Rational spatial arrangement of cropland fallow incorporating connectivity alleviates health-productivity conflict in China

Cropland fallow (CF) is a crucial strategy many developed countries and regions use to balance food supply with the agroecological environment¹. Rather than mere land abandonment, CF is recognized as an improved management process where land users undertake specific measures to promote sustainable land use². Depending on specific degradation contexts, this active ecological restoration involves altering crop types, such as planting green manure or hyperaccumulators to remediate polluted soils, or temporarily ceasing intensive farming through seasonal crop rotation and conservation tillage to restore soil health. The scale and spatial arrangement of cropland fallow (SACF) provide insights into a country or region’s food-production dynamics³, eco-environmental issues⁴, farmers’ willingness to participate⁵, and the influences of international food markets⁶.

Due to global population growth and accelerated urbanization, many countries, especially developing nations, have simultaneously engaged in high-intensity continuous cropland use and applied excessive fertilizers and pesticides. Furthermore, they have implemented fallow measures in inappropriate locations that fail to address ecological degradation while threatening food supply, leading to cropland overuse, soil degradation, severe pollution, and ecological vulnerability^7,8. These problems significantly hinder progress toward the United Nations’ Sustainable Development Goals (e.g., SDG 2, 3, 8, and 15). Therefore, standardizing and quantifying SACF is crucial for promoting the “increasing grain production through soil conservation” strategy and achieving sustainable development.

Previous studies on SACF have primarily focused on several aspects: the dependency of SACF on cropland natural environments⁹; the relationship between the optimization of SACF and cost-effectiveness¹⁰; sustainable utilization methods of fallow cropland in response to changes in food demand and the ecological environment¹¹; and the spatial identification of target areas¹², the hierarchical zoning of fallow priorities¹³, and the regional heterogeneity of these arrangements¹⁰. Although this literature considers both natural and human dimensions of cropland systems, important limitations remain. First, in terms of spatial scale, most analyzes focus on regional or smaller areas, which limits quantitative capacity and generalization to large scales. Second, current SACF research predominantly targets specific environmental issues already designated by national pilot programs, such as heavy metal pollution, groundwater over-extraction, or severe ecological degradation. Consequently, these studies often fail to conduct a holistic evaluation of the intrinsic production and ecological challenges starting from the cropland system itself. As a result, it becomes difficult to effectively identify the pressing issues that croplands face, leading to a lack of prioritization for regions where fallow is urgently needed. These challenges significantly hinder our ability to understand CF’s urgency and diagnostic features on a large scale and within specific timeframes.

Meanwhile, the increasingly prominent issues of frequent extreme weather events, severe pollution, and resource shortages have drawn significant attention to the use of fallow for cropland protection and restoration in sustainable development, land management, and regional planning^14,15,16. There is an urgent demand for accurate information on critical cropland problems, yet such information remains severely insufficient. This lack of accurate information limits the ability to conduct standardized and quantitative analyzes to identify “problematic” croplands with high fallow urgency. Consequently, the scientific references for cropland protection and relevant policy-making remain limited.

Some scholars have diagnosed prominent cropland issues and inferred fallow urgency by integrating evidence on cropland health (CH)¹³. However, in many such studies, cropland health and productivity (CH-P) are proxied mainly by cropland quality or productivity alone. In practice, the theory of CH-P has significantly influenced the goal-oriented approach to cropland fallow (CF), emphasizing the need to balance functions such as crop production, social security, ecological services, and environmental maintenance¹⁷. Accordingly, assessments of fallow urgency should move beyond single-dimension evaluations (cropland quality or cropland productivity (CP) only) toward a comprehensive CH-P appraisal. Furthermore, while a robust CH-P assessment identifies the intrinsic need for fallow, practical implementation requires consideration of the spatial organization of these targeted areas. Concentrated and connected CF can deliver greater benefits and lower costs than fragmented CF. Here, cropland connectivity refers strictly to management-oriented structural connectivity. It describes the spatial contiguity of similar cropland parcels that facilitates efficient agricultural operations, rather than ecological landscape processes such as water movement or biodiversity habitats. However, most existing research on connectivity emphasizes spatial adjacency only, overlooking different post-fallow management needs across cropland types (dryland, irrigated land, and paddy fields) and across CH/CP classes. These considerations underscore the need for a method that jointly captures ecological-environmental, socio-economic, and cost-effectiveness benefits while spatially coupling CH–CP–CC (cropland health, cropland productivity, and cropland connectivity) to map fallow urgency and generate multi-dimensional, decision-oriented results.

Therefore, in order to standardize and quantify the framework for constructing SACF, this study has two objectives: (1) to propose a spatial coupling mapping method for assessing the urgency of cropland fallow, which can reveal the prominent contradictions in cropland and generate a baseline map of fallow urgency across China’s cropland using data from the year 2020; (2) to investigate the trade-off rules for SACF based on CH-CP-CC, and to develop a spatial arrangement plan for CF in China based on this 2020 baseline evaluation. To the best of our knowledge, this study is among the first at the national scale to propose a standardized and quantitative framework for SACF based on the coupling of CH-CP-CC and to deliver reproducible arrangement rules together with a quantified national baseline plan (Fig. 1). We intend this framework to provide a scientific reference for global research on SACF.

Single-factor evaluation of cropland health

Single-factor assessments show wide within-region variation in cropland health (CH) (Fig. 2). Substandard irrigation water is concentrated in the Hai, Songliao, Yellow, and Huai River Basins, reflecting water scarcity in northern China and aligning with previous maps of sewage-irrigated cropland in China¹⁸. For soil heavy metal pollution, conditions are generally better in the north. Outside parts of the North China Plain (NCP), the Huang-Huai-Hai Plain (HHHP), the Loess Plateau (LP), and the Northern Arid and Semiarid Region (NASR), the eight heavy metal (Cd, As, Cr, Hg, Zn, Cu, Ni, and Pb) contents across soil typically meet standards. National shares are light 11.59%, moderate 13.35%, and severe 2.40%. Severe pollution clusters in the southern parts of the Yunnan-Guizhou Plateau (YGP), the south of the Middle Yangtze Plain (MYP), the eastern HHHP, and central Southern China (SC). Moderate pollution is mainly found in YGP, the Sichuan Basin and Surrounding Regions (SBSR), MYP, and SC, with additional occurrences in NASR and smaller areas along the coastal regions of NCP and HHHP.

The potential soil erosion intensity of cropland reflects multiple drivers. Rainfall is a primary force¹⁹, and topography, especially slope and slope length, controls runoff formation and soil loss²⁰. Together, these factors make erosion more severe in the hilly regions of southern China (including SC and southern MYP), on plateaus such as YGP and southwestern LP, and in basin landscapes in the southwestern SBSR. This spatial pattern is inherently consistent with the statistical distribution shown in Fig. 4c, where the area percentage of cropland with severe erosion (“Poor” and “Very Poor” categories) remains very small across all agricultural zones. Consequently, while these areas appear as discrete, highly fragmented pixels in Fig. 2c rather than prominent contiguous clusters, their localized presence underscores acute ecological constraints in southern and plateau landscapes. By contrast, the predominantly flat northern regions experience lower soil erosion intensity due to minimal soil and water loss.

Earthworm density in cropland is shaped by climate, season, soil conditions, and management. As moisture-loving, shade-preferring ectotherms, earthworms perform best at moderate temperatures²¹; accordingly, densities are typically lower in regions constrained by cold, arid, or excessively hot climates. As visually evidenced in Fig. 2d and statistically corroborated by the lower average regional scores in Fig. 4d, these areas primarily correspond to NASR, NCP, and YGP. Furthermore, annual grain production variability (Fig. 2e) shows higher fluctuations primarily in northern and western transitional zones (such as LP and NASR), reflecting areas with more fragile or highly variable farming environments. Soil cation exchange capacity (0–100 cm) is generally lower in SC, MYP, YGP, NASR, and the Qinghai-Tibet Plateau (QTP), and higher in HHHP, NCP, LP, and parts of the eastern SBSR.

Comprehensive evaluation of cropland health

Based on the spatial evaluation of CH (Fig. 5a), and as illustrated by the statistical results in Fig. 5b, we define critical thresholds for CH using the Jenks Natural Breaks method. The threshold for medium to high health is 71.052, and the threshold for low to medium health is 58.816. On this basis, CH is classified into High, Medium, and Low, accounting for 48.15%, 29.97%, and 21.88% of national cropland, respectively. For high-health cropland, key influencing factors are at favorable levels, ensuring the safety of the cropland for a specific period. For medium-health cropland, some factors have gradually degraded due to improper use, becoming potential risks that require preventive measures to stop further deterioration of cropland health. Low-health cropland conditions face serious threats, and the focus should be on identifying the limiting and driving factors affecting cropland use. Targeted fallow management and protection measures should be implemented to improve CH and continuously monitor health evaluation indicators.

Regarding spatial distribution characteristics (Fig. 5a), high- and medium-health croplands are predominantly situated in central and northern regions, the eastern region, and parts of the southwest, such as MYP, HHHP, LP, NASR, NCP, and the central parts of SBSR. High-health croplands are notably concentrated in the Jianghan Plain, the North Jiangsu Plain, the Huanghuai Plain, the Haihe Plain, the Yinwu Plain, the Songnen Plain, and the Chengdu Plain. In contrast, low-health croplands are primarily located in China’s eastern, southern, and southwestern regions, particularly in mountainous or hilly areas like YGP, SBSR, and the southern parts of MYP. Medium-health croplands—transitional between high and low—are the most widespread and occur in mountain and plain landscapes.

As depicted in Fig. 4a–f, the key factors limiting CH vary markedly by agricultural zones, so protection strategies must target each zone’s specific weaknesses. In the SBSR and SC regions, the critical issues for CH are soil heavy metal pollution and poor cation exchange capacity. In the YGP region, the average score for soil heavy metal pollution is below the low-to-medium health threshold. In contrast, the average score for cation exchange capacity falls below the medium-to-high health threshold. In the MYP region, the average scores for soil heavy metal pollution and cation exchange capacity are below the medium-to-high health threshold. Conversely, in most other regions, the indicator averages exceed the medium-to-high threshold. Overall, regions characterized by plateaus, mountains, or hills have lower levels of CH than those dominated by plains, primarily facing health issues such as excessive soil heavy metal content and poor cation exchange capacity. However, each area’s specific cropland health issues must be analyzed based on local characteristics.

Single-factor evaluation of cropland productivity

The results of the single-factor evaluation of CP show a certain regularity in the distribution of values across different regions (Fig. 3). Climate production potential forms distinct bands across China. The high-value zone (100 points) lies mainly in SC and YGP. Medium-high values (60–80 points) occur south of the Yangtze River and east of Yunnan-Guizhou Province, with annual rainfall between 1200 and 1600 mm, including northern SC, MYP, central and northern YGP, southern HHHP, and SBSR. The medium-value zone (40 points) extends north of the Yangtze River to the Greater Khingan Mountains, covering LP, NCP, and central and northern HHHP. In contrast, the inland Northwest and the Qinghai-Tibetan Plateau (NASR and QTP) are classified as low-value zones (20 points) for climate production potential due to annual rainfall and temperature limitations.

Regarding soil properties, tillage texture evaluations show that loam and clay prevail in the south, sandy soils in the north, and loam occurs in small, localized clusters in the central NCP. Soil bulk density (0–100 cm) shows a south-north contrast: lower values occur near the junction of NASR, LP, and NCP. Cropland in southern China—SC, southern MYP, and southwestern YGP—has pH (0–100 cm) closer to agronomic optima than other regions, especially compared with the north. Soil organic matter content (0–100 cm) is higher across the southern middle-lower Yangtze River (MYP, SC, YGP, and SBSR) and NCP, and lower in NASR, LP, HHHP, and QTP.

The level of irrigation guarantee is generally higher in the east and center—notably MYP, HHHP—and lower in the west and northeast. Due to the practical construction of well-facilitated farmland in China in recent years, cropland with adequate irrigation assurance is distributed in patches within the NCP. Flood protection is likewise stronger in eastern China, weaker toward the west, and often low in the southwest and northwest. Agricultural mechanization and cultivation distance indicator scores are notably higher in plain regions compared to mountainous, hilly, or plateau areas. Furthermore, the per capita disposable income of rural residents is generally lower in regions such as YGP, LP, NASR, and NCP. Conversely, MYP and SC show higher rural incomes. Cropping intensity is higher in the HHHP and northern MYP.

Comprehensive evaluation of cropland productivity

According to the Jenks Natural Breaks method, we define the medium-high threshold for CP as 69.886 and the medium-low threshold as 59.667. On this basis, China’s CP is divided into three categories based on these thresholds: High (27.54%), Medium (42%), and Low (30.45%). High-productivity cropland zones combine favorable climate, fertile soils, advanced technology, and strong agroeconomic conditions, making them well-suited to staple-crop production. Medium CP zones suffer from suboptimal utilization of certain factors, leading to decreased production capacity. These areas are gradually becoming a hidden risk to CP, and preventive measures are necessary to avoid further deterioration. Low CP zones face adverse weather conditions, soil infertility, limited technology, and unfavorable economic circumstances; they require targeted diagnosis to identify binding constraints and guide fallow and improvement actions. Based on China’s CP assessment (Fig. 5c), CP exhibits a southeast-high, northwest-low pattern. The eastern belts HHHP, MYP, SC, and NCP—generally show higher CP than the western plateaus and basins—YGP, SBSR, LP, QTP, and NASR. Notably, within east-central SBSR, the Sichuan Basin forms a concentrated distribution of high-productivity cropland.

As shown in Fig. 4g–r, the limiting indicators of CP vary in different regions, and there are also regional common problems. As a limiting indicator of CP, climate production potential is influenced by regional climatic factors, mainly in the northern and western zones such as NCP, NASR, LP, and QTP. Slope limits CP in areas such as QTP, SBSR, and YGP, and cultivation distance shows a similar topographic signal—below the low-medium threshold in QTP, SBSR, and YGP, and below the medium-high threshold in southern NASR, where plateaus meet mountains and populations are sparse. Irrigation is a key bottleneck in NCP, NASR, LP, HHHP, and QTP; in some places, the constraint is insufficient infrastructure, while in HHHP, it is limited water supply. Since over 60% of China is hilly or mountainous, mechanization has historically lagged in complex terrain, further constraining CP. The per capita disposable income of rural residents is below the low-medium threshold in all regions except for SBSR, MYP, and SC, reducing farmers’ ability to invest in tillage, planting, pest management, and harvesting. The cropping intensity is below the low-moderate thresholds in NASR and QTP, due to abandonment/fallow behavior and climatic limits, suppressing CP.

Cropland connectivity evaluation analysis

According to the method and thresholds for measuring cropland connectivity (CC) (see the section “Methods”), which consider cropland type, health, and productivity, China’s CC is categorized as High (50%), Medium (38.92%), and Low (11.08%) (Fig. 5f). Generally, CC is better in China’s plains compared to the mountains, hills, and plateaus, with the HHHP, northern MYP, and NCP regions showing better connectivity than others (Fig. 5e). When we compared the results of the single-factor evaluation of cropland slope, we observed that their spatial distribution characteristics are pretty similar. In areas with predominantly flat topography, CC is high. In contrast, due to the complex and varied topography, croplands tend to be more scattered and disconnected in mountainous, hilly, and plateau regions. The zones with low CC are primarily situated in the MYP, HHHP, SBSR, and YGP areas, where the terrain consists mainly of hills, mountains, or basins, leading to significant cropland fragmentation. Furthermore, low-connectivity croplands are also present in regions with diverse types of cropland. For example, dryland, paddy fields, and irrigated land exist in some areas. Considering the various subsequent fallow measures for each type, despite their spatial proximity, these different cropland types do not fit the definition of relative connectivity with multiple attributes as proposed in this study.

Zoning for the urgency of cropland fallow in China

We classify cropland into 27 fallow urgency types by coupling CH, CP, and CC (Fig. 6) and applying the spatial zoning rules described in “Methods”. Nationally, fallow urgency is higher in northern China than in the south (Fig. 7a). The urgent fallow zone (UFZ) and priority fallow zone (POFZ) concentrate in HHHP, LP, and NASR—particularly at their regional junctions—where High-Low-High (HLH) and High-Low-Medium (HLM) types dominate (Fig. 6a). In these areas, low CP is the primary constraint, shaped by natural factors (climate conditions and tillage texture) and socio-economic limits (irrigation guarantee level, flood protection level, agricultural mechanization level, and per capita disposable income of rural residents). In eastern HHHP, UFZs also occur in groups dominated by LHH—here, low CH is driven primarily by soil heavy metal pollution (Fig. 6b). Prior studies identify severe soil heavy metal pollution in gold mining belts such as Laizhou-Zhaoyuan-Yantai and Muping-Rushan in Shandong Province, and in densely urbanized areas—patterns consistent with our results²². In MYP, UFZ and POFZ appear as fragmented clusters in southern Hunan Province and northern Jiangxi Province—including Changsha-Zhuzhou-Xiangtan and Hengyang-Yongzhou-Shaoyang-Loudi in Hunan Province (Fig. 6d), and the Nanchang-Fuzhou-Yichun-Shangrao-Jiujiang in northern Jiangxi Province (Fig. 6c). These clusters are mainly LHH and LHM, with low CH linked to soil heavy metal pollution and low cation exchange capacity.

Results of cropland fallow in China

Stable cropland area underpins sustainable grain production and thus human well-being, social development, and national security^23,24. We therefore optimize CF under a food-security constraint, setting an upper bound of 20% of national cropland in fallow, following prior work^10,25.

According to CF urgency zoning results, we designate UFZ and POFZ as fallow zones and all others as non-fallow (Fig. 7a–d). The resulting shares are 14.67% fallow and 85.33% non-fallow, well below the 20% cap. Fallow land is concentrated in the north—notably NASR, LP, and HHHP—with NASR contributing the largest fallow area (4.72% of the national cropland). In these northern regions, the dominant pattern is high health and low productivity, driven jointly by natural limits (e.g., climate production potential, slope, tillage texture, and organic matter content) and socio-economic constraints (e.g., irrigation guarantee level, flood protection level, agricultural mechanization level, and per capita disposable income of rural residents), which together depress productivity. In contrast, in southern MYP, cropland is mainly low health and high productivity; soil heavy metal pollution and low cation exchange capacity are the primary key constraints. Here, the fallow goal is to improve CH while maintaining CP.

This study presents a framework for the SACF encompassing the cropland’s natural resources and socio-economic and ecological environment subsystems. This conceptualization fundamentally aligns with the classic models depicting the three pillars of sustainability (biophysical, economic, and social) within complex agricultural systems²⁶. Furthermore, it draws on established land system theories that examine mutual feedback among natural, economic, and managerial systems²⁷. Building upon these tripartite frameworks, our study specifically adapts this logic to focus on the interactions among natural-resource, socio-economic, and eco-environmental factors that affect cropland health, productivity, and connectivity patterns, as well as the uncertainties arising from human utilization. Building on existing evaluation indicator systems and methodologies, we identified patterns in the critical factors influencing cropland health, productivity, and connectivity in China. Through spatial coupling analysis of these three dimensions, we explored their interaction characteristics to prioritize areas with prominent and concentrated contradictions in fallow. This approach aims to provide information support for SACF in different regions, with several aspects requiring further discussion:

The proposed SACF method provides a structured and quantitative approach for identifying fallow-priority areas. Based on the spatial coupling results and spatial trade-off rules of cropland health, productivity, and connectivity, we conducted a standardized and quantitative identification of fallow areas. On the one hand, the evaluation method and indicator system are primarily based on industry/national standards and relevant existing research outcomes, constructing a comprehensive framework for evaluating CH-P from multiple dimensions and functional layers of the cropland system. This approach allows for a comparative analysis of China’s CH and CP levels. On the other hand, many existing studies overlook the differences in fallow management practices required for different types of cropland, such as dryland, paddy fields, and irrigated land. Instead, they only assess connectivity based on spatial adjacency. This approach leads to the identification of cropland within the same concentrated and connected region, where differentiated measures are still needed for fallow management despite being in the same area. Otherwise, achieving uniform management of fallow becomes difficult, potentially increasing costs and reducing work efficiency, making it challenging to maximize fallow’s socio-economic, ecological, and cost benefits. This study proposes a relatively connected algorithm that considers multi-attribute differences to address these issues. The proposed algorithm facilitates the efficient implementation of standardized regional fallow measures while ensuring effective coordination with administrative management departments.

The multi-dimensional fallow spatial trade-off rule, guided by comprehensive benefits, aims to maximize fallow outcomes over a defined policy horizon while respecting budget and food-security constraints. The SACF should jointly consider ecological–environmental, socio-economic, and cost-effectiveness benefits rather than treating them in isolation. In contrast, many previous studies focus on ecological-environmental or socio-economic benefits or combine the two via weighted summation to identify cropland fallow (CF). However, how to operationalize comprehensive benefits in spatial prioritization has been underexplored. We argue that simply labeling areas with the lowest CH or CP as “high-priority fallow” is inappropriate; where both CH and CP are poor, required investments can be prohibitive, and expected gains uncertain. Given finite public budgets and the need to maintain production capacity, the SACF should first target cropland with high urgency, relatively greater expected benefits, and stronger spatial contiguity.

Accordingly, we adopt the principle of “prominent contradiction (i.e., conflict between cropland health and productivity), concentrated and connected (spatially contiguous; high connectivity) “: the CH–CP conflict is the primary decision factor, with cropland connectivity (CC) as a secondary factor to improve implementability and reduce transaction and supervision costs. Practically, the outputs translate into three policy levers. (i) Subsidy/quota targeting: counties with larger shares of UFZ/POFZ and higher CC can receive priority quotas and per-hectare payments, with differentiated rates for contrasting conflict types (e.g., HLH vs. LHH) to reflect distinct management costs. (ii) Instrument choice: CH–CP classes provide a decision pathway between fallow with productivity-enhancing support (typical in HLH/HLM zones) versus remediation fallow or selective medium-term retirement where health deficits dominate (LHH/LHM). (iii) Regional production strategies: consistent with our national pattern, the north emphasizes raising CP (irrigation reliability, mechanization, soil organic matter), whereas the south emphasizes raising CH (pollution control, irrigation-water quality, buffering capacity). For implementation, 30-m grids can be aggregated to administrative units to set eligibility and quotas; CC can guide block design to capture scale economies; and entry/exit criteria can be linked to measurable improvements in CH–P. Future work will calibrate parameters (e.g., connectivity weights, payment coefficients, contract lengths) and evaluate sequencing under multi-year CH–CP–CC data.

Using a unified set of indicators, weights, and connectivity thresholds makes it challenging to reflect the conditions of different regions in China fully. The construction of the indicator system, determination of weights, and grading of connectivity should comprehensively consider regional natural resources, socio-economic, and ecological conditions. Identifying specific limiting indicators and connectivity thresholds for each region is necessary, providing a crucial basis for cropland fallow decisions. However, this study did not sufficiently refine the regional indicator systems and weights, nor did it differentiate the connectivity of cropland across regions. In future research, we will continue to improve the regional evaluation indicator systems, weights, and grading of cropland connectivity to more comprehensively reflect cropland health, productivity, and connectivity.

The dynamic evolution of CH–P, driven by both historical human utilization and future environmental shifts, inevitably impacts the long-term stability and effectiveness of the SACF. Because long-term, continuous observations for several indicators are unavailable, our analysis adopts 2020 as a nationally consistent baseline snapshot and conducts a static, single-year assessment of CH–P. This choice may introduce a one-sided view that emphasizes temporal trends; therefore, our results should be interpreted as a baseline rather than a trajectory. For example, parts of the Northeast black-soil region have experienced sustained, high-intensity farming that depletes high-quality soils; despite higher current productivity, such downward trends in health should not be overlooked. Looking forward, the SACF must also adapt to future scenarios driven by global environmental change. For instance, climate change may alter agricultural climatic resources (e.g., precipitation and temperature suitability), thereby shifting the spatial distribution of high-productivity zones. Similarly, rapid urban expansion and land-use changes could intensify landscape fragmentation, reducing cropland connectivity in peri-urban areas. Therefore, future research should integrate predictive models (e.g., climate scenarios and land-use simulations) into the CH-CP-CC framework. This evolution will transition the SACF from static zoning to a dynamic, adaptive management strategy, allowing policymakers to proactively identify emerging fallow needs and balance long-term food security with ecological resilience.

Our results broadly align with the actual rollout of China’s fallow and rotation policy, though important differences also emerge. In practice, pilot programs have focused on groundwater over-extraction zones (e.g., Hebei’s Heilonggang funnel, Xinjiang’s Tarim Basin), heavy-metal pollution zones (e.g., Changsha–Zhuzhou–Xiangtan in Hunan), and ecologically fragile areas in southwest karst and northwest degraded regions (e.g., Gansu, Guizhou, Yunnan, Guangxi). National-scale research¹⁰ likewise prioritizes these nationally emphasized constraints, so strong concordance is expected. Our results successfully capture many of these regions, particularly those affected by soil pollution and ecological vulnerability. The main difference is that groundwater depletion areas were not highlighted, since hydrogeological stress was not included in our CH–CP index system. Future work will address this limitation by integrating groundwater monitoring into the framework. Our framework also identifies extensive fallow-priority areas in the northern agro-pastoral ecotone, corresponding to ongoing rotation trials in this fragile belt. This convergence underscores the potential of the CH–CP–CC framework to reveal vulnerable zones where fallow or rotation can ease ecological stress and sustain productivity.

Cropland health and productivity evaluation system

Constructing a CH-P evaluation system can provide scientific support for identifying fundamental contradictions in cropland. As research on CH-P continues to evolve, the focus has gradually shifted from the soil system^28,29 towards the cropland system³⁰. The cropland system typically comprises three subsystems: natural resources, socio-economics, and the ecological environment, covering production, livelihood, and ecology³¹. These correspond to the functions of crop production, social security, environmental maintenance, and ecological services provided by the cropland system. Specifically, the cropland system is an integrated “natural resource-socio-economic-ecological environment” complex, consisting of the cropland natural resource system made up of natural elements such as topography, soil, and climate, the cropland socio-economic system made up of socio-economic elements such as agricultural technology levels and agricultural economic conditions, and the cropland ecological environment system made up of ecological elements such as soil heavy metal pollution, irrigation water quality, and soil earthworm density. It serves as a major food production area and one of the sources of socio-economic benefits while also playing a vital role in regulating the regional ecological environment. The combined effects of the cropland natural resource and socio-economic subsystems influence CP, while the cropland ecological environment subsystem reflects CH status. Therefore, based on the natural resource endowments, socio-economic development, and ecological environment conditions of cropland, this study breaks down cropland system services into CP and CH and constructs a CH-P evaluation framework to improve the accuracy of CH-P assessments (Fig. 8).

The health dimension concentrates on assessing cropland’s ecological services and environmental maintenance capability. Such an assessment involves evaluating the quality of the cropland production environment and its resilience in returning to its original health level after being disrupted by external factors^32,33. Guided by soil-health and ecosystem-health perspectives widely used in the literature^34,35 (e.g., the Cornell Soil Health framework based on physical, chemical, and biological indicators²⁸) and by comprehensive CH frameworks^30,36,37, we select six representative indicators that capture both farming-environment status and production elasticity. Specifically, the environmental sustainability capacity of croplands reflects their ability to preserve their own and the surrounding environmental health, including pollution control, potential erosion, and biological characteristics. This study employs four indicators—soil heavy metal pollution, irrigation water quality, potential soil erosion intensity, and soil earthworm density—to gauge the status of the farming environment. Among them, soil heavy metal pollution and irrigation water quality are vital indicators of cropland pollution.

In contrast, the potential intensity of soil erosion can signal erosion risk in cultivated areas²⁰. Earthworms play a vital role in providing various ecosystem functions and services³⁸, and their density can reflect the biological characteristics of cropland, typically measured by the number of earthworms per unit area of soil. The ecological service capacity of cropland demonstrates its ability to provide ecological services, directly influencing its productivity and resilience under different environmental and stress conditions. Therefore, this study aims to assess cropland’s production elasticity by analyzing annual grain production fluctuations and the soil’s capacity to retain and release nutrients. The coefficient of variation of annual grain production is commonly used to gauge food security or the resilience of grain production against natural disasters³⁹. At the same time, soil cation exchange capacity indicates the soil’s ability to retain and release nutrients such as cations, serving as a vital indicator of soil buffering performance and resilience⁴⁰. These six indicators provide a coherent basis for CH evaluation (Table 1).

The productivity dimension represents the cropland’s capacity for crop production, which is crucial for ensuring food security and socio-economic stability⁴¹. This dimension can be characterized by two key aspects: crop production capacity and social security capacity^30,36,37. The crop production capacity denotes the ability of cropland to yield crops, influenced by natural resources and socio-economic factors. These include climate conditions, soil quality, terrain, and technological advancements. Specifically, we chose ten indicators to encompass these elements: climate production potential, slope, tillage texture, soil bulk density, pH, organic matter content, irrigation guarantee level, flood protection measures, cultivation distance, and agricultural mechanization level. Climate production potential is the maximum agricultural yield per unit area achieved under optimal conditions by fully and reasonably utilizing climate resources such as light, heat, and water⁴². The quantity and coordination of these three factors—light, heat, and water—determine the vegetation production potential of a region⁴³. Therefore, we use climate production potential to reflect the climate conditions of cropland. The slope, which impacts cropland’s spatial distribution and quality, represents topographical conditions. Indicators such as tillage texture, soil bulk density, pH, and organic matter content are used to characterize the soil properties of the cropland.

Additionally, irrigation guarantee, flood protection, cultivation distance, and agricultural mechanization levels represent the technological inputs into cropland. Previous research has demonstrated that agricultural mechanization can reduce production costs, improve labor efficiency, and even enhance crop production^44,45. The social security capacity reflects cropland’s crucial role in maintaining economic and social stability, especially in rural areas. Specifically, this capacity is highlighted by the support provided to agricultural income and the assurance of food security. We have chosen two indicators to represent this aspect: the per capita disposable income of rural residents and cropping intensity. The per capita disposable income of rural residents signifies the income available to them over a specific period⁴⁶, reflecting the average income level of rural residents in a particular region or across the country. Cropping intensity, which involves increasing the number of crops grown on a piece of cropland, directly affects the yield of cropland. For example, transitioning from one round of rice planting to two rounds of rice or planting one round of wheat followed by one round of rice can significantly boost farmers’ agricultural income⁴⁷. Furthermore, government subsidies offered to farmers cultivating grain crops further enhance their income and overall economic well-being. Together, these 12 indicators summarize CP (Table 2).

Based on the constructed framework for evaluating CH-P and considering the differing attributes of each evaluation indicator, this study uses a closed interval of [0, 100] to map the attribute values of the indicators to CH or CP evaluation scores. Scores range from 0 to 100, representing the poorest to the best state. The scoring rules for each evaluation indicator are defined using previous research³⁰ or the Jenks Natural Breaks method, converting data from different dimensions to a unified standard⁴⁸. These indicators—including potential annual soil erosion, coefficient of variation of annual grain production, climate production potential, per capita disposable income of rural residents, and agricultural mechanization level—are evaluated using the Jenks Natural Breaks method, while the remaining indicators are referenced from existing literature³⁰. The evaluation unit is the cropland grid cell; all scores are computed for the 2020 baseline year. We employed the Analytic Hierarchy Process (AHP)⁴⁹ method to determine the weights of the evaluation indicators for CH-P. To do so, we first gathered rated importance scores for each indicator from a group of experts. Using these scores, a judgment matrix was constructed for pairwise comparisons, and the relative importance ratios were calculated based on Saaty’s 9-point scale⁵⁰. The weight vector was then derived using the yaahp 10.1 software (https://www.metadecsn.com/yaahp/), which facilitated the computation of the weights based on the pairwise comparison results. Tables 1, 2 show the weights of the CH-P evaluation indicators.

Cropland health indicators calculation

Irrigation Water Quality. As per the “Environmental Quality Standards for Surface Water (GB 3838-2002),” water classified as Class I-V is qualified for irrigation, while water quality below Class V is considered unqualified. Data from more than 2000 water quality monitoring sites for the period 2016–2020 were acquired from the China National Environmental Monitoring Center (https://www.cnemc.cn/). Suppose the percentage of years with water quality below Class V at a monitoring site surpasses 30% of the total recorded years. In that case, the irrigation water quality at that site is deemed unqualified (Qualified: ({Q}_{{water}} le 0.3), unqualified: ({Q}_{{water}} > 0.3)). The calculation formula for irrigation water quality is as follows:

where ({Q}_{{water}}) represents the proportion of unqualified irrigation water quality at monitoring point (i), (U) denotes the number of occurrences of below Class V water at monitoring point (i). from 2016 to 2020, and ({T}_{i}) represents the number of years with recorded data at monitoring point (i). from 2016 to 2020.

Soil heavy metal pollution. We searched for journal articles and theses on soil heavy metals published from 2010 to 2020 on CNKI (https://www.cnki.net/). The search yielded data from nearly 500 soil heavy metal measurement points, encompassing eight types of soil heavy metals: Cd, As, Cr, Hg, Zn, Cu, Ni, and Pb. Pollution levels were determined by comparing the measured concentration of these eight heavy metal elements in cropland to the standard values outlined in the Soil Environmental Quality Standards GB15618—2008 (Clean: ({Q}_{{hv}}le 1), slight pollution: (1 < {Q}_{{hv}}le ,2), light pollution: (2 < {Q}_{{hv}}le 3), moderate pollution: (3 < {Q}_{{hv}}le 5), heavy pollution: ({Q}_{{hv}} > 5)). The calculation formula for the soil heavy metal pollution coefficient is as follows:

where ({Q}_{{hv}}) is the coefficient for soil heavy metal pollution, and ({C}_{i}) stands for the measured concentration of the heavy metal element indicator (i) (As, Cr, Cd, Cu, Hg, Pb, Ni, Zn) in mg/kg. ({S}_{i}) represents the standard value for indicator (i) as specified in GB15618 and is measured in mg/kg. If the measured concentration exceeds the standard value (i.e., ({C}_{i} > {S}_{i})), the evaluation unit’s heavy metal element indicator (i) does not meet the standard; otherwise, it does meet the standard. The soil heavy metal pollution coefficient is obtained by summing these values.

Coefficient of variation of annual grain production. The coefficient of variation for the annual grain production in 2020 is calculated using the actual county-level grain yield data from 2016 to 2020. The formula for calculating the coefficient of variation of the annual grain production is as follows:

where ({{CV}}_{i}) represents the coefficient of variation of the annual grain production in county (i), ({{SD}}_{i}) denotes the standard deviation of the annual grain production in county (i) from 2016 to 2020, and ({{Mean}}_{i}) refers to the mean annual grain production in county (i) during the same period.

Potential soil erosion intensity, soil earthworm density, and soil cation exchange capacity. These three indicator values were obtained from the existing dataset (Table 3).

Based on the calculation and acquisition of the indicators mentioned above, the formula for the CH index is as follows:

where ({{FE}}_{i}) represents the farming environment index of evaluation unit (i), ({M}_{{ij}}) is the score corresponding to indicator(,j) of the farming environment for evaluation unit (i), and ({N}_{{ij}}) is the weight of indicator (j) of the farming environment for evaluation unit (i). ({{PE}}_{i}) represents the production elasticity index of evaluation unit (i), ({X}_{{ij}}) is the score corresponding to indicator (j) of the production elasticity for evaluation unit (i), and ({Y}_{{ij}}) is the weight of indicator (j) of the production elasticity for evaluation unit (i).

Cropland productivity indicators calculation

Climate production potential. The calculation formula⁴³ is as follows:

where ({Y}_{2}) represents the climate production potential, ({Y}_{1}) represents the light-temperature production potential, ({f}_{3}) represents the correction function of the water factor, W represents precipitation, and E represents evaporation.

Irrigation guarantee level. We categorize various types of cropland using the land use dataset: paddy fields, irrigated land, and dry land. Paddy fields are considered to have fully satisfied irrigation requirements, irrigated land generally satisfies irrigation needs, and dry land has no irrigation conditions.

Cultivation distance. The distance between cropland units and residential areas was calculated using the Euclidean method to represent the cultivation distance.

Agricultural mechanization level. The level of agricultural mechanization is determined by the ratio of the total annual agricultural machinery power to the total cropland area in each county. The formula for calculating the agricultural mechanization level is as follows:

where ({{AML}}_{i}) represents the agricultural mechanization level of county (i), (T{{PAM}}_{i}) denotes the total annual agricultural machinery power in county (i), and ({{CLA}}_{i}) represents the total cropland area in county (i).

Slope, tillage texture, soil bulk density, pH, organic matter content, flood protection level, per capita disposable income of rural residents, and cropping intensity. These eight indicator values were obtained from the existing dataset (Table 3).

Based on the calculation and acquisition of the indicators mentioned above, the formula for the CP index is as follows:

where ({{CL}}_{i}) represents the climate condition index of evaluation unit (i), ({Y}_{2,i}) is the score corresponding to the climate production potential indicator for evaluation unit (i), and (w) is the weight of the climate production potential indicator. ({{TC}}_{i}) represents the topographical condition index of evaluation unit (i), ({{Slope}}_{i}) is the score corresponding to the slope indicator for evaluation unit (i), and (h) is the weight of the slope indicator. ({{SP}}_{i}) represents the soil property index of evaluation unit (i), ({I}_{{ij}}) is the score corresponding to the soil property indicator (j) for evaluation unit (i), and ({J}_{{ij}}) is the weight of the soil property indicator (j) for evaluation unit (i). ({{TL}}_{i}) represents the technological level index of evaluation unit (i), ({K}_{{ij}}) is the score corresponding to the technological level indicator (j) for evaluation unit (i), and ({L}_{{ij}}) is the weight of the technological level indicator (j) for evaluation unit (i). ({{AES}}_{i}) represents the agricultural economic status index of evaluation unit (i), ({O}_{{ij}}) is the score corresponding to the agricultural economic status indicator (j) for evaluation unit (i), and ({P}_{{ij}}) is the weight of agricultural economic status indicator (j) for evaluation unit (i). ({{CP}}_{i}) represents the cropland productivity index.

Cropland connectivity evaluation

After evaluating CH and CP, it is still necessary to consider their spatial organization. CH and CP capture intrinsic land quality and productive potential, but do not indicate whether suitable parcels are spatially concentrated or fragmented. Without this dimension, fallow-priority areas may be too scattered to implement efficiently⁵¹. Previous studies show that cropland fragmentation increases production costs and reduces farm efficiency^52,53, raising transaction and supervision costs that weaken ecological and policy outcomes⁵⁴. Empirical evidence in China confirms that fragmentation significantly reduces farmers’ labor productivity⁵⁵. Considering intensive management, implementing fallow management measures on adjacent cropland with similar characteristics can result in higher fallow benefits at a reduced cost. Hence, we propose incorporating CC as a critical dimension in assessing the SACF.

We define CC as management-oriented structural connectivity: two cropland pixels are counted as connected—in an operational sense—only when they are immediate 8-neighbor cells on the 30-m grid and, at the same time, share the same attributes relevant to fallow implementation (cropland type, CH class, and CP class). Because all inputs are gridded at 30-m resolution, 30 m is the smallest spatial link the grid can represent; accordingly, we detect connectivity with 8-neighbor adjacency at 30 m and treat any gap of ≥one 30-m pixel as a break in connectivity. Railways, highways, and large rivers commonly disrupt field connectivity; this one-pixel-gap rule conservatively approximates such interruptions at this resolution and is consistent with the order of magnitude of first-class highway cross-sections (typically > 20 m; JTG B01-2014⁵⁶). While ecological connectivity is often defined through functional linkages such as ecological corridors that facilitate ecological flows⁵⁷, our focus here is on management connectivity, where spatial contiguity supports the operational feasibility of fallow implementation.

However, existing research on CC often considers only spatial adjacency without accounting for differences among cropland types and other aspects^58,59. To align the metric with management needs, we adopt a same-attribute rule (Fig. 9): pixels are considered connected only when cropland type (dryland, irrigated land, and paddy fields), CH, and CP are all of the same categories and the 8-neighbor condition is met. CC is classified into three levels: high, medium, and low, with the corresponding threshold intervals (hm²) as follows: <10, [10, 598.41), and ≥598.41. The thresholds for medium and low cropland connectivity are based on existing studies that define “fragmented cropland” ⁶⁰ as cropland with a spatially contiguous area of less than 10 hm². Our analysis of the distribution of cropland connectivity data revealed the presence of extreme values, and using natural breakpoints or means as classification thresholds may lead to unbalanced classification results. The median is less sensitive to outliers, providing a more accurate reflection of the central tendency of the dataset⁶¹. Therefore, the thresholds for high and medium cropland connectivity are set to the median of CC. Consequently, while the 30 m rule only defines adjacency, the connectivity level is determined by CC classes, which quantify the area of contiguous fallow units. This approach reduces sensitivity to the exact linking radius and ensures that connectivity is interpreted as management-oriented structural contiguity, distinct from ecological functional connectivity emphasized in ecological corridor studies⁵⁷.

Spatial arrangement rules of fallow zones

The cropland’s health, productivity, and connectivity all have distinct focuses, yet they are closely intertwined. CH emphasizes the dominant factors of improper land use in a region; CP depicts the level of productivity resulting from the interaction between humans and the land; CC reflects the spatial distribution of relatively connected cropland, accounting for multiple attributes. Healthy cropland underpins sustained productivity, and well-connected fields make it easier to organize large-scale, efficient operations. Conversely, pushing productivity too hard can degrade health; intensive use also reduces biodiversity, further undermining CH. Over time, this feedback erodes soil fertility and yield stability, weakening the sustainability of CP. Analyzing CH-CP-CC together is essential; their coupled patterns indicate where fallow measures should be prioritized in regional cropland protection. Building on this logic, we believe that implementing targeted fallow in areas with low CH, low CP, and high CC can respectively enhance fallow’s ecological–environmental benefits, socio-economic benefits, and cost efficiency. However, previous studies have generally identified fallow priority zones by highlighting single dominant constraints^62,63,64 or overlaying multiple benefits such as ecological and socio-economic gains^9,10. They seldom distinguished the relative urgency among contradictions within cropland health–productivity (CH–P) coupling. Borrowing from earlier spatial trade-off decision models of cropland fallow based on three-dimensional cubes of cropland use intensity, health, and productivity potential⁶⁵, this study proposes the principle of “prominent contradiction, concentrated and connected”, ensuring that ecological–environmental and socio-economic benefits are maximized while minimizing implementation costs. To refine prioritization, we first further differentiated five CH–P conflict types: Low CH-Low CP (LL), Low CH–Medium CP (LM), Low CH-High CP (LH), High CH-Low CP (HL), and Medium CH-Low CP (ML). Following a conflict-first principle⁶⁵, we ranked them by severity: LH and HL types, where CH and CP are at opposite extremes, represent the sharpest contradictions and are therefore given the highest urgency priority; LM and ML types indicate weaker but still notable mismatches and are assigned secondary urgency priority; and LL types, which reflect marginal cropland with both low CH and low CP, receive the lowest urgency priority. Building on this sequence, we then incorporate CC as a cost-efficiency criterion: blocks with higher CC are more practical to implement, easier to supervise, and more cost-effective than fragmented parcels⁶⁵.

Therefore, using the Jenks Natural Breaks method, this study first categorizes the CH-P evaluation results into high, medium, and low levels. Next, we combine these with the CC classification results to create 27 coupling outcomes of CH-CP-CC (Fig. 10). Based on the principle of “prominent contradiction, concentrated and connected,” we treat CH–CP conflicts as the primary criterion for fallow selection, and CC as a secondary factor to balance spatial feasibility. Finally, the fallow regions are grouped into four categories: Urgent Fallow Zone (UFZ), Priority Fallow Zone (POFZ), Moderate Fallow Zone (MFZ), and Postponed Fallow Zone (PPFZ).

The first category, UFZ (2 combinations, including HLH [High CH-Low CP-High CC] and LHH), mainly contains croplands where CH and CP are at opposite extremes and connectivity is high. These areas combine substantial ecological or production value with urgent deficits in the other dimension, making them the highest-priority fallow targets. The second category, POFZ (2 combinations, including HLM and LHM), represents the same CH–CP conflicts but with lower connectivity, meaning that while fallow is still necessary, efficiency and implementability are somewhat reduced. This hierarchical prioritization follows the principle of giving precedence to the most prominent CH–CP contradictions, consistent with prior fallow zoning studies that ranked pollution and over-exploitation zones as top priorities.

The third category, MFZ (8 combinations, including MLH, LMH, and MLM, etc.), is characterized by less significant conflicts between health and productivity or relatively lower connectivity than the first two categories, making the urgency for fallow lower. The fourth category, PPFZ (15 combinations, including MHH, MHM, and MHL, etc.), represents areas where CH and CP results are similar—either both high, both medium, or both low—indicating minimal conflicts or representing marginal land, such as LLH, LLM, and LLL, and is thus considered a postponed fallow zone.

Based on each area’s rules and characteristics mentioned above, the UFZ and POFZ zones align more closely with our proposed fallow principles, where the conflicts between productivity and health are prominent and connectivity is higher. In contrast, the conflicts in the MFZ and PPFZ zones are less significant, lowering the urgency for fallow. Therefore, this study designates UFZ and POFZ as fallow zones, while MFZ and PPFZ are designated as non-fallow zones.

Data sources

This study covers 31 provinces, municipalities, and autonomous regions of China, excluding Hong Kong, Macao, and Taiwan (Fig. 11). We utilized multi-source data, including land use/land cover, meteorological, statistical, and other existing datasets. The specific data sources are detailed in Table 3. All data were resampled to a spatial resolution of 30 m × 30 m to ensure consistency in this study. We use 2020 as a nationally consistent baseline snapshot because it is the only year with complete and harmonized inputs across China for all indicators. The SACF framework is time-extensible and designed to roll forward as multi-year CH–CP–CC indicators become available.