The Yangtze River Delta (YRD) is located in the eastern coastal region of China (114°50′–122°30′E, 27°10′–35°10′N) and encompasses all of Jiangsu, Zhejiang, Anhui Provinces, and Shanghai Municipality, totaling 41 cities (Figure 1). The topography of the study area slopes from southwest to northeast and is dominated by plains. The total land area is 358,000 km² and cultivated land accounts for 46% of the total area, which is mainly located in northern Anhui and Jiangsu provinces, which are among the main grain-producing areas in China. The YRD contains the largest urban agglomeration in China and is one of the most active, open, and innovative regions in China (Zhong et al., 2024), with a total GDP of CNY 24.5 trillion in 2020. The favorable geographical location has a strong absorption capacity for the population, with a total population of 235 million in 2020, which means that the region has already become a large food-consuming area, and the large population causes the regional ecological pressure to be too large. The imbalance between regional development and population distribution has led to a large difference between regional food supply and demand, and the contradiction is more prominent, which seriously restricts the stable development of the region and affects regional food security.

The data used in this study included land use data, nighttime light (NTL) data, normalized vegetation index (NDVI) data, grain yield data, road network data, and population distribution data (Table 1). Considering data availability and the size of the study area, all spatial datasets were resampled to a resolution of 1,000 m and unified into the Krasovsky_1940_Albers projection coordinate system. We revised the population distribution data of each province according to the 7th census data and combined the cultivated land NDVI data with the grain output of each province to spatialize the grain data.

Based on the ecosystem service flow perspective, a research framework for depicting the potentially optimal food ecosystem service flow patterns within the region was proposed in this paper (Figure 2). First, the supply and demand of regional food services were analyzed. Second, the minimum cumulative resistance (MCR) model was used to analyze the food flow trends and portray the potentially optimal food service flow patterns. Finally, the Bayesian network and service flow trends were used to determine the optimization areas and corresponding optimization strategies were proposed. Compared with existing studies, this framework can effectively identify the spatial mismatch between the supply and demand of food ecosystem services, scientifically reveal the direction and intensity of the potentially optimal food ecosystem service flows, and show the balancing process of regional food ecosystem services (Zhou and Liu, 2024).

In 2020, the food supply in the YRD was 84,454,200 tons, and the food demand was 31,033,800 tons. There was an obvious spatial imbalance between food production and consumption, with stronger spatial heterogeneity (Figure 3). The food supply in the YRD showed an apparent pattern of being high in the north low in the south, and higher in the west than in the east (Figures 3A1,A2). Heifei, Chuzhou, Lu′an, and Yancheng, primarily situated in plain and hilly regions, had strong food supply capabilities. Specifically, the combined food production of these four cities reached 18.1463 million tons, accounting for 21.48% of the total food supply in the YRD. Moreover, high food demand values were distributed in a sporadic pattern, primarily in coastal areas with high population densities (Figures 3B1,B2). In 2020, Shanghai, Wenzhou, Hangzhou, and Suzhou were notable for their significant food demand, with respective demands of 2.7793, 1.4839, 1.4173, and 1.3913 million tons. And the combined food demand of these four cities constituted 22.79% of the total food demand in the YRD for that year.

Spatially, the contradiction between supply and demand for food ecosystem services was significantly smaller in the north than in the south (Figure 4A1), and the eastern coastal areas in the YRD experienced severe food deficits. The overall self-sufficiency of food ecosystem services in the districts and counties of the YRD was optimistic (Figure 4A2), with 64% of the districts and counties experiencing food surpluses, 34% of the districts and counties experiencing food surpluses of more than 100,000 tons, and 7% of the districts and counties experiencing food surpluses of more than 300,000 tons. In addition, there were significant spatial differences in the food supply–demand ratios. The 21 districts with food supply–demand ratios (FSDRs) below 0.1 were primarily clustered in the municipal districts of provincial capital cities (Figure 4B). Among them, Shanghai stood out for its significant reliance on imported food, with an overall FSDR of 0.33 for the municipality, which was significantly below the standard level for supply–demand equilibrium.

The supply of services was widely distributed in the western part of the YRD, while the demand was concentrated in the eastern and southern coastal areas (Figure 5A). The demand areas were scattered in the municipal districts of the western cities, while they were clustered in the east. In terms of cities, Chuzhou City, Lu′an City, Yancheng City, and Xuancheng City were the supply agglomerations, while Shanghai City, Suzhou City, Wenzhou City, and Wuxi City were the major demand areas. As the economic situation obviously affects population agglomerations and arable land protection, the eastern part of the study area had a higher level of economic development, a stronger pulling force on the neighboring population, and a smaller proportion of arable land, so the greater demand for food in the area was accompanied by a weaker food production capacity in the region, leading to more prominent food conflicts in the eastern part, which has become the main food deficit agglomeration area. From a theoretical perspective, using roads as pathways, a close food network between supply and demand was established in the YRD (Figure 5B), and food ecosystem services flow through the road network within the region. There was obvious spatial differentiation in the flows of food ecosystem services, which were susceptible to the influence of spatial distances. The eastern and southern regions had established closer food linkages with neighboring cities, with higher resource mobility, while the western and northern regions had relatively sparse food linkages.

A total of 187 food service flows were identified, all of which followed the principle of the lowest cumulative resistance cost and flowed from supply to demand locations (Figure 6). First, interdistrict and county food flows were dominated by the first tier (1,134–20,000 tons), in which the Wenzhou districts and counties became the main demand areas, the supply areas were more dispersed, and intracity district and county flows accounted for a larger proportion. The demand areas in the fifth tier of food service flow (150,000–21,3,022 tons) were the Putuo District, Yangpu District, Baoshan District and Minhang District, all of which are municipal districts of Shanghai, with most of the food supply coming from Jiangsu Province. Second, the food network in the YRD had formed a network topology that was dominated by districts and counties such as the Pudong New District, Guangde City, and Lucheng District, with decentralized core nodes. In the food network, the demand nodes were dominated by districts and counties such as the Pudong New Area, Lucheng District, and Gulou District and accounted for a larger proportion of the core nodes, while the number of supply nodes such as Guangde City, Jurong City, and Liyang City was relatively small. In addition, the network as a whole was dispersed, and there was a majority of nodes with a low degree of centrality, which means that most nodes were located at the edge of the network and had fewer neighboring nodes. Overall, most of the strong links were concentrated in the geographic neighborhoods of the core nodes, and most of the food flows were concentrated in local areas, with relatively few food flows with large spatial spans.

The imbalance between supply and demand of food ecosystem services was prevalent in the cities of the YRD, where food ecosystem services were closely linked and flowing clearly. Large urban populations were often accompanied by higher demands for ecological services, and Shanghai, Hangzhou and Nanjing became the core node cities for the food service flows in the YRD (Figure 7), as the main food inflows. Theoretically, the three cities had a total food flow of 4,724,300 tons, accounting for 30.70% of the total food service flow in the study area, with a high food throughput capacity. In particular, Shanghai had the largest food service flow inflow, with an inflow of 1.604 million tons and no food outflows. Hangzhou and Nanjing had the dual status of being both supply and demand areas, but the service flows were mostly inward replenishments, and most of the food outflows were remitted to other districts and counties in the city, with smaller outflows to the outside. In addition, the outflows from Nantong City and Xuancheng City became important food service outflows, with a total of 1,612,100 tons of outflows, where the service is mainly pooled to Shanghai and Hangzhou city. Notably, Wenzhou City and Suzhou City in Jiangsu Province also provided the main inflows of food services, with an external inflow of 1,163,500 tons. The two cities had high degrees of economic development, with strong industrial bases and high development potentials, a strong pull of population mobility, and a high demand for food resources. In addition, Zhoushan City was the only city with a unidirectional inflow of services, and Hefei City, Maanshan City, Huai’an City, Wuhu City, and Huainan City had only intracity flows of services. Overall, the food flows were dominated by interprovincial city-to-city flows, with provincial capitals and metropolitan area core regions receiving the main food service inflows. Inter-provincially, the total flow of ecosystem services in the Yangtze River Delta reached 7,693,700 tons. Within intra-provincial movements, Zhejiang Province exhibited the highest frequency, with a flow volume of 2,142,400 tons. And Jiangsu province stood out with the largest flow of food ecosystem services, amounting to 3,416,800 tons, positioning it as the most significant provider of food ecosystem services to Shanghai.

Based on the Bayesian network, we clearly showed the complex linkages among ecological supply, social demand and food ecosystem service supply–demand matching (Figure 8). We defined the nodal FSDR as the target variable and used the variance coefficient reduction (VR) to assess the degree of influence of the drivers on the food supply–demand balance. The results of the study showed that the percentages of variance reductions were as high as 32.87, 26.02, 24.51, and 23.01% for land use, population, NTL and NDVI, respectively. The sensitivities of the remaining variables to the FSDR were all less than 10.00% of the variance reduction percentage, showing that the food supply and demand balance was too slow to reflect the changes in the status of the remaining variables. Overall, due to the presence of land types such as forest and grassland, the influences of population data and NTL on the balance of food service supply and demand were more pronounced than that of the NDVI.

The subset of optimal states for matching food service supply and demand was {Landuse = Cultivated, Population = Low, Night light = Low, and NDVI = High}, and the results are shown in Figure 9A. The areas with optimal state distributions were mainly concentrated in the western part of the YRD, and Chuzhou city, Lu′an city, and Xuancheng city had the best conditions for food service supplies; these areas could serve as the main food production areas in the YRD and provide a guarantee for regional food flow and food security. In addition, the suitable optimization areas were widely distributed in the western and southern parts of the study area (Figure 9B), where ecological optimization can improve the ecosystem service capacity of the region with maximum efficiency to promote the balanced development of ecosystems and social systems. It is worth noting that only Lu′an City, Chuzhou City and Huainan City in the western part of Anhui Province had good service provisioning effects, while most of the remaining areas had poor service provisioning effects and did not reach the ideal situation, which might be related to the lack of food production of arable land (Zhu et al., 2022; Feng et al., 2022).

Based on the flow pattern of food ecosystem services, the suitable optimization areas were classified into two categories, which were important optimization areas and sub-important optimization areas, which aimed to provide guidance for subsequent food ecological optimizations. The important optimization areas were mainly distributed in the southern part of the YRD, with Zhejiang Province occupying an important position (Figure 10). Among them, suitable optimization areas for food services still existed in service demand areas such as the Huangyan District, Pujiang County, and Kecheng District, where ecological restoration initiatives demand prioritized implementation. Strengthening the rectification of non-agricultural and non-grain use of farmland within a region is conducive to increasing the region’s food self-sufficiency rate, thereby reducing the region’s dependence on external food services and alleviating the contradiction of food resource constraints in the region. In addition, the sub-important optimization areas were clustered in the western part of the study area in a ring shape, with Anhui Province accounting for the largest area. Increasing the investments in agricultural science and technology and accelerating the construction of high-standard farmland to enhance the supply capacity of regional food services are highly important to the overall food security in the YRD and can provide the necessary resources to support the stable development of the region in the future, thus promoting the overall high-quality development of the region.

Depicting the supply and demand flow patterns of food ecosystem services and identifying optimization areas are of significant importance for regional ecological management and food security, which also presents a significant challenge. Therefore, this study explores the possible optimal food ecosystem service flow patterns in the YRD by constructing a theoretical framework, which is more scientific and reasonable compared to previous research. Liu et al. (2022) and Zhou and Liu (2024) simulated the ecosystem service flows by using the E2SFCA algorithm, which emphasized the effect of distance on the flows and flow directions. Shi et al. (2020) used the inverse distance weights between ES supply and demand areas as the flow costs and analyzed the flows of ecosystem services within the city of Shanghai using linear programming and the simplex method. Different from that, we introduced the concept of the MCR in this study, considered multiple influencing factors comprehensively, and determined the supply–demand flow relationships and flow paths from the perspective of flow resistance. Then the direction and flow of food ecosystem services under the natural spillover effect were accurately depicted, and the possible optimal food ecosystem services flow pattern was constructed. Ben et al. (2016) used linear optimization to simulate interprovincial grain trade flows and concluded that the Yangtze River Delta is a dense area for grain trading with obvious internal flows. Jiangsu was found to supply a large scale of grain to Shanghai, which is highly similar to the findings of this paper. Notably, to compensate for micro-scale grain trade data limitations, scholars have employed methodologies including the social equity method (Fan et al., 2018), the CHINAGRO model (Yang T. et al., 2022), and input–output tables (Wang et al., 2023) to characterize virtual food flows through virtual water and land resource transfers. Intriguingly, their analytical frameworks and findings exhibit significant parallels with those presented in this study. Su et al. (2023) constructed an ecological compensation method based on the flow of arable land resources, utilizing the relationship between food production and consumption. The study found that most cities in the YRD are of the outflow type, with Shanghai and Hangzhou having the largest inflows. Shen et al. (2023) explored the pattern of agricultural virtual water flows in the YRD region and discovered that Shanghai tends to establish closer trade relations in agricultural products with bordering regions. The study also revealed that Jiangsu Province is gradually becoming the most important source of virtual water for Shanghai within the YRD. The existing studies are in close alignment with the research findings of this paper. The optimization of ecological functions is the purpose of spatial pattern construction. Zeng and Li (2019) and Tang et al. (2019) used Bayesian belief networks to screen a subset of key variables and extract ecosystem service optimization zones based on a realistic perspective; Wu J. et al. (2023) used Bayesian networks to determine subsets of optimal states and analyze them in comparison with the real situation and then extracted optimization areas. These studies confirm the rationality of using Bayesian beliefs for identifying optimized zones in this paper.

Food security is not only affected by the supply and demand of food ecosystem services but also closely related to food consumption and trade, of which food trade has become an important means of balancing regional differences in food supply and demand (Yin et al., 2024). From the actual situation of regional food trade, in the YRD region, the economy is flourishing, the population is dense, and the demand for food is high. It has become the region with the most concentrated food demand and the most active food circulation (Yao, 2021). Interprovincial food linkages play an important role, and Shandong Province, Henan Province, and other regions have become the main sources of food trade in the YRD (Yang Q. et al., 2022). This is because the demand for food mainly consists of four parts: rations, feed grain, industrial grain and seed grain. Although the demand for rations is less than the total regional food supply, the regional food production cannot fully support the amount of food needed for regional development, which leads to the region still in an overall food deficit and needs external replenishment (Yao, 2021). Secondly, the structure imbalance of food varieties is also one of the main reasons for the gap between regional food supply and demand. With the development of the economy, the living standard of people has been greatly improved, and people’s demand for food has gradually tended to be diversified and high-quality (Wu, 2023). On the one hand, rice and wheat are the main types of food production in the region, and the production of beans is relatively small but in high demand, so beans are obviously dependent on the other areas.

There are certain differences between the actual pattern of food trade and the pattern of food ecosystem services supply and demand flows due to their differing focuses and influencing factors. Food trade is a conscious food transaction activity, and the distribution and circulation of food under market mechanisms is a key concern of its research (Wang J. et al., 2021). Driven by a high focus on economic profits, food surplus regions tend to maintain closer ties with economically developed regions, and food transfers tend to flow toward developed regions, impacting a wide range of areas (Dang, 2020). In food trade research, economic factors such as market transactions, prices, quantities, and trade policies of food commodities are the core concerns, to maximize economic benefits (Guan et al., 2022). In contrast, the flow of food ecosystem services emphasizes the connection between natural ecosystems and human wellbeing, focusing on reflecting the possible optimal flow process within a region, and representing the long—term and fundamental ecological and social values. This flow is also subject to practical factors such as transportation conditions, the distance between the supply and beneficiary areas, and the supply–demand relationship of ecosystem services, resulting in a relatively limited scope of influence (Wang and Zhou, 2019). Moreover, the focus of the food ESF is on matching the supply of and demand for ecosystem services within regions. This research highlights the complementarity of ecosystem services across different areas within a region, aiming to contribute to the region’s sustainable development and maximize the overall benefits of ecosystems (Zhu et al., 2019). Overall, the research on ecosystem service flows provides a complementary perspective for analyzing food production and trade from an ecological standpoint, which contributes to a more comprehensive understanding of regional ecosystem services and food security (Guan et al., 2022). In the future management of regional food security, the potential risks and potentials in regional food production and trade can be assessed in the context of regional food ecosystem service flow patterns, to optimize food trade and promote food security and ecological balance (Xiao et al., 2022).

Strengthening sustainable food production capacity is an essential foundation for achieving national food security and ensuring stable economic development (Gao et al., 2024) Identifying priority ecological restoration areas from the perspective of supply–demand balance and exploring the potential of regional food ecosystem services can make ecological interventions generate higher ecological and social values (Qiu et al., 2024). This could improve the efficiency and feasibility of ecological protection and restoration, and implement the food crop production strategy based on farmland management and technological application (Chen Zhen et al., 2024).

The areas without a food supply capacity in the identified optimization areas occupied relatively large areas and were concentrated in the southern and western parts of the study area. The matching of food service supply and demand in important optimization areas was poor, with a large number of locations without a supply capacity and a relatively small number of patches with FSDR > 10, which were scattered in Huainan City, Hefei City, Yancheng City and other places. Since most of the districts and counties in the important optimization areas were in a state of tight balance between supply and demand and were also greatly affected by the service demand areas, the ecological optimization of these areas has a certain degree of urgency and importance, and the ecological restoration of these areas can maximize the stability of the service flows and alleviate the contradiction of the regional food resource constraints. Strengthening the remediation of nonagricultural and nonfood cultivated lands and improving the market adaptability to increase the food supply capacity to a greater extent have become the focus of future developments in this field. In addition, the degree of matching of food ecosystem services in the sub-important optimization areas was relatively good, and the patches with FSDR > 10 occupied a large area and were distributed in a ring-like contiguous patch, which was concentrated in the southern part of Anhui Province, such as Lu′an city, Anqing city and other places. This type of area has a good supply base for food services, and ecological optimization work in this area is conducive to improving the overall food security in the YRD. Increasing investments in agricultural science and technology and accelerating the construction of high-standard farmland to explore the potential for increased food production are priorities for the future development of these areas, which will help to give full play to the advantages of natural resources to create a higher level of Jianghuai Granary and promote the transfer of ecological and social resources.

There is still room for continued depth and refinement in this study. First, remote sensing data, such as NDVI, are influenced by factors like cloud cover and atmospheric conditions, introducing some uncertainty into the results. To enhance the robustness of the data in the future, the reliance on a single data source can be diminished by integrating data from various remote sensing sources (Wang, 2021; Shen et al., 2021). Second, since the assignments of resistance factors have not yet reached a unified standard, this paper, while referencing previous research on resistance assignment in the MCR model, acknowledges that the standard values and assignment results inevitably retain a degree of subjectivity. Therefore, it is essential to optimize the data using more accurate information in the future to enhance the relevance of the research findings. Third, the constraints of regional boundaries greatly simplify the complexity of actual food flows within the region and neglect the close connections the region has with the outside world. Future research should consider the inflow of external ecosystem services, emphasize the openness of the ecosystem, and analyze ecological service flows across a broader area. Additionally, achieving a balance in food ecosystem services involves not only spatial equilibrium but also structural and temporal balances. In future research, incorporating the specific supply and demand volumes of individual crops, as well as historical food surplus data, into the calculations of food ecosystem service supply and demand will more accurately reveal regional food supply and demand patterns. Finally, while this study has explored the potentially optimal flow pattern of food ecosystem services in the YRD by constructing a framework, the research findings still need to be further verified with micro-level empirical data. Furthermore, the key to enhancing research accuracy and practical relevance lies in effectively integrating food ecosystem service supply–demand flows with actual food trade patterns to accurately characterize service flow mechanisms. In future research, building on field investigations to obtain micro-level empirical data should be prioritized to validate the scientific validity and practical applicability of the findings. Concurrently, the integration of interdisciplinary theoretical frameworks and methodological approaches will further refine the research framework, thereby driving forward the continuous advancement of ecosystem service flow studies.