Spatial and Temporal Differences and Influencing Factors of Eco-Efficiency of Cultivated Land Use in Main Grain-Producing Areas of China

Abstract

With global population growth and economic development, the sustainable utilization of arable land resources has become the key to guaranteeing food security and ecological balance. Eco-efficiency in cultivated land use (ECLU)has been increasingly emphasized as an important indicator of the coordinated development of agricultural production and the ecological environment. Studying ECLU in main grain-producing areas (MGPAs) is of great significance for realizing China’s food security guarantee, formulating and implementing scientific land use policies and measures, and safeguarding the long-term healthy development of agriculture. Based on provincial panel data of MGPA from 2008–2021, ECLU is calculated by the super-efficiency slacks-based measure model based on non-desired outputs (SSBM) and non-parametric kernel density estimation. The Dagum Gini coefficient decomposition model was used to explore the spatial non-equilibrium characteristics of ECLU in China, and the geographical and temporal weighted regression (GTWR) model was used to analyze the influencing factors of ECLU. The results showed the following: (1) ECLU in the MGPA showed a fluctuating upward trend, but the overall level was low. (2) In terms of regional disparity, the absolute difference in the development of ECLU among provinces showed a trend of “small-scale expansion followed by reduction”. (3) ECLU showed significant spatial imbalances, with notable internal disparities within the three basins. (4) The effects of economic development level and agricultural irrigation index on ECLU in the MGPA were positively correlated. Based on these findings, this paper suggests implementing region-specific and phased policies tailored to the natural resources and socio-economic conditions of different areas. The aim is to enhance the ecological environment, promote coordinated agricultural development, optimize regional growth, reduce agricultural disparities, and achieve sustainable development for both people and arable land.

1. Introduction

Cultivated land is an essential and scarce resource in global agricultural output, and its utilization is directly linked to food security, economic stability, and the development of ecological security. In China, known for its agricultural prowess since 2004, total grain output has consistently been higher than 650 billion kilograms for eight years in a row; the ration self-sufficiency rate has been above 100%, the cereal self-sufficiency rate above 95%, and the per capita food possession has been roughly 480 kg, which is higher than the internationally recognized 400 kg food safety line. The MGPAs, which contribute over 75% of the country’s grain, assume a pivotal position in agricultural development and the supply of major agricultural products [1]. However, with the dual drive of urbanization and population growth, the competition between the space for agricultural production and the space for urban and rural development is increasing, and a series of problems such as de-farming of cultivated land, non-grain production, non-point source pollution, and agricultural carbon emissions have become more and more prominent. These have led to a consistent decline in the quality of arable land [2], and a reduction in the area of cultivated land.

Based on the primary data from the third national land survey published by China’s Ministry of Natural Resources, it was found that the cultivated land area in the MGPA decreased from 89,102.8 hm² to 88,934 hm², a decrease of 0.189%, from 2013 to 2017. From 2017 to 2019, the cultivated land area decreased by 1568.1 hm², a reduction of 1.763% [3]. This shows a “cliff-type” downward trend, so the protection of the cultivated land of the MGPA remains critical, and cultivated land utilization is still relatively prominent. With this background, China’s “No. 1 Central Document” 2022 focuses on the intervention and containment of the “non-grain” use of cultivated land, strictly implements the priority order of arable land use, and realizes the “trinity” protection of arable land quantity, quality, and ecology. Many scholars have also devoted themselves to the study of cropland governance and protection. Chen and Hu pointed out that the governance of heavy metal-polluted cropland has a significant impact on the resilience of agricultural development [4], which provides empirical support for the ecological protection of cropland utilization. Dang, on the other hand, analyzed the changes in cropland governance strategies from the perspective of policy evolution, and further emphasized the key role of governance measures in responding to cropland utilization problems [5]. At the same time, the efficient utilization of cultivated land resources in the MGPA not only directly influences the future capacity for grain production but also plays a pivotal role in the strategic imperative of national food security. Based on this, scientifically assessing ECLU in the MGPA and excavating its driving mechanism have become important ways to achieve carbon emission reduction and sustainable use of cultivated land, improve the ecological productivity of cultivated land, and optimize the regional development pattern. This is crucial for China to ensure food security and achieve sustainable agricultural development.

ECLU refers to the extension of eco-efficiency in the cultivated land field, which reflects the rational utilization of cultivated land resources in agricultural production activities. It mainly refers to the effective allocation of cultivated land use, which ensures the minimization of resource input and environmental pollution while maximizing the expected output value of agricultural output and grain output [6]. In the setting of constructing an ecological civilization, improving farmland’s productivity and environmental friendliness, ensuring food security, and advancing sustainable agriculture in China depend on incorporating ecological factors into the evaluation of agricultural benefits.

Scholars at home and aboard have made many explorations of the evaluation system of ECLU. In the area of efficiency measurement methods, the connotation of efficiency has extended from the two-dimensional system of “resources, social economy” [7] to the three-dimensional system of “resources, socio-economic, environment” [8]. For example, Li et al. comprehensively considered socio-economic effects and policies and incorporated environmental effects to assess and plan the development of urban green space [9], which provided a more comprehensive analytical perspective for the evaluation system of ECLU. He et al. explored in depth the impacts of intensive land use on carbon emissions in urban areas of China, and revealed heterogeneous correlations between urban land use intensity and carbon emissions [10]. This provides a basis for the subsequent inclusion of carbon emissions into efficiency measurement indicators in this study. In terms of research methodology, the primary use of the SFA model [11] and the DEA model [12] has shifted to the Malmqiust index [13], the DEA hybrid model [12], the SBM model [14], and other methods. For spatiotemporal analysis, techniques such as spatial autocorrelation [15], the Theil index [16], and the kernel density estimation method [17] are primarily used. The research area has gradually been refined from national [18], to provincial [19], to city [12], to county [20]. The micro–meso–macro scales are involved in the research scale. At the microscale, carbon sequestration efficiency and carbon use efficiency have been studied based on soil [21], vegetation [22], and microorganisms [23]. The mesoscale is used to study the spatial and temporal differentiation characteristics [24], influencing factors [25], and trend prediction [26] of ECLU. At the macroscale, the ecological compensation of cultivated land [27] and the zoning of cultivated land use [28] have been proposed. In terms of influencing factors, the methods commonly used in the early stages were mainly multiple regression analysis [29], the Tobit regression model [30], etc. Currently, geographically weighted regression [31] and spatial econometric model [32] are preferred.

For example, Liu et al. use the SBM model to consider both carbon sequestration and carbon emission in ECLU [33]. Spatial auto-correlation and time-series cluster analysis methods were used to determine the spatial and temporal evolutionary patterns of ECLU in the districts and counties. The advantage is that compared with previous studies, this paper reveals how to effectively reduce carbon emissions and increase carbon sequestration while realizing economic growth based on the SBM model, so as to achieve the goal of sustainable development. The disadvantage is that different regions differ greatly in terms of economic growth, policy implementation, and the natural environment, and a unified model may not be able to fully reflect such differences. Ke et al. reveal the spatio-temporal characteristics of ECLU in the Yangtze River Economic Zone using the super-efficiency slacks-based measure (SSBM) and spatial autocorrelation models [34]. The strength of the article lies in the consideration of slack variables and the negative environmental impacts caused by non-desired outputs. However, the article did not consider the eco-efficiency differences in the study area as a whole and within the region, and proposed targeted countermeasures. Yin et al. analyze the driving mechanism of cropland utilization efficiency in the Yangtze River Economic Zone with the help of ordinary least squares (OLS) and geographically weighted regression (GWR) models [6], but the important factor of time evolution is neglected. Therefore, the comprehensiveness in the selection of influencing factors needs to be improved. Other scholars have examined the driving mechanism of ECLU in terms of socio-economic, natural environment, and agricultural development factors. For example, Li et al. used a panel Tobit model to determine the influencing factors of CLUE, and found that the reforestation index (MCI), GDP per capita (GPC), irrigation index (II), and sown area per capita (SAL) are important variables affecting ECLU. MCI, GPC, and STI have a significant positive impact on CLUE, the II negatively affects ECLU, and the SALs favorably increased ECLU [35]. In addition, there have been many useful explorations on the study of regional differences in ECLU. For example, Ke et al. used the kernel density function to characterize the dynamic evolution of regional differences in the ecological efficiency of cropland use, but while the density function method is suitable for visual display and preliminary analysis, it lacks fine quantification and little in-depth research uses this method alone [34].

In general, although current studies consider the ecological impact of farmland land use, and include the resource input required for cultivated land use and its carbon emissions and non-point source pollution in the evaluation index system of ECLU, they lack consideration of the carbon absorption of cultivated land. In fact, as an important ecosystem, cultivated land has been proven to have significant carbon sequestration and carbon sink effects in practice. Therefore, it is necessary to incorporate the carbon sink benefits of cultivated land use to comprehensively evaluate its impact on the ecological environment. The majority of the previously listed research focuses on efficiency measures for space exploration and mechanism analysis. However, its pays very little attention to the degree of spatial differentiation in ECLU and the process of non-stationary change in its influencing factors. Thus, it ignores the targeting errors that may lead to regional policy-making [36]. In addition, the traditional geographically weighted regression model does not take the time effect into account, and the research has certain limitations. Given the aforementioned issues, this research is focused on developing an evaluation index system for the carbon sink of cultivated land use. The study evaluates ECLU in MGPA using the super-efficiency SBM model with a focus on non-desired outputs. Non-parameter kernel density estimation was used to analyze the spatial distribution pattern of ECLU. Based on this, the Dagum Gini coefficient decomposition method and the GTWR model were used to explore the spatial efficiency differences between and within China’s three main grain-producing areas, deeply examining the factors influencing cultivated land use eco-efficiency in the MGPAs. Specifically, (1) the Super-SBM-Undesirable model was used to estimate the eco-efficiency values for the period 2008 to 2021 in China’s main grain-producing areas and analyze their temporal changes. (2) The Dagum Gini coefficient was used to analyze the differences and sources of agricultural eco-efficiency within and among the three main grain-producing basins in China. (3) Non-parametric kernel density estimation was used to analyze the dynamic evolution process of ECLU. (4) A GTWR model was constructed to explore the factors that affect ECLU. A proposed pathway for progress has been put forth to serve as a benchmark for harmonizing the growth of regional farmland resources and safeguarding the integrity of cultivated land.

2. Materials and Methods

2.1. Study Area

The MGPAs are essential for China’s grain production and play an important role in promoting agricultural development. In 2003, the Ministry of Agriculture categorized China’s 13 provinces as major grain-producing areas, using grain production, per capita occupancy, and commercial grain stocks as indicators, including Liaoning, Jilin, and Heilongjiang provinces in the Songhua River basin. The Yellow River Basin encompasses Hebei, Shandong, the Inner Mongolia Autonomous Region, and Henan Province. The Yangtze River basin encompasses Jiangxi, Hubei, Hunan, Jiangsu, and Anhui provinces. As shown in Figure 1, the cultivated land area of the MGPA is 8.81 × 10⁷ hm², accounting for 65.30% of the national total. It is a substantial grain production base and the core area of commercial grain supply in China. The crop cropping structure in the MGPA mainly consists of grain crops such as rice, wheat, corn, and soybeans, etc. With its superior arable land resources and natural resources, it was expected that by 2023, the MGPA would have a planting area of 1.19 × 10⁸ hm², a total grain output of 5.42 × 10⁸ t, accounting for 77.90% of the country’s total grain output, and a total output value of CNY 65.28 trillion, which would steadily shoulder the responsibility of China’s food security.

2.2. Methods

According to the literature review and research objectives, this study employs appropriate research methods, as illustrated in the operational flowchart in Figure 2. Further details on the specific research methods will be provided in the following section.

2.2.1. Super-Efficiency SBM Model Based on Non-Desired Outputs

Tone introduced the SBM model based on unexpected outputs in 2001 [37], and later he refined it to create the super-efficiency SBM model [38]. This model integrates the super-efficiency DEA and SBM models to leverage their respective benefits. Unlike the general SBM model, the super-efficiency SBM model is able to discern efficient DMUs in the frontier plane, making it a more effective comparison tool, and the model is shown as follows:

$$min\rho =\frac{\frac{1}{m}\sum _{i=1}^m \left(\overline{x}∕x_{ik}\right)}{\frac{1}{r_1+r_2}\left({\Sigma }_{s=1}^{r_1}\overline{y^d}/y_{sk}^d+\sum _{q=1}^{r_2}\overline{y^u}/y_{qk}^u\right)}$$

(1)

$$\left\{\begin{array}{c}\overline{x}\ge {\sum }_{j=1,\ne k}^n{x}_{ij{\lambda }_j};\overline{y^d}\le {\sum }_{j=1,\ne k}^n{y}_{sj}^d{\lambda }_j;\overline{y^d}\ge {\sum }_{j=1,\ne k}^n{y}_{qj}^d{\lambda }_j;\overline{x}\ge x_k;\overline{y^d}\le y_k^d;\overline{y^u}\ge y_k^u;\\ {\lambda }_j\ge 0,i=\mathrm{1,2},\cdots ,m;j=\mathrm{1,2},\cdots ,n,j\ne 0;s=\mathrm{1,2},\cdots ,r_1;q=\mathrm{1,2},\cdots ,r_2\end{array}\right.$$

(2)

where $\rho$ is ECLU; $m$, $r_1$, and $r_2$ represent the number of indicators for input, expected output, and unexpected output, respectively; $\overline{x}$, $\overline{y^d}$, and $\overline{y^u}$ present the relaxation variables of the $i$th input element, the $s$th expected output, and the $q$th unexpected output, respectively; $x_{ij}$, $y_{sk}^d$, and $y_{qk}^u$ denote, respectively, the optimal input quantity of factor $i$, the amount of output expected from factor $s$, and the unexpected output quantity of factor $q$ in the $k$th decision modules through the loosening of parameters; $k$ is the number of decision modules; $x_{ij}$, $y_{sk}^d$, and $y_{qk}^u$, respectively, represent the input quantity of factor $i$, the expected output quantity of factor $s$, and the unexpected output quantity of factor q of the $j$th decision modules; and ${\lambda }_j$ is the vector of weight.

2.2.2. Dagum Gini Coefficient and Decomposition

There are several approaches available for examining uneven development within a region. These include the coefficient of variation, the Gini coefficient, the Theil index, and other similar methods. The Dagum Gini coefficient is particularly useful because it takes into account the allocation of sub-groups. Therefore, not only does it identify the root causes of regional differences, but it also addresses the issue regarding overlap within the cluster sample. Its benefits in examining spatially uneven growth are evident. The calculation is based on the following formula:

$$G=\frac{\sum _{j=1}^k\sum _{h=1}^k\sum _{i=r}^{nj}\sum _{r=1}^{nh}\left|y_{ji}-y_{hr}\right|}{2\mu n^2}$$

(3)

where $y_{ji}$($y_{hr}$) is ECLU of unit $i\left(r\right)$. $\mathrm{Y}$ is the average efficiency of all cells; $k$ is the number of subgroups; and $n$ indicates the total number of units. The greater $G$ is, the greater the spatial difference degree of efficiency is. We can further break down the Dagum Gini coefficient into intra-regional variations ($G_w$), the discrepancy between regions ($G_{\mathrm{r}b}$), and the strength of the cross-variances between regions ($G_{\mathrm{t}}$). The formula is as follows:

$$G_w=\sum _{j=1}^k{G}_{jj}{P}_j{S}_j$$

(4)

$$G_{rb}=\sum _{j=2}^k\sum _{h=1}^{j-1}{G}_{jh}\left(P_j{S}_h+P_h{S}_j\right)D_{jh}$$

(5)

$$G_t=\sum _{j=2}^k\sum _{h=1}^{j-1}{G}_{jh}\left(1-D_{jh}\right)\left(P_j{S}_h+P_h{S}_j\right)$$

(6)

2.2.3. Non-Parametric Kernel Density Estimation

Kernel density estimation is a versatile technique that enables us to characterize the distribution mode of a radial vector by means of a smooth, continuously shaped density profile. Assuming the random vector’s density function is $f\left(x\right)$, and there are $n$ independent and identically distributed observations, $y_1$, $y_2$, …, $y_n$ for the random variable $Y$, the expression for the Kernel density function estimate is as follows:

$$f\left(x\right)=\frac{1}{nh}\sum _{i=1}^n K\left(\frac{y_i-\mathrm{y}}{h}\right)$$

(7)

where $n$ represents the number of areas under study; $h$ is the window width (bandwidth); and $K\left(-\right)$ is a stochastic kernel, plus weighting or leveling function, including Gaussian (Normal), Epanechnikov, Triangular, Quartic, and other kernel types. In this study, the kernel density function used a Gaussian kernel distribution, and the window width was $h$ = 0.9 SeN − 0.2 (c = 0.9 Se, Se was the standard deviation of the random variable observations). Kernel density measurements were performed using Matlab 2016 software.

2.2.4. GTWR Model

The addition of the time factor in the geographically weighted regression (GWR) model forms the spatiotemporal geographically weighted regression (GTWR) model, which effectively deals with the spatio-temporal non-stationarity and provides accurate estimation of the factor parameters. The model is expressed as follows:

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

(8)

The spatio-temporal coordinates of the $\mathrm{i}$th sample unit, denoted as ($u_i,v_i,t_i$), correspond to $X$ and $Y$, which represent the explanatory and interpreted variables, respectively; $p$ is the number of explanatory variables; ${\beta }_0\left(u_i,v_i,t_i\right)$ is the intercept term; ${\beta }_k\left(u_i,v_i,t_i\right)$ is the estimated coefficient of the $k$th explanatory variable; and ${\epsilon }_i$ is the model residuals. The application of the GTWR model in this research primarily relies on the ArcGIS 10.5 software, utilizing the GTWR plug-in developed by Huang et al. to conduct the calculations [39].

2.3. Indicator Selection and Data Sources

2.3.1. Indicator Selection

Improvement in ECLU can maximize resource allocation, minimize agricultural resource investment and pollution levels, enhance the benefits of agricultural production from an ecological, social, and economic standpoint, and achieve agriculture’s sustainable development. Many factors have an impact on ECLU. According to Formulas (1) and (2), this study constructed the measurement index of ECLU from three elements of input, undesirable outputs, and desirable outputs, respectively, after fully considering the cultivated land’s non-point source pollution and carbon emissions. The specific instructions are shown in

Table 1. Measuring index of ECLU in the MGPA.

Primary	Secondary Index	Description of the Indicator	Unit
Input	Land	Agricultural acreage	103 hectares
	Labor	Ground average agricultural employees	104 people·hm−2
	Water resources	Ground average irrigated area of cultivated land	103 hm−2
	Mechanical	Ground average total power of agricultural machinery	104 kilowatts·hm−2
	Chemicals	Ground average fertilizer application (pure)	104 t·hm−2
		Ground average amount of pesticide use	104 t·hm−2
		Ground average agricultural film use	t·hm−2
Undesirable outputs	Cultivated land carbon emission	Ground average agricultural carbon emissions	104 t·hm−2
	Non-point source pollution of cultivated land	Agricultural non-point source pollution	t·hm−2
Desirable outputs	Cultivated land carbon sink	Ground average carbon sink	104 t·hm−2
	grain output	Per capita grain output	kg
	Gross agricultural output	Ground average gross agricultural output	100 million RMB·hm−2

Carbon emissions are considered undesirable outputs. The following sources of carbon emissions have been identified along with their respective emission coefficients: fertilizer 0.8956 (kg/kg), pesticide 4.9341 (kg/kg), agricultural film 5.18 (kg/kg), diesel 0.5927 (kg/kg), agricultural sowing 312.6 (kg/km²), and agrarian irrigation 20.476 (kg/km²).

.

In this study, five indicators, including land, labor, water resources, machinery, and chemical products, are selected as input indicators, and seven variables are chosen. Among them, agricultural employees = persons employed in the primary sector × (value of agricultural output/value of agricultural, forestry, livestock, and fishery output). Two indicators of carbon emissions from cultivated land and non-point source pollution emissions from cultivated land are selected for undesired output. Among them, the carbon emission of agriculture refers to the carbon emission model and coefficient of Wang et al. [40], and the six emission sources of chemical fertilizer, pesticide, agricultural film, agricultural diesel, irrigation, and agricultural sowing are selected and multiplied with the corresponding coefficient. The agricultural surface pollution emissions are calculated according to the study of Lai et al. [41], and the amount of fertilizer nitrogen loss, fertilizer phosphorus loss, unproductive use of insecticides, and the residual amount of agricultural film are selected for calculation. For expected output, cultivated land carbon sink and grain yield are selected as indicators, and the cultivated land carbon sink is calculated by referring to Tang’s calculation method [42], which selects wheat, rice, corn, potato, and beans and refers to their corresponding economic coefficient and carbon absorption rate.

2.3.2. Data Sources

Following the guidelines for data accessibility and accuracy, the panel of yearly time series from 2008 to 2021 across 13 provinces produced 182 observations, which were the source of the data used in this study. The “China Statistical Yearbook”, the statistical yearbook of various provinces, the “China Rural Statistical Yearbook”, and the “China Water Resources Bulletin” provided the variable data required to estimate ECLU. Interpolation, or the neighboring two-year mean, was used to fill in certain missing data.

3. Results

3.1. Analysis of the Static Evolution Characteristics of ECLU in the Main Grain-Producing Areas

In this study, MaxDEA8.0 software was used to measure and calculate the annual average ECLU in the MGPA and three river basins from 2008 to 2021, and the specific results are shown in Figure 3.

ECLU in the MGPA generally showed a fluctuating upward trend. From 2008 to 2020, ECLU in the MGPA increased slightly yearly, fluctuating from 0.749 in 2008 to 0.902 in 2020. However, the overall efficiency value still did not reach a practical level. The efficiency value in 2021 reached 1.073, at an adequate DEA level. The excellent utilization level of cultivated land input factors and the strengthening of environmental pollution control along the line have allowed the MGPA to improve the ecological level of cultivated land use in recent years, which has improved ECLU. ECLU somewhat reflects the environmental friendliness of cultivated land use.

From the perspective of the watershed, the efficiency value of the three study areas was ranked as Songhua River Basin (0.987) > Yangtze River Basin (0.783) > Yellow River Basin (0.628), and the regional efficiency differences were apparent. From 2008 to 2013, ECLU in the Yellow River Basin experienced a period of fluctuation. At this time, due to the economic and urban development in China, the water consumption in the Yellow River basin surged, and ecological problems began to appear. Since 2012, it has been emphasized that the overall environmental civilization construction of the Yellow River Basin should be the starting point for improvements. From 2014 to 2021, the efficiency value skyrocketed, with a growth rate of 95.56%, the most significant increase. In the Yangtze River Basin, the State Council has been stepping up efforts to prevent and regulate agricultural non-point source pollution since 2014. This includes cutting back on fertilizer and pesticide use while boosting efficiency. The Yangtze River Basin’s ECLU was expected to reach a feasible level of 1.027 in 2021. The Songhua River basin has the highest overall efficiency level. The three northeastern provinces experienced a severe drought in 2014, which severely damaged agricultural production and reduced grain production. As a result, the efficiency value decreased. It dropped to its lowest point for the entire time frame. But for the majority of the study period, the Songhua River Basin’s efficiency value was at the DEA adequate level. This can be directly attributed to the three eastern provinces’ superior natural environments for agricultural production, scientific management practices, and high levels of mechanization.

3.2. Analysis of Dynamic Evolution Characteristics of ECLU

The research profiles for this study are 2008, 2012, 2016, and 2020. The distribution location, distribution shape, and distribution trend are used to analyze the dynamic evolution process of ECLU in the MGPA, as illustrated in Figure 4. This is carried out using the non-parametric kernel density estimation method. (1) From the perspective of distribution location, the center of the kernel density curve from 2008 to 2020 shifted to the right, which indicates that ECLU in China’s main grain-producing areas has a noticeable trend of improvement. (2) From the perspective of distribution shape, the kernel density curve showed a significant “double-peak” form in 2008 and 2020, indicating that the development of ECLU in the MGPA is unreasonable, and there is a particular difference between high and low values. Specific analysis shows that in 2008, the low value of ECLU in the MGPA was mainly concentrated around 0.4, and the high-value area was mainly concentrated around 1.1, showing a large gap and an apparent two-level differentiation trend. In 2020, the bimodal low-value area of the kernel density curve was concentrated around 0.7, and the high-value area was focused around 1.0. Although the two-level differentiation situation still existed, ECLU in each province improved to a certain extent compared to 2008. (3) In terms of distribution trend, the central peak of the sum density curve of ECLU in the study area exhibited a pattern of initial decrease followed by an increase. The width showed a trend of “narrow–width–narrow”, indicating that the absolute difference in the development of ECLU among the study provinces showed an evolution trend of “first expanding and then narrowing”. In conclusion, from 2008 to 2021, ECLU in the MGPA showed an upward trend, and the internal gap showed a narrowing trend and gradually converged to stability, which was related to the strengthening of the construction of an ecological civilization and the transformation of the agriculture development mode in China, and increasing support for the MGPA.

3.3. Analysis of Spatial Non-Equilibrium Characteristics of ECLU

3.3.1. Overall and Intra-Group Differences in ECLU

In order to investigate the differences in ECLU in the MGPA and river basins in China, this study used the Dagum Gini coefficient and its decomposition method to calculate the Gini coefficient and decomposition value of the cultivated land use ecological efficiency in the MGPA and the three river basins from 2008 to 2021.

Table 2. Gini coefficient of ECLU in the MGPA of China from 2008 to 2021.

Year	Overall Gini Coefficient	Intra-Group Gini Coefficient			Inter-Group Gini Coefficient
		Songhua River	Yellow River	Yangtze River	SR-YR	SR-Y	YR-Y
2008	0.2195	0.0073	0.1786	0.2777	0.1534	0.3366	0.3021
2009	0.2300	0.0281	0.1796	0.2784	0.1779	0.3493	0.3048
2010	0.2075	0.0540	0.1704	0.2157	0.1714	0.3145	0.2551
2011	0.2129	0.0477	0.1828	0.2325	0.1818	0.2946	0.2622
2012	0.1897	0.0435	0.1375	0.1827	0.2041	0.2784	0.2084
2013	0.1963	0.0190	0.1678	0.1731	0.1957	0.3057	0.2131
2014	0.1785	0.0390	0.1415	0.1284	0.1898	0.2951	0.1821
2015	0.1786	0.0087	0.1592	0.1272	0.1814	0.2921	0.1876
2016	0.1609	0.0287	0.1319	0.1207	0.1644	0.2615	0.1707
2017	0.1663	0.0232	0.1549	0.0957	0.1684	0.2763	0.1694
2018	0.1547	0.0017	0.1503	0.1513	0.1605	0.2110	0.1616
2019	0.1191	0.0045	0.1050	0.1199	0.1093	0.1758	0.1356
2020	0.1006	0.0075	0.0975	0.1202	0.0945	0.1078	0.1186
2021	0.0428	0.0224	0.0047	0.0575	0.0376	0.0627	0.0367
Average	0.1684	0.0240	0.1401	0.1629	0.1564	0.2544	0.1934

Notes: “SR-YR” means Songhua River–Yellow River, “SR-Y” means Songhua River–Yangtze River, and” YR-Y” means Yellow River–Yangtze River.

reflects the evolution process of the overall, intra-regional, and inter-regional differences in ECLU in the MGPA.

Figure 5 shows the evolution trend of the Gini coefficient in the major grain-producing area as a whole and the three river basins. The main grain-producing regions’ overall Gini coefficient averaged 0.1684 during the study period, with a specific value decline from 0.2195 in 2008 to 0.0428 in 2021. The overall Gini coefficient showed a downward development trend, indicating that ECLU in the MGPA had prominent spatial imbalance characteristics, and the regional gap was gradually narrowed.

From the perspective of the intra-group Gini coefficient, the overall Gini coefficient of the Songhua River Basin is relatively volatile and experiences the evolution process of “upward–downward–upward" repeatedly, with an average value of 0.0240 and a decline of −28.77%. The overall Gini coefficients of the Yellow River basin and the Yangtze River basin show a downward trend. The Yellow River Basin’s average Gini coefficient is 0.1401, and the basin’s trend decreases by 97.34%, from 0.1786 to 0.0047. The Yangtze River basin has the largest internal difference; the mean Gini coefficient is 0.1629, declining by 79.31% from 0.2777 to 0.0575 over the course of the study. This suggests that the three basins’ cultivated land use ecological efficiency differences are fairly evident.

3.3.2. Inter-Group Differences in ECLU

Table 2 and Figure 6 show the decomposition results of inter-regional differences in ECLU in the MGPA of China. The spatial imbalance degree of ECLU in the Songhua River–Yellow River basin, Songhua River–Yangtze River basin, and Yellow River–Yangtze River basin decreased during the study period. The coefficients decreased from 0.1534, 0.3366, and 0.3021 to 0.0376, 0.0627, and 0.0367, decreasing by 75.49%, 81.37%, and 87.85%, respectively. From the mean Dagum coefficients of each basin, the mean Gini coefficient of the Songhua River–Yellow River basin, Songhua River–Yangtze River basin, and Yellow River–Yangtze River basin are 0.1564, 0.2544, and 0.1934, respectively. Among them, the inter-regional difference between the Songhua River and the Yangtze River basin is the largest, indicating that the cooperative integration form of agricultural economic development and ecological environment of provinces in the Songhua River Basin is good, and a relatively stable alliance is formed within it. Under natural conditions such as climate, topography, and soil, the level of ECLU is much higher than that of the Yangtze River Basin. The difference between ECLU in the middle of the Songhua River and the Yellow River basin is the smallest, indicating that the provinces in the basin are geographically close and the exchange and cooperation of agricultural green production are more convenient and frequent. Hence, the gap between ECLU in different regions continues to decrease.

3.3.3. Spatial Difference and Contribution Rate of ECLU

Figure 7 shows the changing trend of the contribution rate of regionally different sources of ECLU in the MGPA of China from 2008 to 2021. In particular, intra-regional variation measures regional disparities in the ECLU between provinces within the three major river basins, interzone variation measures pure interzone disparities between regions with a high average level of the ECLU and those with a low average level of cropland utilization, and hypervariable density measures the degree of cross-cluster crossover of cluster values between regions. From the perspective of spatial differentiation of regional disparities, the average annual contribution rates of intra-regional differences, inter-regional differences, and hypervariable density of ECLU were 26.16%, 58.96%, and 14.88%, respectively. This indicates that the difference between regions is the primary source of the difference in ECLU in the MGPA of China.

Further analysis from the time dimension showed that the contribution rate of China’s MGPA shows a steady fluctuating trend, from 26.18% in 2008 rising to 27.7% in 2011 falling to 23.97% in 2016 and finally to 31.36% in 2021, with an average annual increase of 0.37%, and the overall change was divided into a “rise–decline–rise” trend. The contribution rate of hypervariable density showed a trend of fluctuating decline, from 14.27% to 7.50%; the overall annual decrease was 0.48%. The contribution rate between regions generally showed significant fluctuating trends. Overall, it rose from 59.55% to 61.14%, with an average annual increase of 0.12%. The specific performance path is a downward trend from 2008 to 2011, a rapid rebound from 2011 to 2014, a gradual decline from 2014 to 2020, and another rise from 2020 to 2021. This indicates that inter-regional differences will be the main source of spatial differences for some time to come. Therefore, coordinating the improvement of ECLU among different basins and reducing the regional differences will become a necessary measure for the green development of agriculture and the improvement of cultivated land eco-efficiency.

3.4. Change in Influencing Factors and Improvement Path of ECLU

3.4.1. Index of Influencing Factors

ECLU in the MGPA is not only affected by internal factors but also by exogenous factors such as economic development level, natural conditions, and technical level. Analyzing the main factors influencing the spatial and temporal evolution of ECLU is of great significance in realizing the efficient green use of cultivated land in the MGPA. Since the internal factors have been considered when measuring ECLU, this study analyzes the exogenous variables such as the level of economic development, agricultural technology, urban and rural development, and the endowment of cultivated land resources. To avoid spurious regression, the standardized variables were tested for multicollinearity, and the variables with a variance inflation factor (VIF) greater than ten were removed. Finally, the resulting six variables were selected from the four indicators.

The disposable income of rural residents is selected as the variable of economic development level. The improvement of residents’ incomes can alleviate the constraint on rural residents’ family funds and affect the input of agricultural production factors to a certain extent. Agricultural mechanization density is selected as the variable of agricultural technology level, which reflects the level of agricultural modernization to a certain extent and is very important for the improvement of agricultural production mode and ECLU. The urbanization level is chosen as the variable for urban and rural areas’ development level. The improvement of urbanization level is conducive to the promotion of agricultural-scale production and agricultural modernization operations, but it may also lead to the outflow of the rural labor force and produce adverse effects. Farmland management scale, agricultural rehabilitation index, and agricultural irrigation index are selected as cultivated land resource endowment variables. In agricultural production, the allocation of production factors has an essential impact on ECLU, and the specific indicators are shown in

Table 3. Index system of factors influencing ECLU in the MGPA of China.

Factor	Variable	Variable Declaration	VIF
Economic development factors	Disposable income of rural residents	Per capita disposable income of rural households	6.23
	GDP per capita	-	11.23
Agricultural technical factors	Agricultural mechanization density	Total power of machinery/planted area	5.48
Urban and rural development factors	Urbanization rate	Total power of machinery/planted area	3.13
Endowment of agricultural resources	Agricultural land management scale	Cultivated area/planted area	4.22
	Agricultural planting structure	Cash crop planting area/food crop planting area	10.56
	Agricultural reforestation index	Crop sown area/Number of people employed in agriculture	3.11
	Agricultural irrigation index	Irrigated area/cultivated area	2.13

.

3.4.2. Analysis of Influencing Factors of ECLU

The spatiotemporal geographic weighted regression model was used to examine the factors influencing the variance of ECLU in China’s main grain-producing regions between 2008 and 2021. The adjusted R² of the GTWR model was 0.93, and the AICC was −196.488, indicating that the model fitting effect was good. Identifying the dominant factors of regional ECLU can be the basis for effective, differentiated management. The main conclusions are as follows:

According to Figure 8, the overall impact factors of the MGPA in China are as follows: agricultural mechanization density (0.290) > agricultural irrigation index (0.154) > agricultural reforestation index (0.083) > urbanization rate (−0.007) > agricultural land management scale (−0.00016) > disposable income of rural residents (0.00003). Among them, the agricultural irrigation index, the agricultural reforestation index, and the disposable income of rural residents showed positive effects. Rural residents’ disposable income level has a weaker influence, falling from 0.0005 in 2008 to 0.00003 in 2021. Economic development can increase the input of agricultural production factors, promote the pace of agricultural modernization, and promote the intensive, large-scale production of cultivated land. In addition, the current agricultural subsidies tend toward water-saving irrigation, ecological agricultural products, etc., which makes the ecological efficiency of cultivated land use continue to improve. The urbanization rate shows a negative impact on the whole, from −0.028 in 2008 to −0.003 in 2021, with an overall decline in influence. The effect of the farmland management scale on ECLU showed an alternating trend between positive and negative, and the absolute values of the coefficient of regression continue to decline.

The average regression coefficient of economic development level on ECLU of the MGPA is 0.00007, showing a positive correlation on the whole. Negative regression coefficient years appeared in 2012 and 2013, during which China’s GDP growth rate was 7.7% in both cases, one of the lowest growth rates since 1999, signaling that China’s economy had moved from a high-speed growth stage to a medium-speed growth stage. Since 2014, the impact of the level of economic development on the ECLU has been increasing. The economic development level and the ECLU showed a weak negative correlation in Heilongjiang, Jilin, and Liaoning. The positive regression coefficients with large values are mainly concentrated in the Yellow River basin (mean value: 0.000049) and the Yangtze River basin (mean value: 0.000037), and the Songhua River basin is relatively less affected by its positive influence. Economic input should be concentrated on agricultural technology, agricultural equipment, and the use of organic fertilizers to ensure efficient grain production while paying more attention to improving ECLU.

The impact of agricultural mechanization density on ECLU was beneficial in Heilongjiang, Liaoning, and Jilin provinces, but negative in the others. Mechanization has no substantial impact on ECLU in Yangtze River Basin provinces, owing to the fact that size is the driving force behind mechanization. The upper reaches of the Yangtze River Basin have diverse geography that makes mechanization challenging. As a result, mechanization solutions should be chosen in accordance with local requirements while promoting agricultural mechanization. On the other hand, the absolute value of the regression coefficient is the largest in Inner Mongolia Autonomous Region (mean value: −0.93), where the comprehensive mechanization rate of crop cultivation, planting, and harvesting reached 86.1% in 2020. However, the high degree of mechanization is often accompanied by a large amount of greenhouse gases that increase the carbon footprint, and the high intensity of chemical fertilizers and agricultural use on ECLU of the negative impact is significant. Therefore, to achieve sustainable agricultural development, mechanization needs to be accompanied by the strengthening of agricultural technology promotion and the rational use of fertilizers and pesticides, among other measures.

A positive regression coefficient of the urbanization rate appears in Liaoning and Jilin provinces. The urbanization rate has a higher impact on ECLU in the Songhua River basin (mean value: −0.0026) and the Yellow River basin (mean value: −0.0073). In Jilin and Heilongjiang provinces, the urbanization rates were 72.81 percent and 63.36 percent in 2021, up 21 percent and 19 percent, respectively, from 2008. Sichuan (mean value: −0.232), Henan (mean value: −0.015, and Hunan (mean value: −0.015) provinces are most negatively affected by the urbanization rate. Urbanization can contribute to the expansion of urban areas, resulting in a decrease in agricultural land and potential effects on the environment and ecosystems. Therefore, the main grain-producing areas should pay more attention to agricultural development in the process of urbanization.

The agricultural irrigation index was positively correlated with the ecological efficiency of cultivated land use as a whole. As shown in Figure 9, the regression coefficients are negative after 2017, with a downward trend in impact intensity. The negative value area is concentrated in the Songhua River basin, and the significant value area is concentrated in Inner Mongolia Province and Hebei Province. The reason is that the level of agricultural water conservancy facilities in China is low, and the supply level is lower than the demand level, which makes the water conservancy equipment as a public good appear to have a “crowding effect” and hinders the development of productivity in the utilization of arable land. The scale of farmland management was negatively correlated with ECLU in the MGPA, and the average regression coefficient was −0.0002. The increase in productivity per unit labor force is beneficial to improve the efficiency of agricultural production. It also leads to the improvement of agricultural production efficiency. The effect of the direction of cultivated land fragmentation on ECLU was affected by the scale of cultivated land management. At the smaller cropland management scale, cropland thinning has a favorable impact on the ECLU. At the larger scale of cropland administration, the fraying of cropland can have a negative impact on land use efficiency. Therefore, the main body of agricultural management should be promoted to determine the appropriate scale of cultivated land management according to local conditions.

The intensity of the impact of the agricultural replanting index on ECLU showed a weakening trend. The average agricultural reforestation index of China’s MGPA in 2021 was 0.76, while the rehabilitation index of Inner Mongolia, Liaoning, Jilin, and Heilongjiang provinces was as high as 1.32, 1.19, 1.20, and 1.14. The regression coefficients of the corresponding land replanting indexes were 0.645, 0.660 and 0.084, respectively, and high values of the regression coefficients were found in the Songhua River Basin, the Inner Mongolia Autonomous Region and Hebei Province in the Yellow River Basin, and Sichuan Province in the Yangtze River Basin. However, the middle and lower plains of the Yangtze River have flat terrain, superior light and heat conditions, and abundant water resources. The relatively superior natural conditions have improved the grain production capacity in these areas, thereby improving ECLU.

4. Discussion

4.1. The Path of Improving ECLU of in China’s MGPA

This study measured the ECLU in China’s MGPA from 2008 to 2021, and analyzed the factors affecting it. It was found that ECLU of the MGPA showed a fluctuating upward trend, with the Songhua River basin having the highest ECLU. Kuang et al. [19] used SBM to measure ECLU to derive that ECLU of each province in China has gradually increased since 2000, with ECLU of the northeastern region being high overall. This is consistent with the conclusion of this study.

Second, through the use of nonparametric kernel density estimation, this paper finds that the differences in ECLU between provinces are gradually shrinking. Similarly, Li et al. [43] analyzed the regional differences in ECLU in China, and concluded that there is distributional extension convergence in the MGPA, implying that the differences between the extremes and the means within the region are gradually shrinking. In addition, Fan et al. [8] analyzed the regional differences in ECLU in the Yangtze River Basin, and Xue et al. [44] showed that the differences in the ecological efficiency of cropland use in the four main grain-producing regions of the Yellow River Basin all expanded significantly. Cui et al. [45] analyzed ECLU in the three northeastern provinces and found that the internal differences in ECLU in the region roughly experienced an evolutionary pattern of narrowing first and then expanding, with spatial imbalances becoming more and more obvious. This is consistent with the conclusion that the internal differences in the ecological efficiency of cropland utilization in the three major river basins are more obvious, as argued in this study.

Finally, the article obtained an overall positive correlation between the effects of economic development level and the agricultural irrigation index on ECLU in the MGPA through GTWR analysis. As a result, the following policy recommendations are put forward.

(1)

The scale of government expenditure should be expanded and the structure of financial support optimized for agriculture. There are differences in ECLU and its influencing factors in China’s three main grain-producing areas. Therefore, it is necessary to take differentiated measures to improve ECLU in China’s three MGPAs and provinces and formulate corresponding strategies according to the actual situation of cultivated land use and economic development level in each province. The impact of financial and regulatory policies in the agricultural sector should be maximized and, consequently, the enhancement of ECLU should be consistently advanced.

(2)

The application of technological innovation should be strengthened and the elements of arable land utilization should be rationally invested in. Scientific and reasonable input factors in the process of arable land use, such as improving the level of agricultural mechanization, will help to improve ECLU. The research, development, and implementation of agro-ecological technology need to be enhanced to reduce carbon emissions and non-point source pollution emissions from ECLU. This can be achieved by utilizing technological approaches such as carbon sequestration, emission reduction on cultivated land, and the efficient utilization of straw resources. While maintaining the desired level of output, it is possible to reduce the wastage of resources and minimize environmental pollution. In terms of institutional innovation, appropriate scale management should be developed through various forms, such as land transfer and land trusteeship, to improve ECLU.

(3)

Regional development should be optimized and the development gap between regions narrowed. ECLU in the MGPA is negatively affected by the urbanization rate as a whole. Therefore, we should pay attention to the occupation of agricultural land by urban land, firmly hold the “red line of cultivated land”, give full play to the development opportunities brought by new urbanization, thoroughly smooth the flow of talents, technology, capital, and other factors between urban and rural areas, and improve agricultural production and ECLU. The improvement of agricultural irrigation facilities should be promoted and ECLU improved. We will increase agricultural education and training in major grain-producing provinces and raise the level of local agricultural labor by promoting the integration of primary, secondary, and tertiary industries.

4.2. Limitations and Future Directions

This article studied the spatial and temporal variation characteristics and spatial differences of ECLU in the MGPA. The article constructs a comprehensive evaluation index system for ECLU based on the perspectives of carbon sinks, carbon emissions, and surface source pollution, which helps to assess ECLU in a more comprehensive way. The SSBM model is used to measure ECLU, which can effectively deal with the problems of multiple inputs and multiple outputs and improve the accuracy of evaluation. The study not only focuses on the spatial and temporal distribution characteristics of ECLU, but also explores in depth its regional differences, dynamic evolution, and other characteristics, which helps to reveal the differences between different regions and their change trends. The GTWR model can be compared with traditional spatial regression models (such as OLS, SAR, SEM, etc.) to demonstrate its advantages in dealing with spatial heterogeneity. Although the study area of this paper is the main grain-producing area of China, the results have potential international reference value. China’s eco-protection measures in the use of cultivated land, such as reducing the use of chemical fertilizers and pesticides and promoting organic agriculture, can reduce greenhouse gas emissions and combat climate change. These practices have a positive impact on global environmental protection, while the experience and policies developed to improve ECLU can be applied by other countries, especially for developing countries, where these technologies and experiences can help to improve their own food production capacity and sustainable agricultural development. At the same time, this study also has the following shortcomings: Improving ECLU is a complex system engineering project. In view of the availability of data, this study only uses objective influencing factors such as economic development level, agricultural technology level, urban and rural development level, and agricultural resource endowment to construct the impact index system. It does not consider the subjective factors that influence farmers’ technology acceptance and ecological farming concepts. Therefore, the following research should focus on expanding the index system of impact factors and constructing an improved index system of impact factor evaluation. This study is based on the provincial scale as the research object, and the evaluation system of ECLU can be constructed based on the micro-scale (city and county) or from the perspective of farmers in the future.

5. Conclusions

This study selected the MGPA of China as the research object; the ecological implications were incorporated into the efficiency of cultivated land usage, and a rating index for cultivated land use efficiency was developed. An evaluation index for ECLU was developed and applied to assess 13 provinces in the MGPA from 2008 to 2021. The research analyzed the spatiotemporal evolution, spatial disparities in efficiency, and factors influencing regional ECLU. The findings indicated that ECLU in the MGPA of China showed a volatile upward trend, but it remained generally low on the whole. ECLU has significant spatial differences, and the differences among the provinces in the three river basins are relatively substantial. Economic development levels and the agricultural irrigation index were found to enhance cultivated land use eco-efficiency, whereas the influence of urbanization rate and rural land management scale was diminishing. The agricultural irrigation index and the agricultural reforestation index showed a change from positive to negative influence.