Spatiotemporal patterns and driving mechanisms of cultivated land multifunctionality evolution: evidence from county-level data in Sichuan Province, China

The pursuit of socioeconomic development, food security, and ecological sustainability has driven the multifunctional evolution of cultivated land. Under the constraint of limited cultivated land resources, achieving synergy among the economic, food security, and ecological functions of cultivated land has become a major challenge for agricultural modernization in Sichuan Province. This study evaluates the spatiotemporal evolution of cultivated land multifunctionality (CLM) and investigates the driving mechanisms underlying the CLM synergy level. The results show that: (1) The overall CLM development level in Sichuan Province increased during the study period, with the economic contribution and food security functions serving as the primary drivers. Nevertheless, the overall quality of CLM development remained relatively low, accompanied by substantial regional disparities, development imbalance, and functional disharmony. (2) The CLM synergy level exhibited pronounced polarization and regional imbalance, together with significant spatial agglomeration. (3) The explanatory power of regional background condition factors for the CLM synergy level gradually weakened, whereas that of economic development condition factors increased. Six factors, regional GDP, rural per capita net income, rural employment population, construction intensity, urbanization rate, and cultivated land aggregation degree, showed significant explanatory power for the CLM synergy level. To promote the synergistic development of CLM in Sichuan Province, policy efforts should focus on ensuring food production and supply, promoting high-quality agricultural development, facilitating the value realization of landscape ecological functions, and implementing differentiated regional strategies. This study provides scientific evidence and practical insights for improving CLM development in China and other regions facing similar cultivated land constraints.

Cultivated land is a fundamental resource for the continuation and development of human civilization and represents one of the complex ecosystems on which human society depends most heavily. As one of the most variable land-use types (Song and Ouyang, 2012), cultivated land possesses inherent ecological and production functions. Cultivated land functions refer to the capacity of cultivated land to provide various products and services required for human survival and development (Yang X. et al., 2024), which contributes to improving land-use efficiency and enhancing the value potential of cultivated land. With socioeconomic development, cultivated land functions have gradually evolved toward diversification (Fan et al., 2023). Human society has shifted from focusing solely on improving individual functions, such as food production or economic output, to comprehensively considering the development of cultivated land multifunctionality (CLM). CLM has consequently become a key concept in cultivated land function research. CLM refers to the capacity of cultivated land systems to provide private and public goods and services through interactions between natural conditions and human activities (Jiang et al., 2020). Rooted in the semi-natural and semi-artificial characteristics of cultivated land, this capacity enables cultivated land systems to simultaneously perform multiple functions, including economic production, food security, ecological conservation, and landscape-cultural services (Song et al., 2015; Liu J. et al., 2023). From the perspective of long-term socioeconomic development, fulfilling cultivated land protection objectives and maximizing the value of cultivated land resources are crucial for national development and people’s livelihoods. Promoting the synergistic development of CLM has become an important pathway for improving cultivated land protection and utilization (Lv et al., 2023).

Sichuan Province is an important agricultural province and major grain-producing region in southwestern China, and its cultivated land resource endowments, utilization characteristics, and multifunctional evolution patterns are representative and distinctive within China. Agricultural productivity growth and grain production have long been subject to tight constraints in both the quantity and quality of cultivated land, making Sichuan a representative case of a region with limited cultivated land resources. Selecting Sichuan as the study area can provide a reference for other agricultural regions, both in China and abroad, that share similar characteristics (Xiang et al., 2024). At the same time, it offers an ideal case for examining how to optimize multiple cultivated land functions under the constraint of limited land resources. Against this background, this study selects 176 counties in Sichuan Province, China, as the research area to quantitatively investigate the development level and regional disparities of CLM, the synergy level of CLM and its spatial agglomeration characteristics, and the driving factors influencing the CLM synergy level. The findings aim to provide practical guidance for promoting the synergistic development of CLM in Sichuan Province and offer scientific references and practical insights for similar regions worldwide to address human–land conflicts and advance cultivated land protection and sustainable utilization.

Literature review

Cultivated land multifunctionality (CLM) has been developed and extended from the concepts and connotations of agricultural multifunctionality and land multifunctionality, undergoing a transformation from single-function orientation to multifunctional development (Zhao and Zhang, 2010). From the perspective of academic inheritance, agricultural multifunctionality provides theoretical support for research on CLM, and the multifunctionality of agriculture determines the multifunctionality of cultivated land (Ning and Liu, 2012). In recent years, scholars have conducted numerous valuable studies on CLM and generated a substantial body of research findings. Existing studies have investigated CLM from various perspectives, which can be summarized into four major aspects.

Research on the types of cultivated land multifunctionality

Existing studies have developed a substantial theoretical foundation for the classification of CLM types. Differences in classification mainly reflect different understandings of the functional roles of cultivated land, and existing classification approaches can be broadly grouped into two mainstream perspectives. One perspective considers the hierarchical relationships among cultivated land functions and divides them into basic functions and derived functions (Wang Q. et al., 2020). Basic functions are regarded as inherent attributes of cultivated land, including production, ecological, and landscape-cultural functions, whereas derived functions emerge from changes in human demands, cognition, and cultivated land utilization capabilities, such as the food security function driven by food security demands (Jiang et al., 2011). Another perspective focuses on cultivated land utilization and distinguishes between basic functions and supplementary functions (Wang and Dong, 2015). Economic and ecological functions are regarded as basic functions, while landscape aesthetics and social security functions are considered supplementary functions (Qian et al., 2022). Representative perspectives also include the ecosystem service perspective (Coyle et al., 2016) and the integrated perspective (Cai, 2001; Wei et al., 2023). Differences in classification results essentially arise from the theoretical inclusiveness of the CLM concept and the heterogeneity of perspectives used to decompose its functional dimensions.

Quantitative assessment of cultivated land multifunctionality

Quantitative assessment of CLM is an important approach for evaluating the development level of CLM and provides a fundamental basis for formulating and implementing strategies for its synergistic development. Existing studies have mainly applied four types of quantitative methods to assess CLM. The material quantity method can be further divided into model-based approaches and field sampling approaches (Tzilivakis et al., 2019). The value-based method mainly evaluates CLM from the perspectives of production, social security, and ecological function values (Liu Y. et al., 2023). The emergy analysis method converts various forms of matter and energy within ecosystems into solar energy equivalents, facilitating comparisons among different products or services (Grassauer et al., 2021). The indicator system method measures CLM comprehensively by selecting evaluation indicators, constructing assessment frameworks, and applying appropriate weighting methods (Wang et al., 2022). The first three quantitative methods mainly rely on indicators and data derived from natural science measurements, whereas the indicator system method generally uses socioeconomic data to construct evaluation frameworks for quantitative analysis and has therefore been widely applied in large-scale spatial differentiation and time-series comparative studies (Zhang et al., 2018a; Gao et al., 2022). Existing studies are often limited by reliance on a single method or insufficient integration among different methodological paradigms (He, 2019), which restricts the accurate quantification of CLM. Within the indicator system method, weighting approaches such as entropy weighting, the analytic hierarchy process, and expert scoring each have their own advantages and limitations, yet most studies still rely on a single weighting approach (Liu et al., 2024). Systematic integration of subjective and objective weights for county-level time-series comparisons remains insufficiently explored. Some studies focus on the direct measurement of individual functions while lacking effective approaches for the comprehensive quantification of complex cultivated land systems (Zhang et al., 2018b).

Research on the spatiotemporal patterns of cultivated land multifunctionality

In terms of research scale, scholars have mostly conducted analyses at large regional levels, such as provincial (Xiao et al., 2022) and urban agglomeration scales (Fan et al., 2026). Methodologically, they have applied a combination of approaches, including the triangular model (Liu et al., 2024), cluster analysis (Wang et al., 2025), and comprehensive evaluation (Chen, 2022), together with spatial autocorrelation analysis, hot and cold spot detection, and landscape pattern indices (Fang et al., 2025). These studies have revealed a clear evolutionary trend in cultivated land multifunctionality: production functions have relatively contracted, while diverse functions have expanded toward suburban areas and regions with better ecological conditions (Wang et al., 2018; Fang et al., 2018). As for spatial differentiation and agglomeration characteristics, existing evidence indicates that high-value areas of multifunctionality are mostly concentrated in traditional agricultural regions with superior production conditions or in urban-rural integration zones. In contrast, ecologically fragile areas, mountainous regions, and rapidly urbanizing fringe zones often exhibit low functional values or even degradation (Zhang et al., 2019; Gao et al., 2022). Such spatial patterns reflect the influence of resource endowments on the evolution of functional configurations. Notably, most of the above studies have been conducted at the prefecture-level city or higher administrative levels. The county-level unit, as a basic administrative division, has not yet been fully explored in the context of cultivated land multifunctionality, and its scale value remains largely underexploited.

Research on the driving mechanisms of cultivated land multifunctionality

The driving mechanisms underlying CLM evolution are essential for understanding its development processes. Existing studies have employed qualitative and quantitative approaches to explore the driving factors and mechanisms of CLM evolution at different spatial scales. CLM evolution is generally considered to result from the combined effects of natural environmental conditions and human socioeconomic activities (Ma et al., 2024). Natural factors, including topography, climate, and soil conditions, provide the fundamental basis for the long-term persistence and evolution of CLM (Qian et al., 2022). Human factors, such as urbanization, economic development, policy interventions, and technological progress, drive short-term transformations and spatial restructuring of CLM by altering land-use patterns (Wu et al., 2024). These driving forces were further classified into two broad categories: endogenous and exogenous (Zhang et al., 2023b). Methods for identifying CLM drivers have evolved from traditional statistical approaches to spatial econometric methods. Early studies mainly applied correlation analysis, multiple linear regression, and Tobit models to identify key influencing factors (Liu Z. et al., 2025). Recent studies have increasingly adopted methods such as the geographical detector model (Niu et al., 2022), geographically weighted regression (Liu J. et al., 2023), and grey relational analysis (Zhang et al., 2025) to examine spatial heterogeneity and interactions among driving factors. The geographical detector model has been widely applied in cultivated land research due to its ability to integrate qualitative and quantitative variables and detect interactions between factors (Xiao et al., 2022; Li et al., 2025). Existing studies have mainly focused on socioeconomic factors in the selection of driving variables (Zhang et al., 2018a). Such studies provide insights into the direct effects of human activities but pay relatively limited attention to the fundamental role of natural conditions and the interactions between natural and human factors. CLM is widely recognized as being shaped by both natural and anthropogenic factors (Kang et al., 2023), while empirical studies that systematically integrate these two dimensions and examine their interactive mechanisms remain limited.

Research gaps and contributions

Existing studies on CLM have provided important references for this study in terms of theoretical development, methodological exploration, and empirical analysis. Regarding the classification of CLM types, research has progressed from binary frameworks such as “basic–derived” and “basic–supplementary” functions toward more diversified and context-specific classification systems. In terms of quantitative assessment, the material quantity method, value-based method, emergy analysis method, and indicator system method each have their own strengths and together constitute a relatively comprehensive methodological toolbox, although effective integration among these approaches remains to be strengthened. Regarding spatiotemporal patterns, macro-level characteristics at the provincial and urban-agglomeration scales have been relatively well documented, with substantial empirical evidence on spatial agglomeration and differentiation. Research on driving mechanisms has also progressed from traditional regression analysis toward spatial econometric methods and geographical detector models, and the joint effects of natural and anthropogenic factors have become widely recognized. Nevertheless, several issues remain to be further explored, which constitute the research focus and marginal contributions of this study. In terms of research scale, existing analyses of the spatiotemporal patterns of CLM have mainly focused on provinces or urban agglomerations, while county-level units, as basic administrative units for CLM research, have received insufficient attention in studies spanning a decade. The limited use of county-level analysis restricts the precise characterization of intra-regional differentiation and makes it difficult to meet the spatially targeted needs of differentiated policymaking. This study takes 176 counties in Sichuan Province as the basic units of analysis and selects 2012, 2017, and 2022 as three observation years to construct a county-level panel dataset spanning 2012–2022, with the aim of revealing the stage-specific characteristics and decadal trends of CLM evolution at a finer spatial scale.

In terms of quantitative methods, although the indicator system method has been widely applied, most studies have long relied on single weighting approaches such as the entropy method, with insufficient attention to the systematic integration of subjective and objective weighting. Although a small number of studies have begun to adopt combined weighting strategies (Tong et al., 2023), mature methodological approaches are still lacking for county-level temporal comparisons in which weights need to reflect both objective data information and policy priorities, such as food security. This study applies a combined weighting–TOPSIS method. Objective weights are first calculated using the entropy method and then adjusted according to the “quantity–quality–ecology” integrated cultivated land protection framework and expert consultation results, with appropriate emphasis placed on the food security function. This approach integrates objectivity with policy orientation in the weighting process.

Regarding driving mechanisms, existing studies have mainly focused on identifying factors affecting the CLM development level, that is, the comprehensive evaluation value (Zhang et al., 2025), while relatively limited attention has been paid to the driving mechanisms of the CLM synergy level. From a policy perspective, synergistic development is more practically oriented than improvement in any single function. A high development level does not necessarily correspond to a high synergy level, and the latter represents a key dimension for assessing whether cultivated land systems achieve high-quality coordinated development. Although previous studies have widely identified interactions between natural and anthropogenic factors (Kang et al., 2023; Ma et al., 2024), empirical research that systematically integrates both types of factors and examines their interactions at the county level remains relatively limited. This study uses the coupling coordination degree of CLM as the dependent variable, selects driving factors from two dimensions—regional background conditions and economic development conditions—and applies the geographical detector model to quantify both the explanatory power of individual factors and the interactions between pairs of factors, thereby examining the differentiated pathways and interactive mechanisms through which the two types of factors influence the evolution of the CLM synergy level.

The marginal contributions of this study can therefore be summarized as follows: (1) using 176 counties as the units of analysis and 2012, 2017, and 2022 as the observation years, this study provides new county-level evidence on the spatiotemporal patterns of CLM over the 2012–2022 period; (2) a combined weighting–TOPSIS evaluation model is constructed to improve the integration of subjective and objective weighting; (3) the analytical focus is shifted from the CLM development level to the CLM synergy level, and the geographical detector model is employed to examine the individual and interactive effects of natural and anthropogenic factors, providing a more comprehensive empirical explanation of the driving mechanisms underlying the synergistic evolution of CLM. These contributions respond to gaps in previous research and provide scientific references for promoting the synergistic development of multiple cultivated land functions under limited cultivated land resources in Sichuan Province and other comparable hilly and mountainous regions.

Study area description and scope

Sichuan Province is located in southwestern China (Figure 1), with a total administrative area of 486,000 km 2. It comprises 18 prefecture-level cities, 3 autonomous prefectures, and 183 counties (districts). The province encompasses diverse geomorphic types, including mountains, hills, plains, and plateaus, accounting for approximately 77%, 13%, 5.5%, and 4.5% of the provincial area, respectively. The unique geographical conditions have resulted in relatively limited cultivated land resources in Sichuan Province, providing a representative case for CLM research. Cultivated land resources are unevenly distributed across the province, with 60% of cultivated land concentrated in Northeastern Sichuan and the Chengdu Plain. The geomorphic pattern, commonly characterized as “one part plain, two parts slopes, and seven parts mountains,” makes Sichuan a representative region of agricultural production in hilly and mountainous areas. Sichuan is also representative of conflicts in cultivated land use. Cultivated land accounts for only 11.37% of the provincial administrative area, and per capita cultivated land is only 447 m 2, less than 50% of the national average. As one of China’s 13 major grain-producing regions and also a major grain-consuming province, Sichuan faces continued pressure on land resources from industrialization and urbanization. Balancing regional food security and economic development under these constraints makes the cultivated land-use conflict in Sichuan an issue of national relevance.

In 2012, China further clarified the policy of “strictly safeguarding the red line for cultivated land protection and strictly controlling land use.” A series of related policies marked a historical shift in cultivated land protection from a focus on quantity control toward integrated management of quantity, quality, and ecology. China began large-scale cultivated land protection and utilization initiatives (Han, 2022), and the dynamic evolution of CLM spatiotemporal patterns became more pronounced (Jiang et al., 2020). After years of cultivated land protection and management, the total cultivated land area in Sichuan Province stabilized at approximately 5.25 million ha in 2025, accounting for about 10.8% of the provincial land area; paddy fields and drylands accounted for 43.3% and 55.6% of cultivated land, respectively (Qi, 2023). The period from 2012 to 2022 was a key phase in the systematic upgrading of China’s cultivated land protection policy system and high-standard farmland construction, providing an appropriate policy window for the evaluation in this study. The years 2012, 2017, and 2022 were selected as equally spaced observation points to reveal the stage-specific characteristics and decadal changes in CLM evolution, while reducing the influence of random annual fluctuations and improving the robustness of the results.

Considering that some county-level administrative units in Sichuan have limited cultivated land resources and small-scale rural economies, these areas are not suitable for CLM research. Following the criteria for selecting research areas proposed by Yang Y. et al. (2024), this study uses a cultivated land economic contribution index greater than 1.5% as the selection criterion. Seven county-level urban districts, including Jinjiang District and Qingyang District in Chengdu, with cultivated land economic contribution indices below 1.5%, were excluded. Finally, 176 counties were selected as the study area. Given Sichuan’s regional heterogeneity and the value of spatial comparison, the province was divided according to differences in topography and socioeconomic development and in accordance with provincial territorial spatial planning documents, including the Territorial Spatial Plan for the Chengdu Plain Economic Zone (2021–2035), into five major regions: the Chengdu Plain Economic Zone, Southern Sichuan Economic Zone, Northeastern Sichuan Economic Zone, Panxi Economic Zone, and Northwestern Sichuan Ecological Demonstration Zone (Figure 1). This division allows the differentiated effects of natural conditions and socioeconomic development on CLM evolution across the five economic zones to be compared within the same provincial administrative framework.

Data sources and processing

To ensure consistency in statistical standards and data collection, this study integrates multiple data sources. Socioeconomic and other observational data were mainly obtained from the Sichuan Statistical Yearbook, China County Statistical Yearbook, and statistical yearbooks of 21 prefecture-level cities (prefectures) in Sichuan Province. Some data were directly obtained through applications to the Sichuan Provincial Bureau of Statistics and local statistical bureaus. Cultivated land data were obtained from the Resource and Environment Science and Data Center (https://www.resdc.cn). Digital elevation data were derived from the ASTER Global Digital Elevation Model (ASTER GDEM), jointly developed by NASA and NIMA. Temperature and precipitation data were obtained from the National Earth System Science Data Center (https://www.geodata.cn). Soil erodibility factor data and construction intensity data were sourced from the Earth Resource Data Cloud Platform (http://www.gis5g.com). The overall accuracy of all vector datasets exceeded 90%. The spatial resolution of temperature and precipitation data was 1 km, while the resolution of other datasets was 30 m. The administrative county-level boundary data were uniformly obtained from the 2023 National County-Level Administrative Division Dataset. For rasterized datasets among the collected spatial data, ArcGIS software was used for processing and conversion into county-level observation units. Missing indicators were supplemented using interpolation or data shifting methods.

Research methods

Construction of the “three functions in one system” theoretical framework for the synergistic development of CLM

Cultivated land multifunctionality (CLM) constitutes a complex, open, integrated, and dynamic system in which the internal subsystems and their constituent elements are interconnected and interact with one another. The synergistic development of CLM is important for improving the efficiency of cultivated land resource utilization (Chen A. et al., 2023). Drawing on previous studies, this study proposes a “three functions in one system” theoretical framework as the theoretical basis for examining the dynamic evolution and synergistic development of CLM (Figure 2). The “three functions” refer to the three core functions of cultivated land: the economic contribution function (FEC), food security function (FIC), and landscape ecological function (FSC). The “one system” refers to the cultivated land multifunctional system formed through the integration of these three core functions. Excessive concentration of cultivated land-related factors in one subsystem may strengthen a dominant function, whereas imbalanced or inefficient factor allocation may lead to poor coordination among functions within the cultivated land multifunctional system. The multiple functions of cultivated land interact, overlap, and integrate with one another. During the evolution of cultivated land functions, a single function may dominate at certain stages, while multiple functions may also jointly assume dominant roles. Functional improvement may therefore take the form of stage-specific optimization of a single function or simultaneous improvement of two or more functions (Yu et al., 2025).

The mechanisms underlying CLM are mainly manifested through the following pathways. (1) Sufficient cultivated land quantity is an important prerequisite for ensuring grain output, while cultivated land quality is a key determinant of production capacity. High-standard farmland construction can simultaneously improve the ecological quality and land leveling of cultivated land (Chen et al., 2021). (2) The formulation and implementation of cultivated land protection policies and land-use regulations directly affect the dominance of different cultivated land functions (Wang W. et al., 2020). For example, designating cultivated land as permanent basic farmland strengthens its food security function (FIC). (3) Agricultural industrial transformation and rural population migration can affect the “non-grain” use and abandonment of cultivated land (Xu and Chen, 2026), while also increasing farmers’ acceptance of agricultural socialized services (Zhang and Li, 2025). The “non-grain” use of cultivated land may therefore reduce regional food security capacity while enhancing the economic contribution of cultivated land. At the same time, reductions in grain-producing cultivated land and the agricultural labor force may promote agricultural mechanization and indirectly stimulate grain productivity.

From the perspective of driving mechanisms, the effects of individual factors differ across cultivated land functions (Chen Q. et al., 2023). Cultivated land economic contribution function (FEC), food security function (FIC), and landscape ecological function (FSC) are not independent of one another but form synergistic relationships through resource allocation and land-use processes. Improvement in a single function may promote other functions by improving cultivated land-use conditions, but may also constrain other functions because of competition for production factors. Therefore, the CLM synergy level does not depend on the absolute level of any single function, but rather on the coordination among the three functions during their joint evolution. Based on this understanding, this study uses the coupling coordination degree to represent the interactions among FEC, FIC, and FSC and the overall synergy level of the cultivated land multifunctional system, thereby transforming the theoretical relationship of “three functions in one system” into a measurable empirical object. The synergistic development of CLM is also jointly driven by external conditions. Considering the formation process of cultivated land functions and the development characteristics of counties in Sichuan Province, this study summarizes these external drivers into two dimensions: regional background conditions and economic development conditions (Ma et al., 2024; Kang et al., 2023). Regional background conditions mainly reflect the basic resource and spatial constraints faced by cultivated land use. Precipitation, relief amplitude, and soil erosion factor represent the natural environmental basis, while patch density reflects the spatial pattern of cultivated land jointly shaped by long-term natural conditions and human use. Economic development conditions influence the evolution of cultivated land functions by changing the allocation of land, labor, capital, and other factors, as well as patterns of land development and utilization. These conditions are represented by regional GDP, rural per capita net income, rural employment population, construction intensity, urbanization rate, and cultivated land aggregation degree (Gao et al., 2022; Zhang et al., 2023a; Fan et al., 2026). The two dimensions of driving factors are not independent. Regional background conditions constitute the underlying constraints on CLM evolution, whereas economic development may strengthen or partially alleviate the constraints imposed by natural conditions through farmland infrastructure construction, land consolidation, agricultural technological inputs, and factor reallocation. The same natural resource conditions may produce different functional outcomes under different levels of economic development, while the same economic input may also generate differentiated effects under different resource endowments. Interactions between regional background conditions and economic development conditions may therefore generate greater explanatory power than individual factors and further influence county-level CLM synergy levels.

Based on the above theoretical analysis, this study develops an analytical logic of “external driving conditions–core cultivated land functions–CLM synergy level.” The development levels of FEC, FIC, and FSC are measured separately through the evaluation indicator system, and the coupling coordination degree model is then used to characterize the synergistic state through which the “three functions” are integrated into “one system.” The coupling coordination degree is used as the specific measure of the CLM synergy level, and driving factors are selected from the two dimensions of regional background conditions and economic development conditions. The geographical detector model is employed to identify the independent explanatory power of each driving factor. Interaction detection is further applied to examine the joint effects among different driving factors, thereby establishing a direct correspondence between the theoretical analytical framework and the subsequent empirical analysis.

Construction and calculation of the evaluation indicator system

Based on the theoretical framework of the “three functions in one system” for cultivated land multifunctional synergy (Zhang et al., 2025), this study constructs an evaluation indicator system for CLM in Sichuan Province (Table 1). The “three functions” refer to the three core functions of cultivated land: economic contribution function (FEC), food security function (FIC), and landscape ecological function (FSC). The “one system” refers to the cultivated land multifunctional system formed through the integration of cultivated land’s core functions, which promotes innovative resource allocation driven by factor integration and facilitates functional interactions and synergistic development. Following the principles of systematicity and scientific validity in constructing evaluation indicator systems, this study identifies three first-level indicators and 16 second-level indicators based on extensive reference to previous studies and consideration of data availability. The evaluation method adopts the combined weighting-TOPSIS method. The specific calculation methods for cultivated land landscape fragmentation, cultivated land aggregation degree, and cultivated land regularity follow standard procedures in landscape ecology. These indicators have been widely used in cultivated land landscape pattern studies, and their calculation formulas and algorithms are detailed in (Xu et al., 2023; Zhang and Zhang, 2022).

The weighting of CLM indicators needs to overcome the limitations of single weighting approaches. The combined weighting method integrates the advantages of expert scoring and the entropy weighting method, producing indicator weights closer to the ideal state. The entropy weighting method calculates indicator entropy values and difference coefficients step by step, and the resulting indicator weight is denoted as W1. The expert-scoring weight is denoted as W2, with its theoretical basis following Tong et al. (2023). Six experts in agricultural economics and related fields were invited to participate in a Delphi consultation, and the consistency ratio was used as the criterion for determining whether the judgment matrix passed the consistency test. W1 was used as the objective weight, while W2 obtained from expert scoring served as a reference for adjusting the indicator weights. The final indicator weights were determined after integrating and adjusting the two sets of weights with reference to the expert-scoring approach. The TOPSIS method ranks alternatives by determining their distances from the “ideal solution” and the “negative ideal solution” in multi-criteria decision-making problems. The sum of the weights of each cultivated land subsystem function is set to 1, and the development level of each subsystem function is evaluated separately. The main calculation procedures of the evaluation method follow those described in previous research (Zhou et al., 2022). The evaluation values of the economic contribution function (FEC), food security function (FIC), and landscape ecological function (FSC) are summed to obtain the CLM development level of each county.

Based on the calculated CLM, development levels, the classification criteria were determined according to the distribution characteristics of the natural breaks method and the classification scheme proposed in previous research (Qin et al., 2018). The classification standards for all cultivated land subsystem functions were defined as low [0, 0.25), moderate [0.25, 0.5), good [0.5, 0.75), and excellent [0.75, 1]. The classification standards for the CLM, development level were defined as low [0, 0.4), moderate [0.4, 0.8), good [0.8, 1.2), and excellent [1.2, 1.6].

Regional disparity model

Sichuan Province contains a large number of counties with extensive spatial distribution. The Dagum Gini coefficient can decompose overall spatial disparities into three components: within-region disparity, between-region disparity, and hypervariable density. It effectively addresses the issue of overlapping subgroups and has been widely applied in studies of spatial differentiation in socioeconomic and resource-environmental fields (Gao et al., 2025). Therefore, this study applies the Dagum Gini coefficient to examine the regional disparities in the development level of CLM in Sichuan Province. The decomposition formula is provided in previous research (Wang and Zhang, 2025).

Coupling coordination degree model

CLM is a complex multifunctional system in which interactions commonly occur among different functions (Chen et al., 2025). Measuring the CLM synergy level can reflect important characteristics of its spatiotemporal evolution. This study employs the coupling coordination degree model to evaluate the CLM synergy level, which can simultaneously characterize the interaction intensity among functions and the overall development level. The decomposition formula of the model is presented in previous research (Wang et al., 2021). Under the “three functions in one system” cultivated land protection policy framework (Zhang and Huang, 2025), most existing studies follow the principle that the weight of the food security function is no lower than that of the economic contribution function, which in turn is no lower than that of the landscape ecological function. Following Liu Y. et al. (2025), this study assigns functional coefficients of 0.3, 0.4, and 0.3 to FEC, FIC, and FSC, respectively. According to the distribution characteristics of the natural breaks classification results and relevant classification criteria, the CLM synergy level was divided into five grades: weak synergy [0, 0.25), low synergy [0.25, 0.5), moderate synergy [0.5, 0.65), good synergy [0.65, 0.8), and high synergy [0.8, 1].

Moran’s I model

The Moran’s I model is an exploratory spatial data analysis method used to examine whether spatial autocorrelation exists among research variables. It mainly includes global spatial autocorrelation analysis and local spatial autocorrelation analysis. This study applies Moran’s I to investigate the spatial association and agglomeration characteristics of CLM synergy level evolution in Sichuan Province. Global spatial autocorrelation analysis based on the global Moran’s I model is used to determine whether an overall spatial autocorrelation pattern exists in the CLM synergy level of Sichuan Province. Local spatial autocorrelation analysis based on the local Moran’s I model is applied to identify the location and extent of spatial agglomeration of CLM synergy levels among counties. The basic decomposition formula follows previous research (Arbia and Nardelli, 2025).

Geographical detector model

The geographical detector is a statistical method used to detect spatial differentiation and identify the underlying driving factors (Wang and Xu, 2017). Its principle is based on the spatial heterogeneity of geographical features. The model can identify general factor associations and is also applicable for exploring the differentiation and impacts of spatial factors. This study applies the geographical detector model to examine whether the selected driving factors are strongly associated with the CLM synergy level in Sichuan Province and to quantify their explanatory power. The theoretical basis, basic formulas, and interpretation of model parameters are described in previous studies (Wang and Xu, 2017; Zhang et al., 2024).

The coupling coordination degree of CLM is used as the dependent variable to investigate the driving factors affecting the CLM synergy level. The explanatory power of each driving factor on the spatial differentiation of the CLM synergy level is quantified using the q-statistic. The q-value represents the explanatory power of driving factors, ranging from 0 to 1; a higher q-value indicates stronger explanatory power. The p-value indicates whether the driving factor has a statistically significant effect on the evolution of the county-level CLM synergy level. The selection of driving factors follows the principles of scientific validity, representativeness, and data availability. Based on literature review and theoretical foundations (Zhu, 2024; Cui and Xu, 2026), ten factors were selected from two dimensions: regional background conditions and economic development conditions (Table 2). Continuous variables were discretized using various methods, including the natural breaks method and quantile classification, in the GD package of R (version 2020) to determine the optimal classification scheme (Li and Yan, 2024). This study adopts an optimal-parameter geographical detector framework to systematically optimize the discretization of the ten continuous driving factors. Each method is initially applied according to the data distribution for which it is suitable, such as uniform, skewed, or exponential distributions. The optimal scheme is selected by maximizing the q-value while satisfying the significance criterion of p < 0.05, ensuring that the selected discretization scheme has the strongest explanatory power for the dependent variable. All calculations are performed using the “GD” package in R.

Spatiotemporal patterns of CLM development level evolution in Sichuan Province

The assessment results of the CLM development level from 2012 to 2022 indicate that the average value of the cultivated land economic contribution function at the county level showed a declining trend followed by a recovery, with an overall decrease of 13.29%. The average value of the cultivated land food security function exhibited an increasing–decreasing trend, with an overall increase of 42.16%. The cultivated land landscape ecological function showed a continuous and gradual upward trend, with a cumulative increase of 1.15%. Compared with the other two functions, the landscape ecological function exhibited more significant polarization among counties. The maximum difference between the highest- and lowest-ranked counties reached 58.83 times in 2012, decreasing to 45 times and 19.83 times in 2017 and 2022, respectively, indicating that the polarization among counties was alleviated overall. After a decade of development, all three cultivated land subsystem functions showed evident improvement. The average CLM development levels at the county level were 0.453, 0.467, and 0.481 in 2012, 2017, and 2022, respectively, demonstrating a continuous increasing trend, with periodic growth rates of 3.16% and 2.89%. Meanwhile, the degree of polarization increased from 3.52 times in 2012 to 4.19 times in 2017 and 6.29 times in 2022, indicating that imbalance and polarization in county-level CLM development have intensified.

The spatiotemporal patterns of the cultivated land economic contribution function were characterized across the study period (Figure 3). Most areas of Sichuan Province remained at the low level of cultivated land economic contribution function ([0, 0.25)) over the study period. The spatial distribution of counties at the moderate level ([0.25, 0.5)) showed a trend of initially decreasing and then expanding. In 2012, counties with moderate economic contribution functions were mainly concentrated in western Ganzi Prefecture, northern Aba Prefecture, Ziyang, Nanchong, and Dazhou. After a decade of development, this concentrated distribution pattern weakened by 2022 and gradually shifted toward a fragmented spatial distribution. The spatiotemporal patterns of the cultivated land food security function were characterized across the study period (Figure 4). The Panxi Economic Zone and Northwestern Sichuan Ecological Demonstration Zone remained at the low level ([0, 0.25)) for a long period. After 10 years of development, most counties in the Chengdu Plain Economic Zone, Southern Sichuan Economic Zone, and Northeastern Sichuan Economic Zone gradually improved from the low level ([0, 0.25)) to the moderate level ([0.25, 0.5)). The distribution range expanded from the Chengdu Plain Economic Zone to the Southern Sichuan Economic Zone and Northeastern Sichuan Economic Zone, showing evident spatial agglomeration characteristics. The spatiotemporal patterns of the cultivated land landscape ecological function were characterized across the study period (Figure 5). Counties with good ecological function levels ([0.5, 0.75)) in 2012 were mainly concentrated in the Northeastern Sichuan Economic Zone, Southern Sichuan Economic Zone, and Chengdu Plain Economic Zone, and this spatial pattern remained largely unchanged until 2022. By 2022, the number and spatial extent of counties with good ecological function levels ([0.5, 0.75)) had increased compared with 2012. Counties with moderate function levels ([0.25, 0.5)) accounted for the largest proportion and were mainly distributed in the Panxi Economic Zone and the western and northern parts of the Chengdu Plain Economic Zone. Counties with low function levels ([0, 0.25)) were mainly concentrated in the Northwestern Sichuan Ecological Demonstration Zone. By 2022, only approximately 50% of counties in the Northwestern Sichuan Ecological Demonstration Zone remained below the moderate level ([0.25, 0.5)), while counties with good ecological function levels ([0.5, 0.75)) exhibited clear spatial agglomeration characteristics.

The spatial pattern of the overall CLM development level (Figure 6) shows that counties with low levels ([0, 0.4)) in 2012 were characterized by fragmented and adjacent distributions, with the smallest number of counties. By 2022, the spatial agglomeration of these counties had increased, mainly concentrated in the Northwestern Sichuan Ecological Demonstration Zone. More than 50% of counties in Sichuan Province were classified at the moderate level ([0.4, 0.8)), mainly distributed west of the Chengdu Plain Economic Zone, including the Panxi Economic Zone and Northwestern Sichuan Ecological Demonstration Zone. These counties showed concentrated and contiguous distributions, and this pattern remained evident from 2012 to 2022. Counties with good CLM development levels ([0.8, 1.2)) were mainly concentrated in the Northeastern Sichuan Economic Zone, Southern Sichuan Economic Zone, and eastern Chengdu Plain Economic Zone, showing highly concentrated and contiguous distributions. From 2017 to 2022, some counties in the Panxi Economic Zone improved from the moderate level ([0.4, 0.8)) to the good level ([0.8, 1.2)). The number of counties with good CLM development levels increased significantly, their distribution expanded, and spatial continuity improved. During the 10-year period, the number of counties with high CLM development levels ([1.2, 1.6]) increased substantially, with broader spatial distribution. However, fragmented distribution remained the main spatial characteristic of this category. Its spatial extent first expanded and then contracted, and these counties were mainly distributed in the Chengdu Plain Economic Zone and Northeastern Sichuan Economic Zone.

Regional disparities in the evolution of CLM development level in Sichuan Province

The measurement results (Table 3) show that the within-region disparities among the five economic zones in Sichuan Province ranged from 0.0610 to 0.1181. In 2012, the largest within-region disparity was observed in the Northwestern Sichuan Ecological Demonstration Zone, while the smallest was found in the Northeastern Sichuan Economic Zone. In 2017 and 2022, the largest within-region disparity occurred in the Northeastern Sichuan Economic Zone, whereas the smallest was observed in the Southern Sichuan Economic Zone. The within-region disparities showed an increasing trend in the Chengdu Plain Economic Zone, Northeastern Sichuan Economic Zone, and Panxi Economic Zone, with the Northeastern Sichuan Economic Zone experiencing the largest increase (80.96%) in the within-region disparity of CLM development level. The Southern Sichuan Economic Zone and Northwestern Sichuan Ecological Demonstration Zone showed declining trends, with decreases of 0.0466 and 0.0065, respectively.

The inter-regional disparity results (Table 4) indicate that regions with relatively high CLM development levels, including the Chengdu Plain Economic Zone, Southern Sichuan Economic Zone, and Northeastern Sichuan Economic Zone, exhibited substantial disparities with the Northwestern Sichuan Ecological Demonstration Zone. The inter-regional disparity values exceeded 0.32 in 2012 and remained above 0.35 in 2017 and 2022. This indicates that regional disparities in CLM development level in Sichuan Province mainly originate from differences between the Northwestern Sichuan Ecological Demonstration Zone and other regions, and such disparities are difficult to eliminate in the short term. The disparities among the Chengdu Plain Economic Zone, Southern Sichuan Economic Zone, and Northeastern Sichuan Economic Zone were relatively small, with values generally below 0.1. Reducing regional disparities in CLM development level in Sichuan Province should therefore focus on addressing regional polarization and improving the development level of CLM across different economic zones.

The decomposition results of regional disparities in CLM development level (Table 5) show that the overall disparities in Sichuan Province were 0.1577, 0.1686, and 0.1653 in 2012, 2017, and 2022, respectively. The overall disparity increased by less than 1%, indicating relatively limited changes during the study period. The within-region disparity values among the five economic zones changed slightly, and their contribution to overall disparity remained between 14.5% and 15%, suggesting that within-region disparities were not the primary source of overall disparities in CLM development level. The between-region disparity values also showed limited variation, but their contribution remained between 67% and 69%, indicating that differences among regions were the dominant factor affecting overall disparities. Therefore, reducing regional disparities in CLM development level in Sichuan Province requires greater attention to differences among regions.

Hypervariable density can be used to identify overlapping characteristics among regions. According to the core–periphery theory (Cui and Li, 2022), regional development does not occur simultaneously across all spatial units. Regions may first form “core areas” through the agglomeration of advantageous factors driven by the siphon effect, after which these core areas gradually diffuse factors to surrounding areas, forming a spatial pattern in which “peripheral areas” depend on “core areas” (Sun et al., 2024). The contribution of hypervariable density was lowest in 2017 (16.16%) and highest in 2022 (17.98%). Although the CLM development level of the Panxi Economic Zone was significantly lower than that of the Chengdu Plain Economic Zone, three counties in Liangshan Prefecture in 2012, thirteen counties in 2017, and ten counties in 2022 exceeded the lowest CLM development level among counties in Chengdu. These results indicate that CLM development does not occur simultaneously across different economic zones. Instead, core areas conducive to the siphon effect gradually emerge and promote factor diffusion, reducing both within-region and between-region disparities. Although CLM development remains constrained by regional natural conditions, overall disparities tend to decrease with socioeconomic development.

Spatiotemporal patterns of CLM synergy level evolution in Sichuan Province

The measurement results of the CLM synergy level at the county level in Sichuan Province show that the coupling coordination values of all 176 counties ranged from 0.228 to 0.713 during 2012–2022. The coupling coordination values ranged from 0.228 to 0.619 in 2012, 0.253 to 0.677 in 2017, and 0.253 to 0.713 in 2022. In 2012, only two counties were classified as weak synergy ([0, 0.25)), while 115 counties were classified as low synergy ([0.25, 0.5)), accounting for 65.34% of all counties, and 59 counties were classified as moderate synergy ([0.5, 0.65)), accounting for 33.52%. The average coupling coordination degree of Sichuan Province was 0.4529, and 45.45% of counties were below the provincial average level. In 2017, the coupling coordination degree of CLM increased slightly compared with 2012. The number of low-synergy counties ([0.25, 0.5)) decreased to 104, accounting for 59.09%, while the number of moderate-synergy counties ([0.5, 0.65)) increased to 71, accounting for 40.34%, representing an increase of 6.82% compared with 2012. One county reached the good synergy level ([0.65, 0.8)). However, 44.89% of counties remained below the provincial average synergy level (0.4672). By 2022, the number of low-synergy counties ([0.25, 0.5)) further decreased to 97, accounting for 55%, while the number of moderate-synergy counties ([0.5, 0.65)) increased to 78, accounting for 44.32%, with an increase of 3.98% compared with 2017. The average coupling coordination degree increased significantly to 0.4807, but 48.3% of counties remained below the provincial average level, and the number of such counties increased compared with 2012 and 2017. Overall, the coupling coordination degree of the 176 counties showed a slight increasing trend, but the overall synergy level remained relatively low. During the study period, the proportion of low-synergy counties continued to decline, but the rate of decline gradually slowed, while the proportion of counties below the provincial average did not decrease correspondingly, indicating that polarization among counties persisted throughout the study period.

The spatiotemporal patterns of county-level CLM synergy levels (Figure 7) indicate that low-synergy counties ([0.25, 0.5)) had the largest spatial distribution and widest coverage in 2012, accounting for approximately 70% of Sichuan Province. Only Ganzi County and Seda County in the northern part of the Northwestern Sichuan Ecological Demonstration Zone were classified as weak synergy ([0, 0.25)). Most counties in eastern Sichuan, particularly those in the Chengdu Plain Economic Zone and Northeastern Sichuan Economic Zone, were classified as moderate synergy ([0.5, 0.65)). The spatial distribution pattern of CLM synergy levels in 2017 was generally similar to that in 2012. The number of counties with moderate synergy ([0.5, 0.65)) increased further and expanded toward western Sichuan. The distribution pattern of moderate-synergy counties exhibited a characteristic of “large-scale concentration with scattered distribution.” In 2022, the major change was the continued increase in the number of moderate-synergy counties ([0.5, 0.65)), which continued to show scattered distribution and gradual expansion in the Panxi Economic Zone. A small number of low-synergy counties ([0.25, 0.5)) were scattered across the Southern Sichuan Economic Zone, Chengdu Plain Economic Zone, and Northeastern Sichuan Economic Zone. In addition, the evolution of CLM synergy levels was closely associated with the development level of CLM. At the county scale, CLM synergy levels exhibited certain spatiotemporal agglomeration characteristics across Sichuan Province, economic zones, and prefecture-level cities. The east–west disparity in CLM synergy levels was more pronounced than the disparity in CLM development levels, showing significant polarization. Many counties in western Sichuan have experienced long-term constraints from insufficient synergy development of CLM, which has substantially restricted functional improvement. Consequently, fluctuations between CLM synergy levels and CLM development levels remain evident.

Spatial agglomeration characteristics of CLM synergy level evolution

Based on the spatiotemporal patterns of CLM synergy level evolution in Sichuan Province, the spatial agglomeration characteristics were further analyzed (Figure 8). The results show that the CLM synergy level in Sichuan Province exhibited strong spatial agglomeration characteristics from 2012 to 2022. The global Moran’s I value of the CLM synergy levels were all greater than 0 and remained around 0.7, passing the significance test at the 1% level. This indicates a significant positive spatial autocorrelation of CLM synergy levels in Sichuan Province. In other words, neighboring regions tended to exhibit similar CLM synergy levels, and county-level CLM synergy levels showed significant spatial dependence. The CLM synergy levels among counties demonstrated mutual promotion and coordinated development characteristics.

From a temporal perspective (Table 6), the global Moran’s I value of CLM synergy level in Sichuan Province decreased from 0.753 in 2012 to 0.691 in 2022, representing a decline of 9%. The Z-value decreased from 16.553 to 15.386, with P = 0.000. These results indicate that although the spatial correlation of CLM synergy levels slightly weakened over the past decade, a strong positive spatial agglomeration pattern remained.

Based on the global spatial autocorrelation results, the local Moran’s I analysis was further conducted to examine the spatial agglomeration characteristics of county-level CLM synergy levels in Sichuan Province and identify statistically significant spatial clusters of hot and cold spots. In the Moran scatter plot, the horizontal axis represents the relative level of CLM synergy in the region compared with the regional average, while the vertical axis represents the spatial lag vector, calculated as the weighted average of observations from neighboring regions (Cui et al., 2024). Using Stata 16.0, local Moran’s I map of county-level CLM synergy levels in Sichuan Province for 2012, 2017, and 2022 were generated (Figure 9). The results show that: (1) Among the 176 counties, 91 counties passed the significance test in 2012, including 57 counties passing the 1% significance test; 88 counties passed the significance test in 2017, including 56 counties passing the 1% significance test; and 81 counties passed the significance test in 2022, including 49 counties passing the 1% significance test. (2) During the study period, the local spatial autocorrelation positions of some counties changed over time, while most counties were concentrated in the (HH) and (LL) quadrants. (3) From 2012 to 2022, the spatial distribution of county-level CLM synergy levels gradually shifted from being dominated by the (HH) and (LL) quadrants toward a pattern characterized by the (HH), (LH), and (LL) quadrants.

The local autocorrelation hotspot maps (Figure 10) reveal that: (1) Counties with High–High (HH) and Low–Low (LL) agglomeration characteristics mainly exhibited localized contiguous spatial clustering. High–High (HH) clusters were mainly distributed in most areas of the Chengdu Plain Economic Zone and Northeastern Sichuan Economic Zone, whereas Low–Low (LL) clusters were concentrated in most areas of the Northwestern Sichuan Ecological Demonstration Zone. (2) The spatial agglomeration characteristics of county-level local autocorrelation were associated with their CLM development levels and synergy levels. For example, counties belonging to the High–High (HH) cluster, including Wenjiang District, Jianyang City, and Xinjin District, generally ranked among the middle and upper levels in Sichuan Province in terms of the three cultivated land functions. These counties exhibited relatively high CLM synergy levels, and neighboring counties at the spatial boundaries also showed similar CLM synergy levels. (3) High–High (HH) clusters represent areas with both high CLM development levels and high synergy levels, where surrounding counties also exhibit high values and form mutual agglomeration patterns, such as Ziyang City and Meishan City. Low–Low (LL) clusters consisted of counties surrounded by neighboring counties with similarly low values, such as Aba Prefecture and Ganzi Prefecture. Overall, the analysis shows that only approximately 50% of counties exhibited significant local spatial autocorrelation. The significance of local spatial autocorrelation gradually declined, and the number of counties with significant local autocorrelation characteristics decreased.

Driving forces of CLM evolution in Sichuan Province

The CLM synergy level is an important basis for assessing whether the functions of regional cultivated land subsystems have been improved and whether the cultivated land multifunctional system has been optimized. The above results show that CLM synergy levels in Sichuan Province exhibit significant regional differentiation and spatial agglomeration. Theoretically, this differentiated pattern is not randomly distributed but results from the joint effects of regional background conditions and economic development conditions. To further test this theoretical inference, this study uses the coupling coordination degree of CLM as the dependent variable and applies the geographical detector model to systematically measure the independent explanatory power and interaction effects of driving factors from the two dimensions of regional background conditions and economic development conditions.

Measurement results of driving factors

According to the detection results of driving factors affecting the CLM synergy level, only the soil erosion factor among the 10 driving factors failed to pass the significance test in 2017 and 2022, while all other factors passed the p < 0.01 significance test. The q-values of the driving factors were mainly distributed between 0.07 and 0.75, and after excluding insignificant factors, the q-values were concentrated between 0.1 and 0.75. This indicates that the selected driving factors exhibited pronounced polarization in their explanatory power for the dependent variable (Figure 11). The distribution ranges of q-values vary among different research contexts. q-values ranging from 0.07 to 0.20 were reported for driving factors of cultivated land spatial evolution in Henan Province (Li et al., 2024a), whereas q-values ranging from 0.30 to 0.90 were reported for the driving mechanisms of rural human–land–industry coordination in Ningxia (Liu J. et al., 2025). Compared with previous studies, the q-value range in this study shows certain polarization characteristics, but remains within the acceptable range of (0, 1], indicating the rationality of the results.

According to the average q-values of the driving factors, the explanatory power ranking was: cultivated land patch density (0.715) > cultivated land aggregation degree (0.713) > rural employment population (0.712) > regional GDP (0.587) > construction intensity (0.569) > precipitation (0.555) > relief amplitude (0.533) > rural per capita net income (0.355) > soil erosion factor (0.188) > urbanization rate (0.182). From the temporal perspective, the explanatory power of regional GDP, rural per capita net income, construction intensity, and urbanization rate increased from 2012 to 2022, whereas the explanatory power of precipitation, relief amplitude, soil erosion factor, cultivated land patch density, cultivated land aggregation degree, and rural employment population declined. Regarding factor categories, three of the four regional background condition factors passed the significance test, with precipitation and relief amplitude showing declining explanatory power over time. All economic development condition factors passed the significance test, and their explanatory power generally increased over time.

Cultivated land patch density (0.715) and cultivated land aggregation degree (0.713) ranked first and second in explanatory power among the 10 driving factors, exceed even core socioeconomic indicators such as regional GDP (0.587) and urbanization rate (0.182). From a measurement perspective, cultivated land patch density and cultivated land aggregation degree are landscape pattern indices whose values jointly reflect multiple influences, including natural topographic fragmentation, land consolidation, urbanization-induced fragmentation, and agricultural management practices. They essentially provide condensed measures of the combined effects of “natural background constraints + long-term human modification” (Chen Q. et al, 2023). Spatial continuity of cultivated land is a basic prerequisite for large-scale farming, mechanized operations, and ecological corridor connectivity. Cultivated land fragmentation constrains improvements in grain yield per unit area, agricultural productivity, and habitat quality, thereby physically limiting the potential for synergistic improvement of the three functions. The high explanatory power of these landscape pattern factors reflects the deep constraints imposed on the CLM synergy level by spatial effects formed through the long-term evolution of cultivated land systems. Because patch density is difficult to change rapidly in the short term, gradually consolidating fragmented plots through land consolidation and high-standard farmland construction represents a fundamental pathway for improving the CLM synergy level.

Interactions among driving factors

Previous studies (Wang and Xu, 2017; Zhang and Qian, 2024) have shown that the effects of driving factors on dependent variables are not independent, and interactions between two factors often produce stronger explanatory effects, including bivariate enhancement and nonlinear enhancement. Therefore, this study applies interaction detection to evaluate the interactions among driving factors of CLM synergy level and examine whether factor interactions enhance or weaken their explanatory power for the dependent variable (Y). The results reflect whether multiple-factor interactions generate enhanced effects on CLM synergy level, indicating that the evolution of CLM synergy level is driven by the combined effects of multiple factors rather than by any single factor independently.

Among the 45 interaction types identified through the interaction detection of driving factors affecting CLM synergy level in Sichuan Province (Figure 12), 38 interactions in 2012 exhibited bivariate enhancement (yellow points), accounting for 84.44% of all interaction types, while seven interactions exhibited nonlinear enhancement (red points), accounting for 15.56%. The strongest bivariate interaction involved rural employment population and cultivated land aggregation degree, with a linear enhancement value of 0.8727. The strongest nonlinear interaction involved cultivated land patch density and urbanization rate, with an interaction value of 0.8534. In 2017, 43 interaction types exhibited bivariate enhancement, accounting for 95.56% of all interactions, while two interaction types showed nonlinear enhancement, accounting for 4.44%. The strongest bivariate interaction was between rural employment population and cultivated land aggregation degree, with a linear enhancement value of 0.8664. The strongest nonlinear interaction involved soil erosion factor and rural per capita net income, with an interaction value of 0.6207. In 2022, 39 interaction types exhibited bivariate enhancement, accounting for 86.67% of all interactions, while six interaction types showed nonlinear enhancement, accounting for 13.33%. The strongest bivariate interaction involved cultivated land patch density and cultivated land aggregation degree, with a linear enhancement value of 0.8223. The strongest nonlinear interaction involved soil erosion factor and rural employment population, with an interaction value of 0.7872.

Further typological identification of the 45 interaction combinations shows that three types consistently exhibited high explanatory power across different years and reflected distinct driving logics. (1) Synergistic enhancement among landscape pattern factors, represented by the interaction between cultivated land patch density and cultivated land aggregation degree. Patch density reflects the degree of cultivated land fragmentation, whereas aggregation degree reflects spatial contiguity. Both are landscape morphology indicators but characterize different dimensions: the former focuses on patch number density, while the latter focuses on spatial structure. When both are at favorable levels, with fewer patches and higher spatial contiguity, the region has a stronger basis for large-scale farming and ecological connectivity, and their joint effect on functional synergy is significantly greater than improving either dimension alone. (2) Cross-dimensional matching between labor and spatial structure, represented by the interactions between rural employment population and cultivated land aggregation degree, and between cultivated land patch density and rural employment population. Agricultural labor represents the quantity of factor inputs, whereas cultivated land spatial structure represents factor allocation efficiency; effective matching is required in both quantity and space. When spatial contiguity is high but labor is insufficient, the advantages of contiguous cultivated land cannot be converted into actual output. When labor is abundant but cultivated land is fragmented, substantial labor input is consumed in movement between plots and boundary management, resulting in diminishing marginal output. The persistently high explanatory power of these combinations indicates that “human–land” matching is a structural issue that cannot be ignored in improving the CLM synergy level in Sichuan Province. (3) Compensatory interactions between natural constraints and economic development conditions, represented by the interactions between soil erosion factor and rural per capita net income, precipitation and construction intensity, and soil erosion factor and rural employment population. Improved economic conditions provide a material basis for compensating for disadvantages in natural endowments. Higher rural income levels indicate a stronger capacity of local governments and farmers to invest in soil and water conservation and irrigation infrastructure improvement, thereby partly offsetting the constraints imposed by steep terrain or uneven precipitation on cultivated land quality. Overall, the three key interaction types show that cross-dimensional interactions are stronger than within-dimension interactions. Except for interactions among landscape pattern factors, cross-dimensional interactions between economic development conditions and regional background conditions generally have higher explanatory power than interactions within economic development conditions. This pattern indicates that improvements in a single dimension have limited effects on the synergistic development of CLM, while cross-dimensional coordinated interventions provide a more effective pathway for substantially improving the synergy level.

Characteristics of CLM evolution in Sichuan Province

The analysis of the spatiotemporal patterns of CLM evolution in Sichuan Province shows that the three cultivated land subsystem functions and CLM development level exhibit significant regional differences, which may be related to variations in topography, geomorphological conditions, and socioeconomic development among regions. The comparison between spatiotemporal patterns and the corresponding subsystem function development levels indicates that cultivated land subsystem functions and CLM development also exhibit spatial agglomeration characteristics, including regional and boundary-based distributions. Western Sichuan should be considered a key region for improving CLM development level in Sichuan Province in the future. The quantitative assessment results and regional disparity analysis indicate that the CLM development level in Sichuan Province has continuously increased, resulting from the combined improvement of the three cultivated land functions. The improvement of the economic contribution function and food security function has been the main driving force behind the increase in CLM development level. This trend may be associated with improvements in cultivated land infrastructure, increases in cultivated land area, enhanced high-standard farmland construction, and advances in agricultural green technologies during 2012–2022. The quality of CLM development in Sichuan Province remains relatively limited, with substantial regional disparities, development imbalance, and functional disharmony. The Northeastern Sichuan Economic Zone, for instance, shows considerable differences in cultivated land resource endowments among counties, resulting in large within-region disparities in CLM development levels. Rural revitalization initiatives and high-standard farmland construction projects have significantly improved cultivated land conditions, enhanced cultivated land function development, and reduced disparities with other economic zones. The Northwestern Sichuan Ecological Demonstration Zone faces constraints associated with poor agricultural production conditions, high difficulty and costs of cultivated land improvement, and ecological conservation responsibilities. CLM development in this region has improved over the past decade, and some counties have achieved relatively higher development levels within the economic zone; however, compared with other economic zones in Sichuan Province, the overall development level remains relatively low and regional disparities remain evident.

The evolutionary characteristics of cultivated land subsystem functions indicate the following mechanisms: (1) The economic contribution function of cultivated land has gradually weakened over time. The evaluation results show a declining trend, and the proportion of the economic contribution function in CLM development level has continuously decreased. This indicates that the role of the primary industry, which is based on cultivated land and dominated by crop production, has gradually weakened within regional economic structures during socioeconomic development. This finding is consistent with previous research suggesting that industrialization and urbanization promote the increasing contribution of secondary and tertiary industries, resulting in a declining contribution of the primary industry (Deng, 2011). (2) Regions with relatively high economic contribution functions of cultivated land can generally be divided into two categories. One category includes counties with strong agricultural foundations and distinctive agricultural industry clusters, where agricultural industries play an important role in regional economic structures and contribute to higher economic contribution functions of cultivated land. Anyue County in Ziyang City, for instance, had a cultivated land economic contribution index exceeding 54% in 2012, supported by characteristic agricultural industries such as Anyue lemon production. The other category includes counties with relatively weak overall development conditions, limited advantages in industrial and service-sector development, and insufficient location conditions, industrial foundations, and resource potential. Although agricultural production conditions are relatively limited, agriculture remains an accessible industry for less-developed counties due to its relatively low entry barriers. These counties therefore often have a relatively high proportion of primary industry output, resulting in higher evaluation values of cultivated land economic contribution function. Hongyuan County in Aba Prefecture had the highest evaluation value of cultivated land economic contribution function in Sichuan Province, while its cultivated land operation capacity index was only 859 million yuan in 2022. The dynamic evolution of cultivated land economic contribution function reflects broader socioeconomic development trends. The proportion of the primary industry is expected to gradually decline and stabilize at a level consistent with local industrial structures and development potential. Regional location conditions, agricultural development potential, and development conditions of other industries should be considered when evaluating future improvements in cultivated land economic contribution function.

The food security function of cultivated land in Sichuan Province has improved significantly, and its periodic increase reflects a direct improvement in function development. This indicates improvements in regional grain production conditions and grain production capacity, which are closely associated with food security policies and high-standard farmland construction. After the 18th National Congress of the Communist Party of China, governments at different levels strengthened efforts in abandoned cultivated land rehabilitation and high-standard farmland construction. Policies such as the Opinions on Establishing and Improving the Provincial Governor Responsibility System for Food Security issued in 2015 further promoted attention to cultivated land food security functions and reflected the implementation of national food security strategies in Sichuan Province (Gan et al., 2024). During 2017–2022, counties with favorable grain production conditions, abundant cultivated land resources, and high agricultural mechanization levels further improved their food security functions. Jianyang City in Chengdu was recognized as a national high-quality grain production base and a key grain production county in Sichuan Province, with grain planting areas remaining above 117 thousand mu and annual grain output exceeding 380 thousand tons.

The evaluation results of cultivated land landscape ecological function show that functional improvement varied among regions, while the overall landscape ecological function of cultivated land improved substantially. Changes in cultivated land carbon emissions and habitat quality are long-term processes, and their variations are relatively limited at the regional scale, indicating that ecological functions require long-term monitoring and assessment. Landscape function indicators are more sensitive to cultivated land improvement activities. In this study, landscape function indicators were calculated using vector data with relatively high accuracy, and improvements in cultivated land regularity and connectivity contributed to functional enhancement. Cultivated land in Sichuan is generally characterized by relatively small patch sizes, and overall spatial variation at the county level is relatively limited. The five economic zones of Sichuan Province exhibit considerable differences in natural geographical conditions. The improvement of cultivated land landscape ecological function in regions such as the Chengdu Plain Economic Zone and Southern Sichuan Economic Zone differed from that in the Northwestern Sichuan Ecological Demonstration Zone. From 2012 to 2022, cultivated land landscape ecological function showed increasing polarization, with reduced disparities within regions and enlarged disparities between regions. These patterns were mainly associated with strengthened high-standard farmland construction, which improved landscape functions, enhanced cultivated land connectivity and regularity, and promoted agricultural green production technologies that reduced chemical pollution. Studies and statistical data also reported the development of high-standard farmland construction (Xiong et al., 2016). Agricultural green transformation was also found to contribute to improvements in cultivated land ecological quality (Tang et al., 2023; Li et al., 2024b). The comparison of cultivated land landscape ecological function with food security function and economic contribution function indicates that landscape-related indicators in the landscape ecological function have stronger effects on the improvement of the other two functions. The improvement of cultivated land landscape ecological function can promote economic contribution and food security functions by optimizing cultivated land conditions, enhancing grain production capacity, and improving agricultural productivity. This finding is consistent with previous studies examining the relationships between cultivated land landscape characteristics, agricultural production efficiency, and grain production capacity (Chen Q. et al., 2023).

Characteristics of CLM synergy level evolution in Sichuan Province

The analysis of the CLM synergy level assessment results and spatial agglomeration characteristics reveals the following findings: (1) The CLM synergy level in Sichuan Province has improved over time, with the overall trend characterized by a transition from low synergy to moderate synergy. However, the number of counties achieving good synergy remains limited. This trend is observed at both county and prefecture-level city scales. (2) The CLM synergy level in Sichuan Province exhibits significant polarization during the study period. Clear regional disparities exist between eastern and western economic zones in terms of CLM synergy level, while the overall spatial evolution pattern remains relatively stable. (3) Regions with higher CLM synergy levels are generally consistent with areas exhibiting significant local spatial agglomeration characteristics. These regions are mainly concentrated in economic zones with relatively high CLM development levels and smaller internal regional disparities. (4) The analysis combining spatial agglomeration characteristics with the dynamic evolution of regional CLM synergy levels indicates that regions with favorable cultivated land conditions, better agricultural production conditions, and stronger socioeconomic development are more likely to improve their CLM synergy levels. The improvement process in these regions generally follows a pattern of initial local agglomeration followed by scattered spatial diffusion, mainly observed in areas surrounding Chengdu.

The spatial agglomeration analysis demonstrates that CLM synergy levels in Sichuan Province exhibit strong spatial correlation in regional distribution. The agglomerated regions are geographically and geomorphologically similar, such as areas with comparable hilly, mountainous, or plain landscapes. These regions also tend to have similar cultivated land production conditions and cultivated land quantities. Meanwhile, neighboring regions often belong to the same prefecture-level administrative units and share similar policy support, fiscal investment, and industrial economic influences, resulting in similarities in socioeconomic development conditions. This inference is consistent with the findings (Liu et al., 2021). Similar conclusions were also reported regarding cultivated land spatial distribution, landscape agglomeration, and protection zoning (Zhang and Qian, 2024). Investigating the driving mechanisms of CLM synergy levels from the perspectives of regional natural conditions and socioeconomic factors is of great significance.

Driving mechanism analysis of CLM synergy level evolution in Sichuan Province

Economic development condition factors reflect the influence of regional socioeconomic development on the CLM synergy level. Regional GDP, rural per capita net income, rural employment population, construction intensity, urbanization rate, and cultivated land aggregation degree all showed significant effects on the CLM synergy level. The mechanisms of these factors are discussed below. Regional economic development may reduce cultivated land quantity and weaken agricultural industrial structures. It may also increase agricultural investment and fiscal expenditure on agriculture, thereby improving modern agricultural development. GDP growth promotes increases in agricultural fiscal investment, and stable economic growth provides stronger support for sustained agricultural investment (Wu et al., 2013).

The influence of rural per capita net income on CLM synergy level is related to changes in rural income structures. Increasing agricultural income may require a reduction in the food security function of cultivated land to some extent. Meanwhile, the shift of rural residents’ income sources from agriculture to non-agricultural activities may reduce cultivated land investment and weaken the economic contribution function of cultivated land. Cash crop cultivation contributes to increases in farmers’ operating income and wage income (Song and Long, 2023).

A larger rural employment population indicates greater rural labor availability and higher labor input into cultivated land. With agricultural labor transfer and modern agricultural development, the rural employment population gradually declines while labor quality improves. Such changes do not necessarily constrain agricultural output, suggesting that the contribution of the rural employment population to the CLM synergy level may gradually decrease. A significant negative relationship between agricultural labor quantity and agricultural labor productivity has been reported (Jiang and Zhang, 2023).

The influence of construction intensity has increased over time, indicating that CLM synergy level has become increasingly affected by land development intensity. Urbanization and industrial expansion reduce cultivated land quantity and alter cultivated land spatial allocation, thereby influencing CLM. Cultivated land quantity is directly affected by construction intensity during land-use transitions (Zhang and Qian, 2024).

The influence of urbanization rate on CLM synergy level has continuously increased. The growth of urban permanent populations promotes changes in urbanization-related factors, which directly affect cultivated land quantity and spatial distribution, thereby influencing CLM synergy level. Urbanization rate directly affects cultivated land utilization during multifunctional cultivated land development (Yang X. et al., 2024).

The influence of cultivated land aggregation degree has gradually weakened. Modern agricultural development reduces cultivated land fragmentation, improves agricultural greening, and strengthens agricultural industrial development, reducing the relative contribution of cultivated land aggregation degree to CLM synergy level. High-standard farmland construction has a stronger effect on improving cultivated land multifunctional utilization among moderately scaled farmers than among small-scale farmers (Wu et al., 2025).

The measured driving factors and their corresponding mechanisms are consistent with previous studies, indicating the reliability of the results and the derived mechanisms. The interaction analysis shows that the explanatory effects of regional GDP, rural per capita net income, construction intensity, and urbanization rate increased during 2012–2022, while those of rural employment population and cultivated land aggregation degree decreased. Economic development condition factors have become increasingly important in influencing CLM synergy level evolution. The urbanization rate exhibited a stronger nonlinear enhancement effect than other driving factors. The strongest interaction occurred between cultivated land aggregation degree and rural employment population. The interaction effects among economic development condition factors were generally weaker than those between economic development condition factors and regional background condition factors. Social factors and natural environmental conditions jointly affect CLM development in the Poyang Lake Plain (Zhang et al., 2023a). Similar findings have been reported in studies on the driving mechanisms of land multifunctionality (Zhang and Zhang, 2022).

Comparison with other typical regions shows that studies of the Huang-Huai-Hai Plain and Northeast Plain have identified socioeconomic factors such as per capita disposable income, effective irrigation rate, and urbanization level as major drivers of CLM synergy level (Zhang et al., 2018a; Gao et al., 2022). The contiguous distribution of cultivated land in these plain regions means that spatial structure factors impose relatively limited constraints on functional synergy. In Sichuan Province, cultivated land patch density and aggregation degree show greater explanatory power than regional GDP and urbanization rate. This difference stems from Sichuan’s geomorphological pattern of “one part plain, two parts slopes, and seven parts mountains,” which results in naturally fragmented cultivated land and makes spatial structure itself a constraint on the CLM synergy level, a feature that is less evident in plain regions. Compared with eastern coastal provinces, the core contradiction in CLM evolution in the Yangtze River Delta and Guangdong is the supply–demand mismatch driven by rapid urbanization (Fan et al., 2026), whereas Sichuan Province faces stronger “background constraints” in areas such as the Northwestern Sichuan Ecological Demonstration Zone. Natural conditions such as high elevation and steep terrain keep the three functions at a relatively low-level equilibrium. The observed decline in the explanatory power of regional background condition factors and the increasing explanatory power of economic development condition factors reflect the gradual easing of these constraints as infrastructure improves. The pathway for improving the CLM synergy level in Sichuan Province is therefore distinctive: land consolidation and high-standard farmland construction are needed to continuously improve cultivated land resource endowments and gradually overcome the structural constraints imposed by natural conditions, providing differentiated experience for similar hilly and mountainous regions.

The driving factors affecting CLM synergy levels include regional background conditions dominated by natural characteristics and economic development conditions characterized by socioeconomic factors. The influence of regional background conditions has gradually weakened, whereas economic development conditions have become increasingly important. Improving CLM synergy level in Sichuan Province requires greater attention to economic development conditions and the interactions among different driving factors. Factors closely associated with cultivated land characteristics, such as cultivated land aggregation degree, and factors related to cultivated land utilization and transformation, such as construction intensity, play important roles in promoting CLM synergistic development.

This study established an evaluation indicator system for CLM development level based on three core subsystem functions: economic contribution function, food security function, and landscape ecological function. Taking Sichuan Province as a case study, the CLM development level of 176 counties was evaluated using data from 2012, 2017, and 2022. The Dagum Gini coefficient was applied to analyze regional disparities in CLM development levels. The coupling coordination degree model was used to characterize the evolution of CLM synergy level, while Moran’s I was applied to examine its spatial association characteristics. The geographical detector model was further employed to identify the driving factors affecting CLM synergy levels. The main conclusions are as follows. The analysis of CLM development level and regional disparities shows that the contribution of cultivated land economic functions to CLM development gradually declined from 2012 to 2022, whereas the development level of the food security function improved significantly. The landscape ecological function promoted the development of other cultivated land functions. The overall CLM development level in Sichuan Province showed a continuous increasing trend, with improvements in the economic contribution function and food security function serving as the main drivers. However, CLM development quality remained relatively limited, with substantial regional disparities and evident development imbalance and functional disharmony.

The analysis of CLM synergy level and spatial agglomeration characteristics indicates that the CLM synergy level in Sichuan Province improved from 2012 to 2022, while polarization and regional imbalance persisted during the study period. CLM synergy level exhibited significant spatial agglomeration characteristics. The agglomerated regions were geographically close, shared similar topographic and geomorphological conditions, and showed relatively small differences in socioeconomic development levels.

The analysis of driving mechanisms of CLM synergy level indicates that six factors, including regional GDP, rural per capita net income, rural employment population, and construction intensity, urbanization rate, and cultivated land aggregation degree, significantly influenced CLM synergy level. The driving factors include regional background conditions dominated by natural characteristics and economic development conditions characterized by socioeconomic factors. The influence of regional background condition factors gradually weakened, whereas the influence of economic development condition factors increased over time. This indicates that future improvements in CLM synergy level should emphasize the role of economic development conditions.

Promoting the synergistic development of CLM does not imply the simultaneous improvement of all system functions at the same level. Instead, it requires optimizing factor allocation, identifying dominant functions within the CLM system, improving disadvantaged functions, and strengthening advantageous functions. Based on the above conclusions, this study proposes the following recommendations. Ensuring food production and supply as the foundation of synergistic development. Regional objectives for cultivated land food security should be clarified, and food security protection boundaries should be established. Cultivated land utilization should ensure an appropriate proportion and spatial distribution of grain-producing cultivated land to optimize the allocation of cultivated land resources.

Enhancing the economic contribution function of cultivated land as a pathway for high-quality agricultural development. Regional agricultural industrial structures should be optimized, agricultural transformation and upgrading should be accelerated in less-developed areas, and distinctive agricultural industries should be strengthened. Investment in agricultural technological innovation should be increased, factor allocation efficiency should be improved, regional agricultural industrial systems should be enhanced, and diversified rural industries and characteristic agricultural clusters should be developed.

Promoting the value transformation of landscape ecological functions as a basis for synergistic development. Permanent basic farmland should gradually be converted into high-standard farmland to reduce cultivated land fragmentation, improve land leveling, and enhance green and ecological agricultural production while balancing grain production objectives and ecological conservation needs. Based on cultivated land landscape characteristics and ecological resources, diversified transformation and utilization of landscape ecological functions should be promoted through the development of modern agricultural tourism parks, ecological agricultural demonstration bases, rural healthcare, and leisure tourism industries.

Based on differences in functional levels and resource endowments among the five economic zones, this study further proposes differentiated regional policy pathways. The Chengdu Plain Economic Zone should prioritize the food security function, strictly safeguard the cultivated land red line in the core area of the “Tianfu Granary,” promote comprehensive high-standard farmland construction, strengthen grain production capacity, and reduce disparities among counties within the region. The Southern Sichuan Economic Zone and Northeastern Sichuan Economic Zone should prioritize the economic contribution function, build on their distinctive agricultural industrial foundations, flexibly implement the strategy of “grain production where suitable and cash crop production where suitable,” promote appropriately scaled farming and innovative factor allocation, and improve the economic output of cultivated land. The Panxi Economic Zone and Northwestern Sichuan Ecological Demonstration Zone should prioritize the landscape ecological function, explore pathways for realizing the value of cultivated land ecological products, improve ecological compensation mechanisms, develop new forms of agriculture–culture–tourism integration according to local conditions, and enhance grain self-sufficiency capacity in selected counties. Through differentiated policies based on regional positioning and functional complementarity, the five economic zones can gradually move from “regional imbalance” toward “coordinated progress,” thereby promoting the overall optimization of the CLM system in Sichuan Province.

The contribution of this study lies in systematically analyzing CLM development level and regional disparities, CLM synergy level and spatial agglomeration characteristics, and the driving mechanisms underlying CLM synergy evolution in Sichuan Province. The study provides a clear analytical framework through the selection and design of quantitative methods, and the findings are supported by comparisons with previous studies. Several limitations remain. (1) The theoretical framework requires further improvement by integrating multidimensional factors more effectively. (2) Due to difficulties in obtaining data for the evaluation indicators, the data used in this study have limitations in terms of timeliness and spatial resolution. Future research should incorporate multi-source remote sensing data and ground observation data to establish a dynamic monitoring system for CLM. These methodological improvements may provide a more comprehensive and dynamic understanding of CLM evolution and more timely scientific support for the precise formulation and dynamic adjustment of cultivated land protection policies. (3) The study period remains relatively short. A 10-year period does not constitute a long time series in the strict sense, and future research should strengthen stage-by-stage comparisons over longer time series.