Cultivated Land Green Use Efficiency and Its Influencing Factors: A Case Study of 39 Cities in the Yangtze River Basin of China

Abstract

In recent years, the Chinese government has been paying more and more attention to agricultural development and ecological protection. Improving cultivated land green use efficiency (CLGUE) is becoming a crucial issue in promoting the sustainable development of agriculture. This study aims to study the current situation and influencing factors of agricultural production from the perspective of green utilization efficiency of cultivated land. It takes 39 cities in the upper, middle and lower reaches of the Yangtze River basin in China as an example. The CLGUE values in those 39 cities from 2011 to 2020 were specifically measured, using the Super-SBM model, kernel density estimation and geographic detector method. Their temporal and spatial heterogeneity was described, and the influencing factors were detected at both single and interactive levels. The results showed that (1) from 2011 to 2020, the green utilization efficiency value of cultivated land in the Yangtze River basin showed an upward trend on the whole; (2) there is clear spatial heterogeneity between CLGUE values in the Yangtze River basin cities, and the distribution is as follows: downstream region > midstream region > upstream region; (3) cultivated land resource endowment, socioeconomic development and agricultural production technology are important factors affecting the variability in CLGUE values. However, there are some differences in the degree and direction of influence of different influencing factors on different sample subgroups.

1. Introduction

At present, sudden challenges such as international trade frictions and extreme climate disasters occur frequently. High-quality agricultural development is an inevitable requirement for adapting to the current economic situation [1] Cultivated land resources are the material basis for human survival and development. However, in the process of urbanization and industrialization, in many countries, including China, cultivated land resources are facing a sharp drop in total area. These problems, such as the idle barren phenomenon and a variety of pollution, seriously restrict the sustainable development of agriculture [2]. The amount of arable land per capita in the world dropped from 0.41 hectares in 1960 to 0.21 hectares in 2019 [3], and China’s arable land area reduced by 6.36 percent in the five years according to the third National Land Survey. The sharp decrease of cultivated land area and the enormous population have brought great challenges to global agricultural development. In addition, with the improvement in science and technology, the input of pesticides, fertilizers and other production used in the agricultural production process also increases total global carbon emissions, thus bringing environmental problems such as global warming, which is contrary to the concept of green and sustainable development [4]. Therefore, on the basis of efficient and reasonable utilization of cultivated land resources, the concept of green and low-carbon production has become the focus of sustainable agricultural development. Scientific and reliable indexing of farmland utilization efficiency can be used as an important decision-making reference to promote the optimization of agricultural layout and sustainable development [5].

As a composite system of the interaction between society, economy and the ecological environment, CLGUE is expected to maximize the comprehensive benefits of society, economy and the ecological environment with reasonable input elements such as pesticides and fertilizers [6,7]. CLGUE values are calculated through the input and output of cultivated land, and the spatial imbalance characteristics of cultivated land utilization and time are revealed, so as to provide decision-making basis for the optimal allocation of cultivated land resources more scientifically, and further promote the high-quality development of China’s agriculture. With the deepening of the concept of sustainable development, the ecological environment index cannot be ignored anymore in the calculation of cultivated land utilization efficiency, which is embodied in the name “green utilization efficiency of cultivated land”. Xie Hualin et al. defined “green utilization efficiency of cultivated land” as “the maximum economic and ecological benefits that can be realized in the utilization process of cultivated land under certain economic and environmental costs” [8].

Existing studies related to land use efficiency can be summarized into several characteristics: (1) In the construction of the index system, the evaluation index of land use efficiency has gradually changed from the simple “input” and “output” indexes [9] to the “non-desirable output” [10,11,12] index with ecological value as the core. (2) In terms of research method selection, the method has gradually transferred from early qualitative analysis to quantitative analysis in recent years. Data envelope analysis (DEA) [5,13,14], random frontier production function [15], SBM [16,17,18] and other methods [19,20] are mostly used for empirical analysis. (3) In terms of the research subjects, most research has focused on urban land [13,15,21], followed by agricultural land [4,22,23] and a small number of studies involving forests and grassland [24]. In general, CLGUE has barely been discussed in existing studies. As for the selection of indicators, there is a problem of incomplete index selection, focusing on economic benefits and ignoring ecological efficiency and social benefits. What is more, most of the research areas are based on traditional provincial administrative regions [11,18,25], and economic coordinated development urban agglomeration [9,26], which is a relatively broad view [14,27]. Meanwhile, those rarely analyzed cultivated land utilization efficiency from the perspective of grain production. Although Liu Mengba [28], Gai Zhaoxue [29] and others studied the spatial and temporal evolution characteristics of cultivated land utilization efficiency in grain production areas, they essentially took the province as the decision-making unit. As for the exploration of the influencing factors of cultivated land utilization efficiency value, the main methods adopted in the existing literature are mediation effect analysis [30], Tobit regression [9,31] and the obstacle identification method [32].

Based on the current research status of the utilization efficiency of cultivated land, this paper makes a further expansion. Firstly, the selection of indicators is more scientific and comprehensive. Regarding input indicators, land, labor, capital and technology, a total of seven indicators are used to comprehensively evaluate the input in the process of farmland utilization. The food security coefficient is selected as the social output measurement index from the output index, while most existing studies ignore this index, or simply make the social output index using total grain output [33]. Compared to total grain output, the index of food security coefficient can better reflect the actual supply level of local grain, so as to better reflect the contribution of cultivated land to the food security of the whole society. In terms of “non-desirable” output indicators, the carbon emissions of various carbon sources in the process of cultivated land utilization are fully considered. Secondly, in terms of the selection of research areas, the Yangtze River basin, an important major grain producing area, is chosen, breaking the boundary of the traditional provincial administrative unit, and measuring the green utilization efficiency value of cultivated land from the perspective of grain production. Combined with the differences in agricultural production structure, economic development level and the number of permanent residents in different regions in the upper, middle and lower reaches of the Yangtze River basin, the heterogeneity of CLGUE values in the three basins is analyzed in space, so as to improve the utilization level of cultivated land according to different local conditions. Thirdly, the detection of the influencing factors of CLGUE is more comprehensive. In addition to the detection of individual influencing factors one by one, as in existing studies [34], this paper also uses geographical detectors to further explore the interaction among the influencing factors based on the realistic consideration that cultivated land utilization efficiency is often influenced by multiple factors. Interaction probing of the influence factors revealed that the interaction influence of any two factors is consistently greater than that of the individual factor.

2. Research Design

2.1. Research Methods

2.1.1. Super-Efficient SBM Models including Non-Desired Outputs

Data envelopment analysis (DEA) is a traditional method for calculating utilization efficiency by using multiple input–output indexes that is widely used in the measurement of cultivated land resources. Traditional DEA models include BCC models with variable returns to scale (Banker, Charnes and Cooper) and CCR models with constant returns to scale (Charnes, Cooper and Rhodes). However, these two models only horizontally compare DMUs at the same time, and the error is relatively large. In 2001, Tone proposed the slack-based model (SBM) on this basis, which set slack variables in the objective function, making it possible to effectively compensate for the deviation. However, in the simple SBM model, the efficiency value of multiple DMUs is 1, which cannot realize the effective evaluation and ranking of DMUs. Therefore, Tone proposed a Super-SBM model based on modified relaxation variables, which can avoid the loss of information of effective DMUs and calculate its efficiency value greater than 1, so that effective DMUs with efficiency values greater than 1 can be compared [18]. The basic principle of the Super-SBM model containing non-desired outputs is as follows: Assuming that $\mathrm{n}$ is the number of decision units in the arable land use process, $\mathrm{m}$ is the number of input factors, ${\mathrm{s}}_1$ is the number of desired outputs and ${\mathrm{s}}_2$is the number of non-desired outputs. $\mathrm{X}$, ${\mathrm{y}}^{\mathrm{a}}$ and ${\mathrm{y}}^{\mathrm{b}}$are the vectors represented by input, output and undesired output, respectively. The expression of this model is as follows:

$$\mathsf{\rho }=\mathrm{m}\mathrm{i}\mathrm{n}\frac{1+\frac{1}{\mathrm{m}}\sum _{\mathrm{i}=1}^{\mathrm{m}}\frac{{\mathrm{D}}_{\mathrm{i}}^{\mathrm{c}}}{{\mathrm{x}}_{\mathrm{i}\mathrm{h}}}}{1-\frac{1}{{\mathrm{s}}_1+{\mathrm{s}}_2}\left(\sum _{\mathrm{r}=1}^{{\mathrm{s}}_1}\frac{{\mathrm{D}}_{\mathrm{r}}^{\mathrm{a}}}{{\mathrm{y}}_{\mathrm{r}\mathrm{h}}^{\mathrm{a}}}+\sum _{\mathrm{k}=1}^{{\mathrm{s}}_2}\frac{{\mathrm{D}}_{\mathrm{k}}^{\mathrm{b}}}{{\mathrm{y}}_{\mathrm{k}\mathrm{h}}^{\mathrm{b}}}\right)}$$

$$\mathrm{S}.\mathrm{T}.\left\{\begin{array}{c}{\mathrm{x}}_{\mathrm{i}\mathrm{k}}\ge {\displaystyle \sum _{\mathrm{j}=1,\mathrm{j}\ne \mathrm{h}}^{\mathrm{n}}}{\mathsf{\lambda }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-{\mathrm{D}}_{\mathrm{i}}^{\mathrm{c}}, \mathrm{i}=1,\cdots \mathrm{m}\\ {\mathrm{y}}_{\mathrm{r}\mathrm{h}}^{\mathrm{a}}\gg {\displaystyle \sum _{\mathrm{j}=1,\mathrm{j}\ne \mathrm{h}}^{\mathrm{n}}}{\mathsf{\lambda }}_{\mathrm{j}}{\mathrm{y}}_{\mathrm{r}\mathrm{j}}^{\mathrm{a}}+{\mathrm{D}}_{\mathrm{r}}^{\mathrm{a}}, \mathrm{r}=1,\cdots {\mathrm{s}}_1\\ {\mathrm{y}}_{\mathrm{k}\mathrm{h}}^{\mathrm{b}}\gg {\displaystyle \sum _{\mathrm{j}=1,\mathrm{j}\ne \mathrm{h}}^{\mathrm{n}}}{\mathsf{\lambda }}_{\mathrm{j}}{\mathrm{y}}_{\mathrm{k}\mathrm{j}}^{\mathrm{b}}+{\mathrm{D}}_{\mathrm{k}}^{\mathrm{b}}, \mathrm{k}=1,\cdots {\mathrm{s}}_2\\ 1-\frac{1}{{\mathrm{s}}_1+{\mathrm{s}}_2}({\displaystyle \sum _{\mathrm{r}=1}^{{\mathrm{s}}_1}}\frac{{\mathrm{D}}_{\mathrm{r}}^{\mathrm{a}}}{{\mathrm{y}}_{\mathrm{r}\mathrm{h}}^{\mathrm{a}}}+{\displaystyle \sum _{\mathrm{k}=1}^{{\mathrm{s}}_2}}\frac{{\mathrm{D}}_{\mathrm{k}}^{\mathrm{b}}}{{\mathrm{y}}_{\mathrm{k}\mathrm{h}}^{\mathrm{b}}}>0\\ {\mathrm{D}}^{\mathrm{a}}\ge 0,{\mathrm{D}}^{\mathrm{b}}\ge 0,{\mathrm{D}}^{\mathrm{c}}\ge 0\end{array}\right.$$

where $\mathsf{\rho }$ is the green land use efficiency index of the decision unit and ${\mathrm{D}}^{\mathrm{a}},{\mathrm{D}}^{\mathrm{b}}$ and ${\mathrm{D}}^{\mathrm{c}}$ represent the desired output, non-desired output, and slack variables of the input variables, respectively, as the weight vector.

2.1.2. The Kernel Density Estimation Method for Exploring Spatial Disequilibrium

In order to study the spatial nonequilibrium characteristics of CLGUE for 39 cities in the upstream, midstream and downstream regions of the Yangtze River basin, kernel density estimation method is selected in this paper, which uses nonparametric estimation to describe the dynamic distribution of data. This method is characterized by not making any prior assumptions about the sample data, but directly studying the distribution characteristics of the data through the data sample itself. The advantage of this method is that it can effectively avoid the subjectivity caused by the function setting in parameter estimation, thus increasing the objectivity and realism of the results. Assuming that ${\prime \mathrm{X}}_1,{\mathrm{X}}_2,\cdots , {\mathrm{X}}_{\mathrm{n}}^{\prime }$ is an independent identically distributed sample of the unit variable $\mathrm{X}$, the kernel density estimate of the probability density function that $\mathrm{X}$ obeys is as follows:

$$\mathrm{f}\left(\mathrm{x}\right)=\frac{1}{\mathrm{n}\mathrm{h}}\sum _{\mathrm{i}=1}^{\mathrm{n}}\mathrm{k}\left(\frac{{\mathrm{X}}_{\mathrm{i}}-\overline{\mathrm{x}}}{\mathrm{h}}\right)$$

$\mathrm{k}\left(\mathrm{x}\right)$ is the kernel function, $\mathrm{n}$is the observed value and $\mathrm{h}$ is the bandwidth as the mean value.

2.1.3. Geographical Detector

Geographic detector is a research method used to reveal spatial heterogeneity and its driving factors. Its core idea is that if an independent variable has an important impact on the dependent variable, then the independent variable and the dependent variable should have similar spatial distribution. The magnitude of influence is measured by $\mathrm{q}$. A larger value of $\mathrm{q}$ indicates a greater effect of the independent variable on the dependent variable, and the smaller the effect otherwise. The expression is as follows:

$$\mathrm{q}=1-\frac{\sum _{\mathrm{h}=1}^{\mathrm{L}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2}{\mathrm{N}{\mathsf{\sigma }}^2}$$

$\mathrm{N}$ and ${\mathrm{N}}_{\mathrm{h}}$ represent the number of regions and the number of subregions, respectively. ${\mathsf{\sigma }}^2$ and ${\mathsf{\sigma }}_{\mathrm{h}}^2$ represent the variance in arable land use efficiency of regions and subregions, respectively.

2.2. Overview of the Study Area

Most of the Yangtze River basin is located in a subtropical monsoon region, with a warm and humid climate. Therefore, the region has superior light, heat, water and soil conditions for agricultural production, which is the most important agricultural production base in China. There are more than 24.6 million hectares of cultivated land, accounting for a quarter of China’s total cultivated land area, and the value of agricultural production accounts for 40% of China’s total agricultural output, as well as 40% of the country’s total grain production. The Chengdu Plain, Jianghan Plain, Dongting Lake area, Poyang Lake area, Chaohu Lake area and Taihu Lake area located in this basin are the main commercial food bases in China.

According to the different geographical conditions and hydrological characteristics of the main stream, the Yangtze River basin is divided into three sections: upper, middle and lower reaches. The upstream Yangtze River basin contains six cities including Chengdu, Yibin, Panzhihua, Luzhou, Chongqing and Yichang. Located between the Qinghai–Tibet Plateau and the Sichuan Basin, the basin has complicated terrain and rapid flow, which makes it difficult to utilize water resources. Midstream is from Yichang to Hukou County, Jiujiang City. The main cities include Wuhan, Xiangyang, Ezhou, Huanggang, Jingzhou, Enshi, Xianning, Huangshi, Yueyang and Jiujiang. There are many tributaries and lakes in the midstream, which flow through Jianghan Plain, Dongting Lake Plain and Poyang Lake plain, and are prone to flood disasters. Below Hukou is downstream, where the main 23 cities are Nanchang, Hefei, Maanshan, Tongling, Anqing, Chizhou, Wuhu, Huangshan, Nanjing, Zhenjiang, Yangzhou, Suzhou, Wuxi, Changzhou, Nantong, Taizhou, Shanghai, Hangzhou, Jiaxing, Huzhou, Ningbo, Shaoxing and Zhoushan. The downstream basin has wide river surface and smooth water flow, which is conducive to the development of shipping. At the same time, the Yangtze River basin spans the east, middle and west parts of China. Due to factors such as geographical location and policy development, there are obvious differences in the development status in the upper, middle and lower reaches of the basin. The study area map is shown in Figure 1.

2.3. Indicator System

2.3.1. Input Indicators

Referring to existing studies on the values of CLGUE by Lu Xinhai [12] and Liang Liutao [35], this paper comprehensively considers the existing research system, classifies the attributes of production factors according to the production function, and selects indicators from four aspects: land, labor, capital and technology. Based on the principles of comprehensiveness, scientificity and operability, seven indexes—cultivated land area, agricultural employees, effective irrigation area, total power of agricultural machinery, amount of pesticide use, amount of fertilizer conversion and amount of agricultural film use—were specifically selected. The selected indicators and their related descriptions are shown in

Table 1. Description of input and output indicators.

Tier 1 Indicators	Tier 2 Indicators	Indicator Description	Abbreviations
Input indicators	Cultivated land area	Year-end cultivated land area/thousand hectares	${\mathrm{I}}_1$
	Workforce	Number of people employed in agriculture/10,000	${\mathrm{I}}_2$
	Irrigation	Effective irrigated area/thousand hectares	${\mathrm{I}}_3$
	Farm machinery	Total power of agricultural machinery/million kilowatts	${\mathrm{I}}_4$
	Pesticides	Pesticide use/million tons	${\mathrm{I}}_5$
	Fertilizer	Fertilizer equivalent pure application amount/million tons	${\mathrm{I}}_6$
	Agricultural film	Amount of agricultural film used/ton	${\mathrm{I}}_7$
Desired output indicators	Economic output	Gross agricultural product/billion yuan	${\mathrm{Y}}_1$
	Social output	Food security factor	${\mathrm{Y}}_2$
Non-desired output indicators	Carbon emissions	Total carbon emissions/ton	E

below.

2.3.2. Desired Output Indicators

In the selection of output indicators, Ke Nan [36], Lu Xinhai [12] and Zang Junmei [18] selected total agricultural output value and total grain production from the two major levels of economic and social benefits, respectively. This paper further expands on this basis by selecting the gross agricultural output value as the indicator of economic benefits and the coefficient of food security as the indicator of social benefits, which can reflect the level of food supply in the region more scientifically than total food production. The food security factor is calculated as total food production/total resident population/400 kg, which is derived from the internationally defined safety line of 400 kg of food possession per capita per year.

2.3.3. Non-Desired Output Indicators

The non-desired output refers to the total carbon emission in the process of cultivated land utilization, such as carbon emission generated in the process of agricultural machinery use, fertilizer, pesticide, agricultural film, etc. The formula for measuring the carbon emission in the process of arable land use is as follows:

$$\mathrm{E}=\sum {\mathrm{E}}_{\mathrm{i}}=\sum {\mathrm{M}}_{\mathrm{i}}·{\mathsf{\sigma }}_{\mathrm{i}}$$

In the above equation, $\mathrm{E}$ represents the total carbon emission of the decision unit. $\mathrm{H}$ represents the carbon emission of the ith carbon source in the process of root utilization. $\mathrm{M}$ represents the base data of the ith carbon source, and $\mathrm{O}$ refers to the carbon emission coefficient of this carbon source. Referring to existing studies, the carbon emission coefficients of individual carbon sources in the process of cultivated land utilization are summarized as shown in

Table 2. Carbon emission factors for each carbon source.

Carbon Source	Carbon Emission Factor
Plowing	$312.6 \mathrm{kg}/{\mathrm{k}\mathrm{m}}^2$
Workforce	0.18 kg/kw
Irrigation	$25 \mathrm{kg}/{\mathrm{k}\mathrm{m}}^2$
Farm machinery	4.9341 kg/kg
Pesticides	0.8956 kg/kg
Fertilizer	5.18 kg/kg

[37].

2.4. Data Source

The data for this study were mainly obtained from the 2011–2020 statistical yearbooks of each prefecture-level city and the statistical yearbooks of the provinces where each prefecture-level city is located. The cultivated land area data of each prefecture-level city for some years were obtained from the national land survey data of the second and third surveys released by the Ministry of Natural Resources of China.

3. Analysis of the Results of Measuring the Values of CLUE

3.1. Time-Series Evolutionary Characteristics of CLUE

In this paper, the Super-SBM model was constructed with the help of Matlab 2020b software to measure the specific values of CLUE in 39 cities in the Yangtze River basin during 2011–2020, and the statistics are shown in

Table 3. The values of GLGUE in 39 cities in the upstream, midstream and downstream of the Yangtze River basin from 2011 to 2020.

Watershed	City	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
Upstream	Chengdu	1.1023	1.1153	1.1196	1.1163	1.1205	1.1067	1.1084	1.1152	1.1114	1.1107
	Luzhou	0.5817	0.5666	0.5361	0.5119	0.6461	0.6705	0.6661	0.6419	0.6518	0.6922
	Panzhihua	0.4316	0.4353	0.4124	0.4149	0.4186	0.4200	0.4119	0.4371	0.4476	0.4536
	Yibin	0.5724	0.5771	0.5629	0.5906	0.5655	0.5905	0.5971	0.5850	0.5841	0.5951
	Yichang	1.1368	1.1067	1.0916	1.0820	1.0586	1.0583	1.3226	1.0524	1.0488	1.0502
	Chongqing	0.1061	0.1075	0.1045	0.0995	0.1556	0.1586	0.1056	0.1637	0.1672	0.1799
Average value		0.6552	0.6514	0.6379	0.6359	0.6608	0.6674	0.7020	0.6659	0.6685	0.6803
Midstream	Ezhou	1.0896	1.1505	1.1252	1.1793	1.2708	1.3141	0.5215	1.3172	1.3114	1.2626
	Enshi	1.0848	1.0961	1.0966	1.0788	1.103	1.1195	1.0199	1.1259	1.1288	1.0427
	Huanggang	0.4515	0.4718	0.4648	0.4536	0.4996	0.5153	0.4894	0.4833	0.5322	0.5363
	Huangshi	0.6631	0.6628	0.5996	0.5909	0.7593	1.0156	1.0453	1.0216	1.0173	1.0298
	Jingzhou	0.4804	0.384	0.4862	0.4813	0.4847	0.4875	0.4704	0.4902	0.4934	0.5127
	Jiujiang	0.2316	0.2406	0.2791	0.2880	0.3626	0.3563	1.2129	0.3522	0.3777	0.382
	Wuhan	1.0138	1.0031	1.0209	1.0364	1.059	1.0716	0.5409	1.0713	1.0624	1.0607
	Xianning	1.1287	1.628	1.1276	1.0473	1.1205	1.2194	1.2603	1.2244	1.1975	1.2093
	Xiangyang	0.6718	0.6444	0.6132	1.0002	0.6077	0.568	0.4312	0.6492	0.6637	0.6424
	Yueyang	0.2665	0.2612	0.2404	0.2452	0.2573	0.2593	0.2920	0.2870	0.3000	0.3185
Average value		0.7082	0.7543	0.7054	0.7401	0.7525	0.7927	0.7284	0.8022	0.8084	0.7997
Downstream	Anqing	0.3541	0.3323	0.2977	0.3144	0.3245	0.3205	0.3562	0.3401	0.3223	0.339
	Chizhou	0.7999	0.9977	1.0133	1.0771	1.1095	1.1584	0.6776	1.0949	1.0769	1.1107
	Hefei	0.2739	0.29	0.2584	0.2687	0.3506	0.348	0.4033	0.3573	0.3508	0.3433
	Huangshan	0.5981	0.5759	0.5309	0.5535	0.5525	0.5681	0.5747	0.5699	0.5235	0.5047
	Jiaxing	0.6349	0.5539	0.5223	0.5113	0.5285	0.5874	0.7915	1.0051	1.0147	1.0265
	Maanshan	0.5433	0.5236	0.5472	0.5611	1.0119	1.0079	1.1401	1.015	1.0234	1.015
	Nanchang	0.3673	0.3385	0.415	0.459	0.5019	0.4804	0.3404	0.5034	0.477	0.4586
	Nanjing	1.1219	1.1366	1.1352	1.1368	1.1363	1.1401	1.195	1.1362	1.1452	1.1545
	Shanghai	1.1458	1.085	1.0468	1.0447	1.0257	1.075	1.1251	1.1372	1.2642	1.2419
	Taizhou	0.4951	0.4495	0.5017	0.5678	0.5667	0.5083	0.454	0.4788	0.4595	0.481
	Tongling	1.3722	1.3191	1.3698	1.4684	1.246	1.1003	0.7115	1.134	1.1416	1.1389
	Wuhu	0.4148	0.4034	0.4079	0.4404	0.4895	0.4482	1.0489	0.4561	0.4526	0.4876
	Yangzhou	0.796	1.0003	0.82	0.7709	0.7957	0.7371	0.6956	0.7314	0.6948	0.7085
	Zhenjiang	1.0023	1.0565	1.0022	1.0088	1.0721	1.0668	1.0262	1.0399	1.0198	1.0083
	Suzhou	1.0012	1.0119	1.0151	1.0109	1.0041	1.0121	1.0086	1.0116	1.1158	1.1445
	Wuxi	1.0056	1.0322	1.0354	1.0391	1.0347	1.042	0.4335	1.0299	1.0092	1.0082
	Changzhou	1.0671	1.058	1.0471	1.0369	1.0345	1.0298	1.0639	1.0104	1.1455	1.1153
	Nantong	0.364	0.3538	0.3772	0.375	0.4559	0.4352	0.4698	0.428	0.4556	0.4567
	Hangzhou	1.0326	1.0501	1.0562	1.0633	1.0583	1.0696	1.0042	1.0798	1.0838	1.084
	Huzhou	1.0586	1.0835	1.0645	1.0464	1.0502	1.0625	1.0118	1.0522	1.0537	1.056
	Ningbo	1.0491	1.0411	1.0351	1.0279	1.0168	1.0113	1.0818	1.0315	1.018	1.0223
	Shaoxing	1.071	1.0584	1.0566	1.057	1.0304	1.0163	1.0384	1.0152	1.0057	0.932
	Zhoushan	1.004	1.0081	1.0098	1.008	1.037	1.074	1.0861	1.0901	1.0768	1.083
Average value		0.8075	0.8156	0.8072	0.8195	0.8449	0.8391	0.8147	0.8586	0.8665	0.8661

below. According to Table 3, the change trend chart of cultivated land use efficiency in the Yangtze River basin was drawn, as shown in Figure 2. The results showed that the cultivated land use efficiency values in the upper, middle and lower reaches of the Yangtze River basin showed an increasing trend during 2011–2020. This is mainly caused by three factors. Firstly, in terms of input, it is mainly reflected in the reduction in labor force and the input of production factors. During the study period, the number of employees in China’s primary industry decreased from 265 million to 177 million, dropping by 33.21 percent. Secondly, at the level of economic output and social output, the continuous improvement in agricultural technology level brings about the improvement in crop output level per unit area. Thirdly, as for non-expected output, with the continuous promotion of the concept of green and low-carbon development in recent years, various industries have actively implemented the new development concept and gradually realized green transformation. The agricultural production process is no exception. With the vigorous promotion of new and renewable energy technologies, carbon emissions in the agricultural production process have also been declining.

There are also differences in the rising trend of cropland use efficiency in the Yangtze River basin. The mean CLGUE of the six cities in the upstream region of the Yangtze River changed more slowly, always fluctuating above and below the mean value. The overall change in the midstream region was more dramatic, and the downstream cities were basically in a stable to slightly increasing trend during the study period. As for the overall varietal trend of the CLGUE values in the three major regions of the Yangtze River basin, there is a slight decline in efficiency values among the six cities in the upper reaches, except Yichang. The efficiency values of the other five cities have increased in fluctuation. The remaining eight cities in the midstream of the Yangtze River, including Ezhou, Huanggang and Huangshi, showed an increase in cropland efficiency during the study period, with Huangshi showing the most significant increase in cultivated land efficiency. Among the 10 midstream cities, Enshi and Xiangyang are among those whose efficiency values have declined overall, but the downward trend is not obvious. The remaining eight cities in the midstream of the Yangtze River, including Ezhou, Huanggang and Huangshi, showed an increase in cultivated land efficiency during the study period, with Huangshi showing the most significant increase. Among the 23 cities in the downstream region, the efficiency values decreased in Anqing, Huangshan, Taizhou, Tongling, Yangzhou and Shaoxing, while the values of CLGUE in other cities showed an increasing trend, with the rate of increase being more obvious in Chizhou, Jiaxing and Maanshan.

3.2. Spatial Distribution Characteristics of CLGUE

The mean value of CLUE in the upstream cities of the study area during the study period is 0.6625, the mean value of efficiency in the midstream cities is 0.7592 and the mean value of efficiency in the downstream cities is 0.8340.

In addition, it was found that there are some differences in the proportion of effective decision-making units among different basins. According to the calculation principle of the Super-SBM model, it is known that an efficiency value greater than 1 is an efficient decision unit, and, conversely, an efficiency value less than 1 is an inefficient decision unit. From Table 3, it can be seen that the only city with a cultivated land use efficiency value greater than 1 in the upstream cities is Yichang, accounting for 16.67%. The midstream cities with efficiency values always greater than 1 are Enshi and Xianning. The efficiency values in Ezhou and Wuhan are less than 1 only in 2017. Huangshi shows an obvious increase in efficiency values, with efficiency values less than 1 in 2011–2015 and greater than 1 after 2015, with a comprehensive assessment of midstream cities accounting for about 40% of effective decision-making units. The cities in the downstream region of the Yangtze River basin with efficiency values always greater than 1 include Nanjing, Shanghai, Zhenjiang, Suzhou, Changzhou, Hangzhou, Huzhou, Ningbo, Shaoxing and Zhoushan, and the cities with efficiency values greater than 1 in most years include Tongling and Wuxi, accounting for about 52.17% of the cities in the downstream region with efficiency values greater than 1. This shows that in the percentage of effective decision-making units, the downstream region is greater than the midstream region, followed by the upstream region.

It is not difficult to see that the spatial distribution of CLGUE values is basically in line with the overall economic development level of the upper, middle and lower reaches of the cities in the Yangtze River basin. In other words, the green efficiency value of cultivated land in areas with a higher economic level is correspondingly higher. Related studies have also concluded that CLGUE values are basically proportional to the level of economic development. The researchers interpreted this conclusion as an increase in non-farm payrolls [38]. Areas with a higher level of economic development will also have higher urban rates, while a higher urban rate indicates that there are fewer agricultural workers. That is, there is less labor input in the process of farmland utilization. In addition to the above general reasons, according to the urban distribution characteristics in the Yangtze River basin, this study proves that this phenomenon should also be attributed to the flat terrain in the region, which is more conducive to the intensive use of cultivated land, so as to reduce the input in labor, irrigation, machinery and other aspects. In addition, economic development can often promote agricultural technological innovation and the transformation of scientific and technological achievements.

3.3. Spatial Evolutionary Characteristics of CLGUE

In order to accurately capture the dynamic evolution characteristics of the CLGUE values in the upstream, midstream and downstream regions of the Yangtze River basin, this study used the software STATA 15.0 to plot the kernel density curves of CLGUE in 2011, 2014, 2017 and 2020, as shown in Figure 3.

According to Figure 3, it is not difficult to see that the spatial distribution of CLGUE in the Yangtze River basin is heterogeneous, and the change characteristics of each basin are quite different. Figure 3a shows the kernel density curve about the CLGUE values in all cities in the Yangtze River basin, with double peaks clearly visible in all study years, indicating a clear polarization of CLGUE values in 39 cities in the Yangtze River basin. In terms of time, from 2011 to 2020, there is a year-by-year shortening of the right tail of the nuclear density curve and a narrowing trend in the extension of the distribution, implying that the spatial gap about the values of CLGUE in the Yangtze River basin is gradually narrowing. At the same time, both peaks of the curve are gradually shifting to the right, which indicates that the values of CLGUE in the study area are increasing and the utilization level is improving. The height of the peak on the right side of the efficiency value greater than 1 is increasing year by year, which indicates that the data here are becoming more and more intensive, representing more and more cities with efficient use of cultivated land. Figure 3a shows the kernel density curves of CLGUE values in the upstream Yangtze River basin cities, and the set of curves as a whole shows a flat and wide feature, which shows that the CLGUE values in the upstream of the Yangtze River basin vary widely among cities. However, the fluctuation in the trailing degree of the right tail of the curve indicates that the degree of difference among cities has been changing over time, gradually increasing from 2011 to 2014. It increases from 2014 to 2017, and then decreases in 2020. However, compared with 2011, there is still a significant trend of reduction in the degree of difference of the CLGUE values among cities in the upstream of the Yangtze River basin. Figure 3c shows the kernel density curves about the CLGUE values in the midstream Yangtze River Basin cities. Similar to Figure 3b, this set of curves also has the same flat and wide characteristics, but the degree of difference in efficiency values is relatively flat for cities in the midstream region compared to the upstream region. The right tail of the curve shortens significantly from 2011 to 2014, showing that the degree of difference in efficiency values among cities has decreased. However, the difference among cities is always relatively smooth from 2014 to 2020. In addition, the overlap of the kernel density curves in each year after 2014 is high, indicating that the distribution of efficiency values of cities in the midstream region has always been relatively smooth with little fluctuation since 2014. Figure 3d shows the kernel density curve about the values of CLGUE in the downstream region of the Yangtze River basin, and the main feature of this graph is the existence of double peaks. However, the effect of the double peaks decreases year by year, the right peak with efficiency values greater than 1 increases significantly and the right tail trails less after 2017. This shows that more and more cities in the downstream region show efficient use of cultivated land, and the degree of difference in efficiency values among cities is becoming smaller.

From the above analysis, it can be seen that the degree of differentiation in CLGUE of each city in the Yangtze River basin is also different, which is manifested in the following aspects: downstream < midstream < upstream. It further indicates that the balance of urban development in the middle and lower reaches is higher than that in the upper reaches of the Yangtze River. Due to factors such as geographical location and industrial type, urban development within the Yangtze River basin is unbalanced. The middle and lower reaches of the Yangtze River have convenient transportation and concentrated urban distribution, which is more conducive to the formation of urban agglomerations with coordinated regional development, such as the Yangtze River Delta urban agglomeration including Nanjing, Suzhou, Wuxi and other cities and the middle reaches of the Yangtze River with Wuhan metropolitan area as the main body. Urban agglomerations have a positive effect on labor flow and industrial coordinated development.

4. Analysis of Influencing Factors

4.1. Selection of Impact Factor Indicators

The temporal evolution and spatial distribution pattern of cultivated land use efficiency are influenced by various factors such as natural conditions, cultivated land resource endowment, economic development level and agricultural production conditions. Combining the existing research results and the current situation of Yangtze River basin as well as the availability of indicator data [39], this paper selects the following indicators: (1) cultivated land resource endowment: per capita cultivated land area, replanting index and the proportion of paddy land area; (2) social economy: per capita GDP, urbanization rate [40] and the proportion of secondary and tertiary industries [23]; (3) agricultural production conditions: chemical fertilizer, pesticide, agricultural film, total power of agricultural machinery and effective irrigation area per unit of cultivated land. In this paper, the values of CLGUE in 39 cities in the Yangtze River basin in 2020 and the above indicators are selected for detection.

4.2. Analysis of the Role of Impact Factors

Combined with the characteristics of the geographical detector, the data of each indicator should be converted from continuous data to category data before detecting the magnitude of the influence of each indicator on the values of CLGUE. In this study, K-means cluster analysis [41] was performed on all the influencing factors using SPSS Statistics 26.0 software, and the influencing factors were classified into five categories. After that, the processed data and efficiency values were imported into GeoDetector 2016 software for detection, and the magnitude of q-values in the detection results was used to determine the degree of influence of the impact factors on the green use efficiency values of cultivated land. The larger the q-value, the greater the influence of the influence factor, and vice versa, the smaller the influence. Q-values of each influence factor in the detection results are organized as shown in

Table 4. Detection results of each impact factor.

Impact Dimension	Impact Factor	Yangtze River Basin-Wide	Upstream	Midstream	Downstream
Cultivated land resource endowment	$\mathrm{Per} \mathrm{capita} \mathrm{cultivated} \mathrm{land} \mathrm{area} {\mathrm{X}}_1$	0.2809	0.3651	0.4279	0.2609
	$\mathrm{replanting} \mathrm{index} {\mathrm{X}}_2$	0.1680	0.0244	0.1840	0.5139
	$\mathrm{the} \mathrm{proportion} \mathrm{of} \mathrm{paddy} \mathrm{land} \mathrm{area} {\mathrm{X}}_3$	0.2620	0.2609	0.3252	0.4254
Social economy	$\mathrm{Per} \mathrm{capita} \mathrm{GDP} {\mathrm{X}}_4$	0.1967	0.2747	0.3088	0.2572
	$\mathrm{urbanization} \mathrm{rate} {\mathrm{X}}_5$	0.2064	0.9332	0.3660	0.1547
	$\mathrm{The} \mathrm{proportion} \mathrm{of} \sec \mathrm{ondary} \mathrm{and} \mathrm{tertiary} \mathrm{industries} {\mathrm{X}}_6$	0.1390	0.0058	0.2550	0.0877
Agricultural production conditions	$\mathrm{Total} \mathrm{power} \mathrm{of} \mathrm{agricultural} \mathrm{machinery} \mathrm{per} \mathrm{unit} \mathrm{of} \mathrm{cultivated} \mathrm{land} {\mathrm{X}}_7$	0.0388	0.2675	0.0357	0.1013
	$\mathrm{Fertilizer} \mathrm{use} \mathrm{per} \mathrm{unit} \mathrm{of} \mathrm{cultivated} \mathrm{land} {\mathrm{X}}_8$	0.0629	0.2682	0.2697	0.2375
	$\mathrm{Pesticide} \mathrm{use} \mathrm{per} \mathrm{unit} \mathrm{of} \mathrm{cultivated} \mathrm{land} {\mathrm{X}}_9$	0.0219	0.3063	0.4310	0.0712
	$\mathrm{Effective} \mathrm{irrigated} \mathrm{area} \mathrm{per} \mathrm{unit} \mathrm{of} \mathrm{cultivated} \mathrm{land} {\mathrm{X}}_{10}$	0.1901	0.4835	0.5168	0.1658
	$\mathrm{Amount} \mathrm{of} \mathrm{agricultural} \mathrm{film} \mathrm{used} \mathrm{per} \mathrm{unit} \mathrm{of} \mathrm{cultivated} \mathrm{land} {\mathrm{X}}_{11}$	0.0149	0.1654	0.2234	0.2446

below.

As can be seen from Table 4, among the 11 influencing factors, the various influencing factors that have some influence effect on the efficiency value in descending order are cultivated land area per capita (0.2809) > paddy land area share (0.2620) > urbanization rate (0.2064) > GDP per capita (0.1967) > effective irrigated area (0.1901) > replanting index (0.1680) > secondary and tertiary industries share (0.1390). Among them, three factors, namely, per capita cultivated land area, proportion of paddy land area and urbanization rate, have the most obvious effect on the values of CLGUE in the whole Yangtze River basin. In addition, four factors, namely, GDP per capita, effective irrigated area, replanting index and the proportion of secondary and tertiary industries, also have some influence, and various influencing factors together cause the spatial heterogeneity of the values of CLGUE in the Yangtze River basin. The four influencing factors of fertilizer, pesticide, agricultural film and total agricultural machinery power per unit of cultivated land under the dimension of agricultural production technology had insignificant effects on the spatial heterogeneity of the values of CLGUE in the Yangtze River basin.

From the above detection results, there are some differences in the factors affecting the values of CLGUE in the upstream, midstream and downstream regions of the Yangtze River basin. The dominant influencing factors in the upstream region of the Yangtze River basin are urbanization rate (0.9332) and effective irrigated area per unit (0.4835). In addition, per capita cultivated land area and pesticide use per unit of cultivated land are also important influencing factors.

From the results of interaction detection in

Table 5. Detection results of the interaction of influencing factors.

${\mathbf{X}}_1$	${\mathbf{X}}_2$	${\mathbf{X}}_3$	${\mathbf{X}}_4$	${\mathbf{X}}_5$	${\mathbf{X}}_6$	${\mathbf{X}}_7$	${\mathbf{X}}_8$	${\mathbf{X}}_9$	${\mathbf{X}}_{10}$	${\mathbf{X}}_{11}$
${\mathrm{X}}_2$	0.2809
${\mathrm{X}}_3$	0.7022	0.1680
${\mathrm{X}}_4$	0.4053	0.4310	0.2620
${\mathrm{X}}_5$	0.3398	0.4880	0.4575	0.1967
${\mathrm{X}}_6$	0.5000	0.6071	0.4949	0.5418	0.2064
${\mathrm{X}}_7$	0.4601	0.4050	0.3365	0.4490	0.3831	0.1390
${\mathrm{X}}_8$	0.4108	0.4621	0.3320	0.4409	0.3745	0.3093	0.0388
${\mathrm{X}}_9$	0.3930	0.3896	0.3718	0.3524	0.3568	0.3786	0.2051	0.0629
${\mathrm{X}}_{10}$	0.4113	0.5044	0.4647	0.3458	0.4433	0.3554	0.2014	0.1964	0.0219
${\mathrm{X}}_{11}$	0.5192	0.5896	0.4396	0.4572	0.6504	0.4377	0.4827	0.6900	0.4300	0.1901

, it is clear that there is a significant interaction between the 11 selected impact factors and the values of CLGUE. The explanatory power of the interaction between any two influencing factors is greater than that of their individual influencing factors. This indicates that the multifactor influence of the values of CLGUE in the Yangtze River basin is better than the single-factor influence.

5. Discussion and Conclusions

5.1. Discussion

This study conducted an empirical analysis of the cultivated land use efficiency and its influencing factors in the Yangtze River basin from 2011 to 2020 and concluded that CLGUE in the Yangtze River basin showed an overall rising trend during the study period. Similarly, Yang et al. used the SBM model to measure the values of CLGUE in China from 2000 to 2020 and found that the average value showed a fluctuating upward trend. Kuang et al. constructed a three-stage DEA–Malmquist model from a dynamic perspective to analyze the increasing state of CLGUE in China during 2003–2017. The results of this study are consistent the conclusions of existing papers, indicating that the cultivated land utilization level in the Yangtze River basin has increased year by year in recent years.

In addition, the spatial heterogeneity of efficiency values in each basin is obvious, and the distribution of efficiency values is as follows: downstream region > midstream region > upstream region. Liu Mengba et al. [28] analyzed the regional differences and spatial convergence of the ecological efficiency of cultivated land use in the main grain producing areas of the middle and lower reaches of the Yangtze River and concluded that there were significant differences in the green use efficiency of local cultivated land in the Yangtze River basin. During the study period, from 2007 to 2018, the internal difference in green use efficiency of cultivated land gradually expanded, and there was a phenomenon of polarization. The research results suggest that we should combine the practical problems of unbalanced development in different regions, adopt policies according to local conditions and promote the balanced development of low-carbon use of cultivated land between regions. For the upper reaches of the Yangtze River, we should accelerate the improvement of the rural land transfer market, vigorously promote the construction of agricultural infrastructure and promote the moderate scale of agricultural operation, so as to improve the green use efficiency of cultivated land from the aspect of intensive land use. For the middle and lower reaches of the region, science and technology should be rationally utilized, and the optimal ratio of various input elements should be scientifically evaluated for different types of cultivated land resource endowment.

Finally, this paper explored influencing factors by using the geographical detector and found that cultivated land resource endowment, social and economic development and agricultural production technology were the important factors affecting the difference in cultivated land use efficiency. The results of interactive detection show that there is a significant interaction relationship between the 11 influencing factors and the green use efficiency of cultivated land, which indicates that the multifactor influence of green use efficiency of cultivated land in the Yangtze River basin is superior to the single-factor influence. Therefore, when formulating policies, it is necessary to jointly promote them according to local conditions and actual conditions, and to improve the overall level of CLGUE from different input–output levels.

5.2. Conclusions

In this study, the Super-SBM model, kernel density estimation and the geographic detector method were used to measure CLGUE in each city from 2011 to 2020, describe the heterogeneity of time and space and detect the single and interactive influencing factors. The main conclusions are (1) from 2011 to 2020, the green utilization efficiency value of cultivated land in the Yangtze River basin showed an upward trend on the whole; (2) there is clear spatial heterogeneity between CLGUE values in the Yangtze River basin cities, and the distribution is as follows: downstream region > midstream region > upstream region; (3) cultivated land resource endowment, socioeconomic development and agricultural production technology are important factors affecting the variability in CLGUE values. However, there are some differences in the degree and direction of influence of different influencing factors on different sample subgroups.

However, this paper also has certain limitations. On the one hand, due to the limitations of the study area, this study only explores the current situation of cultivated land utilization in some cities of the Yangtze River basin in China and can be extended to all cities covered by the Yangtze River basin and even the whole of China in the future. On the other hand, it lacks a certain level of innovation in the form of indicator performance. In future studies, appropriate adjustments can be made based on the actual input–output situation of China’s current cultivated land use, so as to further improve the empirical research system of China’s cultivated land use.