Evolution Characteristics and Control Suggestions for Agricultural Non-Point Source Pollution in the Yellow River Basin of China

Abstract

The Yellow River Basin in China is the region with the most severe agricultural non-point source pollution. The control of agricultural non-point source pollution is an important task for ecological protection and high-quality development in the Yellow River Basin at present and in the near future. This paper takes the eight provinces located along the Yellow River, except Sichuan, as the research object. This study estimates the total amount, intensity, and structure of agricultural non-point source pollution from 2014 to 2023 by adopting quantitative methods such as the pollutant discharge coefficient method, the equivalent pollution load method, and so on. The results reveal that the total amount of non-point source pollution of the Yellow River Basin has risen from approx. 4.94 million tons in 2014 to approx. 7.45 million tons in 2023. However, the growth rate has decelerated over the past five years, and the pollution intensity has decreased by 15~40% on average. The characteristics of agricultural non-point source pollution presents as follows: chemical oxygen demand (COD) emissions have become the most significant pollutant, accounting for 90% of the total pollution; livestock and poultry breeding has become the main source of pollution; and the key areas of pollution have shifted from the lower reaches to the middle and upper reaches, but the regional differences have been narrowing, as measured by the Gini coefficient. An analysis of the Kuznets curve indicates that most of the provinces in the Yellow River Basin still depend on an extensive growth model characterized by high input, high emission, and low output. Finally, this paper proposes a classified governance and measurement system for regions and sources, aiming to enhance the agricultural non-point source pollution prevention and control system. It also advocates for accelerating the green transformation of agricultural production in the Yellow River Basin to achieve the rapid decoupling of pollution emission from economic growth.

1. Introduction

Agricultural non-point source pollution represents a significant challenge in water quality management faced by nations worldwide. For example, agricultural practices have been directly responsible for over 50% of water quality degradation incidents across Europe [1]. In Sweden, approximately 97% of nitrogen and 90% of phosphorus pollution in its rivers are attributed to runoff from agricultural lands [2]. The eutrophication observed in the Mekong Dam Reservoir in Thailand is predominantly driven by agricultural non-point source pollution [3].

Since the reform and opening up period, China’s rapid modernization has spurred remarkable growth in its agricultural economy; however, this progress has also resulted in substantial environmental challenges primarily linked to agricultural non-point source pollution. This form of pollution has emerged as one of the principal types of water pollution for China. According to the estimation based on the statistical data in the “2023 China Ecological Environment Statistical Annual Report” published by the Ministry of Ecology and Environment of China, emissions from agricultural sources accounted for significant proportions of total pollutants, including chemical oxygen demand (COD) at 64.2%, ammonia nitrogen (NH₃-N) at 24.5%, total nitrogen (TN) at 52.7%, and total phosphorus (TP) at 70.6%. With the exception of NH₃-N, the shares of each of the other three pollutants in total water pollution exceeded 50%. In recent years, the Chinese government has enacted a series of policies aimed at mitigating agricultural non-point source pollution, including initiatives such as the “Implementation Plan for the Control and Supervision of Agricultural Non-Point Source Pollution (Trial)”. Meanwhile, key documents such as No. 1 Central Document have consistently underscored the imperative to “promote green agricultural development” and “strengthen governance over rural ecological environments”. Effectively managing agricultural non-point source pollution has become crucial for enhancing agricultural productivity, increasing farmers’ incomes, and fostering greener rural communities for China.

The Yellow River Basin plays a crucial role in China’s agricultural development and is often referred to as the “granary of China”. The provinces of Henan and Shandong, located in the lower reaches of the Yellow River Basin, are significant grain-producing regions within the country. In addition, the Hetao Plain and the Fenwei Plain, which belong to the key grain-producing areas of China, are located in the middle and upper reaches of the Yellow River. In recent years, the Yellow River Basin has maintained a stable grain output, accounting for approximately 35% of the national total. Its sown grain area constitutes roughly 36% of the national total. Specifically, wheat and corn production in this region represent 57.1% and 42.0% of the national totals.

Nevertheless, compared with the national average, the Yellow River Basin exhibits a more pronounced agricultural-oriented industrial structure, characterized by relatively extensive agricultural production models. As a result, agricultural non-point source pollution is particularly severe in the Yellow River, which has become the primary bottleneck impeding water quality improvement for the basin. According to the findings of the second census of national pollution sources, the emissions of COD, TN, and TP from agricultural sources in the provinces located along the Yellow River accounted for 63.29%, 42.67%, and 64.25% of the national totals, respectively [4]. In 2021, the Central Committee of the Communist Party and the State Council of China issued the “Outline of the Plan for Ecological Protection and High-Quality Development in the Yellow River Basin”, explicitly designating “enhancing the comprehensive management of agricultural non-point source pollution” as a key priority to promote ecological protection and high-quality development in the Yellow River Basin [5]. Continuously reinforcing and optimizing the control of non-point source pollution in the Yellow River Basin are of critical importance for improving the ecological and environmental protection and governance standards in the region and advancing the green and high-quality development of agriculture.

Currently, research on agricultural non-point source pollution in China, particularly within the Yellow River Basin, has become increasingly abundant. Research related to this topic can be categorized into four primary areas: (1) Investigating the sources of agricultural non-point source pollution, which include chemical fertilizer application [6,7], livestock and poultry breeding [8,9], aquaculture [10,11], agricultural solid waste [12,13], and rural domestic sewage [14]. These are recognized as significant contributors to agricultural non-point source pollution. (2) Quantifying agricultural non-point source pollution at various regional scales, such as provincial or lower administrative scales, river basins, and regions. Common methodologies employed for quantitative assessment include list analysis [15,16], production/discharge coefficient methods [17,18], and unit investigation (case analysis) [19]. (3) Examining the influencing factors of agricultural non-point source pollution, including economic growth [20,21], production practices [22,23], fertilization strategies [24,25], and natural conditions such as soil erosion [24,26]. (4) Analyzing the relationship between agricultural non-point source pollution and economic growth. The majority of studies indicate that in the Yellow River Basin, as well as in typical provinces and regions, agricultural non-point source pollution and economic growth remain in a stage where they have not yet decoupled or exhibit weak decoupling [27,28].

However, the majority of the aforementioned studies focus on specific dimensions to investigate non-point source pollution, lacking a comprehensive and multi-faceted characterization of the evolution of non-point source pollution in the Yellow River Basin. To address this gap, this study aims to compensate for the limitations of existing research by systematically analyzing the characteristics of non-point source pollution in the Yellow River Basin. Specifically, this paper first employs the production/discharge coefficient method to quantify both the overall and provincial-level agricultural non-point source pollution in the Yellow River Basin, providing a detailed analysis of the evolution trends in terms of total amount, intensity, and structure from 2014 to 2023. Secondly, through the application of the Gini coefficient method and the environmental Kuznets curve framework, this study identifies regional disparities in agricultural non-point source pollution within the Yellow River Basin and explores the coupling relationship between pollution and economic growth. Finally, based on a thorough examination of the evolutionary characteristics of agricultural non-point source pollution in the Yellow River Basin, targeted policy recommendations for future non-point source control are proposed, aiming to provide robust support for enhancing environmental and ecological governance and promoting green and sustainable agricultural development for the Yellow River Basin.

2. Data and Methods

2.1. Research Subjects

Since the provinces located along the Yellow River exhibit significant differences in natural ecology, resource endowment, and economic development, these provinces are regarded as the research object in this paper. There are a total of nine provinces (or autonomous regions) located in the Yellow River Basin, including Sichuan, Qinghai, Gansu, Ningxia, Inner Mongolia, Shaanxi, Shanxi, Henan, and Shandong. Considering that only Aba County and Ruoergai County in the Aba Prefecture of Sichuan province belong to the basin, Sichuan province is not taken into account. In addition, according to the theory of basin geography and economy, the remaining eight provinces are categorized into the upper reaches (Qinghai, Gansu, and Ningxia), middle reaches (Inner Mongolia, Shaanxi, Shanxi), and lower reaches (Henan, Shandong).

The agricultural non-point source pollution sources studied in this paper include three categories: pollution from the planting industry, livestock and poultry breeding, and solid waste in farmland. Aquaculture in the Yellow River Basin contributes minimally to national pollution levels and is not considered a major source. Agricultural non-point source pollution typically involves the diffusion of nitrogen, phosphorus, organic pollutants, etc., into water bodies through surface runoff, underground seepage, or atmospheric deposition due to the improper use of agricultural inputs like chemical fertilizers, pesticides, and mulching films. This paper primarily addresses the pollution caused by the application of chemical fertilizers in crop cultivation. It also includes the diffusion of manure and sewage from livestock and poultry breeding into water bodies without effective treatment or utilization. Additionally, the improper storage or disposal of crop straw leads to the loss of nitrogen, phosphorus, organic matter, etc.

The types of agricultural non-point source pollutants in this article align with ecological environment statistical reports in China, which include COD, NH₃-N, TN, and TP.

2.2. Analysis Methods

2.2.1. Calculation of Agricultural Non-Point Source Pollutant Emissions

Agricultural non-point source pollutant emissions are calculated utilizing both the pollutant discharge coefficient method and the equivalent pollution load method.

For the three types of pollution sources, non-point source pollution resulting from crop farming and livestock and poultry breeding is calculated by adopting the relevant methodologies and coefficients outlined in the “Manual of Agricultural Source Pollution Calculation Methods and Coefficients”, which was released during the second national survey of pollution sources in China in 2017. The assessment of non-point source pollution stemming from solid waste on farmland is based on Guo Fengshan’s research paper [22].

The specific calculation formula for non-point source pollution resulting from fertilization in crop farming is presented as Equation (1). According to the “Manual of Accounting Methods and Coefficients for Agricultural Source Production and Discharge”, the characteristic pollutants associated with non-point source pollution for crop farming include NH₃-N, TN, and TP.

$$NP{P}_m=A\times e_m\times {\displaystyle \frac{q_t}{q_{t_0}}}\times {10}^{-3}$$

(1)

where, NPP_m refers to the emissions of class m non-point source pollution resulting from crop farming (measured in tons), where m corresponds to specific types of pollutants; NPP_m denotes the area dedicated to crop cultivation within a particular province during a certain year (in hectares); e_m refers to the coefficient for pollutant loss in a certain province for a given year (in kg/hectare); q_t represents the application rate per unit area of the nitrogen-containing fertilizer or phosphorus-containing fertilizer used for planting in a certain province in 2017 (in kg/hectare); q_t0 refers to the application per unit area of the nitrogen-containing fertilizer or phosphorus-containing fertilizer used for planting in a certain province in other years (in kg/hectare). The amount of nitrogen-containing fertilizers applied refers to the converted pure quantity of nitrogen fertilizers and nitrogen-containing compound fertilizers, which are used for calculating TN and NH₃-N. Similarly, the amount of phosphorus-containing fertilizer applied represents the conversion quantities of phosphorus fertilizers and phosphorus-containing compound fertilizers, which are used for calculating TP.

The specific formula for non-point source pollution resulting from livestock and poultry breeding is presented as Equation (2). The characteristic pollutants associated with non-point sources in this context include COD, NH₃-N, TN, and TP.

$$NP{R}_m={\displaystyle \sum _i\left[Q_i\times \left(p{l}_i\times f{l}_{i,m}+p{s}_i\times f{s}_{i,m}\right)\right]}\times {10}^{-3}$$

(2)

where NPR_m refers to the emissions of class m non-point source pollutants resulting from livestock and poultry breeding (measured in tons). Here, m denotes the four characteristic pollutants associated with livestock and poultry breeding; Q_i refers to the number of livestock and poultry animals raised in a specific province in a certain year (by head or individual counts), where i indicates four types of livestock and poultry animals: pigs, dairy cows, beef cattle, and poultry. pl_i refers to the proportion of livestock and poultry raised on a certain scale within a certain province for a given year (%); ps_i is defined as the percentage raised by farmers in a specific province during a certain year (%); fl_i,m and fs_i,m refer, respectively, to the pollutant discharge coefficient (in kg/head or kg/individual) for class i livestock and poultry raised by large-scale farms and farmers on a small scale.

The specific calculation formula for non-point source pollution resulting from farmland solid waste is presented as Equation (3). The characteristic pollutants associated with non-point source pollution from farmland solid waste are COD, TN, and TP.

$$NP{S}_m={\displaystyle \sum _j\left(P_j\times E{U}_j\times {\rho }_{m,j}\right)}\times \left(1-{\eta }_m\right)\times e_m$$

(3)

where NPS_m refers to the emission volume of non-point source pollutants of class m solid waste originating from farmland (measured in tons). Here, m denotes various types of pollutants associated with solid waste from farmland; P_j refers to the crop yield within a specific province in a certain year (measured in tons), where j represents the seven major crops: rice, wheat, corn, beans, tubers, oilseeds, and vegetables. EU_j indicates the ratio of crop straw to grain; ρ_m,j represents the product coefficient (in kg/ton) for a particular pollutant related to a certain type of crop; 1-η_m refers to the nutrient loss rate from crop straw (%); e_m refers to the pollutant emission coefficient (%).

The equivalent pollution load method was employed to aggregate the emissions of various pollutants. As an indicator that represents the total volume of pollutant emissions, the equivalent pollution load serves as a crucial metric in environmental assessment. For different pollutants, this method is also a tool for standardizing and unifying. As illustrated in Equation (4), the equivalent pollution load for four non-point source pollutants is expressed as the equivalent volume of wastewater that contains these pollutants. The concentration utilized in this equation is based on the Class III standards outlined in China’s “Environmental Quality Standards for Surface Water” [29].

$$TPL={\displaystyle \sum _m\raisebox{1ex}{$\left(NP{R}_m+NP{R}_m+NP{S}_m\right)$}\!\left/ \!\raisebox{-1ex}{${\rho }_m$}\right.}\times {10}^3$$

(4)

where TPL denotes the total pollution load (in m³) of agricultural non-point source pollution in a specific province in a certain year. ρ_m refers to the Class III water quality standard concentrations where COD is 20 mg/L, TN is 1 mg/L, and TP is 0.2 mg/L.

2.2.2. Regional Disparities in Agricultural Non-Point Source Pollution

The Gini coefficient serves as a quantitative measure of income distribution inequality, derived from the Lorentz curve. Mathematically, it represents the ratio of the area between the line representing absolute equality and the Lorentz curve to the area between this line and the absolute non-equality line within the context of the Lorenz curve. In this study, the Gini coefficient is employed to assess disparities in agricultural non-point source pollution among the provinces located along the Yellow River.

The calculation method is outlined as follows: The provinces within the Yellow River Basin are arranged in ascending order based on their agricultural non-point source pollution emissions, with each province assigned a rank denoted by i (i = 1, 2, …). Subsequently, the proportion of agricultural non-point source pollution emissions corresponding to each rank is calculated along with the cumulative percentage, referred to as E_iC. Finally, the formula for calculating the Gini coefficient (G_R) for year R is represented as Equation (5).

$$G_R=1-\left({\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$n$}\right.}\right)\left(2\times {\displaystyle \sum _{i=1}^n{E}_{iC}-1}\right)$$

(5)

The time series results from 2014 to 2023, which were derived from all the aforementioned equations at both the provincial level and the Yellow River Basin level, were analyzed and visualized using Excel.

2.3. Data Sources

The data concerning the total agricultural output value, sown area of crops, yield of major crops, and the application of chemical fertilizers for the eight provinces located along the Yellow River from 2014 to 2023 were sourced from the China Rural Statistical Yearbook published annually in China. Statistics on the number of major livestock and poultry in stock were derived from the statistical yearbooks released by each province. The nitrogen and phosphorus discharge (loss) coefficients related to major crop sowing in each province, as well as those for large-scale livestock and poultry breeding and small-scale farmers, were obtained from the “Manual of Agricultural Pollution Source Production and Discharge Coefficients” compiled during China’s second national census of pollution sources. Notably, there is a significant lack of data regarding the proportion of large-scale livestock and poultry breeding among different species across various provinces; therefore, this information is represented by the national average proportion for large-scale livestock and poultry breeding. which means that the national average proportion data are used for 2015, 2020, and 2023, while Linear interpolation was adopted to fill in data for the rest years (see

Table 1. Proportion of large-scale livestock and poultry breeding on national scale in 2015, 2020, and 2023.

Livestock Species	Pigs	Cows	Poultry
2015	47%	53%	61%
2020	57%	70%	70%
2023	65%	80%	80%

).

Since there are no official statistics for the key parameters utilized in the calculation of non-point sources for solid waste originating from farmland, the straw-to-grain ratio, pollution generation coefficients for various pollutants, the nutrient loss rate of straw, and the pollutant emission coefficient are all derived from case studies illustrated in the relevant literature [22,30]. The crop straw-to-grain ratios and pollution coefficients for different pollutants are presented in

Table 2. Crop straw-to-grain ratios and pollutant production coefficients of different pollutants.

Crop Type	Crop Straw-to-Grain Ratio			Pollution Coefficient (kg/ton)
	Average	max.	min.	COD	TN	TP
Rice	0.99	1.01	0.97	5.63	5.82	0.42
Wheat	1.1	1.17	1.03	6.39	5.15	0.9
Corn	1.28	1.52	1.04	11.23	10.69	2.39
Beans	1.58	1.71	1.44	69.61	22.23	2.24
Tubers	0.59	0.61	0.57	2.26	1.83	0.67
Oilseeds	2.94	3	2.87	20.57	45.43	3.06
Vegetables	1.54	1.63	1.44	5.1	0.92	0.45

. The nutrient loss rates associated with crop straw for COD, TN, and TP were found to be 11.67%, 10.83%, and 10.83%, respectively, while the corresponding pollutant emission coefficients were recorded at 0.5967%, 0.2014%, and 0.1847%.

3. Results and Discussion

3.1. The Characteristics of Agricultural Non-Point Source Pollution Emissions in the Eight Provinces Located Along the Yellow River

The total amount of agricultural non-point source pollution in the eight provinces located along the Yellow River continues to rise. The total amount of agricultural non-point source pollutant emissions and the equivalent pollution load increased from 4.9387 million tons and 829.775 billion m³ in 2014 to 7.4508 million tons and 1087.525 billion m³, respectively, with average annual growth rates of 4.7% and 3.1%. Compared to the results in the relevant literature on agricultural non-point source pollution in the Yellow River Basin, the magnitude of the findings in this study is essentially comparable (5.5824 million tons in 2022 [16] and a total COD of 9.05 million tons in 2021 [27]).

Both the total amount of agricultural non-point source pollution and the equivalent pollution load exhibited a similar trend, with a slight increase observed between 2014 and 2019; however, there was a relatively rapid escalation from 2019 to 2023. By 2023, the total amount of pollutants had risen by 36% compared to that in 2014, while the equivalent pollution load saw an increase of 25%.

Despite this continuous rise in agricultural non-point source pollution in the Yellow River Basin over the past five years, it is noteworthy that the average annual growth rate has declined each year. As illustrated in Figure 1, the year-on-year growth rates for equivalent pollution load during 2019–2023 were recorded at 12.0%, 7.0%, 3.7%, and 3.4%, respectively, while the year-on-year growth rates for total pollutant emissions in recent years were 12.2%, 8.1%, 5.9%, and 5.5%. These data indicate a significant downward trend in both indicators. This findings suggest that since the introduction of the “Outline of the Plan for Ecological Protection and High-Quality Development of the Yellow River Basin”, there has been a gradual curtailment in the growth trajectory of agricultural non-point source pollution within this region, marking positive progress towards its control.

The intensity of agricultural non-point source pollution in the eight provinces located along the Yellow River showed a downward trend. As illustrated in Figure 2, the equivalent pollution load per unit output value in these provinces decreased from 0.330 m³/CNY in 2014 to 0.285 m³/CNY in 2023, representing a cumulative reduction of approximately 13 percent. The variations in pollution intensity also displayed distinct phased characteristics: from 2014 to 2017, there was a slight increase accompanied by fluctuations; from 2017 to 2019, the intensity declined rapidly to below 0.3 m³/CNY; and from 2019 to 2023, it experienced a rebound in 2023 following a slight decline.

The intensity of pollution resulting from the planting industry decreased significantly. Figure 3 exhibits the trend in the equivalent pollution load per unit area of the planting industry from 2014 to 2023. It is evident that the pollution intensity showed a clear linear downward trajectory, declining from 2197.22 m³/ha in 2014 to 1337.00 m³/ha in 2023, representing a cumulative decrease of approximately 40 percent.

As illustrated in Equation (1), the observed decline can be attributed to a gradual reduction in fertilizer application per unit area sown across each province located along the Yellow River. As depicted in Figure 4, the amount of fertilizer applied per unit area sown decreased in nearly all provinces. Notably, Qinghai experienced the most rapid rate of decline, with an average annual decrease of 9% for nitrogenous fertilizers and 15% for phosphorus fertilizers. Shaanxi and Ningxia exhibited the slowest rates of decline in phosphorus-containing fertilizers (see

Table 3. Average annual decrease rate of fertilizer application per unit area sown for eight provinces.

Provinces	Average Annual Decrease Rate
	Phosphorus-Containing Fertilizers	Nitrogen-Containing Fertilizers
Shanxi	8.6%	7.5%
Inner Mongolia	10.6%	4.1%
Shandong	5.0%	5.0%
Henan	6.0%	4.9%
Shaanxi	1.7%	3.4%
Gansu	4.4%	4.6%
Ningxia	2.8%	4.0%
Qinghai	15.6%	9.3%

).

3.2. Characteristics of Agricultural Non-Point Source Pollution by Provincial Region

By province, there were notable differences in the trend in total agricultural non-point source pollution from 2014 to 2023 across the eight provinces. It was observed that agricultural non-point source pollution continued to increase in the vast majority of provinces. As illustrated in Figure 5, the provinces that showed a significant increase are Shanxi, Inner Mongolia, Gansu, Qinghai, and Ningxia. Among these provinces, Ningxia exhibited the highest average annual growth rate for total pollutant emissions at 8.6%, followed closely by Inner Mongolia and Shanxi, at 8.1% and 8.4%, respectively. Correspondingly, their average annual growth rates for equivalent pollution load were recorded at 7.3%, 7.1%, and 6.4%. Conversely, Shaanxi was identified as the only province demonstrating a significant decline in both parameters. It experienced average annual decreases of 1.1% in total pollutant emissions and 1.7% in equivalent pollution load. Additionally, Shandong and Henan displayed considerable fluctuations within their trends: Shandong experienced a decline characterized by variability with average annual reductions of 0.73% for total pollutant emissions and 0.67% for equivalent pollution load; meanwhile, Henan showed slight increases amidst fluctuations with average annual growth rates reaching 3% for total pollutant emissions and approximately 0.6% for equivalent pollution load.

From the perspective of the pollution share by province (see Figure 6), in 2014, Henan, Shandong, Inner Mongolia, and Gansu were identified as the four provinces with the largest proportion of pollution emissions. However, this pattern shifted over time. The shares of pollution from Shandong and Henan experienced significant declines, while Qinghai’s contribution to overall pollution increased markedly, positioning it among the provinces that substantially contributed to total emissions. By 2023, the top four provinces contributing the most significantly to total agricultural non-point source pollutant emissions were Inner Mongolia at 26.5%, Gansu at 17.9%, Henan at 17.4%, and Qinghai at 13.0%. In terms of equivalent pollution load contributions, Inner Mongolia led with 23.1%, followed by Henan at 21.6%, Gansu at 14.6%, and Shandong at 13.2%.

The regional disparities in pollution emissions were analyzed using the Gini coefficient index. Significant differences in the Gini coefficient for various types of pollutants were observed between 2014 and 2023. As illustrated in

Table 4. The Gini coefficients of agricultural non-point source pollution emissions in the eight provinces located along the Yellow River from 2014 to 2023.

Year	Equivalent Pollution Load	Total Pollution Emissions	NH3-N	COD	TN	TP
2014	0.335	0.281	0.395	0.281	0.390	0.344
2015	0.338	0.292	0.403	0.292	0.394	0.342
2016	0.327	0.283	0.401	0.283	0.385	0.334
2017	0.348	0.307	0.427	0.307	0.400	0.351
2018	0.345	0.307	0.432	0.307	0.397	0.349
2019	0.333	0.309	0.425	0.309	0.387	0.325
2020	0.310	0.298	0.411	0.298	0.369	0.324
2021	0.307	0.300	0.415	0.300	0.366	0.311
2022	0.306	0.313	0.410	0.313	0.364	0.299
2023	0.311	0.335	0.396	0.335	0.373	0.282
10-year average	0.326	0.302	0.411	0.302	0.382	0.326

, the Gini coefficients for equivalent pollution load, TN, and TP showed a downward trend, dropping from 0.335, 0.390, and 0.344 in 2014 to 0.311, 0.373, and 0.282 in 2023, respectively. Conversely, the Gini coefficients for total pollutant emissions and COD gradually increased from 0.281 in 2014 to 0.335 in 2023. Given that COD contributes over 90% of total pollutant emissions, regional disparities in overall pollutant emissions are closely linked to COD levels; thus, their values align accordingly. The Gini coefficient for NH₃-N initially rose before declining again, peaking at a value of 0.432 in 2018 while reaching minimums of 0.395 and 0.396 in both 2014 and in 2022. When analyzing the 10-year average Gini coefficients for various indicators, it was found that NH₃-N pollution displayed the most significant regional variation, whereas COD pollution showed minimal regional variation overall. In summary, regional differences concerning agricultural non-point source pollution within the Yellow River Basin have been diminishing—a trend closely associated with national initiatives from recent years aiming at enhancing infrastructure and improving pollution control measures particularly within upstream provinces.

3.3. Characteristics of Pollutant Emissions from Various Types of Pollutants

During the period from 2014 to 2023, emissions from the four major agricultural non-point pollutants exhibited distinct phased characteristics, closely mirroring the variation pattern of total agricultural non-point source pollution shown in Figure 1. As depicted in Figure 7, before 2016, the emissions of all four types of pollutants experienced a slight increase; however, there was a significant reduction in NH₃-N, TN, and TP levels while COD levels remained stable. After 2019, the emissions of all four pollutants began to see an upward trend again. Although the overall growth rate of pollutant emissions has accelerated compared to the pre-2016 upward phase, there has been an obvious decrease in the average annual growth rate over the past five years. The growth rates of NH₃-N, COD, TN, and TP were 9.9%, 12.4%, 9.1%, and 15.7%, respectively, in 2020 compared with 2019. The growth rates of these four pollutants then dropped sharply to −0.6%, 5.7%, 3.2%, and 1.8% in 2023 compared with 2022. This indicates that substantial progress has been achieved regarding NH₃-N and COD treatment.

With respect to the emission intensity per unit of output value—excluding COD—the other three pollutants demonstrated significant reductions: NH3-N decreased by 25%, TN by about 23%, and TP by roughly 16% in the years of analysis (2014–2023). The emission intensity for COD in 2023 stood at 1800 tons per billion CNY, essentially unchanged from the 1799.5 tons per billion CNY recorded back in 2014.

In terms of the share of each pollutant in total emissions (Figure 8), there is a notable difference for the results that are calculated based on total emissions and based on equivalent pollution load. Regarding total emissions, COD constitutes the vast majority, with its proportion increasing year by year. In 2014, COD accounted for 91.7% of total emissions, rising continuously to 93.5% by 2023. Following COD is TN, which represented 6.3% in 2014 but has since declined to 4.9%. NH₃-N and TP each accounted for less than 1 percent. With respect to equivalent pollution load, NH₃-N maintained a relatively low share ranging from 4.6% to 5.6%, while the shares of the other three pollutants hovered around 30%. The highest proportion was observed for TN, which constituted 34.3% in 2023, which was followed by COD at 32.4% and TP at 28.7%. It is evident that when considering both emission volume and the impact of emissions on water quality, equal attention should be given to COD, TN, and TP.

3.4. Characteristics of Pollutant Emissions Resulting from Different Sources

The trends in the three types of agricultural non-point source pollution within the Yellow River Basin exhibit significant differences (Figure 9). From 2014 to 2023, emissions from crop fertilizer sources consistently declined at a rapid pace, decreasing at an average annual rate of 5.0% from 81,500 tons and 102.878 billion m³ in 2014 to 51,700 tons and 64.681 billion m³ in 2023. This decline underscores the substantial effectiveness of China’s initiatives for reducing pesticide and fertilizer usage for pollution control over recent years. In contrast, emissions from livestock and poultry breeding have experienced considerable growth over the past five years, increasing from 5.4318 million tons and 726.31 billion m³ in 2019 to 7.3948 million tons and 1022.139 billion m³—an average annual increase of about 8 percent—primarily driven by Chinese residents’ dietary patterns that favor a high consumption of meat and eggs. Furthermore, emissions originating from farmland straw have increased at an annual rate of approximately 2%, increasing from 3600 tons and 586 million m³ in 2014 and reaching levels of around 4400 tons and 704 million m³ by the end of this period.

In terms of the share of total pollutants attributed to each source, livestock and poultry breeding has accounted for the largest proportion, which has been steadily increasing over the past decade. Regarding equivalent pollution load, its share rose from 89% in 2014 to 94% in 2023. Conversely, the contribution from crop fertilizer sources has significantly decreased, with their equivalent pollution load dropping from 11% in 2014 to 6% in 2023. The impact of crop straw sources is minimal, contributing only 0.07%, which is negligible in quantitative terms. However, considering the relatively rapid growth of total emissions discussed earlier, it is essential to focus on optimizing methods for storing and utilizing crop straw effectively.

3.5. The Relationship Between Agricultural Non-Point Source Pollution and Economic Growth in the Yellow River Basin

The environmental Kuznets hypothesis posits an inverted “U” relationship [31] between environmental pollution and economic growth, which was proposed by economists Gorssman and Krueger based on panel data from 42 countries. On the left side of the inverted “U” curve, environmental pollution is positively correlated with economic growth, indicating that economic development grows at the expense of expanded environmental pollution. Conversely, on the right side of the inverted “U” curve, as the economy continues to grow beyond a certain point, pollution begins to decline after reaching its peak level. This reflects a decoupling phase where economic growth becomes less associated with environmental harm. This study employs this hypothesis to analyze the relationship between agricultural non-point source pollution and agricultural output value in eight provinces along the Yellow River.

From the perspective of the Yellow River Basin as a whole, there exists a strong positive correlation between the equivalent pollution load/total pollution emissions from agricultural non-point sources and agricultural output value. Specifically, as agricultural output value increases, non-point source pollution rises correspondingly (Figure 10). Drawing on the environmental Kuznets curve, it can be inferred that the current growth pattern of agricultural output in the Yellow River Basin remains relatively crude. This growth is primarily dependent on the expansion of the agricultural production scale, which has a significantly adverse impact on pollution levels.

Examining the data by province levels, it is evident that in the vast majority of provinces, agricultural non-point source pollution presents a strong positive correlation with agricultural output value. As illustrated in Figure 11, an increase in total agricultural output value corresponds to a rise in both equivalent pollution load and total pollution emissions from agricultural non-point sources in Shanxi, Inner Mongolia, Gansu, Ningxia, and Qinghai provinces. In contrast, Henan has maintained relatively stable levels of equivalent pollution load and total pollution emissions. Only Shandong and Shaanxi have demonstrated a downward trend. It was revealed that these three provinces possess more advanced capabilities for pollution control. They have achieved increased scales of agricultural production and output value while maintaining stable or even reduced pollution emissions. This suggests that agricultural development has entered a decoupling stage from environmental degradation. The remaining provinces continue to experience growth in agricultural production at the expense of degrading the environmental quality to some extent. There remains significant potential for improvement regarding the management of agricultural non-point source pollution across these areas.

3.6. Main Conclusion in Terms of Evolution Characteristics for Agricultural Non-Point Source Pollution in Yellow River Basin

From 2014 to 2023, the total amount of non-point source pollution in the Yellow River Basin, encompassing eight provinces, has continued along an upward trajectory. However, the growth rate has significantly slowed over the past five years. Notably, there has been a marked decline in intensity indicators such as equivalent pollution load per unit of output value and total pollution emissions per unit area from agricultural activities. In 2023, these two indicators showed reductions of 15% and 40%, respectively, compared to their levels in 2014. The evolving trends in both the scale and intensity of agricultural non-point source pollution underscore the progress and achievements made in managing this issue within the Yellow River Basin over the past five years. Nevertheless, it is important to note that total emissions from agricultural non-point sources have yet to be reversed. Additionally, fluctuations and recurrences remain evident for certain pollutants and sources of contamination. Therefore, controlling agricultural non-point source pollution will continue to be a critical task for effective pollution management and ecological governance within the Yellow River Basin.

The focus of non-point source pollution control in the Yellow River Basin has evolved over the past decade. In terms of pollutant types, COD constitutes the largest share of total pollutants, exceeding 90 percent, with this proportion continuing to rise. Consequently, COD has emerged as the primary agricultural non-point source pollutant that requires the most attention. Regarding pollution sources, livestock and poultry breeding has increased by 6 percentage points from 88% in 2014 to 94% in 2023, while contributions from crop fertilizers have halved from 12% to 6% during the same period. This shift indicates a gradual transition in pollution control targets from crop fertilizer sources towards livestock and poultry manure. Examining the eight provinces along the Yellow River reveals that the contribution of non-point source pollution to total pollution levels in downstream provinces such as Shandong and Henan has decreased by 3 to 7 percentage points. Conversely, midstream provinces like Shanxi and Inner Mongolia have experienced a significant increase in their contributions. Upstream provinces including Ningxia, Gansu, and Qinghai are witnessing a rapid escalation in non-point source pollution levels. As a result, there is an observable spatial shift in pollution patterns moving from the middle and lower reaches toward the upper reaches of the Yellow River Basin.

There exists a strong positive correlation between the increase in agricultural output value and non-point source pollution in the Yellow River Basin. Notably, in Shandong, Shaanxi, and Henan provinces, a relationship has begun to emerge where non-point source pollution emissions have stabilized at certain levels or even declined alongside increases in agricultural output. However, in other provinces, the growth of agricultural output continues to rely heavily on an extensive production model characterized by high input, high emission, and low output, resulting in low production efficiency, a negligible scale effect, and inadequate pollution control capacity. Economic growth is partially based on the cost of environmental damage. In the future, it is imperative for the Yellow River Basin to enhance the added value of agricultural products while improving agricultural mechanization levels. Additionally, promoting recycling practices and the green development of agricultural production resources is crucial. It is essential to expedite the transition towards the decoupling stage as swiftly as possible.

4. Control Suggestions

It is recommended that, based on a thorough understanding of the key characteristics of non-point source pollution control in the Yellow River Basin, efforts to manage agricultural non-point source pollution and optimize prevention measures should be continuously strengthened in critical areas and weak links.

4.1. Measures Should Be Tailored to the Contribution Levels and Pollution Characteristics of Various Sources

In terms of non-point source pollution resulting from crop farming, although its contribution to overall non-point source pollution is evidently declining, the issues related to the excessive and improper use of pesticides and fertilizers in the Yellow River Basin remain significant. In light of the domestic evaluation standard that stipulates that total fertilizer application intensity should be less than 250 kg/ha, it is concerning that the total fertilizer application intensity in Ningxia, Inner Mongolia, Shaanxi, Shanxi, Henan, and Shandong exceeds the standard [32].

Therefore, it is imperative to encourage and guide farmers towards adopting more scientific, reasonable, and precise fertilization methods. Actively promoting techniques such as soil testing combined with formula fertilization, staged fertilization, and slow-release fertilization can help mitigate nitrogen and phosphorus loss pollution resulting from excessive or inappropriate fertilization practices. Additionally, the moderate substitution of chemical fertilizers with organic and bio-fertilizers should be advocated without compromising agricultural production levels. A study conducted in the Yellow River Delta [33] indicated that for every 10% increase in chemical fertilizers used, there is an approximate 17% rise in nitrogen leaching load; conversely, for every 10% increase in organic fertilizers applied, there is a corresponding decrease of about 4% in nitrogen leaching load.

Regarding non-point source pollution arising from animal husbandry—which currently represents the largest contributor to pollution—it is essential to enhance control measures across all facets. This includes improving recycling processes for livestock and poultry breeding as well as upgrading facilities related to collection, storage, transportation, and resource utilization. Large-scale livestock farms should prioritize promoting clean production facilities while strengthening oversight on comprehensive manure management practices along with sewage treatment efforts. Furthermore, small-scale farmers are encouraged to adopt the production model of “distributed disposal and centralized treatment”. It is also important to establish a support incentive mechanism for manure transportation and utilization.

Although the pollution contribution from straw as a solid waste source is nearly negligible, the indiscriminate burning of straw remains prevalent in certain provinces along the Yellow River, resulting in significant pollution to both water and air. In addition to enforcing strict supervision over unauthorized straw burning, it is essential to promote the return of straw to fields based on local conditions. Furthermore, we should encourage the utilization of straw for biomass production and other forms of energy utilization. It is also important to enhance the supporting system for straw collection, storage, and transportation.

4.2. Measures Should Match the Agricultural Non-Point Source Pollution Characteristics of Each Province and Region

In the upper reaches of Qinghai, Gansu, and Ningxia provinces, livestock and poultry breeding constitutes the primary source of non-point pollution, accounting for over 99%. The proportion of small-scale farmers in these provinces is relatively significant, representing approximately 50% of the agricultural sector. These small-scale farmers exhibit a limited willingness to mitigate pollution issues and deploy the practices of low effiency for pollutant reduction in wastewater management. This presents a critical challenge in tackling non-point source pollution.

According to the findings from the second census of pollution sources in Qinghai province [16], the reduction rate of wastewater pollutants among small-scale farmers of livestock and poultry breeding ranges from only 75.22% to 86.69%, which falls below the provincial average. To mitigate this issue, small-scale farmers should first assess their livestock farming scale based on the ecological carrying capacity specific to their region while promoting a “decentralized collection and centralized treatment” model. This approach would establish a solid foundation for enhancing the purification process and optimizing resource utilization. Secondly, they can also actively pursue an integrated production model that combines agricultural farming with livestock farming to achieve balanced planting and breeding practices. The scale should be determined according to the actual carrying capacity of surrounding soil conditions as well as local agricultural practices concerning manure absorption and sewage management.

In the middle and lower reaches of the Yellow River, particularly in provinces such as Shaanxi, Henan, and Shandong, there is a significant prevalence of non-point pollution originating from agricultural activities. In Shaanxi and Henan, this proportion exceeds 20%, while in Shandong, it approaches 7%. These figures are notably higher than the overall contribution of crop farming sources to pollution across the eight provinces studied. To address this issue effectively, these three provinces must enhance their systematic governance of non-point source pollution stemming from planting activities. Measures include rationally optimizing planting structures and increasing the cultivation of crop varieties characterized by low nitrogen and phosphorus output coefficients. Additionally, it is essential to refine tillage, planting, and fertilization systems with a primary focus on controlling nitrogen and phosphorus runoff while strengthening end-of-pipe controls for non-point source pollution.

In other provinces located in the middle and lower reaches such as Inner Mongolia, Shanxi, and Shandong, efforts should be directed towards improving both the quality and efficiency of pollution control measures at key large-scale livestock farms. Measures include optimizing breeding scales and flushing practices, promoting wet cleaning methods and aerobic composting, and implementing buffer zones from pollution to water [34]. Furthermore, it is crucial to study and establish minimum requirements for mandatory farmland allocation while encouraging the recycling of manure and sewage back into agricultural fields.

4.3. The Blocking and End-of-Pipe Purification System for Controlling Agricultural Non-Point Source Pollution Needs Optimization

Both crop farming and livestock and poultry breeding near farmland can cause non-point source pollution in runoff due to rainwater erosion. To address this, it is essential to strengthen source control and implement process blocking and end-of-pipe purification measures. Research indicates that these technologies can reduce nitrogen loads in runoff by 15% to 40% and phosphorus loads by 14% to 42% [35].

Recently, various regions in the Yellow River Basin have explored ecological solutions such as ditches, constructed wetlands, hedges, and vegetation filter belts to create buffer zones for mitigating farmland runoff into receiving water bodies. These buffers have proven effective at intercepting and filtering pollutants. Moving forward, it is important to summarize best practices; optimize multiple strategies; establish a diversified treatment system integrating ditches, canals, and ponds; and enhance the effectiveness of process blocking and terminal purification.

Moreover, within the irrigation districts along the Yellow River Basin—characterized by complex diversion systems—the lack of effective soil isolation promotes intricate pathways for non-point source pollution transport. Hence, there is an urgent need to improve the understanding of pollution migration patterns in key irrigation areas while enhancing real-time monitoring capabilities for nitrogen and phosphorus levels in fields.

4.4. It Is Essential to Develop Green Agriculture Production Processes That Expedite the Decoupling Between Agricultural Production and Non-Point Source Pollution Emissions

Agricultural green production plays a significant role in the decoupling between economic growth and environmental pollution. It is essential to implement agricultural production methods that prioritize water conservation, reduce pollutants, enhance product value, and showcase distinctive features.

Improving production efficiency is a fundamental approach. Relevant measures include the promotion of agricultural mechanization, enhancing the scientific and precise measurement and monitoring of fertilization, and exploring the application of information technology and artificial intelligence in agricultural production.

The transition to a greener and more circular economy is also crucial: green circular ecological agriculture and organic farming systems should be developed. It is necessary to study environmental standards for planting and breeding that align with regional characteristics and current pollution conditions. Stricter environmental emission and quality standards should be established in areas where ecological function positioning and cultivated land productivity are significant.

Furthermore, there is an urgent need to strengthen a series of mechanisms for pollution reduction and agricultural green development. This can include adopting a government-led model for centralized collection and storage; creating a unified platform for trading; exploring emission trading schemes for pollutants as well as ecological products; promoting interaction and mutual assistance in the development of green agriculture among adjacent provinces [36] within the Yellow River Basin; and encouraging industrial enterprises, banks, financial institutions, and other social capital to jointly participate in non-point source pollution control efforts.