Unveiling the Spatiotemporal Drivers of Green Utilization Efficiency of Cultivated Land in China: A PEST-GTWR Framework

Abstract

Promoting green utilization of cultivated land is the key to balancing resource use and ecological capacity. However, its working mechanisms are still unclear. This study attempts to address this empirical research gap through a three-stage cyclic system (Input-State-Output). It employed the PEST framework (Politics, Economy, Society, Technology) to identify external drivers. Using advanced methods, including the Super-SBM model, Dagum Gini coefficient, and Kernel density estimation, this paper mapped the spatiotemporal drivers of China’s green utilization efficiency of cultivated land (GUECL) between 2000 and 2020. The results indicate that despite some variation, the GUECL exhibited a distinct upward tendency over the study period. Spatially, efficiency was highest in northeastern China, while eastern and western China indicated moderate efficiency, and it was the lowest in central China. Regional differences generally narrowed, with trans-variation remaining the primary source of differences. External drivers varied across regions. At the national level, fiscal support and the R&D staff reduced GUECL, while economic growth increased it. In contrast, at the regional level, environmental regulation helped in western China, while income disparity boosted it in central China. Moreover, farm size and machinery use promoted GUECL in the eastern, central, and northeastern China, while cropping intensity and farmer education had positive effects in the central and eastern regions. This study provides a scientific foundation for developing region-specific strategies to promote the green utilization of cultivated land. It provides a valuable Chinese case for global research on sustainable land use.

1. Introduction

Cultivated land serves as a cornerstone of human survival and development, fulfilling multiple functions such as grain production, social stability, and ecological maintenance [1,2,3]. However, there remains a conflict between short-term exploitation and long-term conservation. To address this, the green utilization of cultivated land approach has been introduced [4].

This approach employs sustainable practices such as organic fertilization, and eco-friendly pest control to sustain yields and lower environmental impact [5,6], thereby improving resource utilization efficiency and enhancing ecosystem services. It is primarily quantified through the Green Utilization Efficiency of Cultivated Land (GUECL) [7,8]. Recently, the unsustainable intensive farming and short-sighted yield targets have significantly compromised the cultivated land ecosystem, particularly in developing countries [9]. Current challenges, such as excessive chemical use and high carbon emissions, are damaging land quality and creating ecological pressure, hindering improvements in GUECL [10,11,12]. Therefore, understanding and measuring this efficiency is essential for harmonizing economic and environmental outcomes of cultivated land system, ensuring that cultivated land can be used sustainably.

As a major global agricultural producer, China has increasingly integrated the green utilization of cultivated land into its national strategies in recent years [13,14]. Several key policies have since been launched, such as the 2015 Zero-Growth Action Plan for Fertilizer and Pesticide Use, the 2021 Black Soil Protection Project, and the 2024 Opinion on Accelerating Green Transformation. These initiatives are designed to promote the reduction and efficiency improvement of agricultural inputs such as chemical fertilizers and pesticides, ultimately synergizing agricultural production and ecological protection. According to the Ministry of Agriculture and Rural Affairs, fertilizer and pesticide use decreased by 6.19 million tons and 37,000 tons, respectively. The greenhouse gas intensity of crop production dropped by 16% from 2015 to 2020 [15], reflecting initial progress in green utilization of cultivated land. Despite this national progress, the level of GUECL varies considerably across regions. Driven by multiple external factors such as differences in development paths, fiscal capacity, and policy implementation, different regions exhibit distinct priorities during the process of cultivated land green utilization. For instance, northeastern China prioritizes land conservation and mechanization, while western China focuses on ecological restoration and compensation [16,17]. These divergences result in varying performance levels, dynamic trends, and influencing mechanisms in the GUECL, highlighting the need for context-specific policies to support regional sustainable land use.

Existing studies on GUECL have primarily focused on measurement, regional differences, and influencing mechanisms. Specifically, existing studies have advanced the evaluation system by integrating the environmental dimension like carbon emissions into the traditional resource–economy two-dimensional evaluation framework [18,19]. This effort has enriched the evaluation system of GUECL. In terms of measurement methods, early studies widely used the Data Envelopment Analysis (DEA) to calculate the efficiency [20]. However, the traditional DEA model struggles to effectively address undesirable environmental outputs, which potentially leads to biases in efficiency measurement of decision-making units (DMUs) [21]. Hence, the Super-SBM model was developed to solve this problem [22]. It treats environmental outputs as undesirable outputs and incorporates slack variables to deal with them, making it especially suitable for measuring complex green utilization efficiency. Subsequently, existing studies have used tools such as the coefficient of variation (CV) and Dagum Gini coefficient to examine regional differences and their underlying source. For example, the Dagum Gini coefficient has been employed to explore the difference in cultivated land use efficiency across three major river basins (Songhua, Yellow, and Yangtze) [19]. Based on the analysis of the regional differences of GUECL, scholars have further investigated the driving factors behind these differences. Previous studies have used a wide and inconsistent range of factors to explain GUECL, such as irrigation area, the number of agricultural scientists, per capita GDP, financial support and precipitation [23,24,25,26,27]. The analytical methods have shifted from static techniques (e.g., OLS, Tobit) that only identify average effects factors, to more dynamic models like Geodetector and GTWR, which are better at uncovering the varying impact of factors in different regions [28,29,30,31].

Despite substantial progress in measurement approaches and empirical analysis, existing studies are still limited by an underdeveloped theoretical framework and fragmented influencing indicators. First, most studies adopted a simplistic input–output framework, overlooking the critical intermediary “state” stage—the ecological transformation process where inputs are converted into outputs. Second, regarding external drivers, the identification of factors has primarily relied on fragmented, experience-based variable selection, lacking a unified theoretical framework to capture their complex impacts on GUECL. To fill these gaps, this study reconceptualized GUECL as a three-stage cyclic process (input-state-output), explicitly incorporating the internal transformation mechanisms that prior work has largely ignored. We also incorporated the PEST analytical framework (Politics, Economy, Society, and Technology) to analyze the external factors of GUECL [32]. This framework creates a structured set of indicators to solve the problem of fragmented factors and build a more coherent theory of what drives GUECL. Based on these theoretical foundations, our study presents three objectives. First, conduct a comprehensive measurement of GUECL. Second, the study explores the regional differences, and third, uncovers its driving forces. The results will provide critical information and a scientific foundation for creating effective, region-specific land use policies.

This study is organized as follows: Section 2 introduces this study’s theoretical basis and research framework; Section 3 explains the methodologies and data; Section 4 offers the results of the study; Section 5 provides discussion and limitations; and Section 6 concludes the main findings and policy implications.

2. Theoretical Basis and Research Framework

A comprehensive understanding of GUECL requires considering both internal and external drivers (Figure 1). Internally, the cultivated land use system operates through three interconnected stages, including input, state, and output. These stages coordinate the flows of material, energy, and value to align resource utilization and socio-economic development with environmental sustainability. The input stage is the system’s starting point. At this stage, production factors, including land, agricultural inputs, and labor, are introduced into this system [33]. These inputs not only determine the productive potential of cultivated land but also generate ecological burdens [34,35,36]. The main goal of this stage is to optimize the input structure to achieve rational resource allocation and ecological pressure control [37]. The state stage is the core of material and energy transformation within the system. It acts as a critical intermediary where material inputs are transformed into green outputs, typically featuring enhanced carbon sequestration and reduced non-point source pollution. This stage reflects the system’s ecological regulation capacity through improvements in ecological functions and production efficiency. The output stage presents the final results of the system operation, manifested as a trade-off between desirable outputs (economic and ecological benefits) and undesirable outputs (environmental pollution) [38]. This trade-off is achieved through market and policy mechanisms. For instance, green and organic-certified products command price premiums due to their safety and health attributes, which effectively translate ecological advantages into economic returns [39]. This stage also serves as a feedback mechanism that guides future input decisions through adaptive regulation.

Externally, the operation of this system is also affected by multi-dimensional external factors. This study used the PEST framework, categorizing the external factors influencing GUECL into four dimensions: Politics, Economy, Society, and Technology [40,41]. These dimensions influence the flow and transformation of elements within the internal system. Specifically, the politics dimension provides institutional foundations and regulatory direction. It influences input structures through policy instruments such as input guidelines, incentive mechanisms, and environmental standards [42]. The economic dimension reflects regional financial capacity and structural conditions, determining the feasibility of supporting green investments and technology adoption [43]. The society dimension captures behavioral preferences and value orientations, including farmers’ green willingness to adopt sustainable practices, directly affecting policy implementation and technological diffusion. Technology permeates all system stages and serves as a core driver for optimizing input structures, regulating ecological processes, and improving output capacity. From a systemic perspective, politics and economy constitute the external constraints and resource base, society provides behavioral responses and implementation support [44]. At the same time, technology acts as the central nexus linking external conditions to internal system performance [45]. Through the interaction between the internal cycles and external drivers, the system evolves toward a more orderly and efficient operation, thereby promoting the long-term advancement of GUECL.

3. Methodology and Materials

3.1. Study Area

This study focuses on mainland China, covering 31 provinces, municipalities, and autonomous regions across China, excluding Hong Kong, Macao, and Taiwan. China exhibits a general topographic trend of higher elevation in the west and lower elevation in the east, leading to many different types of landscapes across the country (Figure 2). This geographical complexity contributes to significant regional variations in hydrothermal conditions, soil fertility, and farming practices. This study adopted a widely used division approach, classifying mainland China into northeastern, eastern, central, and western China [46,47]. This division approach is particularly suited for analyzing GUECL due to its ability to capture the pronounced heterogeneity in cultivated land endowment, ecological conditions, and socio-economic development levels [48]. Specifically, northeastern China has black soil and high mechanization but suffers from soil degradation (0.22 cm/year of erosion). Eastern China is economically developed but faces the encroachment of cultivated land and a wide urban–rural income disparity. Central China contributes 29.2% of the country’s grain production but struggles with severe soil erosion problems. Western China, characterized by fragile ecosystems and limited fiscal capacity, encounters complex challenges in achieving green transformation. China’s regional diversity offers a typical and insightful context for analyzing regional differences in GUECL and uncovering the influencing mechanisms.

3.2. Research Methods

To provide a systematic understanding of GUECL, this study focused on three objectives, namely, measuring GUECL, revealing regional differences, and exploring the spatiotemporal drivers of GUECL (Figure 3). This sequence is logically structured, as an accurate measurement of efficiency is a prerequisite for diagnosing regional differences, which thereby forms the essential context for investigating the underlying driving mechanisms.

Firstly, the Super-SBM model was employed to measure the level and spatiotemporal distribution of GUECL accurately. This approach integrated resource inputs, desirable outputs, and undesirable outputs within the analysis framework. Secondly, to explore the regional differences and their dynamics, the Dagum Gini coefficient was applied to decompose the contributions of intra-regional differences, inter-regional differences, and the trans-variation density. Moreover, kernel density estimation was used to visually capture the dynamic evolution of the GUECL distribution, including the spatial positioning, distribution pattern, polarization analysis, and tailing phenomena of GUECL. Thirdly, to unveil the spatiotemporal drivers of GUECL, the influencing factors were selected based on the PEST analysis framework, and the GTWR model was used to quantify their impacts. Notably, this study revealed the core drivers of GUECL at both the national level and across four major regions, which can provide theoretical basis and empirical evidence for the formulation of differentiated policy strategies.

3.2.1. Measurement of GUECL: Super-SBM Model

To accurately evaluate the GUECL, this study employed the Super-SBM model. It comprehensively considers the efficiency of multiple inputs and outputs, enabling the differentiation between effective and ineffective decision-making units [25]. The model can be seen in Equation (1):

$$\begin{array}{l}minp={\displaystyle \frac{\frac{1}{m}{\sum }_{i=1}^m\left(\stackrel{\_}{x}/x_{ik}\right)}{\frac{1}{s_1+s_2}\left[{\sum }_{s=1}^{s_1}\left({\stackrel{\_}{y}}^g/y_{sk}^g\right)+{\sum }_{q=1}^{s_2}\left({\stackrel{\_}{y}}^b/y_{qk}^b\right)\right]}}\\ \left\{\begin{array}{c}\stackrel{\_}{x}\ge {\sum }_{j=1,\ne k}^n{x}_{ij}{\lambda }_j;{\stackrel{\_}{y}}^g\le {\sum }_{j=1,\ne k}^n{y}_{sj}^g{\lambda }_j;{\stackrel{\_}{y}}^b\ge {\sum }_{j=1,\ne k}^n{y}_{qj}^b{\lambda }_j;\\ \stackrel{\_}{x}\ge x_k;{\stackrel{\_}{y}}^g\le y_k^g;{\stackrel{\_}{y}}^b\ge y_k^b\\ {\lambda }_j\ge 0,i=1,2,\dots ,m;j=1,2,,\dots ,n;\\ j\ne 0,s=1,2,,\dots ,s_1;q=1,2,,\dots ,s_2\end{array}\right.\end{array}$$

(1)

where $p$ represents the GUECL, $\stackrel{\_}{x}$, ${\stackrel{\_}{y}}^g$, and ${\stackrel{\_}{y}}^b$ represent the slack variables for inputs, desirable outputs, and undesirable outputs; $\lambda$ is the weight vector.

3.2.2. Regional Differences of GUECL: Dagum Gini Coefficient

In this study, the Dagum Gini coefficient is applied to measure the regional differences of the GUECL among the four regions of China [49]. The formula is as follows:

$$G={\displaystyle \frac{\sum _{j=1}^k\sum _{h=1}^k\sum _{i=1}^{n_j}\sum _{r=1}^{n_h}|y_{ji}-y_{hr}|}{2n^2\bar{y}}}$$

(2)

where $G$ is the overall Gini coefficient of GUECL, with a higher value indicating more significant regional differences in China’s GUECL, $k$ and $n$ represent the number of regions and provinces, respectively, while $j$($h$) and $i$($r$) indicate each region and each province. $y_{ji}$($y_{hr}$) denotes the GUECL of province $i$($r$) within region $j$($h$), and $\bar{y}$ represents the mean value of GUECL.

Additionally, the intra-regional Gini coefficient $G_{jj}$ and the inter-regional Gini coefficient $G_{jh}$ can be calculated, indicating the differences among provinces within region $j$ and the differences between region $j$ and region $h$. The detailed calculation formulas are as follows:

$$G_{jj}={\displaystyle \frac{\sum _{i=1}^{n_j}\sum _{r=1}^{n_j}\left|y_{ji}-y_{jr}\right|}{2n_j^2\bar{y}}}$$

(3)

where $y_{ji}$ and $y_{jr}$ represent the GUECL of province $i$ and province $r$ within region $j$, respectively.

$$G_{jh}={\displaystyle \frac{\sum _{i=1}^{n_j}\sum _{r=1}^{n_{\mathrm{h}}}\left|y_{ji}-y_{hr}\right|}{n_j{n}_h({\bar{y}}_j+{\bar{y}}_h)}}$$

(4)

where $y_{ji}$ and $y_{hr}$ represent the GUECL of province $i$ in region $j$ and province $r$ in region $h$, respectively.

The Gini coefficient is further divided into three components: intra-regional differences ($G_w$), inter-regional differences ($G_{nb}$), and trans-variation density ($G_t$), as follows:

$$G=G_w+G_{nb}+G_t$$

(5)

3.2.3. Dynamic Evolution of GUECL: Kernel Density Estimation

We employed the Kernel Density Estimation (KDE) method to reveal the distribution features and dynamic evolution of GUECL in different regions. KDE is a non-parametric method that effectively captures the distribution characteristics of random variables [50]. The model is as follows:

$$f\left(x\right)={\displaystyle \frac{1}{NH}}{\sum }_{i=1}^{N}K\left({\displaystyle \frac{X_i-\bar{x}}{H}}\right)$$

(6)

where $N$ is the number of observations; $K$(·) is the kernel function; $H$ is the bandwidth; $X$ denotes the GUECL in different regions at various time points; $\bar{x}$ is the mean value.

3.2.4. Driving Mechanisms of GUECL: Geographically and Temporally Weighted Regression

The study compared OLS, TWR, GWR, and GTWR models to identify the most suitable one for analyzing the influencing factors and finally employed the GTWR model [51]. This model effectively addresses the issue of spatial heterogeneity by incorporating factors of spatial variation. It is capable of delivering more precise and spatially specific insights, which can pinpoint the key factors influencing GUECL at different spatial locations [26]. The model is as follows:

$$y_i={\beta }_0\left(u_i,v_i,t_i\right)+{\sum }_{k=1}^n{\beta }_k\left(u_i,v_i,t_i\right)x_{ik}+{\epsilon }_i$$

(7)

where $y_i$ denotes the GUECL value of the $i$-th observation; ($u_i,v_i,t_i)$ denote the longitude, latitude, and observation time point of the $i$-th sample point, respectively; ${\beta }_0\left(u_i,v_i,t_i\right)$ represents the intercept for the $i$-th sample point across different periods; $x_{ik}$ is the value of the $k$-th explanatory variable for the $i$-th region; and ${\epsilon }_i$ is the error term.

3.3. Selection of Variables and Data Description

3.3.1. Variables Used to Measure GUECL

Combining the aforementioned theoretical framework and existing studies [35,52,53], this study selected thirteen variables to construct an evaluation system of GUECL (

Table 1. Evolution system of GUECL.

Dimensions	Layer of Factors	Description of the Indicators
Inputs	Land	Sown area of grain crops (hm2)
	Irrigation resources	Effective irrigation area (hm2)
	Production materials	Fertilizer use per unit of land area (t·hm−2)
		Agricultural plastic film use per unit of land area (t·hm−2)
		Pesticide use per unit of land area (t·hm−2)
	Energy consumption	Agricultural diesel consumption per unit of land area (t·hm−2)
		Agricultural machinery total power per unit of land area (kW·hm−2)
	Labor	Agricultural labor force per unit of land area (person·hm−2)
Desirable outputs	Crop output	Grain yield per unit of land area (t·hm−2)
	Economic output	Agricultural output value per unit of land area (10,000 yuan·hm−2)
	Carbon sink	Carbon sequestration per unit of land area (t·hm−2)
Undesirable outputs	Non-point Source Pollution	Non-point Source Pollution Intensity (t·hm−2)
	Carbon emission	Carbon emission intensity per unit of land area (t·hm−2)

). The input factors were chosen from the resource perspective, including land, irrigation resources, production materials, energy consumption, and labor, which comprehensively reflect the actual input of various factors. The desirable output indicators are selected from socio-economic and environmental perspectives, including economic output, crop output, and carbon sink. Carbon sink is calculated: $C_i=\sum _i^n\left[0.45\times Q_i\times \left(1-W_i\right)/H_i\right]\times {\displaystyle \frac{44}{12}}$, where $C$ represents the total amount of carbon sink, $Q$ denotes the economic yield of crops such as rice, wheat, corn, beans, and tubers, $W$ represents the moisture content of each crop, and $H$ stands for the economic coefficient of each crop [19]. The undesirable outputs are selected from an environmental perspective, including agricultural non-point source pollution and carbon emissions [54,55]. Agricultural non-point source pollution is calculated based on nitrogen and phosphorus loss during cultivated land use, as well as residual amounts of agricultural plastic film and pesticides [56]. Carbon emission is calculated: $E=\sum E_i=\sum T_i\times {\delta }_i$, where $T$ denotes the specific values of agricultural carbon sources, including fertilizers, pesticides, agricultural plastic film, diesel, and agricultural irrigation, and $\delta$ denotes the carbon emission coefficient of agricultural carbon sources [57].

3.3.2. Influencing Factors of GUECL

This study selected nine factors from four dimensions based on the theoretical framework (

Table 2. Variables representing influencing factors of GUECL.

Dimensions	Variables	Short Form	References
Politics	Financial support for agriculture	FSA	[58]
	Environmental regulation intensity	ERI	[59]
Economy	Urban-rural income disparity	U-RID	[60]
	Economic development level	EDL	[61]
	Operational scale	OS	[62]
Society	Multiple cropping index	MCI	[63]
	Farmers’ average educational attainment	FAEA	[25]
Technology	Agricultural machinery input intensity	AMII	[64]
	R&D staff	R&D	[53]

). Politically, financial support for agriculture and environmental regulation intensity are represented by policy regulation intensity and ecological constraint rigidity. Economically, urban-rural income disparity, economic development level, and operational scale are reflected in urban-rural flow potential, regional support capacity, and economies of scale. Socially, a dual-channel model of multiple cropping index and farmers’ average educational attainment captures farming tradition inertia and agent transformation awareness. Technologically, agricultural machinery input intensity and R&D staff are selected to form a two-dimensional framework covering technology application and technology creation.

3.4. Data Source

This study used panel data from 2000 to 2020. To verify data authenticity, the sources of the data include the National Bureau of Statistics of China (https://www.stats.gov.cn/, accessed on 13 January 2025), as well as the China Statistical Yearbook, the China Rural Statistical Yearbook, the China Environmental Statistical Yearbook, and statistical yearbooks of various provinces.

4. Results

4.1. Temporal and Spatial Evolution of GUECL

From 2000–2020, the evolution of GUECL can be categorized into three distinct stages, namely initial growth (2000–2007), fluctuating decline (2007–2015), and stable growth stage (2015–2020) (Figure 4). During the initial stage, driven by rapid efficiency improvements across all regions, particularly in central China, GUECL increased from 0.926 to 1.065, representing a 15.01% rise. Provinces like Henan and Hubei experienced growth rates of 146.94% and 92.42%, respectively. Meanwhile, Guangdong in eastern China and Qinghai in western China also recorded notable increases of 70.21% and 159.66%. The fluctuating decline stage witnessed a slight decline in GUECL from 1.065 to 1.044, indicating a slowdown in growth momentum. The decline trend was particularly evident in western China, with Inner Mongolia and Xinjiang showing reductions of 37.5% and 11.18%, respectively. From 2015 to 2020, GUECL entered a stable growth stage, rising from 1.044 to 1.103. Growth rebounded most strongly in central China, where provinces such as Hubei and Hunan recorded increases of 28.43% and 20.56%, respectively. Overall, China’s GUECL exhibited a steady upward trajectory over the two decades, reflecting continuous improvements in green and efficient cultivated land use.

China’s GUECL underwent progressive spatial restructuring from 2000 to 2020 (Figure 5). Most provinces experienced significant improvements, and the overall spatial distribution became more balanced, gradually narrowing regional differences. In 2000, GUECL exhibited a clear spatial gradient with higher efficiency in peripheral regions, such as Tibet, Xinjiang, and Heilongjiang, and lower efficiency in most eastern and central provinces. After 2010, central China, such as Shanxi, Henan, and Hubei, experienced rapid growth in efficiency, narrowing the regional gap and reshaping the spatial pattern. Over the two decades, the spatial pattern of GUECL transitioned from peripheral dominance and central underperformance towards greater national convergence. The period prior to 2010 was characterized by widening gaps, partially offset by strong catch-up effects in central China. In contrast, the decade following 2010 witnessed a broader diffusion of high-efficiency zones and enhanced spatial coordination, reflecting a transition from polarization to relative equilibrium.

4.2. Regional Differences of GUECL

From 2000 to 2020, the overall difference in China’s GUECL showed a significant declining trend (Figure 6a). The national Gini coefficient declined from 0.202 to 0.127, indicating improved regional coordination in GUECL. At the regional level, different trends emerged across the four major regions. In northeastern China, the Gini coefficient enhanced slightly from 0.046 to 0.049, reflecting growing provincial differences. The most notable increase occurred between 2007 and 2008, when GUECL improved in Heilongjiang and Jilin but declined sharply in Liaoning. The Gini coefficient in eastern China manifested a V-shaped evolution, first decreasing and then increasing. After 2012, it rose consistently, suggesting the intra-regional differences gradually widened. Western China saw a decline in intra-differences, marked by a fall in the Gini coefficient from 0.198 to 0.082. A sharp reduction occurred between 2017 and 2018, mainly attributed to substantial GUECL improvements in Inner Mongolia and Yunnan, which helped narrow the gap with other provinces. In central China, the Gini coefficient exhibited a generally declining trend, decreasing from 0.193 to 0.123. This indicates improved coordination among provinces.

The diminishment of hatched zones over time indicates a substantial reduction in inter-regional differences across the six combination patterns (Figure 6b). From 2000 to 2020, the Gini coefficients for northeast–east, northeast–west, northeast–central, east–west, east–central, and west–central regions decreased by 0.047, 0.065, 0.089, 0.074, 0.042, and 0.121, respectively. This trend demonstrates synchronized progress in agricultural green development and enhanced inter-regional collaborative development.

The decomposition results show that the trans-variation density contributed more than inter-regional and intra-regional density in most years (Figure 6c). This suggests that the overlap problem in GUECL among different regions was relatively prominent. Therefore, it is imperative to formulate and implement regional coordination policies to ensure the common improvement of the GUECL nationwide.

4.3. Dynamic Evolution of GUECL

The dynamic evolution of GUECL at the national scale exhibits several notable structural features (Figure 7a). Regarding spatial positioning, the rightward shift of distribution curve center indicates an increase in national GUECL. The distribution pattern shifts from wide to narrow, indicating the gradual convergence of GUECL. The absence of significant tailing phenomena also confirms this. Polarization analysis reveals consistent multi-modal configurations, with dominant primary peaks and secondary peaks showing pronounced elevation differentials, evidencing coexisting bipolar and multipolar GUECL development trajectories.

From a regional perspective, the distribution curve center of the four major regions all exhibit consistent rightward shifts, reflecting the national improvement trend (Figure 7b–e). The distribution pattern shows increased peak heights and narrowed distribution ranges in eastern, central, and western China, signifying reduced absolute interprovincial differences. In contrast, northeastern China exhibited a decreased peak value with slight width expansion, suggesting a moderate increase in interprovincial differences, consistent with earlier findings on intra-regional differences. The distribution curve in central China and western China shows a leftward tailing phenomenon, indicating that some provinces significantly lag behind regional GUECL averages. In contrast, the other two regions show negligible tailing phenomena. Polarization analysis indicates transitional patterns from a multipolar to a non-polar state in northeastern and eastern China, showing remarkable mitigation of multipolar differentiation. Meanwhile, central and western China persistently maintain remarkable multipolar characteristics, reflecting prominent intra-regional differences in GUECL.

4.4. Influencing Mechanism of GUECL

4.4.1. Model Selection

According to

Table 3. Diagnostic information of the four models.

Indicators	OLS	TWR	GWR	GTWR
R2	0.169	0.283	0.573	0.781
Adjusted R2	-	0.273	0.567	0.778
AICc	0.374	−53.621	−322.776	−519.166
Sigma	-	0.222	0.171	0.123
RSS	36.984	31.947	19.027	9.775

, the GTWR model demonstrates superior performance with the highest R² value of 0.781. The adjusted R² value of the GTWR model is also higher than that of the other three models. Moreover, the GTWR model has the smallest residual sum of squares (RSS) and residual standard deviation (Sigma), reflecting the optimal fitting performance. Furthermore, AICc, an essential indicator for testing model quality, indicates that a smaller value implies better fitting of observed data and stronger explanatory power. Therefore, we selected the GTWR model as the most appropriate tool for analyzing the influencing factors of GUECL.

4.4.2. Driving Factors of GUECL Based on GTWR

Before constructing the GTWR model, we used the Variance Inflation Factor (VIF) to test for multicollinearity among driving factors (

Table 4. Descriptive statistics and regression coefficients of GTWR.

	FSA	ERI	EDL	OS	U-RID	MCI	FAEA	AMII	R&D
N	31	31	31	31	31	31	31	31	31
Min	−4.045	−7.366	−0.027	−0.862	−0.419	−0.489	−0.070	−0.006	−0.086
Max	0.570	10.649	0.178	0.363	0.163	0.258	0.078	0.007	0.012
Mean	−1.821	0.599	0.055	0.023	−0.095	0.001	0.012	0.002	−0.007
Std.	1.132	4.433	0.042	0.255	0.154	0.162	0.038	0.003	0.016
VIF	1.28	1.13	2.37	1.46	2.17	1.28	1.73	1.08	1.89
Eastern Mean	−2.301	−2.507	0.041	0.161	−0.150	−0.030	0.045	0.002	−0.003
Western Mean	−1.650	4.453	0.070	−0.146	−0.068	0.010	−0.021	−0.001	−0.010
Northeastern Mean	−0.645	−2.197	0.017	0.156	−0.198	−0.082	−0.011	0.004	−0.008
Central Mean	−1.951	−0.535	0.067	0.064	−0.006	0.076	0.034	0.004	−0.006

). The VIF values are all below 3, with an average of 1.60, meeting the basic requirements for using GTWR regression.

At the national level, the regression results in Table 4 reveal that the coefficients for FSA (−1.821), R&D (−0.007), and U-RID (−0.095) are all negative, indicating that these three factors generally have negative effects on GUECL. In contrast, the coefficient for EDL is 0.055, indicating that higher regional GDP per capita generally enhances GUECL.

From the regional perspective, each factor exerted varying influence on GUECL across the four regions, resulting in significant spatial differences in GUECL (Figure 8). In northeastern China, results show that OS and AMII emerge as the primary drivers of GUECL (Figure 8d,h), as the contiguous cultivated land and mechanized operations help optimize resource allocation and promote land use efficiency. However, ERI exerts a negative impact on GUECL (Figure 8b). The inadequate regulation of agricultural practices worsened environmental pollution and accelerates soil carbon loss, progressively influencing the enhancement of GUECL. In contrast to northeastern China, ERI in western China exerts a significantly positive effect on GUECL, particularly in Guizhou, Ningxia, and Tibet (Figure 8b). This policy-induced heterogeneity demonstrates that the spatial variation in governance outcomes directly reflects the diversity of regional land systems. However, OS, AMII, and FAEA in this region are generally negative (Figure 8d,g,h). Rugged terrain severely limits land consolidation and mechanization, impeding economies of scale. Moreover, inadequate adaptation of machinery to complex topographies and the relatively low FAEA hinder the promotion of green agricultural technologies. These constraints collectively influenced both input structure and output performance, thereby reducing GUECL. In eastern China, OS, MCI, FAEA, and AMII (Figure 8d,f–h) show generally positive effects, reflecting the synergistic influence of multiple factors on GUECL. Specifically, efficiency is driven by larger farms, productive land use, educated farmers using sustainable methods, and machinery for standardized, cleaner production. However, U-RID (Figure 8e) has a significantly negative impact, especially in economically advanced provinces such as Shanghai, Jiangsu, and Zhejiang. In central China, OS, MCI, FAEA, and AMII (Figure 8d,f–h) also show significant positive impacts on GUECL, indicating a similar driving mechanism to eastern China. Interestingly, U-RID shows a significant positive effect on GUECL in provinces such as Shanxi, Henan, and Hubei (Figure 8e). The opposite effects of U-RID demonstrate how this economic driver is reconfigured by regional development stages, thereby creating a distinct spatial heterogeneity in its impact on GUECL.

In summary, the significant differences in driving factors across China’s four major regions provide compelling empirical evidence for the theory of spatial heterogeneity. The unbalanced endowments in natural capital, developmental stages, and institutional environments across regions give rise to distinct, path-dependent, and self-reinforcing socio-ecological configurations. This inherent heterogeneity compels a shift in policy focus from seeking a single optimal pathway to implementing targeted, region-specific governance strategies that are precisely aligned with the unique characteristics of each regional land system, thereby fostering synergistic improvements in the green functionality of cultivated land through diverse, context-appropriate pathways.

5. Discussion

5.1. Interpretation of Regional GUECL Patterns

During the study period, the spatial pattern of GUECL in China shifted from pronounced regional differences toward enhanced macro-level coordination. The national Dagum Gini coefficient declined from 0.202 to 0.127 (Figure 6a), and the kernel density curves showed higher peaks and narrower spans (Figure 7a), indicating a trend of overall convergence. This accords with findings in food functional areas, which also demonstrate national convergence trend [65]. This phenomenon also aligns with spatial equilibrium and regional convergence theories.

While national-level convergence is evident, substantial intra-regional heterogeneity and persistent localized inefficiencies remain prevalent. For instance, in northeastern China, a traditionally high-efficiency grain production region, the intra-regional Gini coefficient rose and the kernel density curve widened. These results are mainly from growing disparities between Liaoning and Heilongjiang Although both provinces have actively advanced agricultural development, their industrial development priorities diverge, leading to a significant spatial disparity in GUECL. Liaoning leans more toward industrialization, while Heilongjiang focuses on agricultural advancement as its core development direction [66,67]. Additionally, intra-regional differences narrowed in western and central China, yet their kernel curves show long tails, revealing the continued lag of structurally constrained provinces such as Shanxi, Anhui, and Gansu. Despite marginal improvements, these provinces remain in the low-efficiency level, confirming the emergence of structural spatial inertia. Furthermore, trans-variation intensity has become the main source of differences, suggesting that spatial imbalance increasingly stems from divergent development trajectories rather than internal variations alone. For example, eastern China relies on human capital and economic strength (FAEA, EDL), while northeastern China depends more on operational scale and mechanization (OS, AMII). Although similar in outcomes, these divergent paths reflect fundamentally different institutional and economic conditions. In summary, while GUECL converges at the macro level, improvement mechanisms remain spatially heterogeneous. This underscores the urgent need for region-specific approaches and a cross-regional collaborative governance framework.

5.2. Interpretation of Driving Mechanisms from the PEST Perspective

This study applied the PEST framework to identify the external drivers of GUECL, which are integrated with the internal “input-state-output” system. By influencing this system’s operational mechanisms, the drivers shape its performance trajectory and efficiency, accounting for the spatial heterogeneity in GUECL (Figure 9).

Politically, institutional regulations primarily affect the input and state stages of the system by directly influencing the preference for the use of production factors. For instance, the negative impact of FSA suggests that traditional output-oriented policies may unintentionally encourage farmers to overuse environmentally harmful inputs, such as chemical fertilizers and pesticides. This finding aligns with Liu et al. [68], who concluded that such fiscal support can negatively affect grain eco-efficiency. Economically, external factors also influence both the input and state stages. At the input stage, the regional economic situation affects farmers’ affordability of green production inputs [69]. At the state stage, economic development supports GUECL by fostering green product markets and price regulation mechanisms, providing a critical external impetus for sustained system functioning. Notably, disparities in economic development could lead to spatial variations in U-RID. In central China, moderate income gaps help stabilize rural labor supply and encourage factor returns, indirectly promoting a greener input foundation. Conversely, in eastern China, widening income gaps accelerate rural labor outmigration, increasing reliance on chemical inputs among remaining farmers, and weakening ecological regulatory capacity. Socially, social capital serves as the micro-foundation linking external policies with the internal operation of the system by shaping farmers’ behavioral decisions. At the input stage, farmers’ environmental awareness determines whether production factors are optimized [70], thereby influencing the resilience of the system at the state stage. For instance, in central and eastern China, higher FAEA enhances farmers’ adoption of green utilization practices, corroborating the well-established role of human capital in facilitating pro-environmental agricultural technologies [71]. In contrast, limited education in western China hinders such adoption. Technologically, innovation capacity serves as a cross-cutting driver throughout the three stages, influencing both resource use efficiency and ecological transformation quality. AMII shows a significantly positive effect across most regions, reflecting its role in reducing redundant inputs, standardizing farming practices, and facilitating pollution control. However, the negative impact of R&D staff at the national level indicates a disconnect between scientific investment and the actual needs of green agriculture, revealing the absence of an effective loop linking technological supply, demand, and field application.

5.3. Limitations

Due to the limited availability of data, the evaluation system of GUECL did not include straw return to the field, which probably influenced the environmental dimensions of GUECL. Future research should incorporate such relevant factors into the indicator system, which would help uncover deeper driving mechanisms of GUECL and provide more accurate support for policy formulation. In addition, while GTWR model expertly revealed the spatially heterogeneous associations between drivers and GUECL, we noted an inherent inferential limit common to econometric analyses: establishing causality. This study provided the critical first step by pinpointing where these relationships are significant. Future research could build on these findings using methods like instrumental variables or difference-in-differences designs, to move from correlation to causation.

6. Conclusions and Policy Implications

6.1. Conclusions

This study reveals the spatiotemporal differences, dynamic evolution, and driving mechanisms of China’s GUECL. The main results indicate that from 2000 to 2020, China’s GUECL demonstrated significant progress and distinct spatial patterns. At the national scale, GUECL showed a clear upward trend, growing at an average annual rate of 0.88%. However, this growth was not uniform, revealing a distinct regional hierarchy. The northeastern region consistently had the highest efficiency, followed by the eastern, western, and central regions. Analysis of regional differences showed a positive trend toward greater spatial coordination. Both the national and inter-regional Gini coefficients declined significantly over these two decades, indicating a reduction in the overall gap between different areas. While most regions saw their internal differences narrow, northeastern China experienced a slight increase. The primary source of these differences was trans-variation density. This dynamic trend was further illustrated by kernel density estimation, which showed a rightward shift in the national efficiency curve, confirming overall growth. The northeastern region’s curve flattened and widened, signaling internal divergence. In contrast, the eastern, central, and western regions developed sharper, narrower peaks, pointing to a convergence of efficiency levels. The drivers behind GUECL were found to be complex and spatially heterogeneous. At the national level, FSA, U-RID, and R&D generally had a negative impact, while EDL was a consistent positive force. At the regional level, these influences varied dramatically. Scale-mechanization primarily drove the northeastern. Eastern China benefited from multi-factor synergy but was constrained by urban-rural disparities. Furthermore, central China gained a positive boost from these same disparities, reflecting dynamics of return migration, and western China’s progress relied almost exclusively on environmental investment, with other factors playing a minimal or even negative role. In a nutshell, this study highlights the spatial complexity of GUECL improvement in China and provides a scientific basis for region-specific green transformation strategies. It also contributes a “Chinese case” to the global pursuit of sustainable cultivated land use.

6.2. Policy Implications

The findings suggest that the current output-oriented agricultural subsidy system may hinder the structural adjustment toward green inputs. This practice not only exacerbates the misallocation of agricultural resources but also undermines the long-term sustainability of cultivated land. Moreover, fragmented governance mechanisms have resulted in significant spatial differences in GUECL, posing serious challenges to coordinated national efforts. To address these issues, a comprehensive and multi-dimensional policy framework is needed, which should mainly include the following three aspects:

(1) Restructure current subsidies and incentives to prioritize ecological outcomes over production targets. Targeted support should be directed toward farmers and enterprises that adopt green production techniques, such as reducing chemical input use, implementing soil protection practices, and applying integrated pest management to enhance ecological outcomes and guide behavioral change toward sustainable farming.

(2) Establish cross-regional cooperation mechanisms to promote complementarity and coordinated development of regions with comparative advantages. Northeastern China can share its technologies and land management experience with other regions. Similarly, eastern China can play a leading role in disseminating green agricultural technologies. Meanwhile, all regions should develop differentiated strategies tailored to their specific resource endowments and development contexts.

(3) Build an integrated support system integrating “monitoring-incentives- innovation” to ensure the long-term functionality of collaborative governance. This includes the deployment of technologies such as satellite remote sensing to build a unified national monitoring platform for real-time tracking of land input intensity, output efficiency, and pollution emissions. Meanwhile, GUECL should be incorporated into local government performance evaluations and linked to fiscal transfer mechanisms, creating a reward-and-penalty system that strengthens local accountability. Additionally, reforming the agricultural research and extension system is necessary to align innovation with production demands and enhance farmer adoption of green technologies through more effective training and dissemination channels.