Merits of dietary patterns for China’s future food security satisfying socioeconomic development and climate change adaptation

Summary

Food security depends on food production exceeding consumption, which are influenced, respectively, by climate change and socio-economic development. We adopted a hybrid approach for predicting features of future food security in China. Specifically, complex interactions in food security were examined using copula-based Markov Chain Monte Carlo simulation. Crop yields per unit area were simulated with a denitrification-decomposition (DNDC) model under four climate-change scenarios. A high shortage risk for soybean and lower shortage risks for maize and wheat were predicted. Compared with the extent of food security in China under other dietary patterns, the performance advantage of the Chinese dietary pattern was identified, which could mitigate crop shortage risks. The shortage risk of core crops would not be influenced greatly under the different dietary patterns. In the planning years 2025 and 2030, the shortage risks for soybean and rice would be pronounced (i.e., more than 50%), whereas those for maize and wheat would not be prominent.

Highlights

• •Food security of China was measured based on four staple food and feed crops
• •The uncertainties of food production and consumption were identified
• •The influence of dietary shifts and planting structures was focused on
• •Food security of China was compared with that under other dietary patterns

Environmental science; Energy policy

Introduction

Increase in incomes has led to major transformations in dietary patterns.^1,2,3 However, dietary changes pose challenges for food security, that is, the availability of sufficient food to meet the global demand in the context of planetary limits and resource restrictions.^4,5 In particular, recent trends in socio-economic development have been accompanied by an increase in consumption of animal-sourced foods (ASFs).^6,7,8 For example, calories derived from ASFs (e.g., meat, dairy, and eggs, see Table S1) consumed in the daily diets of inhabitants of developed countries are triple those consumed in the daily diets of inhabitants of developing countries (Table S2). Moreover, crop production is threatened by global climate change.⁹ Precipitation and temperature fluctuations in the context of climate change may affect the global hydrological cycle,^9,10 which is of critical importance for crop production in arid and semi-arid regions.^11,12 Thus, there is an urgent need to evaluate the influence of dietary transitions and crop production on food security against a background of advancing socio-economic development and climate change.¹³

Food security is associated with food production and consumption. Given socio-economic developmental trends, further increase in ASF consumption will be an important factor in rising demand for feed crops.¹⁴ When competition for food and feed is intensified, food security will be closely linked to dietary preference.¹⁵ Previously, food security analysis mainly focused on the dietary structures, measured by their calories, with consideration of healthy nutrition.¹⁶ For example, the planetary health diet, proposed by the EAT-Lancet Commission, is described as a dietary preference with more plant-based foods and fewer animal-sourced and processed foods to improve health and environmental benefits.¹⁷ In addition, given that food security analysis necessarily has a future-focused perspective, the yields of food and feed crops should be simulated with consideration of the influence of climate change. With regard to tools available for the study of food security, nexus approaches can be used to highlight complex interactions between food consumption and production.¹⁸ Detailed information on specific elements (e.g., consumers' food preference as a function of socio-economic development and variation of crop yield with agricultural practices) can be derived from quantitative analyses of statistical or meteorological data.¹⁹ Many studies have demonstrated diversity of dietary structures in relation to socio-economic factors (e.g., income level).^20,21

Climate change may affect crop yields by altering the characteristics of precipitation and temperature.^12,16 Many previous studies have focused on complex issues associated with adaptive strategies in agriculture under the influence of climate change.²² For example, an increase in temperature by 1°C–3°C may lead to a reduction in wheat yields by 8.5%–28.5% in southern Africa.²³ It is also important to address socio-cultural issues associated with climate change, such as farmers’ attitudes toward maintaining land in production.²⁴ In simulations of the relationship between meteorological factors and crop yields, crop growth models [e.g., denitrification-decomposition [DNDC]) have been widely used to analyze the variability of plant-sourced food production in the context of climate change.^25,26,27,28 Given that a large proportion of feed crops is consumed in the process of livestock breeding, the impacts of climate change extend to ASF production.^24,25 Furthermore, climate change directly influences animal growth through the reduction of their feed intake and associated metabolic effects.²⁹ Socio-economic influences of climate change include a probable increase in the cost of food production against a background of unforeseen climate-change events and reduced access to food, especially for vulnerable households.^30,31

Dietary preferences and their environmental impacts are inextricably intertwined within a dynamic socio-economic system.³² As influenced by differences in dietary customs, carbon emissions associated with dietary guidelines range between 687 kg CO_2e capita⁻¹ yr⁻¹ in India and 1,579 kg CO_2e capita⁻¹ yr⁻¹ in the United States.^33,34,35 Chinese and Indian dietary customs characteristically entail low water and carbon footprints and are thus sustainable.³⁶ Given current socio-economic developmental trends, dietary preferences relating to ASF or processed foods are an important factor influencing rising demand for feed crops.^16,35 With the intensification of competing demands for food and feed, the issue of food security is of mounting concern.¹⁵ Previously, sustainable diets were considered to have low environmental impacts, while also ensuring food security and healthy nutrition.^37,38 Dietary pattern is also considered an important adaptive strategy for mitigating the impacts of climate change.³⁹ For example, reduction in the intake of ASF (e.g., beef) can potentially mitigate 0.7–8 Gt CO_2e yr⁻¹ of greenhouse gas emissions globally.⁴⁰

Household incomes directly influence food consumption, as evidenced by socio-economic developmental trends.^21,41,42 In particular, household income strongly influences ASF consumption but exerts a weak effect on the consumption of plant-sourced food.⁴³ Food consumption is also influenced by dietary preferences in different countries, as indicated by shifts in ASF consumption in these countries (Table S3). In light of variation in dietary preferences among countries and lower dietary intake of beef and milk in China, the impacts of per capita food consumption are less severe in China compared with those in some Western countries.⁴⁴ Thus, dietary shifts and the socio-economic aspects of food security merit further study.⁴⁵ Scenario analysis is regarded as an effective tool for exploring the potential average levels of dietary structures in the long term.⁴⁶ Multimarket models (e.g., International Model for Policy Analysis of Agricultural Commodities and Trade [IMPACT]) can simulate variation in food demand as a result of socio-economic development.⁴⁷

Multiple uncertainties arise from the variability in food production and consumption processes, as evidenced, for example, by the inherent randomness of dietary patterns and crop yields.^48,49 Although previous studies have investigated the relationship between socio-economic characteristics and diet-related environmental footprints,⁵⁰ socio-economic heterogeneity in dietary patterns across the world may have been neglected. The consequences resulting from such uncertainties further influence robust decision-making relating to food security.⁵¹ Therefore, a nexus approach should be applied to illuminate the complex interactions between food consumption and production, and to identify uncertain features of planting structures as well as dietary preferences.¹⁸ Quantitative statistical or meteorological data analyses can yield insights into the qualitative aspects of food systems (e.g., consumers’ food preferences in the process of social development, farming patterns associated with variation in crop yields, and crop yields under multiple climate-change scenarios).⁵²

In light of our review of previous studies on future food security, which considered dietary transitions and crop production, we identified the following knowledge gaps: (1) correlated features of dietary patterns and socio-economic development impacting food security; (2) interactions between agricultural practices and crop yields under a changing climate; and (3) comparison of the influence of dietary shifts on the food security of China in countries with a similar socio-economic baseline. Therefore, this study had the following objectives: (1) to explore complexities and interactions within food production and dietary structures against a background of climate change and advancing socio-economic development, and (2) to identify dietary patterns effective for ensuring food security through a comparative study of variation in food security with consideration of multiple typical dietary habits in high- and middle-income countries. We projected China’s food security in 2025 and 2030 on the basis of shortage risks pertaining to four staple food and feed crops: wheat, maize, rice, and soybean. We considered 12 dietary habits, incorporating animal-sourced and processed foods (eggs, pork, mutton, beef, milk, soybean oil, and cake) and plant-sourced foods (wheat, maize, rice, and soybean). We then simulated and predicted crop yields according to four climate-change scenarios under shared socioeconomic pathways (SSPs 1–26, 2–45, 3–70, and 5–85). Thirty-one administrative regions were selected for an analysis of food demand, and 104 cities distributed across 11 provinces were chosen as the main crop-planting districts.

Results

Correlations between crop structures and yields

Significant correlations were detected between planting structures and crop yields within food production systems across the provinces in China (Figure 1A and Table S10). The most highly significant correlations were observed for maize planting. These results indicated that planting structures for the crop would continue to evolve in relation to yields obtained during the previous year, especially in Shandong, Henan, and Heilongjiang provinces. Conversely, the planting structures for soybean and wheat in Hebei Province and Inner Mongolia were not strongly correlated with crop yields.

Figure 1.

Study areas considered in the food production and consumption analysis

Correlations between dietary structure and socio-economic development

In the baseline scenario, dietary structures in the 31 administrative regions in China showed significant correlations with socio-economic development in these regions (Figure 1B and Table S6). Indicators with low root-mean-square error (RMSE) values and high Nash-Sutcliffe efficiency (NSE) values pointed to a strong correlation associated with the intake of beef, mutton, pork, edible vegetable oil, eggs, and cake. The results indicated that Chinese dietary patterns would continue to evolve as a result of socio-economic development. Thus, demand for ASF and processed foods would likely rise with increase in gross domestic product (GDP) per capita. Conversely, the demand for grain was likely to decrease. Considering the striking variation in dietary preferences between higher-income and lower-income countries under a comparative scenario,⁵³ we aimed to compare the extent of food security in China under multiple dietary preferences at a global scale. Accordingly, we incorporated dietary patterns of ASF and processed foods from 11 representative countries, of which seven were high-income countries (HICs), three were middle-income countries (MICs), and one was a low-income country (LIC). Dietary patterns in most HICs were predicted to continue to evolve as a result of socio-economic development. Significant correlations with socio-economic development in most countries were indicated by intakes of poultry, soybean oil, pork, and milk (Table S8).

Demand for staple crops

To assess food shortage risks, we converted demand for different ASF and processed foods (Tables S7 and S8) into demand for associated staple crops in accordance with feed use for livestock production (Table S9). Our projections indicated that in 2025 and 2030, urban residents in southern China would become the largest consumers of staple crops, annually consuming 34–40 Mt of maize, 37–42 Mt of soybean, 35–41 Mt of wheat, and 74–81 Mt of rice (Table 1). Maize would also be used for biofuel production. On average, 3.3 t of maize are needed to produce 1 t of ethanol. The China Chemistry Industrial Association (CCIA) has indicated its intention to produce no more than 1.5 billion liters of maize-based ethanol in 2020 in its biofuel development plan.⁵⁴ We estimated that 4 Mt and 5 Mt of maize would be consumed in biofuel production in 2025 and 2030, respectively.

Table 1.

Total demand for maize, soybean, wheat, and rice in China under the baseline scenario projected for 2025 and 2030

Crops	Regions		2025		2030
			Mean values	Standard deviation	Mean values	Standard deviation
Maize	Urban residents	Southern region	33917.00	974.80	39514.00	1814.50
		Northern region	21227.00	368.88	22906.00	380.43
		Western region	1505.60	91.27	1661.50	87.63
	Rural residents	Southern region	12509.00	495.77	11756.00	456.42
		Northern region	8298.50	260.60	7692.60	235.53
		Western region	728.61	114.03	723.02	97.43
Soybeans	Urban residents	Southern region	37099.00	1260.50	42336.00	1521.60
		Northern region	21828.00	2403.30	23683.00	2202.40
		Western region	1981.90	133.01	2181.70	154.45
	Rural residents	Southern region	13254.00	700.40	12342.00	647.61
		Northern region	8482.20	375.53	8258.30	418.19
		Western region	1172.50	137.69	1115.40	115.72
Wheat	Urban residents	Southern region	34560.00	1303.80	40589.00	1809.60
		Northern region	34558.00	4270.40	37110.00	3917.30
		Western region	3625.60	221.09	3916.70	208.07
	Rural residents	Southern region	12817.00	571.83	11784.00	504.63
		Northern region	17167.00	618.26	15400.00	685.93
		Western region	5105.10	233.88	4709.90	211.65
Rice	Urban residents	Southern region	73668.00	2558.20	81210.00	2936.40
		Northern region	31311.00	6650.50	33039.00	5996.40
		Western region	2075.20	125.94	2244.40	120.46
	Rural residents	Southern region	57557.00	2405.30	49939.00	2023.70
		Northern region	13146.00	467.76	12075.00	795.16
		Western region	1182.90	83.82	1107.50	75.50

Unit: 1000 tonnes.

We also projected that demand for staple crops would vary to a greater degree if the dietary patterns of other countries were adopted in China (Tables 2 and S8). Demand for staple crops would increase if the dietary patterns of most HICs and MICs were adopted (Figure 2). Compared with dietary patterns in China, more staple crops would be consumed under the dietary patterns in most HICs, except for Japan. For example, the total demand for soybean, rice, wheat, and maize under the dietary pattern of the United States would be 1.4–1.7, 1.5–1.9, 1.9–2.5, and 2.1–2.5 times, respectively, as much as the demand under the local dietary pattern of China. In addition, more staple crops would be consumed under the dietary patterns of some MICs. For example, the total maize demand under the dietary patterns of Brazil, Thailand, and Vietnam would exceed that under the local dietary pattern of China.

Table 2.

Total demand for maize, soybean, wheat, and rice in China under the comparative scenarios projected for 2025 and 2030

		Maize		Soybeans		Wheat		Rice
		2025	2030	2025	2030	2025	2030	2025	2030
China	Mean values	7.82	8.42	8.38	8.99	10.78	11.18	17.89	17.88
	Standard deviation	0.12	0.19	0.29	0.28	0.46	0.44	0.76	0.70
Australia	Mean values	15.81	16.32	9.67	10.37	35.84	37.56	32.12	32.06
	Standard deviation	0.45	0.46	0.43	0.47	1.09	1.25	1.01	1.12
UK	Mean values	15.20	15.83	10.19	11.03	26.08	28.06	29.39	29.38
	Standard deviation	0.13	0.15	0.17	0.18	0.31	0.32	0.48	0.47
Canada	Mean values	18.35	18.89	11.81	12.63	34.23	35.84	33.51	33.38
	Standard deviation	0.22	0.24	0.30	0.32	0.63	0.70	0.79	0.83
France	Mean values	16.44	17.58	11.20	12.14	29.77	32.64	33.03	33.75
	Standard deviation	0.17	0.19	0.43	0.46	0.73	0.78	1.17	1.24
Germany	Mean values	17.68	18.66	10.55	11.40	25.50	27.82	29.05	29.61
	Standard deviation	0.28	0.29	0.15	0.17	0.38	0.41	0.49	0.49
USA	Mean values	19.98	20.51	13.98	15.50	35.49	37.60	33.13	33.37
	Standard deviation	0.14	0.17	0.31	0.40	0.48	0.57	0.56	0.68
Japan	Mean values	7.36	8.58	10.44	11.25	19.02	20.56	30.20	30.69
	Standard deviation	0.07	0.08	0.68	0.72	1.11	1.18	1.88	2.00
Brazil	Mean values	14.14	17.13	13.94	16.64	21.74	26.84	22.89	24.92
	Standard deviation	0.28	0.28	0.30	0.30	0.48	0.58	0.49	0.50
Thailand	Mean values	7.70	8.40	7.42	8.32	12.74	13.44	21.11	21.26
	Standard deviation	0.14	0.14	0.31	0.31	0.53	0.53	0.87	0.87
Vietnam	Mean values	20.97	25.41	16.37	19.43	20.94	23.49	31.59	33.36
	Standard deviation	1.10	1.24	0.80	0.90	0.54	0.64	0.66	0.70
Laos	Mean values	5.67	6.37	4.69	5.08	13.54	14.33	20.96	21.16
	Standard deviation	0.11	0.11	0.39	0.39	0.66	0.67	1.10	1.12

Unit: 10⁴ kt.

Figure 2.

Demand for staple crops in China under the baseline and comparative scenarios projected for 2025 and 2030

Food security analysis

Considering changes in food demand and production, climate conditions, and dietary preferences, the shortage risks for maize and wheat were predicted to be low, whereas the shortage risk for soybeans would exceed 95% under China’s current dietary pattern. The shortage risks for maize and wheat would be low under the baseline scenario from 2025 to 2030 over the next 5 years (Figure 3; Table S12), consistent with the forecasted increase in grain yield reported by Cao et al.⁵⁵ If the current dietary patterns were replaced by those of other countries, the risks of shortages of the aforementioned crops would vary. The risks of shortages of rice, maize, wheat, and soybean would be exacerbated under the current dietary patterns of some HICs and MICs. Moreover, decreases in the shortage risks for staple crops would not necessarily occur if the dietary patterns prevailing in MICs were adopted in China. For example, shortage risks for soybean, rice, wheat, and maize would increase by 0%–0.4%, 0%–9.8%, 0.6%–91.5%, and 1.1%–48.6%, respectively, if the dietary patterns of Brazil and Vietnam were applied. Some dietary patterns (e.g., that of Japan) would reduce the occurrence of soybean shortages by 7.2%–16.8% in 2025 and by 1.8%–2.6% in 2030. Thus, the Chinese dietary pattern was indicated to be effective in ensuring the security of wheat and maize.

Figure 3x.

Risks of crop shortages under multiple dietary patterns and climate-change scenarios projected for the planning years of 2025 and 2030 (unit: 10⁷ tonnes)

Discussion

Five copula functions (i.e., Frank, Gumbel, Clayton, Student’s t, and Gaussian copulas) and Monte Carlo Markov Chain (MCMC) simulation were used to analyze the following interactive features (Figure 4). First, to determine the relationship between dietary structures and social development, we incorporated food demand under multiple dietary patterns in China and other countries. Strong correlations were detected between food demand and social development in most regions of China and in other countries (Tables S5 and S6). Second, a distinct relationship was observed between crop yield per unit area during the previous year and the planting area during the following year (Table S10). We subsequently simulated the distributed features of dietary patterns and planting areas using the copula sampling method in accordance with the fitted copula functions.

Figure 4.

Analytical framework for evaluation of food security

Prediction of food consumption in China

The present findings indicated that, in light of China’s population and dietary pattern, and considering the total food demand, urban residents in the southern part of the country would consume the largest quantities of beef and mutton, poultry, milk, soybean oil, and pork compared with residents in other regions. Demand for milk, beef and mutton, and eggs in China would dramatically increase by 36%–63%, 28%–35%, and 13%–19%, respectively, from 2025 to 2030 compared with the total demands for these foods in 2015 (Table S7). Moreover, southern China would account for the greatest consumption of ASF and processed food products in the country. Considering variation in dietary preferences, the dietary patterns of HICs would influence the total demand for eggs, beef and mutton, milk, and pork, whereas the dietary patterns of MICs and the LIC would influence demand for soybean oil and poultry.

Production of staple crops in 2025 and 2030

With regard to the food production analysis, the climate-change scenarios were downscaled to the main planting districts of the staple crops in China. Crop yields per unit area were simulated with the DNDC model under four climate-change scenarios. We derived the total yields of the staple crops after incorporating distribution data for the planting areas (Table S11). The total yields of crops in China were estimated on the basis of regional contributions. Staple crops are important sources of materials for fodder and biofuel production.⁵⁶

Food shortage risk

To analyze the food shortage risk, the demands for meat, milk, eggs, seafood, and soybean oil were converted into demands for staple crops (i.e., maize, wheat, rice, and soybean) in accordance with requirements for these crops in food production. The results indicated that, in 2025 and 2030, quantities of these staple crops consumed by urban residents would be, respectively, 1.56–2.20, 0.90–1.69, 0.34–0.95, and 1.60–2.15 times higher than the quantities consumed by rural residents. Among rural residents, residents in southern China would consume the largest quantities of maize, rice, and soybean, whereas those in northern China would consume the largest quantities of wheat. Among urban residents, those in southern China would consume the largest quantities of all staple crops. In addition, shortage risks for the staple crops were determined according to their demand and supply distribution features. The findings indicated that shortage risks for the staple crops were highly dependent on dietary patterns, with that for soybean exceeding 95% for six of the dietary patterns (Figure 5).

Figure 5.

System boundary in the food security analysis

Policy implications

Advocacy of a balanced diet policy based on varying dietary habits

In China, the shift in dietary patterns (Figure S1 and Data S3) indicated that per capita ASF demand in the northern, southern, and western regions of China would increase by 16.40%–74.65%, 16.39%–87.74%, and 34.18%–84.47%, respectively, from 2025 to 2030. Thus, considering the urbanization trend and dietary habits in different regions, a policy to promote balanced diets is needed. The intake of certain types of ASF (e.g., pork and eggs), which would increase greatly consistent with the trend for socio-economic development in 2025 and 2030, should be targeted in the national dietary guidelines. For example, rural residents in Tianjin City and in Guangdong Province should pay attention to their energy and protein intakes, respectively (Data S8). Urban residents in the south-western region (e.g., Chongqing City and Sichuan Province) would have the greatest fat intake. The present comparison of different dietary patterns revealed that those in Japan, China, Laos, and Thailand would effectively reduce the energy and fat intakes associated with the Australian and Vietnamese diets by 45%–76% and by 51%–64%, respectively.

Dietary variation between urban and rural residents in the context of urbanization should be considered in efforts to address food security

The urbanization rates, as indicated by the ratios of urban residents, would exceed 70% in eight and 20 provinces in 2025 and 2030, respectively. When we considered differences in the dietary patterns of urban and rural residents (Data S3), the increasing number of urban residents as a result of the urbanization trend would influence food demands, especially for grain and milk. For example, the annual grain demand of urban residents would decrease by 22.20%–69.33% and 14.47%–71.45% in 2025 and 2030, respectively, relative to the demands of rural residents. Conversely, the annual milk demands of urban residents would increase by 52.49%–87.74% and 51.12%–84.49% in 2025 and 2030, respectively. The protein intake of urban residents through daily ASF consumption (Data S8) would increase by more than 50% in nine provinces compared with that of rural residents. According to government planning documents, the annual average growth rate of China’s population is projected to increase by 4.86% and 7.49% in 2025 and 2030, respectively (Table S4), relative to the population in 2020. Concurrently, the urban population would increase by 16.72% and 27.11% in 2025 and 2030, respectively. Dietary variation between urban and rural residents would inevitably influence food security. Therefore, to mitigate food shortage risks, the recommended dietary allowance should be promoted according to variation in the food demands of urban and rural residents.

Climate change and farming structures would jointly influence crop production

Per unit yields of staple crops were directly influenced by the effects of climate change. Additionally, the planting structures of staple crops were correlated with their per unit yields during the previous year. Thus, crop production would be influenced by the effects of climate change as well as farmers’ decision-making. Our projections of planting structures for the staple crops in 2025 and 2030 showed an increase in soybean and wheat planting areas, and a decrease in maize planting area, in most provinces compared with the corresponding structures in 2020. Considering the effects of climate change, decision-makers should extensively increase the soybean planting area to ensure food security. Considering China’s policy to secure food supply,⁵⁷ the planting areas of soybeans in 2025 and 2030 would be expected to increase by 20.54% and 19.49%, respectively, in the 20 provinces compared with the planting areas in 2018. In addition to considering the effects of climate change, analyses of food security should focus on the coupled effects of the government’s and farmers’ decision-making in relation to crop trading and planting.

Limitations of the study

The study has several limitations. First, it should be noted that, in addition to the level of social development, residents’ food consumption is influenced by other important factors (e.g., age, gender, and career). The variation in dietary structure with consideration of these factors may be sufficient to influence the effectiveness of the prediction of food demand. Second, government support would also play an important role in decision-making for crop cultivation under the background of food security. The foregoing factors were not incorporated in the current study owing to the non-availability of data.

Conclusions

To fulfill the 2nd Sustainable Development Goal by 2030, which includes achieving food security, encompassing interactions between food demand and production within the planetary limits and resource restrictions is critical for a country, such as China, with a large population. The variation in food production and consumption, caused by climate change and dietary transitions, poses challenges for food security. For example, the intake of certain types of ASF greatly increases consistent with the trend for socio-economic development. Crop production is threatened by global climate change and varied planting structures.

In this study, the food security of China in 2025 and 2030 was projected on the basis of the gap between food consumption and crop production. According to feed use for livestock production, four staple food and feed crops (i.e., rice, wheat, maize, and soybean) were selected to measure food security. Performance of Chinese dietary patterns was compared with that of 11 dietary patterns to identify advantages that can effectively ensure food security in China. A hybrid approach was adopted for identification of complex interactions in food security based on copula-based MCMC simulation. Crop yields per unit area were simulated using a DNDC model under four climate-change scenarios (i.e., SSPs 1–26, 2–45, 3–70, and 5–85).

The predicted mean total yields for each crop in China would be 285–319 Mt for maize, 39–44 Mt for soybean, 256–290 Mt for wheat, and 181–207 Mt for rice in the planning years (2025 and 2030). Shortage risks of these core crops were estimated on the basis of the distribution features for the demand and supply of the crops. To compare food demands under different dietary patterns, the relationship between per capita food consumption and GDP in China and other typical dietary habits in high-, middle-, and low-income countries was analyzed, so as to identify relatively effective dietary patterns to ensure food security through a comparative study of variation in food security. The shortage risk of the core crops would be influenced greatly under the different dietary patterns. In the planning years 2025 and 2030, the shortage risks of soybean and rice would be pronounced (i.e., more than 50%), whereas the shortage risks of maize and wheat would not be pronounced. Therefore, as influenced by variation in food consumption, dietary structures, and crop production under changing climatic conditions in the future, stronger conflicts in supply and demand would occur in soybean and rice than in maize and wheat in China.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Meirong Su (sumr@gdut.edu.cn).

Materials availability

This study did not generate new materials.

Data and code availability

This study did not generate any datasets.

All data reported in this paper will be shared by the lead contact upon reasonable request.

This paper does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon reasonable request.

Acknowledgments

This work was supported by the Basic Science Center Project of the National Natural Science Foundation of China (52388101), National Science Foundation of China (Nos. 72122004 and 41801203), and Youth Foundation for Humanities and Social Sciences of the Ministry of Education of China (No. 22YJCZH228). The authors greatly appreciate the editor and the anonymous reviewers for their constructive comments and suggestions, which have helped to improve the paper.

Author contributions

All authors contributed to and collaborated in this research. W.Y., M.S., and Z.Y. developed the methods. W.Y., Y.C., Q.R., and Y.H. provided the case-study data for application of the methods. C.X., J.L., S.Y., D.C., Z.L., and Z.T. analyzed the data. All authors contributed to paper preparation, and have seen and approved the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Software and algorithms
DNDC 9.5	Li et al.58	http://www.dndc.sr.unh.edu
Copula functions: Multivariate Copula Analysis Toolbox (MvCAT)	Sadegh et al.59	https://www.mathworks.com/matlabcentral/fileexchange/69217-multivariate-copula-analysis-toolbox-mvcat?s_tid=srchtitle
Deposited data
Climate change scenarios: CMIP6 (Coupled Model Intercomparison Project Phase 6)	Eyring et al.60	https://esgf-node.llnl.gov/search/cmip6/\
Daily Downscaled Projections of CMIP6	Thrasher et al.61	https://www.nccs.nasa.gov/services/data-collections/land-based-products/nex-gddp-cmip6

Experimental model and study participant details

System boundary

The system boundary for the analysis of food security was developed using input and output features within the framework of life-cycle analysis. Figure 5 depicts the successive stages considered in the system boundary.

The stage of food production

Maize, wheat, rice, and soybean were selected as staple crops in the food production analysis. Six provinces, which collectively contribute 62% and 49% of China’s maize and soybean supplies, respectively, were selected as the main planting districts of maize and soybean. Five administrative regions (Inner Mongolia Autonomous Region and the provinces Heilongjiang, Hebei, Henan, and Shandong), which provide 56% of China’s wheat supplies, were selected as the main planting districts of wheat (Figure 1A). A DNDC model was used to simulate the crop output per unit area during the planning years (i.e., 2025 and 2030) under four climate-change scenarios. Feed or staple crops required for the production of animal-sourced and processed foods (eggs, pork, mutton, beef, milk, soybean oil, seafood, and cake) were converted into the relevant staple crops using coefficients provided by the National Development and Reform Commission.⁶² The transfer coefficients of fodder and the proportions of grain components are shown in Table S9. The crop planting structures varied according to the expectation yields of decision-makers. Thus, the co-relationship between planting structures and crop yields was incorporated into the system boundary of the food security analysis. Although the future development of crop-based biofuels in China remains unclear, given their potential influence on food supplies,⁶³ the CCIA plans to use a limited proportion of maize to produce biofuels in response to the market’s sizeable uptake of maize-based ethanol.⁵⁴

The stage of food demand

We focused on the demands of urban and rural residents at various levels of socio-economic development for grains and ASF. With reference to the Thirteenth Five-Year Plan for Economic and Social Development of the People’s Republic of China (2016–2020), we estimated the urban and rural populations and GDP per capita for the years 2025 and 2030 (Table S4) in 31 administrative regions, including 22 provinces, five autonomous regions, and four municipalities (the cities of Beijing, Tianjin, Shanghai, and Chongqing; Figure 1B). The source data for food consumption per capita, and GDP per capita in China are described in Data S1. The source data for GDP per capita and food consumption per capita under the comparative scenario were obtained from Roser¹ and Ritchie and Roser² (Data S2).

Method details

Food security, which encompasses interactions among food demand and production, climate change, and socio-economic development, is critical for a country, such as China, with a large population. We assessed China’s food security on the basis of projected food production and demand in the context of continuing population growth, changes in dietary preferences, and climate change. Fluctuations in dietary and planting structures were verified according to their influence on food security. Drawing on relevant studies that analyzed the influence of livestock systems (e.g., Herrero et al.⁶⁴), we selected staple crops (i.e., maize, wheat, rice, and soybean), which are used as fodder in the production of ASF, as indicators of food shortage risks. To assess shortage risks for the staple crops, we used a hybrid approach that considered the joint probabilities of correlated variables. The correlations between food production and demand were examined using copula functions and MCMC simulation. To assess food demand, correlations between the GDP and food demand per capita considering multiple dietary preferences were investigated. To investigate food production, we examined the interactions between planting structures—the proportion of a planting area covered by a crop—and crop yields. Climate-change scenarios for SSPs under low, medium, and high forcing levels (i.e., SSPs 1-26, 2-45, 3-70, and 5-85, respectively) were incorporated into a crop simulation model to forecast crop yields.

Analytical framework for evaluation of food security

Our analysis of food security (or the likelihood of a food shortage) in the context of climate change comprised several components (Figure 4): (a) collection of data on planting structures, crop yields, food demand, and meteorological variables under different climate-change scenarios; (b) life-cycle analysis, entailing statistical downscaling methods, and DNDC model simulations; (c) inventory analyses of temperature, precipitation, food consumption, socio-economic development, planting structures, and crop yields; (d) analyses of uncertainties and joint distributions using copula functions and MCMC simulations; and (e) food shortage risk analysis entailing an assessment of food production and demand in the context of advancing socio-economic development and climate change.

Uncertainty analysis

Five copula functions (Frank, Gumbel, Clayton, Student’s t, and Gaussian) were used to examine correlations between food demand and socio-economic development under two scenarios: a) a baseline scenario that described the dietary patterns in eight types of food for urban and rural residents in China, and b) a comparative scenario that described the dietary patterns in seven types of food for typical developed and developing countries. Association parameters for the copula functions were estimated based on MCMC simulation. The indicators RMSE and NSE were used to assess the performance of the copula functions. Under the baseline scenario, 31 administrative regions of China were divided into four regions according to dietary preferences. Association parameters for the fitted copulas for the baseline and comparative scenarios (Tables S5 and S6) indicated that dietary patterns were correlated with socio-economic development.

Copula functions can combine correlations between two univariate distributions within a joint probability distribution.⁶⁵ Given the correlations in food security analysis,^62,63 we used copula functions to derive joint probability distributions for dietary patterns and socio-economic development ($f_{fs}$), and planting structures and crop yields ($f_{pc}$). Assuming that X and Y are the variables for food demand and production with cumulative distribution functions $F_{X_1}\left(x_1\right)$, $F_{Y_1}\left(Y_1\right)$, $F_{X_2}\left(x_2\right)$, and $F_{Y_2}\left(Y_2\right)$⁶⁶, and referring to Sklar,⁶⁷ we derived a unique copula function (C) and joint probability density functions of the two variables [i.e., $f_{fs}\left(x_1,y_1\right)$ or $f_{pc}\left(x_2,y_2\right)$] as follows:

$$f_{fs}\left(x_1,y_1\right)=C_{{\theta }_1}\left(F_{X_1}\left(x_1\right),F_{Y_1}\left(y_1\right)\right)f_{X_1}\left(x_1\right)f_{Y_1}\left(y_1\right)$$ (Equation 1)

$$f_{pc}\left(x_{2t},y_{2\left(\mathrm{t}-1\right)}\right)=C_{{\theta }_2}\left(F_{X_{2t}}\left(x_{2t}\right),F_{Y_{2\left(\mathrm{t}-1\right)}}\left(y_{2\left(\mathrm{t}-1\right)}\right)\right)f_{X_{2t}}\left(x_{2t}\right)f_{Y_{2\left(\mathrm{t}-1\right)}}\left(y_{2\left(\mathrm{t}-1\right)}\right)\text{,}$$ (Equation 2)

where x₁ denotes per capita food consumption, y₁ denotes per capita GDP, x_2t is the ratio of the crop planting area in the tth year, y_2(t−1) is the crop output per unit area in the (t−1)th year, $f_{fs}\left(x_1,y_1\right)$ and $f_{pc}\left(x_{2t},y_{2\left(\mathrm{t}-1\right)}\right)$ are joint probability density functions of correlated variables (i.e., x₁ and y₁, as well as x_2t and y_2(t−1)), and θ₁ and θ₂ are association parameters in the copula functions. Many copula models (e.g., Gaussian, Archimedean, and Student’s t copulas) can be used to capture the dependence features of the above-mentioned correlated parameters.⁶⁸

Quantification and statistical analysis

The Multivariate Copula Analysis Toolbox (MvCAT) was used to approximate the predictive uncertainties of the fitted copulas, and estimate the posterior distribution of the copula parameters using a hybrid-evolution MCMC approach. The indicators RMSE and NSE were calculated and used to assess the performance of the copula functions.⁵⁹

$$RMS{E}_{fs}=\sqrt{\frac{\sum _{i=1}^n{\left[{\stackrel{⌢}{f}}_{fsi}-f_{fsi}\left(\theta \right)\right]}^2}{n}}\text{,}$$ (Equation 3)

$$RMS{E}_{pc}=\sqrt{\frac{\sum _{i=1}^n{\left[{\stackrel{⌢}{f}}_{pci}-f_{pci}\left(\theta \right)\right]}^2}{n}}\text{,}$$ (Equation 4)

$$NS{E}_{fs}=1-\frac{\sum _{i=1}^n{\left[{\stackrel{⌢}{f}}_{fsi}-f_{fsi}\left(\theta \right)\right]}^2}{\sum _{i=1}^n{\left[{\stackrel{⌢}{f}}_{fsi}-{\overline{\stackrel{⌢}{f}}}_{fsi}\right]}^2}$$ (Equation 5)

$$NS{E}_{pc}=1-\frac{\sum _{i=1}^n{\left[{\stackrel{⌢}{f}}_{pci}-f_{pci}\left(\theta \right)\right]}^2}{\sum _{i=1}^n{\left[{\stackrel{⌢}{f}}_{pci}-{\overline{\stackrel{⌢}{f}}}_{pci}\right]}^2}\text{,}$$ (Equation 6)

where RMSE ∈ [0, +∞] and NSE ∈ [-∞, 1] denote the extent of the variation in observed and predicted values, ${\stackrel{⌢}{f}}_{fs}$ is the joint probability distribution of observed dietary patterns and socio-economic development, ${\stackrel{⌢}{f}}_{pc}$ is the joint probability distribution of observed planting structures and crop yields, $f_{fs}$ is the joint probability distribution of dietary patterns and socio-economic development predicted with the copula functions, and $f_{pc}$ is the joint probability distribution of planting structures and crop yields predicted with the copula functions.

To quantify the risk of food shortage, the probability of food demand exceeding food production was calculated. Food production and demand were estimated using the joint probability distributions derived from the copula functions for crop yields and planting structures, and from dietary patterns and socio-economic development. Copula and Latin-hypercube sampling, whose sample size were 1 million and 100 thousand for the prediction of food demand and production, were used to superpose multiple distributions in staple-crop demand and production.

Published: June 9, 2025