wheat PC increased extensively in NC, which is consistent with the results of Long et al. [10]. In contrast, wheat PC decreased in YRB relative to its historical period. It is associated with differential variation of key climate variables under each scenario (Table 5). The impact of climate change on wheat yield also affected protein yield. According to IPCC [37], wheat yields increased under all SSP scenarios with adaptation measures, while they decreased when no adaptation measures were applied. Due to the changes in wheat yield being greater than the variation in wheat PC, wheat protein yield increased over a large area (85.43–99.76%) in China under all SSP scenarios in the future under adaptation measures. However, wheat protein yield and production decreased mainly without adaptation measures. This suggests that the application of management adaptations can contribute to the improvement of wheat protein production by improving wheat production.

4.2. Other Factors Affect Wheat Protein Content and Yield

Other environmental factors, such as soil water content and atmospheric CO₂ concentration, also have important effects on wheat grain PC. Soil water content has a strong impact on N uptake [55], and an increase in total sunshine hours before wheat jointing is beneficial to the grain PC [56], while higher total sunshine hours between flowering and harvest maturity are associated with lower grain PC. In addition, elevated CO₂ concentration contributes to wheat protein production but decreases wheat grain PC [11,18]. Potential benefits from elevated atmospheric CO₂ concentration are negated by the impacts of precipitation and increasing air temperature on global wheat protein yields [57]. When CO₂ is emitted into the atmosphere, it will rapidly diffuse and mix uniformly, leading to relatively consistent atmospheric carbon dioxide concentrations across different locations globally within the same year, with minimal interannual variations. Additionally, the limited availability of widely accessible observational data further restricts the consideration of CO₂ concentration in this study. Therefore, the influence of CO₂ concentration was not taken into consideration. In future research, the impact of CO₂ concentration on wheat grain PC could be further explored through field experiments and controlled studies.

Factors such as wheat varieties, field cultivation management, pesticide and fertilizer application [58], and the influence of sociopolitical and economic conditions [59] also have important effects on wheat grain PC. In this study, the PC data were collected only for those experiments in which conventional tillage or only N fertilizer was applied, to reduce the effect of other experimental variables. However, there may be potential bias because the environment and field management methods of the experiments in different studies still differ. This may lead to potential uncertainty in the projections of wheat PC [56,60]. Modeling wheat PC under SSP scenarios remains a challenge for future research.

4.3. Uncertainties in Predicting Wheat PC and Protein Yield

The changes in wheat grain PC were analyzed with a statistical model, but the mechanism has not been strictly explained. Lobell and Burke [61] pointed out that all process models contain empirical or statistical rules to some extent, while many assumptions relating to crop development and mechanisms in statistical models are also included [62,63]. The number of variables affecting the PC of wheat grains may pose a potential challenge and bias to the results. Only some of the variables that had the greatest effect and were independent of each other were included in the model. Statistical models can be used to project the response of crops to climate change [64,65,66]. Sunshine hours and soil water content were not fully considered due to the limits of future data accuracy and availability. While performing future projections, the field management adaptations were simply divided into “with adaptation” and “no adaptation”, according to IPCC. The effects of specific adaptations on wheat PC were not discussed. More factors should be further considered. To further elucidate the profound relationship between climate change and wheat PC, crop growth models can be applied in future research.

GCM is one of the important methods for humans to investigate and project future climate changes. It is indicated that GCMs are the second-largest source of error in future crop yield projections [67]. Variations in model performance at different temporal and spatial scales may also affect the accuracy of crop yield and quality predictions [64,65]. It is reported that EC-Earth3 and EC-Earth3-Veg-LR models demonstrated excellent simulation capabilities for air temperature and DTR [26,27,28]. More focus should be given to enhancing GCM simulation performance to reduce uncertainties in crop yield and protein content (PC) projections [29].

Furthermore, while investigating the relationship between wheat PC and climate variables, the independent wheat PC data collected for this study were obtained from published literature, and only approximately half of the studies provided corresponding wheat yield data. This could potentially restrict the accuracy of the estimation results to some extent. Based on the available yield data, the corresponding wheat yield data were extracted from the dataset of wheat yield and harvested area, located by latitude and longitude information. This extracted data was then validated with the available yield data. The RMSE between them is 0.036 t/hm² (N = 100), which indicates that this dataset can well represent the real wheat yield regional variations. Considering the data availability and the accuracy of the dataset, the wheat yield and harvested area data were used in this study. It should be noted that the wheat yield and harvested area data are generated based on the MODIS land cover product data and statistical information. As a result, some inherent errors may be present, which could potentially influence the estimation results [32]. Further enhancing the accuracy of the wheat yield dataset can help reduce the uncertainty of the projection results.

Additionally, our study only projected the changes in wheat grain PC and protein yield in the mid-21st century, and more accurate simulations in the 2050s and beyond can be conducted in future studies under additional climate change scenarios.

5. Conclusions

Climatic variables during the period from the heading to the maturation period of wheat have important effects on wheat grain PC. Mean maximum air temperature and mean diurnal temperature range have the greatest positive effects on wheat PC, while total precipitation has a negative correlation.

Under the condition of maintaining the existing agronomic level and cultivation areas, the average wheat grain PC increased significantly under all SSP scenarios from 2041 to 2060 compared to the historical period of 1991 to 2010. Particularly, the SSP1-2.6 scenario exhibits a much greater magnitude of change than the SSP2-4.5 and SSP5-8.5 scenarios. Differences in key climatic variables during the key crop growing periods resulted in variations in wheat grain PC. The grain PC and protein yield of wheat in different cultivation areas showed different variation patterns. The wheat grain PC in the North China Plain and Guanzhong Plain will significantly increase, while that in the Yangtze River Basin will decrease. Under all SSP scenarios, the grain protein yield of wheat with adaptation measures increased gradually, while the yield without adaptations decreased significantly. The changes under the SSP5-8.5 scenario were much greater than those under other scenarios. The application of management adaptations can contribute to the improvement of wheat protein production by improving wheat production.

In conclusion, global warming will increase the wheat grain PC but will also lead to a reduction in wheat grain protein yield and total protein production. Implementing appropriate adaptation measures can help alleviate the impact of global warming-induced wheat yield reduction on wheat grain PY.