Spatial-Temporal Evolution Characteristics of Agricultural Intensive Management and Its Influence on Agricultural Non-Point Source Pollution in China

Abstract

The influencing mechanism of agricultural non-point source pollution under intensive agricultural management is complicated. This paper adopted provincial panel data from 2008 to 2020 to estimate the level of agricultural intensive management, the agricultural chemical oxygen demand (COD), total nitrogen (TN) and total phosphorus (TP) emissions and emission intensity of agricultural non-point source pollution in different regions of China and analyze the spatial-temporal differentiation characteristics. Moreover, the mediation effect model and spatial spillover effect model were adopted to further explore the influence mechanism of agricultural intensive management on agricultural non-point source pollution. The results show that (1) The total emissions and emission intensity of agricultural non-point source pollution both showed an increasing trend, and these areas with high levels of agricultural non-point source pollution are roughly consistent with those areas with high-level of agricultural intensive management. (2) At the overall level, there were mediating effects of natural ecology, agricultural land management, planting and rearing structure and pollution control investment between the relationship of agricultural intensive management and agricultural non-point source pollution, among which agricultural land management was the largest. Additionally, there was significant spatial heterogeneity in the influencing mechanism of agricultural intensive management on non-point source pollution. (3) There were significant spatial agglomeration characteristics in both agricultural intensive management and agricultural non-point source pollution, which showed a fluctuating trend of “rise-decline-rise-decline”. (4) Agricultural intensive management has a significant positive spatial spillover effect on COD, TN and TP emissions of agricultural non-point source pollution. However, environmental regulation could cause agricultural non-point source pollution to be transferred nearby. Scientific understanding of the spatio-temporal differentiation characteristics and influencing mechanism of agricultural non-point source pollution under the agricultural intensive management model is conducive to providing reference for policy regulation.

1. Introduction

With the rapid development of China’s urbanization and the reform of rural land separation from powers, the peasant household management of traditional scattered agricultural production is changing to intensive management. Accordingly, the consumption of rural resources in China is constantly expanding, while the self-purification capacity of agrarian ecological environment is comparatively inadequate. Moreover, China had adopted an extensive agricultural development mode based on fertilizer and pesticide inputs for a long time, thus agricultural non-point source pollution in China is increasingly serious [1,2]. According to the bulletin of the first National Survey of pollution sources, 40 percent of the main water pollutants in China originate from agricultural pollution source. The COD, TN and TP emissions amounts to 1,240,900 tons, 2,704,600 tons and 284,700 tons, respectively, showing an increasing trend year by year (Data from the first national pollution survey bulletin in China). Therefore, scientifically appreciating the impact of agricultural intensive management and executing the emission reduction target of agricultural pollutants are the strategic requirements for achieving agriculture green development of and rural revitalization.

As a large-scale agricultural operation mode of modernization, agricultural intensive management may play both positive and negative roles on agricultural non-point source pollution. On one hand, agricultural intensive management may result in non-point source pollution because of its production scale, increasing resource consumption, deficiency of ecological environment self-purification ability. On the other hand, it may reduce agricultural non-point source pollution with its scale effect and technology spillover effect by sharing resources and technological breakthroughs to lower green agricultural development cost and improve the utilization efficiency of agricultural resources [3]. Thus, it will have a profound impact to verify the spatial and temporal evolution trend and effects of agricultural non-point source pollution under the influence of agricultural intensive management in China more systematically and scientifically.

Research on agricultural non-point source pollution mainly focuses on the following three aspects. Firstly, in terms of the quantitative accounting of agricultural non-point source pollution, there are many measurements of the total emissions of COD, TP and TN, including on-site monitoring [4], data analysis, equal-standard pollution load method [5], and prediction based on time series model [6], etc. Secondly, as for the analysis of temporal and spatial characteristics of agricultural non-point source pollution, current studies are mostly focused on a certain province or region. For example, Chen and Yang [7] analyzed the spatial difference and distribution characteristics of agricultural non-point source pollution in Hunan province and Xizang autonomous region, respectively. Additionally, Zhou et al. and Qiu et al. [8,9] studied the temporal and spatial differences of agricultural non-point source pollution and its inherent composition in China. Thirdly, Islam and Vincent [10] discussed the relationship between economic development and agricultural non-point source pollution and found that economic development would influence agricultural non-point source pollution from the three aspects of scale effect, structural effect and pollution reduction effect. Liang et al. [11], based on the research of Steve et al. [12] on economic growth, technology garden and public goods, pointed out that both scale effect and structure effect would aggravate agricultural non-point source pollution, and technological progress and economic development would play a certain role in reducing pollution. Furthermore, Deng et al. [13] estimated the agricultural agglomeration level by the panel data of 337 prefecture-level cities in China from 2007 to 2015 and explored its threshold effect on agricultural non-point source pollution emissions. Thus, it can be seen that the influence mechanism of agricultural planting structure, economic development and technological progress on agricultural non-point source pollution has been well verified. In addition, natural ecology, e.g., climate, sunshine and rainfall situation [14,15] has also been demonstrated to be related to agricultural non-point source pollution. As such, this article would further the recent research on the influence theoretical framework of agricultural intensive management on agricultural non-point source pollution by considering the natural ecology as an influencing factor together with other influencing factors of planting structure and pollution treatment investment.

Thus, this paper focuses on the following issues: (1) What is the level of agricultural intensive management at provincial level in China? What are the characteristics of its spatial-temporal evolution? Based on the provincial panel data from 2008 to 2020, we would further the existing research to clarify the characteristics of agricultural non-point source pollution including COD, TP and TN with the consideration of three agricultural sectors of planting, livestock and poultry breeding and fishery from the perspective of input and output. (2) What is the influence mechanism of agricultural intensive management on agricultural non-point source pollution in China? We would construct the theoretical framework of the influence mechanism of agricultural intensive management on agricultural non-point source pollution in China, and explore whether there are mediating effects of natural ecology, agricultural land management, planting and rearing structure and pollution control investment? (3) Is there spatial spillover effect of agricultural non-point source pollution under agricultural intensive management? As such, compared with previous research, the contributions of this paper are mainly manifested in three aspects. Firstly, this paper calculated the level of agricultural non-point source pollution in China by adopting the inventory analysis method in which the total emissions and emission intensity of agricultural pollution according to the pollution emission coefficient of each region are determined. Secondly, the influence mechanism and spatial spillover effects of agricultural intensive management on agricultural non-point source pollution were evaluated from the spatial-temporal evolution characteristics. Thirdly, the mediating effects of natural ecology, farmland management, planting and rearing structure and pollution control investment between intensive management and agricultural non-point source pollution in China was analyzed.

2. Theories and Hypotheses

2.1. Direct Effect Analysis and Hypothesis

In the transforming process of agricultural production mode and green agricultural development, the relationship between agricultural intensive management and agricultural non-point source pollution should not be ignored [16,17]. There are three different views on the impact of intensive management on agricultural non-point source pollution. The first view is that intensive management has a significant positive impact on agricultural non-point source pollution. This is because agricultural industrial agglomeration would reduce the cost of agricultural production and operation, as well as the adoption of pesticides and fertilizers, thus achieving the purpose of agricultural non-point source pollution control [16]. The second view holds that there is a non-linear relationship between agricultural intensive management and non-point source pollution. For example, Xu and Xue [18] demonstrated that the relationship between agricultural industrial agglomeration and agricultural non-point source pollution shows an N-shaped trend, in which agricultural non-point source pollution shows rise first, then decline and rise last with the increase in agricultural industrial agglomeration. Deng et al. [13] found that the relationship between agricultural agglomeration and agricultural non-point source pollution showed two stages of decline and then rise. According to the hypothesis of environmental Kuznets curve (EKC), the relationship between economic growth and pollution emission shows the characteristic of an inverted U curve, that is, economic growth would inevitably be accompanied by an increase in environmental pollution before economic development reaches a certain level, and then environmental pollution would be improved with economic growth [19]. The third view holds that the negative externalities play a dominant role in the impact of intensive management on agricultural non-point source pollution. Agricultural intensive management is a large-scale production and operation mode of modern agriculture, mainly relying on mechanization, chemicals and other means, which might bring a series of ecological problems while improving the yield of farmland [20,21]. Based on the above analysis, we argue that agricultural intensive management has a direct impact on agricultural non-point source pollution, and the effects varies in different levels of agricultural intensive management.

2.2. Mediating Effect Analysis and Hypothesis

Under intensive management mode, agricultural non-point source pollution may be affected indirectly by the production and management behavior of agricultural production operators and other objective factors brought by large-scale production. For example, the scale effect produced by agricultural agglomeration could lead to the variation of production mode by changing the utilization efficiency of production means, thus aggravating agricultural non-point source pollution [13]. As such, Steve et al. [12] found that the ecological environment variation driving force could be divided into direct driving force (e.g., climate change, land cover change, etc.) and indirect driving force (e.g., population, economy, science and technology, social politics and culture etc.). Moreover, Qiu et al. [9] emphasized that direct and indirect driving forces have different effects on ecological environment evolution and agricultural non-point source pollution. Specifically, agricultural intensive management may have an indirect impact on agricultural non-point source pollution through the following aspects:

• (1). Natural ecology. Soil erosion and surface runoff are the main conditions for agricultural non-point source pollution. Many climatic factors could affect soil erosion, among which precipitation is the most closely related to agricultural non-point source pollution [22]. Additionally, with the increase in agricultural intensive management in China, the pollutants discharged by agricultural activities and rural life would cause persistent negative impacts on water, soil, air, biodiversity and agricultural ecosystems [23].

• (2). Agricultural land management. On the one hand, the expansion of planting and breeding will increase the input of production materials such as chemical fertilizers, pesticides and feed, as well as the increase in agricultural waste, which would increase the total emission of agricultural non-point source pollutants. On the other hand, large-scale production and scientific management under agricultural intensive management could also fully leverage the potential of agricultural means of production and resource utilization, thus reducing pollution [4]. Therefore, how the scale of agricultural land management under agricultural intensive management mode affects agricultural non-point source pollution should be further clarified.

• (3). Planting and rearing structure. An important factor in improving the ecological environment is the upgrading of industrial structure [24]. The sustainable development of agricultural intensive management will lead to the adjustment of agricultural industrial structure, which mainly involves the change of the proportion of the planting industry, livestock and poultry breeding industry and aquaculture industry as well as the change of the internal structure of planting industry. Meanwhile, the popularization and adoption of modern agricultural technology and large machinery may strengthen the substitution of other agricultural production elements, so as to alleviate agricultural non-point source pollution [25]. While intensive use of agricultural production elements may lead to excessive input of chemicals and thus aggravate agricultural non-point source pollution [22].

• (4). Pollution control investment. Besides scientific and technological investment, pollution control investment also includes capital investment. However, studies showed that pollution control investment has both positive and negative effects. On the one hand, the overuse of chemical technology and irrigation technology and other modern science and technology would have negative impact on water and soil environment. On the other hand, the intensive operation and production mode would enhance the input of advanced facilitates, production technology and the implementation of excellent management means [26], which is a benefit for controlling pollution from the source, reducing the risk of pollutants and improving the treatment rate of existing pollutants.

Based on the above analysis, agricultural intensive management could indirectly affect agricultural non-point source pollution through natural ecology, agricultural land management, planting and rearing structure, and pollution control investment.

2.3. Spatial Spillover Effect Analysis and Hypothesis

As agricultural non-point source pollution is characterized by dispersion, hysteresis, uncertainty and duality, and has a wide range of influence [22]. Moreover, agricultural production conditions and agricultural resource endowment vary greatly among regions, and there may be mutual transfer of agricultural non-point source pollution between different regions [27]. It becomes particularly important to explore whether agricultural intensive management has a spatial spillover effect on agricultural non-point source pollution. As Lei and Su [28] analyzed the spatio-temporal evolution of fertilizer non-point source pollution in China and found that the adjustment and transfer of industrial structure promoted the spatial spillover effect of fertilizer non-point source pollution between regions. Furthermore, scholars found that the spatial dependence of agricultural environment is one of the influencing factors of agricultural non-point source pollution. For example, Guo and Huang [24] found that agricultural non-point source pollution has high spatial heterogeneity at county level in Huaihe River Ecological Economic Belt. Wang et al. [29] studied agricultural non-point source pollution intensity and its spatial convergence characteristics in China from 1999 to 2017 by adopting inventory analysis method. It is found that the convergence rate is faster with consideration of the spatial factors, while lower than the convergence rate of regional economic growth. Thus, we propose that the spillover effect of intensive agricultural management on agricultural non-point source pollution would be different because of difference in the agricultural production conditions, operation modes and agricultural technology levels in different regions.

In conclusion, based on the above analysis, this paper constructs the following theoretical analysis framework for the influence mechanism of agricultural intensive management on agricultural non-point source pollution in China (Figure 1).

3. Materials and Methods

3.1. Study Area and Data Sources

The study area in this manuscript are 31 provinces in mainland China, excluding Hong Kong, Macao, and Taiwan, Diaoyu island, Sansha city and other regions with the consideration of the lack in statistical data, the time horizon is 2009–2020. The data mainly come from “China’s Statistical Yearbook”, “China’s Environmental Statistical Yearbook”, “China’s Rural Statistical Yearbook”, “China’s Animal Husbandry and Veterinary Yearbook”, “National Cost and Income of Agricultural Products”, “Regional statistical yearbook in China” and Agricultural statistical data.

3.2. Model Setting

3.2.1. Measure of Agricultural Intensive Management Level

Currently, there are many evaluation indexes, methods and standards on the level of agricultural intensive management, among which the most common evaluation index is referred by the Food and Agriculture Organization of the United Nations (FAO), which is calculated by the total amount of labor and production input per unit of cultivated land. However, because of the difficulties in data collection, the measurement object of agricultural intensive management level is mostly limited to the agricultural planting industry [30]. In fact, driven by high economic extra value, livestock, poultry and aquaculture have become the pillar industry of China’s rural production development in recent years, in which intensive production has become very popular [31]. Based on this, from the broader connotation of agriculture, this paper considers the input-output of the planting industry, livestock and poultry breeding industry and fishery, and constructs the index of agricultural intensive management level (Table 1). Furthermore, as entropy is an objective description of uncertainty without any artificial factors, which can ensure the authenticity and scientific nature of evaluation results, this paper adopts the entropy value in objective weight method to measure the agricultural intensive management level on the basis of the previous research [32].

Table 1

Indicators of agricultural intensive management level.

Variables	Industry Sector	Variables Content	Explanation of Variables	Variables Source
Level of agricultural intensive management	Planting industry	Land inputs	Average area of cultivated land per economically active agricultural population(hm2/person)	UNFAO Database
		Power inputs	Average tractor use per 1000 hectares of arable land(Ministry/thousand hm2)
		Fertilizer inputs	Average fertilizer application per 1000 hectares of arable land(t/1000 hm2)
	Livestock and poultry industry	Cost of pig inputs	Average material and service consumption costs per large-scale pig farm(million yuan/pc)	[33]
		Cost of Poultry inputs	Average material and service consumption costs per large-scale farm(million yuan/each)
		Pig outputs	Average pig slaughter per large-scale pig farm(10,000 heads/pc)
		Poultry outputs	Average number of poultry slaughtered per large-scale farm(10,000 pcs/pc)
	Fishery industry	Fish outputs	Average fish output per 1,000 ha of aquaculture area (t/1000 hm2)	[34]
		Intermediate Inputs	Average intermediate consumption costs per thousand hectares of aquaculture area (billion yuan/thousand hm2)

The inventory analysis method has been mainly adopted to measure the level of agricultural non-point source pollution in large-scale regions, in which the basic data are from publicly available statistical data with the advantages of simple data acquisition and high authority [35]. Thus, this paper adopts the inventory analysis method to calculate the level of agricultural non-point source pollution in China. Firstly, the main sources of agricultural non-point source pollution, namely pollution producing units, are identified and the amount of agricultural non-point source pollution is determined. Then, according to the pollution emission coefficient of each region, the total emissions and emission intensity of agricultural non-point source pollution are finally calculated. As such, this paper mainly analyzes five types of pollutions consisting of farmland chemical fertilizer, livestock, poultry and aquaculture, farmland solid waste and rural life, including COD, TN and TP based on the specific situation of China’s agricultural production and the research of Duan et al. [36]. The calculation formula is as follows:

(1)$E={\sum }_{i}E{U}_i{\rho }_i\left(E{U}_i{S}_{}\right)={\sum }_{i}P{E}_i{}_{}{\rho }_i\left(1-{\eta }_i\right)C_i\left(E{U}_{i}S\right)$

(2)$EI=\frac{E}{AL}$

In Equations (1) and (2), $E$ is denoted as the total emissions of agricultural non-point source pollution, mainly including COD, TN and TP emissions. $E{U}_i$ is noted as the statistical number of pollutant indicators in unit $i{\rho }_i$ and ${\eta }_i$ represents the pollution intensity coefficient of unit i pollutant and the coefficient of related resource utilization efficiency, respectively. $P{E}_i$ is denoted as the pollution yield of each unit; $C_i$ is the emission coefficient of pollutants in unit i, which is determined by unit and spatial characteristics ($S$) and represents the comprehensive impact of regional environment, rainfall, hydrology and various governance measures on pollution. The reduction amount (i.e., the amount of P₂O₅) is adopted in the calculation of phosphate fertilizer, thus it is multiplied by 43.66%. $EI$ is denoted as the emission intensity of agricultural non-point source pollution; $AL$ is the agricultural land area in the research.

3.2.2. Mediating Effect Model

Based on the analytical framework constructed above, the influence mechanism of agricultural intensive management on agricultural non-point source pollution in China could be mediated by natural ecology, agricultural land management, planting and rearing structure and pollution control investment. Then, the mediating effect models are set as follows:

(3)$E_{it}={\alpha }_0+{\alpha }_1{A}_{it}+{\sum }_1^k{\alpha }_k{X}_{kit}+f_t+u_i+{\epsilon }_{it}$

(4)$M_{it}={\beta }_0+{\beta }_1{A}_{it}+{\sum }_2^k{\beta }_k{X}_{kit}+f_t+u_i+{\delta }_{it}$

(5)$E_{it}={\gamma }_0+{\gamma }_1{A}_{it}+{\gamma }_2{M}_{it}+{\sum }_3^k{\gamma }_k{X}_{kit}+f_t+u_{it}+{\delta }_{it}$

where, $E$ represents the total emissions of agricultural non-point source pollution, $A$ represents the level of agricultural intensive management, $M$ is the mediating variable, and $X$ is the control variable. Eit represents the logarithm of total emissions of agricultural non-point source pollution in the year t of region i; $A_{it}$ represents the level of agricultural intensive management in year t in region i; $M_{it}$ represents the mediating factors in the year t of region i, including four variables: natural ecology, agricultural land management, planting and rearing structure and pollution control investment. $X_{kit}$ represents the control variable k affecting agricultural non-point source pollution emissions and mediating factors in the year t in the i region; $F_t$ is the time fixed effect that cannot be observed in t year; $u_i$ represents the unobservable individual fixed effect in province i. ${\epsilon }_{it}$, ${\delta }_{it}$ and $u_{it}$ represent the perturbation terms of corresponding models, respectively.

3.2.3. Spatial Spillover Effect Model

Moran I is usually adapted to test whether the variables have spatial agglomeration characteristics, and its calculation formula is as follows:

(6)$I=\frac{n{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}{\sum }_{i=1}^n\left(x_i-\overline{x}\right)}$

The value of Moran I is between (−1, 1). When Moran meets the condition of I > 1, it indicates that the variable has spatial autocorrelation; otherwise, there is negative spatial correlation; when Moran I = 0, it indicates that there is no spatial correlation.

This paper constructs the spatial lag model (SLM) and the spatial error model (SEM) for further research on the influence mechanism of agricultural intensive management on agricultural non-point source pollution according to the basic form of spatial econometric model. The spatial lag model is as following:

(7)$ln{E}_{it}={\alpha }_0+{\alpha }_{1}lnAI{M}_{it}+{\alpha }_{2}lnpgd{p}_{it}+{\alpha }_{3}lne{r}_{it}+{\alpha }_{4}lnle{l}_{it}+\rho Wln{E}_{it}+{\epsilon }_{it}$

In Equation (7), $\rho$ is denoted as the spatial autoregressive coefficient, $W$ is the spatial weight matrix, is noted as the spatial lag variable, and ε is the random error term. The spatial error model (SEM) is as following:

(8)$ln{E}_{it}={\beta }_0+{\beta }_{1}lnAI{M}_{it}+{\beta }_{2}lnpgd{p}_{it}+{\beta }_{3}lne{r}_{it}+{\beta }_{4}lnle{l}_{it}+\rho Wln{E}_{it}+{\epsilon }_{it}$

(9)${\epsilon }_{it}=\mathsf{\lambda }W{\epsilon }_{it}+u_i$

In Equation (8), λ represents the moving average coefficient of spatial error, $W{\epsilon }_{it}$ is the error term of spatial lag, and other parameters are the same in the SLM model.

3.3. Variable Measurement

According to the theoretical framework of the influence mechanism of agricultural intensive management on agricultural non-point source pollution constructed above, indicators are selected as shown in Table 2. (1) As the effect of natural ecology on agricultural non-point source pollution is usually manifested in the variation of land resources, water resources and climate, this paper adopts the entropy method to integrate agricultural water consumption, annual average temperature, annual precipitation and annual sunshine duration into one index to measure the natural ecology. (2) Agricultural land management is mainly expressed by the degree of arable land utilization and fertilizer application, thus the land reclamation rate, arable land irrigation rate and fertilizer application rate per unit of arable land are integrated into one index to measure the agricultural land management. (3) The influence of planting and rearing structure on agricultural non-point source pollution is mainly reflected by the changes in agricultural structure and internal structure of planting industry. As such, this paper integrates the proportion of livestock and poultry breeding and the proportion of food crops to cash crops into a comprehensive index to measure the planting and rearing structure. (4) Pollution control investment is denoted as the total power of agricultural machinery and total investment in environmental pollution control. (5) The control variables mainly include economic development level; environmental regulation and labors’ education level, which are represented by per capita GDP; the proportion of agricultural energy conservation and environmental protection expenditure in regional GDP; and the average years of education in rural areas.

Table 2

Variables description.

Variables	Definition	Symbols	Variables Description	Variables Source
Explained variables	COD	COD		[36,37]
	TN	TN		[36,37]
	TP	TP		[36,37]
Core explanatory variable	Degree of Agricultural Intensive Management	AIM		Show in Table 1
Mediating variables	Natural Ecology	NE		[13,32]
	Agricultural Land Management	ALM		[38,39]
	Planting and Rearing structure	PRS		[13,39]
	Pollution Control Investment	PCI		[40,41]
Control variables	Economic Development Level	Pgdp	GDP per capita	[41,42]
	Environmental Regulation	Er	The proportion of agricultural energy conservation and environmental protection expenditure in the region’s GDP	[23,42]
	Labors’ Education Level	Lel	Average years of education in rural areas	[13]

4. Results

4.1. Spatio-Temporal Characteristics of Agricultural Non-Point Source Pollution under the Influence of Agricultural Intensive Management

4.1.1. The Spatial and Temporal Evolution of Agricultural Intensive Management

The average level of agricultural intensive management in China was 0.2291 in 2008, which has been rising steadily since then, reaching to be 0.4764 in 2020, with a cumulative increase of 108%. Due to the limited space of this article, only the spatial and temporal characteristics of provincial agricultural intensive management level in 2008 and 2020 are shown in Figure 2.

The evolution process of agricultural intensive management in China 2008 to 2020 could be further divided into three stages. The first stage is from 2008 to 2012, and the evolution process of agricultural intensive management showed a rising trend. The level of agricultural intensive management changed from 0.2991 in 2008 to 0.3472 in 2012, which varied from 0.2991 in 2008 to 0.3472 in 2012, with an average annual growth rate of 3.79%. The second stage is from 2013 to 2015, and the agricultural intensive management rise steadily with an average annual growth rate of 1.3%. The third stage is from 2016 to 2020, with the rising trend of agricultural intensive management accelerated by an average annual growth rate of 3.89%.

Then, based on the evolution process of agricultural intensive management (taking the data of the year 2020 as an example), regions could be divided into three types. Among them, the high-level of agricultural intensive management area mainly concentrated in the eastern regions which include the province of Sichuan, Henan, Shandong, Guangdong, Hunan, Hebei, Hubei, Anhui, Jiangsu, Shanghai, Yunnan and Jiangxi. Then, the medium level of agricultural intensive management areas are mainly distributed in the regions of Beijing, Liaoning, Zhejiang, Fujian, Heilongjiang, Jilin, Tianjin, Chongqing, Gansu, Inner Mongolia, Xinjiang and Tibet. The low level of agricultural intensive management areas is mainly located in the western regions including provinces of Shanxi, Hainan, Ningxia, Guangxi, Guizhou, and Qinghai. Compared with 2008, Shanxi province had developed into high-level of agricultural intensive management area in 2020, while Liaoning province had dropped to the medium level of agricultural intensive management area.

4.1.2. Temporal Evolution of Agricultural Non-Point Source Pollution

The total emission and emission intensity of COD, TN and TP in China from 2008 to 2020 could be evaluated by the Equations (1) and (2), and the data of agricultural non-point source pollution in China could be summed up seen in Figure 3.

As seen in Figure 3, the COD, TN and TP emissions of agricultural non-point source pollution in 2008 was 144.6 million tons, 548.83 tons and 1.1731 mil-lion tons, respectively, which were consistent with the results of the first national pollution survey bulletin in China in 2007, among which the COD, TN and TP emissions of agricultural non-point source pollution in 2007 was 124.09 million tons, 270.46 tons and 2.84 million tons, respectively. As such, by combining the research results with relevant literature [7,13], the accounting method and production-discharge coefficients on the COD, TN and TP emissions of agricultural non-point source pollution adopted in this paper are reliable. The average COD, TN and TP emissions of agricultural non-point source pollution is 15.044 million tons, 5.7655 million tons and 1.375 million tons, respectively. From 2008 to 2020, the COD emissions showed an inverted “U” shape on the whole, that is the year of 2008 to 2015 and the year of 2016 to 2020, among which the level was roughly the same between 2008 and 2020 and the peak appeared in 2015. The emissions of TN showed an overall trend of growth with some fluctuations. It could be divided into three periods, 2008 to 2013 and 2013 to 2017, which both showed an inverted u-shaped development trend, while the period from 2017 to 2020 showed an increasing trend. The emissions of TP increased slowly from 2008 to 2018, while they increased by 25% since 2018. Additionally, the total emissions of agricultural non-point source pollution in China showed a declining trend in 2013 compared with 2012, which mainly due to the reduction of rural population. In 2013, the rural population in China was 575.55 million, which was down by 56.08 million compared with 2012.

The average COD, TN and TP emission intensity of agricultural non-point source pollution is 23.07 kg/hm², 8.84 kg/hm² and 2.11 kg/hm², respectively, each showing rising trend. The emission intensity of COD increased from 22.13 kg/hm² in 2008 to 22.56 kg/hm² in 2020, and the emission intensity of TN from 2008 to 2020 increased from 8.32 kg/hm² to 9.05 kg/hm². Moreover, the emission intensity of TP from 2008 to 2020 increased from 1.78 kg/hm² to 3.12 kg/hm², with annual growth rates of 3.92%, 5.62% and 10.31%, respectively.

In summary, the total emissions and emission intensity of agricultural non-point source pollution in China showed an overall growth trend in the past decade, while the growth rate has slowed down gently since 2017.

4.1.3. Spatial Heterogeneity of Agricultural Non-Point Source Pollution

To further describe the spatial differentiation of agricultural non-point source pollution, the average COD, TN and TP emissions of agricultural non-point source in each region from 2008 to 2020 are adopted. Based on the cluster analysis, COD, TN and TP emissions and emission intensity in different regions of China are divided into five levels (Figure 4).

As a whole, the spatial distribution of China’s agricultural non-point source pollution is significantly different. Furthermore, these areas with a high level of agricultural non-point source pollution are roughly consistent with those areas with a high level of agricultural intensive management, such as the regions of Shandong, Sichuan, Henan, Hunan, Guangdong, Jiangsu, Anhui, Hubei and other provinces (seen in Figure 4a-1,b-1,c-1). These areas with the high level of emission intensity of agricultural non-point source pollution is also with the high level of agricultural intensive management, such as Beijing, Henan, Shandong, Jiangsu and other regions (Figure 4a-2,b-2,c-2). The areas with low emissions and emission intensity of agricultural non-point source pollution are mainly distributed in Xinjiang, Tibet, Gansu, Ningxia, Shaanxi, Qinghai, Inner Mongolia and other western regions with low-level of agricultural intensive management.

4.2. Effects of Agricultural Intensive Management on Agricultural Non-Point Source Pollution

4.2.1. Test for Stationarity of Variables

To avoid the limitations of test methods and to ensure the robustness of test results, the LLC test, IPS test, ADF-Fisher test and PP-Fisher test are adopted to conduct the unit root test in this paper. The test results are shown in Table 3. The results show that there is no unit root, and the data is stable, which could be further analyzed.

4.2.2. Mediating Effect Test

Testing of influence mechanism on the whole. The regression result of the effect of agricultural intensive management on COD, TN and TP emissions on the whole is shown in Table 4, Table 5 and Table 6, respectively. According to models (1), (6) and (11), the direct effect of agricultural intensive management on agricultural non-point source pollution is significant at the level of 1%. As Wen and Ye [43] proposed the determination process of mediation effect, there is the mediation effect of agricultural intensive management on agricultural non-point source pollution. Among which, there are positive mediating effects of natural ecology, agricultural land management and planting and rearing structure, and negative mediating effects of pollution control investment. The mediating effect of agricultural land management in the relationship of agricultural intensive management with COD, TN and TP emissions is 0.0429, 0.0332 and 0.217, respectively. The mediating effect of pollution control investment, natural ecology, planting and rearing structure in the relationship of agricultural intensive management with COD, TN and TP emissions is 0.0316, 0.0298, 0.0206, 0.0027, 0.0021, 0.0018, and 0.0104, 0.0067, 0.0052, respectively. According to models (2), (7) and (12), agricultural intensive management has a positive but insignificant impact on natural ecology, while the variation of natural ecology will aggravate agricultural non-point source pollution. The reason may be that agricultural intensive management is one of the agricultural development modes adapting to climate change, and an external dynamic factor affecting agricultural non-point source pollution including rainfall and its surface runoff caused by climate change [32]. In addition, temperature, sunshine duration and other factors related to climate change also affect agricultural non-point source pollution emissions by changing pollution transfer conditions. According to models (3), (8) and (13), agricultural intensive management aggravated agricultural non-point source pollution by promoting the development of agricultural land management. With the advancement of urbanization, the amount of cultivated land in China is limited and decreasing year by year. Thus, only by strengthening the input then the output and comprehensive utilization benefit could be improved. Previous studies showed that in recent years, the intensity of chemical fertilizers and pesticides on a national scale was on the rise trend, which would have a negative impact on agricultural non-point source pollution. However, it is worth noting that agricultural land management has the largest mediating effect on agricultural non-point source pollution. Therefore, reasonable control of fertilizer input, maintaining the red line of 1.8 billion mu of cultivated land, and transforming the extensive agricultural production mode are the key measurements to promote the sustainable and green agriculture in China.

On one hand, according to models (4), (9) and (14), planting and rearing structure plays a mediating role in the positive impact of agricultural intensive management on agricultural non-point source pollution. This indicates that with the increase in the proportion of livestock and poultry farming, the emissions of agricultural non-point source pollution will increase accordingly. As the accounting results of agricultural non-point source pollution show that livestock and poultry farming contribute more than 40% to the emission of three kinds of pollutants. It can be seen that the continuous expansion of livestock and poultry farming intensifies agricultural non-point source pollution under the influence of agricultural intensive management. This is mainly because China’s livestock and poultry farming production and operation scale is increasing, leading to the agricultural ecological environment bearing overloaded capacity. On the other hand, the results of models (5), (10) and (15) show that agricultural intensive management could reduce agricultural non-point source pollution through agricultural pollution control investment, which indicates that reasonable technological and financial investment can reduce the negative impact of agricultural production on the environment in China’s agricultural production. The reason may be that agricultural intensive management and large-scale operation has promoted the application and promotion of agricultural green production technology, machinery and management means, which is conducive to cost saving, energy consumption and pollution emission reduction. In addition, a large amount of investment in pollution control has obviously suppressed agricultural non-point source pollution in recent years. However, it should be noted that agricultural non-point source pollution in China is still on the rise, which indicates that the current investment in pollution control is far from enough, which should be promoted by the investment of human, financial, resource, technology and others.

4.2.3. Heterogeneity Test of Influencing Mechanism in Different Agricultural Intensive Management Areas

To explore the spatial heterogeneity of the effects of agricultural intensive management on agricultural non-point source pollution, the mediation analysis was conducted on the high level, medium level and low level of agricultural intensive management area, respectively. The results are shown in Table 7.

Table 7 show that the mediating effect of natural ecology on COD, TN and TP emissions of agricultural non-point source pollution in the medium and low level of agricultural intensive management area is insignificant, while only significant in high level of agricultural intensive management area. The results show that the impact of natural ecology effects on agricultural non-point source pollution in China is mainly driven by highly intensive operation area. According to the previous results, the high-level areas of agricultural intensive management are distributed in Beijing, Henan, Shandong, Jiangsu and other regions which are mainly in eastern and central China. However, during 2010–2015, the cultivated land area in western China increased by 5.2 × 103 km² from 2010 to 2015, while the cultivated land area in eastern and central China decreased by 5.5 × 10³ km² and 5.2 × 10³ km², respectively. As such, to alleviate the pressure brought by the decrease in cultivated land area and improve the grain yield per unit area, the increase in input of various production factors will also promote the increase in agricultural non-point source pollution in the high-level areas of agricultural intensive management. Additionally, there are significant mediating effects of planting and rearing structure, pollution control investment in high and medium level areas of agricultural intensive management, which is positive and negative, respectively. The absolute value of the mediating effect is the largest in the high-level area, followed by the medium level area. There are no significant mediating effects of planting and rearing structure, pollution control investment in low level areas of agricultural intensive management. The possible reason is that the high and medium level areas of agricultural intensive management are mainly with developed economic and scientific level and can effectively apply agricultural scientific and technological achievements to agricultural production, thus alleviating agricultural non-point source pollution. Previous research showed that the performance level of agricultural scientific and technological achievements transformation in eastern China was as high as 74.29 [44]. On the contrary, the low-level areas of agricultural intensive management are mainly developing areas in western China, and their transformation and application of agricultural scientific and technological achievements would be restricted by the lack of agricultural infrastructure, leading to the insignificant reduction effect of pollution control investment on agricultural non-point source pollution [22,45].

4.2.4. Testing of Spatial Spillover Effects

Based on the tools of Geoda10.1, the global Moran I of agricultural intensive management and COD, TN and TP emissions of agricultural non-point source pollution in China from 2008 to 2020 are shown in Table 8.

As seen in Table 8, the global Moran I of agricultural intensive management and COD, TN and TP emissions of agricultural non-point source pollution from 2008 to 2020 are all greater than 0, indicating that both agricultural intensive management and agricultural non-point source pollution have significant spatial correlation characteristics. The global Moran I of agricultural intensive management from 2008 to 2020 all passed the significance level test within the range of 0.1412–0.3603. The global Moran I generally shows a trend of “rise-decline-rise-decline”. During the period of 2008 and 2012, the global Moran I index of agricultural intensive management shows an increasing trend, indicating that the global spatial correlation of agricultural intensive management in this period is constantly strengthened. During 2012 to 2015, it decreased from 0.3308 to 0.2206, indicating that the agricultural intensive management was spatially dispersed. Then, from 2015 to 2019, the global Moran I of agricultural intensive management increased from 0.2206 to 0.3603, showing a trend of continuous strengthening, while slightly decreased to 0.3307in 2020.

The global Moran I of COD, TN and TP of emissions of agricultural non-point source pollution in China are all significant at the level of 1%, showing a consistent fluctuating trend of “rise-decline-rise-decline”. Taking the COD emissions of agricultural non-point source pollution as an example, from 2008 to 2012, the Moran I of COD increased from 0.4021 to 0.4719, indicating that the fluctuation trend of agricultural non-point source pollution at this period has a significant positive spatial correlation. From 2012 to 2015, the Moran I of COD decreased from 0.4719 to 0.4056, indicating that the spatial correlation of agricultural non-point source pollution is weakened. From 2015 to 2019, the global Moran I of COD increased from 0.4056 to 0.5021, reaching the maximum value in the study period, indicating that the global correlation of agricultural non-point source pollution in this period is increasing, and the global Moran I of agricultural non-point source pollution decreased to 0.4293 in 2020.

Thus, it can be concluded that there is a significant spatial correlation of agricultural non-point source pollution in China consistent with Xu et al. [23]. By contrast of the results with no fixed effect model, spatial fixed effect model, time fixed effect model and bidirectional fixed effect model, and the results of R² and log-likelihood, the estimation result of the spatial-temporal double fixed model is the most appropriate for further research. Therefore, the spatial-temporal double fixed model in the spatial error model SEM, namely model 4, is selected as the decision model to analyze the effect of agricultural intensive management on COD, TN and TP emissions of agricultural non-point source pollution. The results of the spatial econometric model are shown in Table 9, Table 10 and Table 11.

From the perspective of the overall results of spatial econometric model and spatial error model SEM or SLM spatial lag model, the regression coefficients are all positive at the significance level of 1%, indicating that agricultural non-point source pollution has significant spatial spillover effects. Therefore, agricultural non-point source pollution could not only be influenced by the local factors, but also by its surrounding area. Model 4 shows that when COD, TN and TP emissions of agricultural non-point source pollution increase by 1% in its surrounding area, the COD, TN and TP emissions of agricultural non-point source pollution in the local area will increase by 0.9568%, 0.8264% and 0.7734%, respectively.

The results of other control variables on COD, TN and TP emissions of agricultural non-point source pollution are analyzed with spatial econometric models in Table 9, Table 10 and Table 11. The regression coefficient of per capita GDP on COD, TN and TP emissions of agricultural non-point source pollution is −0.0176, −0.0133 and −0.0104, respectively, indicating that economy development has significant “pollution reduction” effect on agricultural non-point source pollution. The effect of quadratic term of per capita GDP on agricultural non-point source pollution is not significant. Additionally, the regression coefficient of environmental regulation on COD, TN and TP emissions of agricultural non-point source pollution is 0.0276, 0.0184 and 0.0115, respectively, which is significant and consistent with the previous research. The results indicate that environmental regulations would cause the nearby transfer effect of agricultural non-point source pollution. However, the regression coefficients of laborers’ education level on COD, TN and TP emissions of agricultural non-point source pollution are negative but insignificant. Therefore, the inhibiting effect of laborers’ education level on agricultural non-point source pollution cannot be verified. This means that the improvement of laborers’ education level seldom inhibiting agricultural non-point source pollution.

5. Discussion

The influencing mechanism of agricultural non-point source pollution under agricultural intensive management are more complex compared with industrial agglomeration and industrial environmental pollution [13,40]. However, agricultural intensive management is the critical way to develop modern agriculture [46]. As such, clarifying the influence mechanism of intensive management on agricultural non-point source pollution is an important means to realize agricultural modernization [47]. Therefore, this paper adopts the entropy value and inventory analysis method with provincial panel data to calculate the agricultural intensive management level, COD, TN and TP emissions and emissions intensity of agricultural non-point source pollution in 2008–2020 in China. Accordingly, the spatial-temporal differentiation characteristics are analyzed and the mediation effect model and spatial spillover effect model are adopted to explore the influence mechanism of agricultural intensive management on agricultural non-point source pollution. On the basis of the research results, management strategies for agricultural non-point source pollution control under the influence of agricultural intensive management in China are suggested.

First, this paper adopts the panel model to empirically analyze the effect and mechanism of intensive management on agricultural non-point source pollution. The results show that the average level of agricultural intensive management has been steadily increasing from 2008 to 2020, which is consistent with the finding of Zhao and Zhou [48]. The total emissions and emission intensity of agricultural non-point source pollution in China showed an overall growth trend in the past decade. It is also found that these areas with a higher level of agricultural intensive management are also those with higher emissions and emission intensity of agricultural non-point source pollution, which is consistent with the research of Jiang et al. [49].

Second, our study further discusses the mediation effect of natural ecology, farmland management, planting and rearing structure, and pollution control investment between intensive management and agricultural non-point source pollution in China, among which the mediating effect of agricultural land management is the largest and the mediating effect of pollution control investment is negative. As Jiang et al. [49] pointed out, the increase in farmers’ income and the adjustment of the agricultural industrial structure have consolidated the inhibitory effect of appropriate scale operation on agricultural non-point source pollution, which is also supported by our research. However, our study further conducts a heterogeneity test of influencing mechanism in different agricultural intensive management areas in which the division of agricultural intensive management is different with Yang and Wei [16]. Third, spatial spillover effects are explored to further test the influence of intensive management on agricultural non-point source pollution. On the whole, there is spatial heterogeneity of agricultural non-point source pollution, which is also supported by the previous research [42,50]. Moreover, there is a significant spatial spillover effect of agricultural non-point source pollution, as well as the spatial spillover effect of agricultural intensive management on COD, TN and TP emissions of agricultural non-point source pollution which shows a fluctuating trend of “rise-decline-rise-decline”. The results are partially consistent to the finding of Deng et al. [13] for which there is a spatial correlation between agricultural agglomeration and local areas of agricultural non-point source pollution. Qin et al. [51] found that environmental regulation was an important way to suppress agricultural non-point source pollution. However, in our study, environmental regulation is regarded to result in agricultural non-point source pollution transferring nearby, which is consistent with the research of Wan et al. [52].

We also have related prospects for future research. This paper is limited to an empirical analysis of spatial-temporal evolution characteristics and effects of agricultural non-point source pollution under the agricultural intensive management. Other policies or factors on agricultural non-point source pollution were not considered. In addition, other mediating effects such as environmental decentralization and agricultural facility construction are also worth discussing.

6. Conclusions

The main conclusions are as follows: (1) From the perspective of time series, COD, TN and TP emissions and emission intensity of agricultural non-point sources in China all show an increasing trend from 2008 to 2020. These areas with higher emissions and emission intensity of agricultural non-point source pollution are also with the higher level of agricultural intensive management. (2) The overall mediating effect test results show that natural ecology, agricultural land management, planting and rearing structure and pollution control investment all play mediating roles. The absolute value of the mediation effect of agricultural land management is the largest, followed by the mediation effect of pollution control investment, planting and rearing structure, and natural ecology. It is noteworthy that the pollution control investment plays a negative mediating role. Furthermore, the divisional mediating effect test results show that the mediating effects of natural ecology, agricultural land management, planting and rearing structure, and pollution control investment are all significant in the high-level intensive management area. The effect of natural ecology in the medium-level intensive management area is insignificant, and only agricultural land management plays a mediating role in the low-level intensive management area. (3) According to the results of the spatial econometric model, both agricultural intensive management and agricultural non-point source pollution show significant spatial agglomeration characteristics from 2008 to 2020, accompanied by a fluctuating trend of “rise-decline-rise-decline”. Moreover, agricultural intensive management has a significant positive spatial spillover effect on COD, TN and TP emissions of agricultural non-point source pollution. However, environmental regulation may cause agricultural non-point source pollution to transfer nearby.

Author Contributions

Conceptualization, J.J. and L.X.; methodology, J.J.; software, J.J.; validation, J.J., L.X., M.L. and J.D.; formal analysis, J.J.; J.J. and L.X. conceived and designed the research question. J.J.; M.L. constructed the models and wrote the paper; J.D. reviewed and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available within the article.

Acknowledgments

The authors are grateful to the anonymous referees who provided valuable comments and suggestions to significantly improve the quality of the paper.

Conflicts of Interest

The authors declare no conflict of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. Analysis framework of the influence mechanism of agricultural intensive management on agricultural non-point source pollution.

Figure 2. Spatial distribution of China’s regional agricultural intensive management degree in the year of 2008 (a) and 2020 (b).

Figure 3. Total emissions and emission intensity of agricultural non-point source pollution in China from 2008 to 2020.

Figure 4. Spatial distribution characteristics of agricultural non-point source pollution in China. Note: (a-1,a-2) shows the average COD emissions and emission intensity of China’s regional agricultural non-point source pollution from the year of 2008 to 2020, respectively. (b-1,b-2) shows the average TN emissions and emission intensity of China’s regional agricultural non-point source pollution from the year of 2008 to 2020, respectively. (c-1,c-2) shows the average TP emissions and emission intensity of China’s regional agricultural non-point source pollution from the year of 2008 to 2020, respectively.

Table 3

Robustness test of variables.

Variables	LLC	IPS	ADF-Fisher	PP-Fisher
AIM	−17.824 ***	−5.369 ***	1003.200 ***	1174.040 ***
COD	−40.066 ***	−13.079 ***	1432.740 ***	1308.610 ***
TN	−32.470 ***	−10.740 ***	1332.893 ***	1190.398 ***
TP	−33.290 ***	−10.290 ***	1287.392 ***	1039.492 ***
NE	−113.415 **	−34.547 **	1156.872 ***	1263.270 ***
ALM	−54.653 ***	−30.689 ***	1509.990 ***	1065.400 ***
PRS	−12.457 ***	−15.796 ***	850.706 ***	949.600 ***
PCI	−17.890 ***	−10.427 ***	1264.350 ***	1222.510 ***

Note: *** p < 0.01, ** p < 0.05.

Table 4

The effects of agricultural intensive management on agricultural non-point source pollution (COD emissions as explained variable).

Explanatory Variables	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)
AIM	0.0795 *** (22.40)	0.0062(24.15)	0.0682 **(27.82)	0.0655 *(28.19)	0.0734 *** (24.25)
NE		0.1567 *(8.76)
ALM			0.4754 **(6.80)
PRS				0.3046 ***(0.67)
PCI					−0.6574 *** (−10.36)
Control variables	YES	YES	YES	YES	YES
Constant term	4.8321 *** (6.97)	3.4238 *** (12.67)	2.6554 ***(8.02)	3.2438 ***(6.93)	4.0025 *** (10.04)
Intermediary Effect		0.0027	0.0429	0.0104	−0.0316
Individual/time fixed effects	YES	YES	YES	YES	YES
Goodness of fit (R2)	0.6954	0.5723	0.6008	0.5987	0.6275
Wald chi2	493.81 ***	601.32 ***	522.36 ***	566.45 ***	634.78 ***
Pesaran’s test	0.2157	0.1473	0.1392	0.1012	0.0924

Note: the numbers in brackets are t values, *** p < 0.01, ** p < 0.05, * p < 0.1; model 1 is the direct effect model, model 2 is the natural ecology mediating effect model, model 3 is the agricultural land management mediating effect model, model 4 is the planting and rearing structure effect model, model 5 is the pollution control investment mediating effect model.

Table 5

The effects of agricultural intensive management on agricultural non-point source pollution (TN emissions as explained variable).

Explanatory Variables	Model (6)	Model (7)	Model (8)	Model (9)	Model (10)
AIM	0.0669 *** (25.45)	0.0044(23.05)	0.0736 **(25.98)	0.0495 *(27.09)	0.0756 *** (22.13)
NE		0.1167 *(9.45)
ALM			0.4468 **(9.70)
PRS				0.3021 ***(0.52)
PCI					−0.6652 *** (−9.86)
Control variables	YES	YES	YES	YES	YES
Constant term	5.3216 *** (7.26)	4.9632 *** (13.86)	3.4322 ***(8.26)	3.7832 ***(7.24)	4.2280 *** (10.01)
Intermediary Effect		0.0021	0.0332	0.0067	−0.0298
Individual/time fixed effects	YES	YES	YES	YES	YES
Goodness of fit (R2)	0.5003	0.5910	0.6381	0.5902	0.6164
Wald chi2	502.35 ***	621.94 ***	545.89 ***	605.42 ***	640.34 ***
Pesaran’s test	0.2001	0.1536	0.1295	0.0827	0.0822

Note: the numbers in brackets are t values, *** p < 0.01, ** p < 0.05, * p < 0.1; model 6 is the direct effect model, model 7 is the natural ecology mediating effect model, model 8 is the agricultural land management mediating effect model, model 9 is the planting and rearing structure effect model, model 10 is the pollution control investment mediating effect model.

Table 6

The effects of agricultural intensive management on agricultural non-point source pollution (TP emissions as explained variable).

Explanatory Variables	Model (11)	Model (12)	Model (13)	Model (14)	Model (15)
AIM	0.0508 *** (20.58)	0.0031(22.61)	0.0529 **(25.83)	0.0407 *(26.84)	0.0693 *** (22.53)
NE		0.1006 *(6.06)
ALM			0.2058 **(6.82)
PRS				0.2067 ***(0.58)
PCI					−0.5627 *** (−11.93)
Control variables	YES	YES	YES	YES	YES
Constant term	4.2201 *** (7.01)	3.0264 *** (15.68)	2.6649 ***(7.30)	3.1189 ***(7.04)	4.1037 *** (9.48)
Intermediary Effect		0.0018	0.0217	0.0052	−0.0206
Individual/time fixed effects	YES	YES	YES	YES	YES
Goodness of fit (R2)	0.5012	0.5798	0.6728	0.5860	0.6325
Wald chi2	502.32 ***	637.36 ***	504.46 ***	529.59 ***	627.10 ***
Pesaran’s test	0.1186	0.1005	0.1258	0.1203	0.0876

Note: the numbers in brackets are t values, *** p < 0.01, ** p < 0.05, * p < 0.1; model 11 is the direct effect model, model 12 is the natural ecology mediating effect model, model 13 is the agricultural land management mediating effect model, model 14 is the planting and rearing structure effect model, model 15 is the pollution control investment mediating effect model.

Table 7

Mediating effects in different agricultural intensive management area (COD emissions, TN emissions, TP emissions as explained variable).

Explanatory Variables	High Level of Agricultural Intensive ManagementEstimated Parameters			Medium Level of Agricultural Intensive ManagementEstimated Parameters			Low Level of Agricultural Intensive ManagementEstimated Parameters
	COD	TN	TP	COD	TN	TP	COD	TN	TP
NE	0.1480 *(7.34)	0.1328 *(9.36)	0.1003 *(8.29)	0.0193(0.43)	0.0098(0.39)	0.0057(0.23)	0.0031(0.15)	0.0020(0.11)	0.0009(0.10)
ALM	0.6008 ***(10.96)	0.5732 ***(11.36)	0.398 ***(11.32)	0.6792 ***(7.68)	0.6004 ***(6.93)	0.6832 ***(9.46)	0.4814 ***(9.45)	0.3876 ***(10.32)	0.3329 ***(10.37)
PRS	0.3129 **(5.79)	0.3017 **(6.39)	0.3178 **(6.38)	0.3347 **(7.75)	0.2983 **(7.67)	0.2764 **(5.38)	0.1037(1.04)	0.0994(1.39)	0.0773(0.97)
PCI	0.3328 ***(−14.25)	0.3076 ***(15.15)	0.2975 ***(13.56)	0.3047 ***(−11.20)	0.3102 ***(13.26)	0.2586 ***(15.22)	−0.0327(−1.09)	−0.0487(−1.02)	−0.0219(−0.09)

Note: the numbers in brackets are t values, *** p < 0.01, ** p < 0.05, * p < 0.1. The same below.

Table 8

Moran I test of agricultural intensive management and agricultural non-point source pollution in China from 2008 to 2020.

Year	Agricultural Intensive Management		COD		TN		TP
	Global Moran I Value	p Value	Global Moran I Value	p Value	Global Moran I Value	p Value	Global Moran I Value	p Value
2008	0.141	0.000	0.402	0.000	0.398	0.000	0.367	0.000
2009	0.181	0.059	0.443	0.000	0.403	0.000	0.389	0.000
2010	0.214	0.000	0.459	0.000	0.424	0.000	0.391	0.000
2011	0.304	0.001	0.461	0.000	0.446	0.000	0.401	0.000
2012	0.330	0.012	0.471	0.000	0.453	0.000	0.438	0.000
2013	0.326	0.030	0.465	0.000	0.430	0.000	0.401	0.000
2014	0.281	0.028	0.443	0.000	0.422	0.000	0.395	0.000
2015	0.220	0.000	0.405	0.000	0.413	0.000	0.391	0.000
2016	0.232	0.001	0.410	0.000	0.419	0.000	0.396	0.000
2017	0.254	0.001	0.430	0.000	0.423	0.000	0.405	0.000
2018	0.291	0.000	0.489	0.000	0.446	0.000	0.413	0.000
2019	0.360	0.000	0.502	0.000	0.459	0.000	0.439	0.000
2020	0.330	0.013	0.429	0.000	0.429	0.000	0.413	0.000

Table 9

Results of spatial econometric model (COD emissions as explained variable).

Variables	SEM				SLM
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Cons	2.447 ***(8.845)				2.213 ***(7.123)
lnAIM	−0.048 **(−2.038)	0.013(0.928)	−0.066 ***(−2.554)	0.031 **(2.356)	−0.035(−1.732)	0.044(1.243)	−0.038 *(−1.778)	0.043 **(1.786)
lnagdp	−0.022 *(−1.922)	−0.013(−1.156)	−0.021 **(−2.168)	−0.017 *(−2.081)	−0.065 **(−2.071)	−0.013(−0.607)	−0.054 **(−3.054)	−0.043 ***(−2.954)
ln2agdp	−0.001(−0.16)	−0.000(−0.367)	−0.002(−0.058)	−0.001(−0.028)	−0.003(−0.27)	−0.000(−0.003)	−0.002(−0.056)	−0.001(−0.167)
lner	0.267 ***(4.766)	0.066 **(2.345)	0.253 *** (15.412)	0.027 **(3.297)	0.377 ***(0.755)	0.088 **(2.408)	0.386 ***(7.543)	0.040 * (2.743)
lnlel	−0.333 ***(−18.076)	0.078(1.220)	−0.234 ***(−15.342)	−0.743(−25.452)	−0.305 *** (−21.452)	0.067(1.046)	−0.282 *** (−18.657)	−0.080 ***(−27.543)
$\rho /\lambda$	0.376 ***(7.949)	0.655 ***(28.763)	0.435 ***(8.964)	0.843 ***(26.343)	0.677 ***(18.654)	0.799 ***(35.643)	0.603 ***(14.569)	0.512 ***(17.273)
R2	0.623	0.932	0.644	0.956	0.611	0.923	0.632	0.943
Log-likelihood	69.125	630.57	60.087	678.62	64.25	620.67	40.32	640.71

Note: Model 1, model 2, model 3 and model 4 are the non-fixed effect model, spatial fixed effect model, time fixed effect model and bidirectional fixed effect model in spatial error model SEM, respectively. Models 5, 6, 7 and 8 are the non-fixed effect model, spatial fixed effect model, time fixed effect model and bidirectional fixed effect model of spatial lag model SLM, respectively. T value in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

Table 10

Results of spatial econometric model (TN emissions as explained variable).

Variables	SEM				SLM
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Cons	2.132 ***(8.099)				1.491 ***(6.382)
lnAIM	−0.032 **(−3.293)	0.011(0.974)	−0.054 ***(−3.284)	0.022 **(2.493)	−0.022(−1.642)	0.033(1.045)	−0.029 *(−1.649)	0.033 **(1.528)
lnagdp	−0.021 *(−2.036)	−0.013(−1.046)	−0.019 **(−3.297)	−0.013 *(−2.467)	−0.049 **(−3.290)	−0.012(−0.403)	−0.044 **(−4.392)	−0.043 ***(−1.456)
ln2agdp	−0.001(−0.12)	−0.000(−0.238)	−0.002(−0.052)	−0.000(−0.01)	−0.002(−0.26)	−0.000(−0.002)	−0.002(−0.049)	−0.001(−0.158)
lner	0.231 ***(3.294)	0.047 **(3.294)	0.206 *** (13.26)	0.018 **(4.324)	0.347 ***(0.747)	0.073 **(3.299)	0.416 ***(6.483)	0.030 * (3.497)
lnlel	−0.312 ***(−14.289)	0.055(1.045)	−0.203 ***(−12.483)	−0.557(−23.462)	−0.294 *** (−18.375)	0.057(1.058)	−0.229 *** (−16.364)	−0.066 ***(−22.467)
$\rho /\lambda$	0.293 ***(6.489)	0.558 ***(30.283)	0.332 ***(9.374)	0.762 ***(28.393)	0.583 ***(17.389)	0.732 ***(34.273)	0.583 ***(13.473)	0.499 ***(15.583)
R2	0.598	0.706	0.583	0.826	0.573	0.612	0.553	0.695
Log-likelihood	56.349	583.543	51.239	660.342	54.321	594.203	435.239	632.124

*** p < 0.01, ** p < 0.05, * p < 0.1.

Table 11

Results of spatial econometric model (TP emissions as explained variable).

Variables	SEM				SLM
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Cons	1.985 ***(6.321)				1.231 ***(5.912)
lnAIM	−0.036 **(−1.465)	0.013(0.884)	−0.053 ***(−3.294)	0.017 **(2.495)	−0.032(−1.354)	0.0333(1.219)	−0.022 *(−1.463)	0.033 **(1.443)
lnagdp	−0.010 *(−1.432)	−0.012(−1.354)	−0.013 **(−1.465)	−0.010 *(−1.994)	−0.055 **(−2.033)	−0.014(−0.668)	−0.057 **(−3.356)	−0.0338 ***(−3.789)
ln2agdp	−0.000(−0.143)	−0.000(−0.321)	−0.001(−0.053)	−0.0006(−0.012)	−0.0029(−0.116)	−0.000(−0.003)	−0.001(−0.054)	−0.001(−0.153)
lner	0.232 ***(5.458)	0.058 **(2.342)	0.217 *** (13.432)	0.011 **(4.291)	0.321 ***(0.642)	0.066 **(5.395)	0.312 ***(5.372)	0.021 * (1.356)
lnlel	−0.286 ***(−16.438)	0.063(1.043)	−0.209 ***(−11.764)	−0.503(−43.291)	−0.289 *** (−22.245)	0.064(2.543)	−0.221 *** (−15.346)	−0.077 ***(−25.34)
$\rho /\lambda$	0.321 ***(6.374)	0.532 ***(25.384)	0.442 ***(9.372)	0.600 ***(24.485)	0.532 ***(16.321)	0.685 ***(33.284)	0.593 ***(15.483)	0.465 ***(16.47)
R2	0.581	0.649	0.541	0.773	0.473	0.531	0.442	0.597
Log-likelihood	52.463	539.203	48.293	580.463	55.483	533.200	33.332	593.218

*** p < 0.01, ** p < 0.05, * p < 0.1.

References

1. Zhang, Q.; Razzaq, A.; Qin, J.; Feng, Z.; Ye, F.; Xiao, M. Does the Expansion of Farmers’ Operation Scale Improve the Efficiency of Agricultural Production in China? Implications for Environmental Sustainability. Front. Environ. Sci.; 2022; 10, 918060. [DOI: https://dx.doi.org/10.3389/fenvs.2022.918060]

2. Jiang, G. How Does Agro-Tourism Integration Influence the Rebound Effect of China’s Agricultural Eco-Efficiency? An Economic Development Perspective. Front. Environ. Sci.; 2022; 10, 921103. [DOI: https://dx.doi.org/10.3389/fenvs.2022.921103]

3. Lu, Y.; Tan, Y.; Wang, H. Impact of Environmental Regulation on Green Technology Adoption by Farmers Microscopic Investigation Evidence From Pig Breeding in China. Front. Environ. Sci.; 2022; 10, 885933. [DOI: https://dx.doi.org/10.3389/fenvs.2022.885933]

4. Liu, C.; Liu, Y.; Feng, C.; Wang, P.; Yu, L.; Liu, D.; Sun, S. Distribution Characteristics and Potential Risks of Heavy Metals and Antimicrobial Resistant Escherichia Coli in Dairy Farm Wastewater in Tai’an, China. Chemosphere; 2020; 262, 127768. [DOI: https://dx.doi.org/10.1016/j.chemosphere.2020.127768] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32777611]

5. Xu, J.; Xie, J.; Cheng, Z.; Zhu, S.; Wang, B. Source Apportionment of Pulping Wastewater and Application of Mechanical Vapor Recompression: Environmental and Economic Analyses. J. Environ. Manag.; 2021; 292, 112740. [DOI: https://dx.doi.org/10.1016/j.jenvman.2021.112740]

6. Ni, X.; Parajuli, P. Evaluation of the Impacts of BMP and Tail Water Recovery System on Surface and Groundwater Using Satellite Imagery and Swat Reservoir Function. Agric. Water Manag.; 2018; 210, pp. 78-87. [DOI: https://dx.doi.org/10.1016/j.agwat.2018.07.027]

7. Chen, W.; Yang, S. Measurement and Prevention of Agricultural Non-point Source Pollution-a Case Study of Human Province. System Engin.; 2015; 6, pp. 95-100.

8. Zhou, F.; Jin, S.; Zhang, H. Spatial Differences and Distribution Characteristics of TN and TP Emissions from Agricultural Non-point Sources in Tibet. Chin. J. Agric. Res. Reg. Plan.; 2019; 40, pp. 35–41+67.

9. Qiu, W.; Zhong, C.; Li, Z.; Pan, Y. Temporal and Spatial Patterns of Agricultural Non-point Source Pollution Emissions in China. Chin. J. Agric. Res. Reg. Plan.; 2019; 40, pp. 26-34.

10. Islam, N.; Vincent, J. Unveiling the Income-environment Relationship: An Exploration into the Determinants of Environmental Quality. Harvard Institute for International Development Discussion Paper; Harvard University: Cambridge, MA, USA, 1999; pp. 1344-1356.

11. Liang, L.; Qu, F.; Feng, S. Economic Development and Agricultural Non-point Source Pollution: Decomposition Model and Empirical Study. Res. Environ. Yangtze Basin; 2013; 22, pp. 1369-1374.

12. Steve, R.C.; Prabhu, L.P.; Elena, M.B. Ecosystem and Human Well-Being: Scenarios; Island Press: Washington, DC, USA, London, UK, 2005; Volume 2.

13. Deng, Q.; Li, E.; Ren, S. Effects of Agricultural Agglomeration on Agricultural Non-point Source Pollution: Threshold Effect Analysis Based on Panel Data of Prefecture-level Cities in China. Geogr. Res.; 2020; 39, pp. 212-231.

14. Zhao, X. Quantitative Assessment of Impacts of Climate Change on Agricultural Non-point Source Pollution in Yingkou. Applied Tech. Soil Water Conserv.; 2019; 4, pp. 16-18. [DOI: https://dx.doi.org/10.3969/j.issn.1673-5366.2019.04.06]

15. Cheng, Y.; Mou, X.; Liu, G. Change Trend of Precipitation on Agricultural Non-point Source Pollution in Wanzhou Region. Cq. World Acad. Edit.; 2013; 30, pp. 10-12. [DOI: https://dx.doi.org/10.3969/j.issn.1007-7111.2013.09.003]

16. Yang, X.; Wei, X. Agricultural Industrial Agglomeration, Agricultural Non-point Source Pollution and Agricultural Green Development Based on Spatial Heterogeneity Perspective. Jiangsu Agric. Sci.; 2022; 50, pp. 244-252.

17. Moges, M.; Schmitter, P.; Tilahun, S.; Steenhuis, T. Watershed Modeling for Reducing Future Non-point Source Sediment and Phosphorus Load in the Lake Tana Basin, Ethiopia. J. Soil Sediment.; 2018; 18, pp. 309-322. [DOI: https://dx.doi.org/10.1007/s11368-017-1824-z]

18. Xu, C.; Xue, L. Agricultural Industrial Agglomeration and Agricultural Non-point Source Pollution: Based on Spatial Heterogeneity Perspective. Sci. Financ. Econ.; 2019; 8, pp. 82-96.

19. Deng, H.; Li, C.; Wang, L. The Impact of Corporate Innovation on Environmental Performance: The Moderating Effect of Financing Constraints and Government Subsidies. Sustainability; 2022; 14, 11530. [DOI: https://dx.doi.org/10.3390/su141811530]

20. Rohila, A.; Kumar, A.; Ghanghas, B.; Mahalle, S.; Nimbrayan, P. Eco-friendly Agricultural Practices and Their Assent by Farmers in Haryanastate of India. Appl. Ecol. Environ. Res.; 2019; 17, pp. 5383-5400. [DOI: https://dx.doi.org/10.15666/aeer/1703_53835400]

21. Tang, Q.; Feng, S.; Liang, G.; Ding, S. Landscape Preferences of Different Stakeholders in Highly Intensive Agricultural Landscapes. Chin. J. Appl. Ecol.; 2018; 29, pp. 2633-2640. [DOI: https://dx.doi.org/10.13287/j.1001-9332.201808.015]

22. Shou, C.; Du, H.; Liu, X. Research Pogress of Source and Mechanism of Agricultural Non-Point Source Pollution in China. Appl. Ecol. Environ. Res.; 2019; 17, pp. 10611-10621. [DOI: https://dx.doi.org/10.15666/aeer/1705_1061110621]

23. Xu, B.; Niu, Y.; Zhang, Y.; Chen, Z.; Zhang, L. China’s Agricultural Non-point Source Pollution and Green Growth: Interaction and Spatial Spillover. Environ. Sci. Pollut. Res.; 2022; 29, pp. 60278-60288. [DOI: https://dx.doi.org/10.1007/s11356-022-20128-x]

24. Guo, L.; Huang, Z. Spatial Distribution and Control of Agricultural Non-point Source Pollution in Huaihe River Ecological Economic Belt. Res. Environ. Yz Basin; 2021; 30, pp. 1746-1756.

25. Wu, Y.; Miao, L. Research on the Environmental Effects of Agricultural Support Policies in China: Theory and Demonstration. Modern Econ. Discus.; 2017; 9, pp. 101-107.

26. Degife, A.; Zabel, F.; Mauser, W. Land Use Scenarios and Their Effect on Potential Crop Production: The Case of Gambella Region, Ethiopia. Agriculture; 2019; 9, 105. [DOI: https://dx.doi.org/10.3390/agriculture9050105]

27. Zheng, Y.; Lei, G.; Yu, P. Eco-compensation Schemes for Controlling Agricultural Non-point Source Pollution in Maoli Lake Watershed. Water; 2021; 13, 1536. [DOI: https://dx.doi.org/10.3390/w13111536]

28. Lei, J.; Su, S.; Yu, W. Spatial-temporal Pattern Change and Group Prediction of Fertilizer Non-point Source Pollution in China. Chin. J. Eco. Agric.; 2020; 28, pp. 1079-1092.

29. Wang, W.; Li, Z.; Shi, P.; Zhang, Y.; Pan, B.; Li, P.; Ding, S.; Li, J. Vegetation Restoration and Agricultural Management to Mitigate Nitrogen Pollution in the Surface Waters of the Dan River, China. Environ. Sci. Pollut. Res.; 2021; 28, pp. 47136-47148. [DOI: https://dx.doi.org/10.1007/s11356-021-13905-7]

30. Yu, L.; Wang, Z.; Zhang, H.; Wei, C. Spatial-Temporal Differentiation Analysis of Agricultural Land Use Intensity and Its Driving Factors at the County Scale: A Case Study in Hubei Province, China. Int. J. Environ. Res.; 2020; 17, 6910. [DOI: https://dx.doi.org/10.3390/ijerph17186910]

31. Liu, J.; Zhao, M.; Yang, Y. Impacts of Government Subsidies and Environmental Regulations on Green Process Innovation: A nonlinear Approach. Technol. Soc.; 2020; 63, 101417. [DOI: https://dx.doi.org/10.1016/j.techsoc.2020.101417]

32. Li, L.; Gou, M.; Wang, N.; Ma, W.; La, L. Landscape Configuration Mediates Hydrology and Nonpoint Source Pollution under Climate Change and Agricultural Expansion. Ecol. Indic.; 2021; 129, 107959. [DOI: https://dx.doi.org/10.1016/j.ecolind.2021.107959]

33. Guo, L.; Lin, Y. Research on the Driving Force, Path and Influence of Pig Breeding Scale in China. Rural Econ.; 2020; 8, pp. 126-135.

34. Jiang, Q.; Zhao, W. Input-output Analysis of Different Varieties of Aquaculture in China. China Fish. Econ.; 2018; 36, pp. 90-96.

35. Chen, M.; Chen, J.; Rice, Y. Inventory Analysis and Spatial Feature Identification of Agricultural and Rural Pollution in China. Chin. Environ. Sci.; 2006; 26, pp. 751-755. [DOI: https://dx.doi.org/10.3321/j.issn:1000-6923.2006.06.025]

36. Duan, H.; Liu, D.; Yang, G.; Dong, Y. Estimation of Agricultural Non-point Source Pollution Intensity Based on Inventory Analysis. Environ. Sci. Manag.; 2009; 34, pp. 58–61+74. [DOI: https://dx.doi.org/10.3969/j.issn.1673-1212.2009.12.016]

37. Lai, S.; Du, P.; Chen, J. Investigation and Assessment of Non-point Source Pollution Based on Unit Analysis. J. Tsinghua Univer.; 2004; 44, pp. 1184-1187. [DOI: https://dx.doi.org/10.3321/j.issn:1000-0054.2004.09.009]

38. Wang, H.; He, P.; Shen, C.; Wu, Z. Effect of Irrigation Amount and Fertilization on Agriculture Non-point Source Pollution in the Paddy Field. Environ. Sci. Pollut. Res.; 2019; 26, pp. 10363-10373. [DOI: https://dx.doi.org/10.1007/s11356-019-04375-z]

39. Chen, T.; Miao, J.; Zhu, Z. Measuring Green Total Factor Productivity of China’s Agricultural Sector: A Three-stage SBM-DEA Model with Non-point Source Pollution and CO₂ Emissions. J. Clean Prod.; 2021; 318, 128543. [DOI: https://dx.doi.org/10.1016/j.jclepro.2021.128543]

40. Acosta-Nassar, M.; Morell, J.; Corredor, J. The Nitrogen Budget of a Tropical Semi-intensive Freshwater Fish Culture Pond. J. World Aquacult. Soc.; 2010; 25, pp. 261-270. [DOI: https://dx.doi.org/10.1111/j.1749-7345.1994.tb00189.x]

41. Hu, H.; Dai, J.; Zhang, L.; Wang, X.; Ma, Y. Direct Effect and Spatial Spillover Effect of Rural Human Capital on Agricultural Non-point Source Pollution. Glob. Nest J.; 2021; 22, pp. 642-648. [DOI: https://dx.doi.org/10.30955/gnj.003402]

42. Zhang, W.; Li, G.; Uddin, M.; Guo, S. Environmental regulation, Foreign Investment Behavior, and Carbon Emissions for 30 Provinces in China. J. Clean. Prod.; 2020; 248, 119208. [DOI: https://dx.doi.org/10.1016/j.jclepro.2019.119208]

43. Wen, Z.; Ye, B. Mediating Effects Analysis: Method and Model Development. Adv. Psych. Sci.; 2014; 22, pp. 731-745. [DOI: https://dx.doi.org/10.3724/SP.J.1042.2014.00731]

44. Liu, X.; Shen, Q.; He, Z. Empirical Study on Fund Performance of Agricultural Scientific and Technological Achievements Transformation in China. J. Geotech. Econ.; 2015; 4, pp. 74-81.

45. Yi, F.; He, M. Impact of Changes in Labor Resources and Transfers of Land use Rights on Agricultural Non-point Source Pollution in Jiangsu Province, China. J. Chromatog.; 2017; 207, pp. 134-140.

46. Yu, L. Constructing New Agricultural Management System and Steadily Promoting the Innovation of the Management System. Anhui Agric. Sci. Bull.; 2014; 20, pp. 16–18+21. [DOI: https://dx.doi.org/10.16377/j.cnki.issn1007-7731.2014.20.037]

47. Zhang, Y.; Long, H.; Li, Y.; Tu, S.; Jiang, T. Non-point Source Pollution in Response to Rural Transformation Development: A Comprehensive Analysis of China’s Traditional Farming Area. J. Rural Stud.; 2021; 83, pp. 165-176. [DOI: https://dx.doi.org/10.1016/j.jrurstud.2020.10.010]

48. Zhao, D.; Zhou, H. Agricultural Production Agglomeration: How to improve grain production efficiency: A Review based on different development paths. Agric. Technol. Econ.; 2020; 8, pp. 13-28. [DOI: https://dx.doi.org/10.13246/j.cnki.jae.2020.08.002]

49. Jiang, S.; Zhou, J.; Qiu, S. Whether Moderate Scale Operation Can Inhibit Agricultural Non-point Source Pollution-Empirical Evidence Based on Dynamic Threshold Panel Model. Agric. Technol. Econ.; 2020; 7, pp. 33-48. [DOI: https://dx.doi.org/10.13246/j.cnki.jae.2021.07.003]

50. Zou, L.; Liu, Y.; Wang, Y.; Hu, H. Assessment and Analysis of Agricultural Non-point Source Pollution Loads in China: 1978–2017. J. Environ. Manag.; 2020; 263, 110400. [DOI: https://dx.doi.org/10.1016/j.jenvman.2020.110400]

51. Qin, T.; Peng, J.; Deng, Z.; Wang, J. Impact of Environmental Decentralization and Regulation on Agricultural Non-point Source Pollution. China Popul. Res. Environ.; 2021; 14, pp. 61-70.

52. Wan, W.; Han, Y.; Wu, H.; Liu, F.; Liu, Z. Application of the Source–sink Landscape Method in the Evaluation of Agricultural Non-point Source Pollution: First Estimation of an Orchard-dominated Area in China. Agric. Water Manag.; 2021; 252, 106910. [DOI: https://dx.doi.org/10.1016/j.agwat.2021.106910]