Synergistic Reduction and Common Driving Forces of Agricultural Pollution and Carbon Emissions Based on Agricultural Grey Water Footprint

Abstract

Managing agricultural water pollution (AWP) and agricultural carbon emissions (ACE) together is crucial for addressing the global water resources crisis and climate challenges. Traditional water quality indicators are limited in large-scale evaluations of AWP. The common trends of ACE and AWP, as well as the spatial heterogeneity of their common driving factors also remain unclear. This study introduces a novel framework for analyzing the synergistic reduction of AWP and ACE from the perspective of agricultural grey water footprint (AGWF) and examines disparities in common driving factors across areas with differing levels of economic development and pollution intensities in Zhejiang Province. The results indicate that ACE and AGWF in Zhejiang showed an upward trend from 2010 to 2012, followed by a significant decline from 2013 to 2020. A consistent synergistic reduction trend in grey water footprint and carbon emissions was identified in both the planting and livestock husbandry sectors across Zhejiang. Socio-economic factors jointly influenced the reductions in ACE and AGWF, with technological level and the labor-to-research-and-development (labor-R&D) ratio being the primary drivers, accounting for 79.41% and 78.38% of these reductions, respectively. The impact of agricultural R&D expenditure intensity on AGWF and ACE exhibited spatiotemporal heterogeneity and sectoral disparities. The key to promoting the synergistic reduction of AGWF and ACE lies in advancing the research, development, and application of green agricultural technologies especially in regions where such technologies are not yet fully developed. The results provide a theoretical framework and practical implementation for the integrated management of AWP and ACE, as well as sustainable agricultural policy formulation.

1. Introduction

Agriculture is the largest global water consumer, significantly contributing to greenhouse gas emissions (GHG) and pollution [1,2]. It consumes 85% of human blue water usage, accounts for 25% of global GHG emissions, and utilizes between 80 and 90% of global nitrogen and phosphorus resources [3]. Agricultural activities have fueled the global warming trend through GHG emissions [4], while the entry of excessive nitrogen and phosphorus into water bodies has caused water pollution, exacerbating the scarcity of water resources [5]. Due to the highly homogeneous pollution and carbon emission (CE) sources in agricultural production processes [6], examining the concurrent reduction in agricultural carbon emissions (ACE) and pollution holds paramount importance for improving ecological quality and mitigating climate change.

The synergistic effect between pollution and carbon reduction refers to the achievement of CE reduction through pollution control efforts or the reduction of pollutants through implementing carbon reduction policies and measures. Most studies have investigated agricultural water pollution (AWP) and ACE separately [7,8]. Research on the synergistic effects of AWP and ACE reduction is still in its early stages. For instance, Huang and Yang [9] found that the zero-growth policy for chemical fertilizers and pesticides has significant effects on ACE and non-point source pollution (NSP) reduction. Liu and Liu [6] confirmed the existence of synergistic effects in AWP and ACE reduction from the perspective of agricultural NSP, using a two-way fixed effects model and a moderating effects model. Based on measuring the synergistic effects of agricultural NSP and ACE reduction, Hou et al. [4] further analyzed the regional differences, dominant factors, and the spatiotemporal heterogeneity of the synergistic effects of AWP and ACE reduction. In summary, existing studies have explored the synergistic reduction levels of regional agricultural NSP and ACE using the coupling coordination degree (CCD) model [4] and the synergistic control cross-elasticity method [10]. However, traditional water quality indicators predominantly focus on the concentration of specific pollutants when evaluating AWP, failing to fully capture its comprehensive impact on water resources. Grey water footprint (GWF) serves as a metric to quantify the extent of water pollution, considering the volume of freshwater required to dilute water pollutants and adequately attain prescribed water quality requirements [11]. The GWF incorporates water consumption into water pollution assessments, enabling the evaluation of water pollution at global, national, provincial, and municipal levels [12,13,14,15]. By comparison, the agricultural grey water footprint (AGWF) exhibits stronger spatiotemporal scalability when characterizing AWP, facilitating comparisons of water pollution across different periods and regions, and is more suitable for studying the synergistic mitigation between AWP and ACE on a large scale [7,16].

Scholars have examined the factors influencing agricultural NSP [17], AGWF [13], ACE [18], and the driving factors behind the synergistic effects of agricultural NSP and ACE mitigation [4]. For instance, Han et al. [8] utilized the CCD model, the Tapio model, and the LMDI model to examine the interaction between ACE and agricultural economic growth, and to explore the coupling and decoupling effects of ACE and its driving factors. They emphasized that agricultural development scale was the primary factor driving the rise in ACE, whereas advancements in agricultural technology were the key contributors to its reduction. In contrast, Kong et al. [13] used the SDE model and the generalized Divisia index method to explore the spatiotemporal variations and driving factors of AGWF. Specifically, they found that AGWF intensity, i.e., the ratio of AGWF to agricultural gross domestic product (AGDP), which reflected the technological level of agricultural water pollution control, had the greatest inhibitory effect on the growth of AGWF, whereas AGDP exerted the greatest driving effect. Hou et al. [4] employed the CCD model, Dagum Gini coefficient, and GeoDetector model to analyze the regional disparities and driving factors underlying the synergistic effects of agricultural NSP and ACE mitigation. They found that factors such as agricultural economic scale, cropping structure, and agricultural machinery predominantly influenced the spatial differentiation of these synergistic effects. However, few studies have examined the common driving factors of AWP (including NSP and AGWF) and ACE, particularly considering different spatial scales. Furthermore, various studies have reached differing conclusions regarding the drivers of AGWF and ACE.

Moreover, the government has implemented a series of pollution and carbon reduction policies, including the “Zhejiang Province 2012 Total Pollutant Reduction Plan” and the “Five Waters Treatment” action [19,20]. These policies aim to reduce pollution and CE and promote coordination between socio-economic development and resource environmental protection through specific measures, such as the transformation and upgrading of the agricultural industrial structure, the upgrading of agricultural irrigation and livestock manure management technologies, and the enhancement of agricultural production efficiency. Due to the varying impacts of different policy factors on the reduction of AWP and ACE, some policies might result in synergies, while others could lead to trade-offs. Therefore, the exact influence of various socio-economic and policy factors on promoting synergies in the co-control of AWP and ACE remains unclear. Additionally, regional differences in economic conditions and pollution levels hinder our understanding of how these factors collectively influence the observed synergistic trends between AWP and ACE. The variation in the influence of the same socio-economic factors across different administrative units has yet to be thoroughly investigated. Therefore, it is essential to thoroughly investigate the synergies in trends in ACE and AGWF and their common driving mechanisms from the perspective of partitioning.

In view of this, we innovatively utilize AGWF to analyze the synergistic reduction of AWP and ACE, as well as their common drivers from the perspective of partitioning. This study, taking Zhejiang as a case, aims to: (1) investigate the regional differences in the synergistic trends of ACE and AGWF; (2) examine the factors influencing ACE and AGWF and their spatial heterogeneity at the provincial, regional, and city scales, and propose a common driving mechanism for ACE and AGWF; (3) quantify the levels of synergistic reduction of ACE and AGWF at both regional and sub-regional scales.

2. Theoretical and Research Framework

Liu and Liu [6] pointed out that the excessive use of agricultural chemicals can lead to their loss and dispersion into the surrounding environment of cultivated land, thereby causing NSP. These chemicals can also produce GHG through decomposition. Therefore, there is a synergistic effect between agricultural NSP and ACE. Hou et al. [4] also mentioned that traditional agricultural production practices, such as the improper use of fertilizers and inadequate treatment of agricultural waste, often result in significant amounts of both NSP and GHG, and a synergistic effect similarly exists between the two. As shown in Figure 1, ACE and AGWF also share a common root and origin. Planting carbon emissions (PCE), mainly resulting from applying fertilizers, pesticides, and agricultural plastic films, as well as tillage operations, irrigation practices, and diesel fuel consumption, are significant contributors to overall GHG [8]. During agricultural activities and farming operations, the excessive use of chemical fertilizers and pesticides tends to cause their runoff and dispersion, leading to nitrogen and phosphorus pollution, which, in turn, increases the planting grey water footprint (PGWF) [21]. Meanwhile, fertilizers and other agricultural chemicals release greenhouse gases such as CO₂, N₂O, and CH₄ (converted into carbon equivalents) during decomposition. Therefore, PGWF is closely linked to PCE. Approaches to reducing the application of agricultural fertilizers and pesticides to lower PGWF may also contribute to a decrease in PCE. Both livestock grey water footprint (LGWF) and livestock carbon emissions (LCE) originate from livestock and poultry [22,23]. Yan et al. [24] pointed out that the treatment of livestock and poultry manure (a solid–liquid mixture of animal excreta), as well as intestinal fermentation in livestock and poultry, generates CE. If manure and urine flow into water bodies without special treatment, they will cause water pollution and subsequently lead to the formation of LGWF [13]. Therefore, LGWF and LCE are closely interconnected. Measures aimed at reducing livestock and poultry manure emissions or managing manure more effectively to lower LCE may simultaneously contribute to reducing LGWF.

The framework outlined for this research is displayed in Figure 2. We analyzed the yearly characteristics of AGWF and ACE across Zhejiang province, respectively. Based on their differing pollution and economic levels, we subsequently categorized the cities within Zhejiang into three regions. Our objective was to examine the correlation and synergistic trends between GWF and CE within the two agricultural sub-sectors across the province. Next, we investigated the common driving forces behind AGWF and ACE from the perspectives of regional, city, and provincial scales. Furthermore, we explored the regional heterogeneity of drivers for AGWF and ACE at the city scale, focusing on how these factors vary across different cities within Zhejiang. On this basis, we proposed a common driving mechanism for AGWF and ACE, which elucidates their interconnectedness. We also offered specific policy recommendations aimed at promoting synergistic reduction in both AGWF and ACE. Lastly, we examined the synergy levels of AGWF and ACE under the implementation of comprehensive measures across Zhejiang, evaluating how effectively these measures work together to mitigate both types of environmental impacts.

3. Materials and Methods

3.1. Study Area

Zhejiang is located in the southeastern coastal region of China (118°01′ E–123°10′ E, 27°02′ N–31°11′ N) (Figure 3). Zhejiang, comprising 11 prefecture-level cities (here, city, in China’s administrative division, refers to an urban area directly under the jurisdiction of a prefecture-level city, encompassing both urban and rural regions), is an important agricultural province and a major contributor to China’s agricultural production. In 2020, Zhejiang established the first national innovation zone for pollution and CE reduction, serving as a demonstration area for collaborative governance of pollution and CE reduction. The GDP of Zhejiang in 2020 reached CNY 6.46 × 10¹², with agriculture contributing CNY 0.21 × 10¹² to the total. In 2020, the usage of fertilizers, pesticides, and diesel in Zhejiang was as high as 6.96 × 10⁸ kg, 0.37 × 10⁸ kg, and 8.86 × 10⁸ kg, respectively.

3.2. Data Sources

The data utilized in this study primarily include (1) the model parameters, specifically the CE coefficients for planting and livestock, as well as the AGWF accounting coefficients detailed in Tables S1–S4 of the Supplementary Material, are sourced from Yan and Zhang [24], Song et al. [22], and Han et al. [8]. These coefficients are widely used in the calculation of ACE and AGWF across various nations and regions [8,22,24]. According to Yan and Zhang [24], the breeding cycle of poultry in ACE accounting is set at 55 days. (2) Agricultural statistics mainly cover the usage of pesticides, agricultural film, and chemical fertilizers, as well as the breeding and stock numbers of cattle, sheep, pigs, and poultry in livestock husbandry. These data are derived from the Statistical Yearbooks of Zhejiang Province and its cities (2011–2021). (3) The main socio-economic indicators, including GDP, AGDP, the number of agricultural employments, and total population, are also collected from the China Urban Statistical Yearbook and Statistical Yearbooks of Zhejiang Province and its cities from 2011 to 2021.

3.3. Methods

3.3.1. ACE Accounting

ACE is primarily derived from the production processes involved in crop and livestock farming. PCE is generated from farmland irrigation, cultivation, and the utilization of agricultural supplies [8]. LCE involves the release of CO₂, CH₄, and N₂O (converted into carbon equivalents) during the digestion process of livestock and manure disposal. Existing studies commonly select six primary sources of PCE for their calculations: fertilizers, pesticides, agricultural plastic films, tillage operations, irrigation, and diesel fuel [8]. LCE mainly comes from pigs, cattle, sheep, and poultry. The calculation formulas [8,24] are provided in Equations (1)–(3).

$$PCE=\sum C_i\times {\epsilon }_i$$

(1)

where $C_i$ represents the consumption of the i-th carbon source; ${\epsilon }_i$ denotes the emission coefficient of the i-th carbon source.

$$LCE=\sum {GEI}_1\times T_i\times {\delta }_{1i}+\sum {GEI}_2\times T_i\times {\delta }_{2i}+\sum {GEI}_3\times T_i\times {\delta }_{3i}$$

(2)

where ${GEI}_1$, ${GEI}_2$, and ${GEI}_3$ represent the conversion coefficients for enteric fermentation (CH₄), manure fermentation (CH₄), and manure fermentation (N₂O), respectively. ${\delta }_{1i}$, ${\delta }_{2i,}$ and ${\delta }_{3i}$ represent the carbon equivalents coefficients corresponding to different carbon sources.

$$T_i=\left\{\begin{array}{c}{Alive}_i\times {\displaystyle \frac{M_i}{365}},{Alive}_i<365\\ {\displaystyle \frac{\left(C_{it}+C_{i(t-1)}\right)}{2}},{Alive}_i\ge 365\end{array}\right.$$

(3)

where $T_i$ is the breeding cycle. Livestock breeding and slaughter can lead to changes in the number of livestock during the year; therefore, adjustments need to be made to the livestock breeding quantity. If the livestock and poultry production cycle (Alive_i) is under one year (365 days), the annual yield is employed to derive the average yearly feed intake (Formula (3)). Conversely, for production cycles equal to or exceeding one year, the average annual feed intake is approximated using the livestock and poultry year-end inventory [25].

3.3.2. AGWF Estimation

AGWF encompasses the GWF arising from both the planting industry and livestock husbandry. The PGWF mainly refers to the volume of freshwater required to dilute pollutants that enter groundwater or surface runoff through leaching during rainfall or irrigation due to fertilizers and pesticides that crops fail to fully absorb and utilize [21]. Given the considerable contribution of nitrogen fertilizer to AWP, it has been adopted as a key indicator for assessing the PGWF [22]. However, poultry manure and urine constitute the primary sources of COD and TN in livestock water pollution [13]. The calculation methods [13] are shown in Equations (4)–(7).

$$PGWF={\displaystyle \frac{\alpha \times N}{C_{max\left(N\right)}-C_{nat}}}$$

(4)

where $\alpha$ is the nitrogen fertilizer leaching rate, set as 7% [26]; $N$ represents the application rate of nitrogen fertilizer (t); $C_{\mathrm{m}\mathrm{a}\mathrm{x}\left(N\right)}$ represents the maximum allowable concentration of N in water, which is set as 0.015 kg/m³ [27]; $C_{nat}$ denotes the natural concentration of pollutants in water, assumed to be 0 kg/m³ [28].

$$LGWF=max\{{LGWF}_{COD},{LGWF}_{TN}\}$$

(5)

$${LGWF}_i={\displaystyle \frac{L_i}{C_{max\left(i\right)}-C_{nat}}}$$

(6)

$$L_i={\sum }_{j=1}^4{NUM}_j\times {DAY}_j\times \left(f_j\times n_j^f\times {\beta }_j^f+p_j\times n_j^p\times {\beta }_j^p\right)$$

(7)

where $C_{\mathrm{m}\mathrm{a}\mathrm{x}\left(i\right)}$ denotes the maximum permissible concentrations of COD and TN in water, set at 0.06 kg/m³ and 0.015 kg/m³, respectively [22]. $L_i$ stands for the pollutant discharge from the i-th livestock type (t); ${NUM}_j$ represents the count of the j-th livestock type (heads); ${DAY}_j$ indicates the breeding cycle of the same (days); $f_j$ and $p_j$ are the fecal and urine excretion coefficients for the j-th livestock type (kg/d), respectively; $n_j^f$ and $n_j^p$ denote the TN and COD coefficients in the feces and urine of the j-th livestock type (kg/t); ${\beta }_j^f$ and ${\beta }_j^p$ represent the respective loss rates of TN and COD in these excretions (%).

3.3.3. K-Means Clustering

K-means clustering is a technique that groups samples based on their similarity, with each cluster characterized by its mean value, which is widely used in environmental science and is highly regarded as an effective classification tool [29]. To examine the synergistic trends of ACE and AGWF across Zhejiang, as well as the differences in common driving factors in regions with varying pollution and economic levels, we employed the K-means clustering method using R version 4.3.2. Specifically, we leveraged the ‘cluster’ and ‘factoextra’ packages, which are essential for performing clustering analysis and visualizing cluster results, to categorize Zhejiang’s cities based on their AGWF, ACE, AGDP, and GDP. The fundamental principles and detailed methodologies underlying this analysis are comprehensively outlined in the referenced literature [30].

3.3.4. Decomposition of Driving Forces

Index decomposition analysis (IDA) includes methods such as LMDI, SDA, and IPAT. Among them, the LMDI model features independent decomposition characteristics, which ensure unbiased results and prevent the introduction of residual terms, thus avoiding uncertainty in the results [31]. Due to its robust design characteristics, LMDI is considered the optimal IDA method [32]. Based on the synergistic characteristics of AGWF and ACE, representative and commonly influential variables were selected to decompose ACE and AGWF [8,21,33,34]. According to Ang [31], Equation (8) is the multiplicative LMDI decomposition method:

$$D={\displaystyle \frac{D}{AGDP}}\times {\displaystyle \frac{AGDP}{GDP}}\times {\displaystyle \frac{GDP}{P}}\times {\displaystyle \frac{AL}{ARD}}\times {\displaystyle \frac{ARD}{AGDP}}\times {\displaystyle \frac{AGDP}{AL}}\times P$$

(8)

where D represents ACE or AGWF, P represents population size, AL represents agricultural labor population, and ARD represents agricultural research and development (R&D) funding.

${\displaystyle \frac{D}{AGDP}}$ represents the ACE or AGWF generated per unit of AGDP, indicating the intensity of ACE or AGWF. It serves as a measure of agricultural technological level [8,21].

${\displaystyle \frac{AGDP}{GDP}}$ denotes the ratio of regional AGDP to total GDP, indicating the contribution of the agricultural sector’s GDP to the overall GDP. It is one of the important factors for measuring the relative development status of the agricultural sector.

${\displaystyle \frac{GDP}{P}}$ represents the per capita GDP, reflecting the level of the regional economy.

${\displaystyle \frac{AL}{ARD}}$ is the labor-to-research-and-development ratio (labor-R&D ratio) [34], representing the relationship between labor and R&D investment and embodying labor-intensive and technology-intensive development models.

${\displaystyle \frac{ARD}{AGDP}}$ represents the ratio of regional agricultural R&D investment to AGDP, indicating agricultural R&D expenditure intensity [33]. It highlights the scale and level of investment in regional agricultural technological innovation.

${\displaystyle \frac{AGDP}{AL}}$ represents the AGDP generated per unit of the agricultural labor force, defined as agricultural labor productivity. It reflects the economic value created by workers in the agricultural sector.

P represents the agricultural population.

The additive LMDI model can be written as follows:

$$∆D={∆D}_t-{∆D}_{t-1}=∆a+∆b+∆c+∆d+∆e+∆f+∆g$$

(9)

$$∆X_i={\displaystyle \frac{D_t-D_{t-1}}{ln{D}_t-ln{D}_{t-1}}}\times ln{\displaystyle \frac{X_{it}}{X_{it-1}}}$$

(10)

where $D_t$ and $D_{t-1}$ represent the ACE or AGWF in the t-th and t−1-th years, respectively; a, b, c, d, e, f, and g denote technological level, industrial structure, per capita GDP, labor-R&D ratio, agricultural R&D expenditure intensity, agricultural labor productivity, and population, respectively; $X_i$ represents the i-th impact factor; $X_{it}$ and $X_{it-1}$ represent the values of the i-th impact factor in year t and year t−1, respectively.

3.3.5. Synergistic Level Calculation

The synergistic control cross-elasticity method, originally proposed by Mao et al. [35], has garnered widespread adoption in recent years for investigating the synergistic effect of pollution and carbon reduction [36]. As such, this paper utilizes the synergistic control cross-elasticity method to measure the synergistic degree (S) between ACE and AGWF changes (including planting and livestock husbandry):

$$S={\displaystyle \frac{∆ACE/ACE}{∆AGWF/AGWF}}={\displaystyle \frac{({ACE}_t-{ACE}_{t-1})/{ACE}_{t-1}}{({AGWF}_t-{AGWF}_{t-1})/{AGWF}_{t-1}}}$$

(11)

where ${ACE}_t$ and ${ACE}_{t-1}$ represent the ACE in the t-th year and the (t−1)-th year, respectively; ${AGWF}_t$ and ${AGWF}_{t-1}$ represent the AGWF in the t-th year and the (t−1)-th year, respectively. The classification of S is shown in

Table 1. Level of S and its descriptions.

Range	Value	Level of S	Descriptions
ΔACE/ACE < 0, ΔAGWF/AGWF > 0	S ≤ 0	/	No synergy
ΔACE/ACE > 0, ΔAGWF/AGWF < 0
ΔACE/ACE < 0, ΔAGWF/AGWF < 0	0 ≤ S ≤ 1	I	Synergistic reduction, the reduction in AGWF is greater than that of ACE
	S ≥ 1	II	Synergistic reduction, the reduction in ACE is greater than that of AGWF
ΔACE/ACE > 0, ΔAGWF/AGWF > 0	0 ≤ S ≤ 1	III	Synergistic increase, the increase in AGWF is greater than that of ACE
	S ≥ 1	IV	Synergistic increase, the increase in ACE is greater than that of AGWF

. The increases and decreases in ACE and AGWF are driven by seven factors decomposed using the LMDI method. Therefore, the effect after the implementation of comprehensive measures described in this study reflects the combined effects of these seven factors on the increases or decreases in ACE and AGWF.

4. Results

4.1. Trends in ACE and AGWF

ACE and AGWF from 2010 to 2020 were calculated, with the clear trends illustrated in Figure 4. From 2010 to 2012, ACE and AGWF exhibited an upward trend, increasing by 2.62 × 10⁸ kg and 0.36 × 10⁸ m³, respectively. From 2013 to 2020, ACE and AGWF underwent notable declines, with ACE reducing by 35.11 × 10⁸ kg and AGWF shrinking by 25.71 × 10⁸ m³. The LCE and LGWF increased by 2.23 × 10⁸ kg and 0.95 × 10⁸ m³, respectively, from 2010 to 2012. However, between 2013 and 2020, they declined significantly by 28.97 × 10⁸ kg and 15.93 × 10⁸ m³, respectively. The PCE was relatively stable from 2010 to 2020, while the PGWF decreased by 10.72 × 10⁸ m³. Therefore, the change in ACE and AGWF is dominated by livestock husbandry.

The variations in ACE and AGWF among cities in Zhejiang were significantly different. Specifically, Quzhou and Jiaxing exhibited a diminishing contribution rate to ACE. In contrast, Taizhou, Wenzhou, and Ningbo demonstrated an escalating contribution rate (Figure 4b). Concurrently, the proportion of AGWF in Jiaxing declined, while those in Shaoxing and Wenzhou progressively increased (Figure 4d).

As illustrated in Figure 5, from 2017 to 2020, a relatively steep decline in PGWF was observed in Jiaxing and Shaoxing. In contrast, the PCE and PGWF in the other cities in Zhejiang remained relatively stable. Specifically, Taizhou and Ningbo emerged as the primary contributors to the PCE, while Jiaxing and Shaoxing made significant contributions to the PGWF. However, the PCE of the remaining cities was below 4 × 10⁸ kg and their PGWF was under 3 × 10⁸ m³. In the livestock industry, LCE and LGWF across various cities generally followed a trend of rising initially and then declining, with peak emissions mainly occurring between 2011 and 2013. Notably, Zhoushan, due to its island geography and small-scale farming practices, exhibited significantly lower LCE and LGWF than other mainland cities.

4.2. Synergies in the Trends in ACE and AGWF

A clustering analysis was performed on Zhejiang Province, utilizing AGWF, ACE, GDP, and AGDP as key indicators. To evaluate the robustness of the clustering results, the average silhouette coefficient was calculated (see Figure S1 in the Supplementary Material). The analysis revealed that the average silhouette coefficient attains its maximum when the data are divided into three clusters. As shown in

Table 2. Results of K-means clustering.

Clusters	Levels 1				Member Cities
	AGWF	ACE	GDP	AGDP
H-H-H-H area	H 1	H	H	H	Hangzhou and Ningbo
H-H-L-L area	H	H	L	L	Wenzhou, Shaoxing, Taizhou, Jinhua, Jiaxing, and Quzhou
L-L-L-L area	L	L	L	L	Huzhou, Lishui, and Zhoushan

¹ L and H represent low and high levels, respectively, as values below and above the mean of Zhejiang. Moderate levels are denoted for AGWF, ACE, GDP, and AGDP of Zhejiang.

, the clustering results are as follows: (1) Region 1 (region specifically refers to a collection of multiple cities) includes Hangzhou and Ningbo, where the average values of AGWF, ACE, GDP, and AGDP are all higher than the provincial averages, identifying it as a high-pollution and high-economic area; (2) Region 2 consists of Wenzhou, Shaoxing, Taizhou, Jinhua, Jiaxing, and Quzhou. This region is characterized by average AGWF and ACE values exceeding the provincial averages. In contrast, GDP and AGDP averages fall below the provincial levels, identifying it as a high-pollution, low-economic area; and (3) Region 3 consists of Huzhou, Lishui, and Zhoushan, where the average values of AGWF, ACE, GDP, and AGDP are all lower than the provincial averages, categorizing it as a low-pollution, low-economic region.

Given the marked disparities in CE and GWF between planting and livestock husbandry, these two distinct agricultural sub-sectors were examined to uncover synergies in the trends in their CE and GWF. As shown in Figure 6, ACE and AGWF across Zhejiang exhibit a clear linear relationship, indicating a strong correlation between regional ACE and AGWF. As depicted in Figure 6a, from 2010 to 2013, the LCE and LGWF in Zhejiang and Region 3 showed a consistent upward trend, while the LCE and LGWF in Region 1 and Region 2 fluctuated. This is due to Region 3 transitioning from traditional high-emission livestock farming models to green and sustainable practices between 2010 and 2013. This transition drove the growth of LCE and LGWF in Zhejiang and also led to synergies in the trends in these indicators in Zhejiang and Region 3. During this period, the livestock farming models in different cities within Region 1 and Region 2 completed their transitions ahead of schedule. As a result, LCE and LGWF reached their respective peaks at different times, leading to fluctuating changes. LCE and LGWF across Zhejiang decreased together significantly from 2014 to 2020. In addition, resulting from the ongoing reduction in pesticide and fertilizer application across Zhejiang, the PCE and PGWF showed a synchronized downward trend from 2010 to 2020.

4.3. Common Driving Forces of ACE and AGWF

The changes in ACE and AGWF were decomposed into the combined effects of seven influencing factors based on the LMDI model. Figure 7, Figures S2–S4 (see the Supplementary Material), and Figure 8 demonstrate the synergistic and driving effects of various factors on ACE and AGWF. Within the same period and sector, each factor exerted a consistent effect on both ACE and AGWF, with the direction of its influence on ACE aligning with that on AGWF. The effects of agricultural labor productivity, per capita GDP, population, industrial structure, technological level, and the labor-R&D ratio on ACE and AGWF show no spatial, temporal, or sectoral heterogeneity (the green bars in the first three columns and the orange bars in the last three columns in Figure 7 and Supplementary Figures S2–S4). Agricultural labor productivity and per capita GDP together can explain 98.01% and 98.03% of the increase in ACE (Figure 7a,b) and AGWF (Figure 7c,d), respectively, in Zhejiang from 2010 to 2020 (the green bars in the first two bars). Technological level and labor-R&D ratio together can explain 78.38% and 79.41% of the decrease in ACE (Figure 7a,b) and AGWF (Figure 7c,d), respectively, in Zhejiang from 2010 to 2020 (the yellow bars in the last two bars). This indicates that agricultural labor productivity and per capita GDP were the main drivers of ACE and AGWF increase across Zhejiang from 2010 to 2020, while technological level and the labor-R&D ratio were the main factors for their reduction. The decline in the labor-R&D ratio has progressively strengthened its inhibitory effect on ACE and AGWF across Zhejiang. For example, the labor-R&D ratio in Zhejiang decreased from 0.192 capita per CNY 10,000 in 2010 to 0.095 per CNY 10,000 in 2019 and 0.037 capita per CNY 10,000 in 2020. Meanwhile, ACE experienced a substantial reduction of 48.78 × 10⁸ kg between 2010 and 2019, followed by a decrease of 44.72 × 10⁸ kg from 2019 to 2020. In parallel, AGWF declined by 29.28 × 10⁸ m³ from 2010 to 2019, with a drop of 24.05 × 10⁸ m³ from 2019 to 2020. Population growth increased ACE and AGWF across Zhejiang from 2010 to 2020 (the green bars in the third column). For instance, from 2010 to 2015, population growth in Zhejiang contributed to increases of 1.43 × 10⁸ kg in ACE and 45.67 × 10⁸ m³ in AGWF (Figure 8a,b). This indicates that ACE and AGWF across Zhejiang will likely continue to rise with population growth. In addition, industrial structure contributed to the joint reduction of ACE and AGWF across Zhejiang from 2010 to 2020 (the green bars in the third bar). For example, changes in industrial structure reduced ACE and AGWF by 15.62 × 10⁸ kg and 9.16 × 10⁸ m³, respectively, in Zhejiang from 2015 to 2020 (Figure 8c,d). This suggests that as the proportion of AGDP in total GDP continues to decline, ACE and AGWF across Zhejiang will continue to decrease.

The impact of agricultural R&D expenditure intensity on ACE and AGWF exhibits spatial-temporal heterogeneity (Figure 8a,c and Figure 8b,d) and sectoral differences (Figure 8a,b and Figure 8c,d). From 2010 to 2015, this factor decreased ACE and AGWF in Zhejiang and Region 2 (high-pollution and low-economic areas), while increasing ACE and AGWF in Region 1 and Region 3 (Figure 8a,b). For instance, it was estimated to reduce PCE by 1.75 × 10⁸ kg and PGWF by 1.30 × 10⁸ m³ in Zhejiang from 2010 to 2015 and to increase PCE by 2.17 × 10⁸ kg and PGWF by 1.24 × 10⁸ m³ in Region 1 (Figure 8a). However, from 2015 to 2020, agricultural R&D expenditure intensity increased ACE and AGWF across Zhejiang (Figure 8c,d).

4.4. Regional Heterogeneity of Drivers at the City Scale

The disparities in driving forces are further analyzed at the city level (Figure 9). From 2010 to 2020, the impact patterns (whether positive or negative) of various factors on GWF and CE in the planting industry and livestock husbandry exhibit consistency across cities. Several cities, including Zhoushan, Wenzhou, Taizhou, and Shaoxing, exhibit unique characteristics. In Zhoushan, Wenzhou, Taizhou, and Shaoxing, agricultural R&D investment intensity contributed to reductions in ACE by 48.60%, 20.41%, 6.67%, and 20.18%, respectively, and AGWF by 32.51%, 20.56%, 8.31%, and 20.31%. Notably, in Wenzhou, Taizhou, and Shaoxing, the relatively low initial AGDP but rapid economic growth (see Supplementary Material Figure S5), combined with increased agricultural R&D investment, significantly reduced ACE and AGWF associated with economic development. However, in Zhoushan, due to fluctuations in R&D investment and the rapid increase in AGDP, the impact of agricultural R&D expenditure intensity on ACE and AGWF showed a fluctuating upward trend, with emission reductions outweighing emission increases.

In addition, the impact of Zhoushan’s industrial structure and labor-R&D ratio on ACE and AGWF exhibited unique characteristics compared those in to other cities (Figure 9 and Figure 10). The primary reason is that Zhoushan is mainly composed of islands, with a high degree of land fragmentation. The arable land is relatively scattered, and the scale of agriculture is comparatively small, which leads to fluctuations in agricultural output value. Therefore, the contribution of AGDP to total GDP and its impact on ACE and AGWF are prone to fluctuations. However, changes in the industrial structure tend to have a greater positive impact on ACE and AGWF than negative, further resulting in an industrial structure that increases ACE and AGWF (unlike in other cities with more stable industrial structures, where the effect is the opposite) [37]. Furthermore, the labor-R&D ratio and agricultural R&D expenditure indicators signify the proportional relationships between the agricultural labor force and AGDP, and between these factors and agricultural R&D investment, respectively. All cities in Zhejiang have experienced a decline in labor availability, while AGDP has shown an upward trend (Figures S5 and S6 in the Supplementary Material). As a result, disparities in agricultural R&D investment among cities are the key factor driving the spatial heterogeneity in the impact of the labor-R&D ratio and agricultural R&D expenditure on ACE and AGWF. Zhoushan’s R&D investment has shown fluctuating growth, whereas other cities have experienced steady growth (Figure S6 in the Supplementary Material). Therefore, the influence of the labor-R&D ratio and agricultural R&D expenditure on ACE and AGWF in Zhoushan is continuously shifting. In contrast, the labor-R&D ratio has a relatively stable negative impact on ACE and AGWF in other cities.

4.5. Change in Synergistic Degree

The synergistic change of AGWF and ACE has been quantified by utilizing S and combining the accounting data of AGWF and ACE across Zhejiang (Figure 11). The synergistic levels of AGWF and ACE were evenly distributed across Zhejiang from 2010 to 2020 (except that the synergistic level of the plantation was not III), with S values ranging from −3.82 to 3.84, covering three levels of synergy: no synergy (16.25%), synergistic reduction (72.50%), and synergistic increase (11.25%).

There was no synergy between PGWF and PCE across Zhejiang for two to three years during the study period, primarily from 2010 to 2013, with S values less than 0. This is attributed to the fact that, during this period, certain cities within Zhejiang enforced effective pollution control strategies (such as the “11th and 12th Five-Year Plans”), with notable advancements in water resource management. Although PCE has slightly increased due to the expansion of agricultural production, PGWF has gradually decreased through the enhancement of agricultural water use efficiency and the rational utilization of water resources. These two indicators did not exhibit a synergistic trend. However, the synergistic levels of PGWF and PCE were at level I across Zhejiang in most years (mainly from 2013 to 2020), with S values ranging from 0.07 to 0.96, suggesting that comprehensive measures in most years had a greater impact on reducing PGWF than on reducing PCE. Furthermore, the synergistic level of PGWF and PCE reached IV in Region 2 during 2011–2012, with an S value of 2.36. In contrast, the synergistic level in Zhejiang and Region 2 was II during 2016–2017, with S values of 1.19 and 1.82, respectively. Commencing in 2013, PGWF and PCE across Zhejiang began demonstrating varying pollution and carbon reduction effects.

There was no synergy between LGWF and LCE during 2011–2012 in Region 2 and Region 3, and during 2012–2013 in Region 1. The synergistic level of LGWF and LCE was at level I across Zhejiang after 2012, with S values ranging from 0.07 to 0.99, indicating that comprehensive measures in most years had a greater impact on reducing LGWF than on LCE. The synergistic level of LGWF and LCE was at level II across Zhejiang for one to three years during the study period, with S values ranging from 1.01 to 1.84, suggesting that comprehensive measures in a few years had a greater impact on reducing LCE than on LGWF. The synergistic level of LGWF and LCE reached levels III and IV across Zhejiang during certain years, with most occurrences in 2010–2013, except for Region 3 in 2019–2020.

5. Discussion

5.1. Common Driving Mechanism of ACE and AGWF

As a major agricultural country, China is responsible for approximately 17% of global ACE and around 30.5% of global chemical fertilizer consumption [22,38]. To mitigate these environmental impacts, China has set ambitious climate goals, including carbon peaking by 2030 and carbon neutrality by 2060, collectively referred to as the “Dual Carbon” targets, which is China’s goal of promoting carbon dioxide absorption while reducing carbon dioxide emissions and achieving a transition from a high-carbon to a low-carbon or even carbon-free energy structure. This strategy is part of a broader vision to build an ecological civilization, encapsulated in the concept of “Beautiful China,” which aims for substantial improvements in environmental quality.

In line with these goals, the Chinese government is working to reduce AWP and ACE through a combination of policy measures and technological innovations. For example, in 2022, the State Council and the Ministry of Agriculture and Rural Affairs introduced the “Implementation Plan for Synergistic Reduction of Pollution and Carbon Emissions” and the “Implementation Plan for Agricultural Emission Reduction and Carbon Sequestration.” These national-level initiatives reflect China’s commitment to achieving its environmental targets. However, the effectiveness of such policies is also contingent on their application at the local level.

Zhejiang, as a case in point, has made significant strides in reducing both ACE and AWP. Local government efforts, including the “Zhejiang Province 2012 Total Pollutant Reduction Plan” under the “Twelfth Five-Year Plan” (2012), which aimed to reduce pollution emissions across various sectors through a series of measures, and the “Five Waters Treatment” action introduced in 2013 which sought to use water treatment as a breakthrough to force industrial transformation and upgrading and reduce water pollution, have contributed to measurable improvements. From 2013 to 2020, Zhejiang witnessed a downward trend in ACE and AGWF, which can be attributed to the province’s shift from a traditional economic model to a more sustainable development approach. These policies, alongside changes in socio-economic factors, have created a direct impact on AWP and ACE, providing a model for other regions aiming to balance economic growth with environmental preservation. The following will delve deeper into the specific mechanisms behind these impacts, exploring how the combination of policy changes, technological advancements, and socio-economic shifts has influenced both AWP and ACE in Zhejiang.

Advancements in technology, such as precision fertilization and manure collection and treatment technologies, have improved resource use efficiency in agricultural production (including the efficiency of fertilizer and pesticide use, as well as the resource utilization of livestock manure) [39]. These improvements not only reduce environmental pollution caused by agricultural production but also suppress the generation of ACE and AGWF [40,41].

An unreasonable agricultural industrial structure, characterized by unbalanced planting structures and small-scale and fragmented livestock farming, can lead to excessive reliance on fertilizers and pesticides in crop production, as well as insufficient waste management in livestock farming [37]. Consequently, industrial structure contributes to increased regional ACE and AGWF, as seen in Zhoushan. The optimization of industrial structure, such as promoting low-carbon production methods like ecological and organic agriculture, adjusting crop planting structures, and transitioning from small-scale, energy-intensive, and highly polluting decentralized planting and livestock farming models to intensive, large-scale planting and livestock farming systems, can achieve water and energy savings while increasing production [42,43]. This also contributes to reducing regional ACE and AGWF, as observed in cities like Hangzhou, Ningbo, and Wenzhou.

Population growth is a significant driver of agricultural production expansion and economic development. However, it has also led to the rapid expansion of the planting industry, accompanied by the excessive use of fertilizers and pesticides, increasing PCE and PGWF [8,44]. Rising income levels have shifted dietary preferences from grains to livestock and poultry products in the livestock sector. This change has fueled the growth of livestock and poultry breeding to accommodate population expansion and escalating consumer needs [45,46], but it has also resulted in a corresponding increase in LCE and LGWF.

The labor-R&D ratio is a key factor in curbing the growth of ACE and AGWF. This is due to the agricultural labor force in Zhejiang showing a general declining trend, while agricultural R&D investment continues to rise. The reduction in agricultural labor has objectively accelerated the process of agricultural mechanization, partially substituting human and animal labor [47]. This substitution effect contributes to improved agricultural production efficiency, reducing regional ACE and AGWF [48].

Agricultural R&D expenditure intensity reflects the regional technological level from an economic perspective. Therefore, due to varying levels of alignment between regional pollution and economic development, agricultural R&D expenditure intensity may have a dual effect on ACE and AGWF. In regions with high pollution and medium or low economic levels (e.g., Region 2 in 2010–2015), the increase in agricultural R&D expenditure intensity reduced ACE and AGWF by improving agricultural production efficiency. In regions with high pollution and high economic levels (e.g., Region 1), as well as regions with high economic levels and medium or low pollution levels (e.g., Region 2 in 2015–2020), an increase in agricultural R&D expenditure intensity may trigger a rebound effect. This phenomenon aligns with the Jevons Paradox and the Khazzoom–Brookes Postulate [49]. Specifically, improvements in agricultural production efficiency could reduce production costs and cycles for residents and agricultural enterprises, leading to an expansion in cultivated areas and livestock production. This, in turn, could increase emissions, offsetting or even surpassing the reductions achieved [50,51]. The fundamental cause of this rebound effect lies in the disconnect between policy design and the actual regional conditions, specifically when policies fail to fully consider the complex feedback mechanisms between local pollution levels and economic development status [52].

Agricultural labor productivity is a major factor driving the growth of ACE and AGWF. While enhancements in agricultural labor productivity aid in decreasing labor time and costs, they often result in an enlargement of agricultural production scale, subsequently increasing the use of fertilizers, pesticides, and plastic films [47], thereby increasing ACE and AGWF.

5.2. Comparison with Other Literature

The PCE calculated in this paper across Zhejiang is consistent with the results of Fang et al. [53], while the LCE, as calculated, is close to those of Cheng and Yao [54]. The total AGWF, as calculated, falls within the range of results reported by Kong et al. [13] and Zhang et al. [21]. Certain studies have also predominantly investigated the synergistic effects of AWP and ACE reduction [4]. Furthermore, inter-provincial studies by Han et al. [8] and Kong et al. [13] have confirmed that advancements in technological levels are the main contributors to the reduction of regional ACE and AGWF. In comparison, the innovation and main contributions of this study are: (1) proposing a novel research framework for exploring the synergistic reduction of AWP and ACE, (2) investigating the synergistic trends of AWP and ACE and their regional differences from the perspective of AGWF and partitioning, (3) identifying the common dominant factors affecting ACE and AGWF across provincial, regional, and city scales and quantifying their contributions and regional heterogeneity on ACE and AGWF, and (4) proposing a common driving mechanism of ACE and AGWF. Due to the good spatiotemporal extensibility of GWF in water pollution assessment, the framework proposed in this paper possesses great application potential for analyzing the synergistic effects between AGWF and ACE at global, national, provincial, and municipal scales. Consequently, it offers guidance for emission reduction strategies in other regions and contributes to the formulation and implementation of environmental policies worldwide.

5.3. Limitations and Future Work

Consistent with most previous studies, this research primarily focuses on ACE and AGWF from the perspectives of planting and livestock husbandry. While these two sectors are the main contributors to ACE and AGWF, forestry and fisheries were not included due to data availability constraints. However, to more comprehensively reveal the potential for synergistic reductions in ACE and AGWF and their driving mechanisms, future studies should incorporate forestry and fisheries into the analysis when data becomes available. Additionally, given data limitations and the complexity of factor decomposition, this study did not examine all potential factors (such as rural population, intensity of application of agricultural fertilizers and pesticides) influencing ACE and AGWF. Future research could integrate a wider array of variables to provide a more thorough examination.

6. Conclusions and Policy Implications

6.1. Conclusions

This study analyzed the synergistic reduction of AWP and ACE from the perspective of AGWF. Utilizing K-means clustering, 11 cities in Zhejiang were divided into three sub-regions. Through the index decomposition and synergistic degree, this study revealed the drivers of AGWF and ACE, as well as the synergistic changing trends between them across Zhejiang. The main results are as follows:

(1)

The ACE and AGWF indices in Zhejiang initially exhibited an upward trend and subsequently experienced a steep drop. Specifically, from 2010 to 2012, ACE increased by 2.62 × 10⁸ kg, while AGWF rose by 0.36 × 10⁸ m³. However, from 2013 to 2020, both experienced a significant decline, with ACE dropping by 35.11 × 10⁸ kg and AGWF decreasing by 25.71 × 10⁸ m³, dominated by livestock husbandry.

(2)

A significant linear correlation exists between ACE and AGWF across Zhejiang. From 2010 to 2020, PCE and PGWF exhibited a synchronized declining trend, whereas LCE and LGWF demonstrated a similar downward pattern from 2014 to 2020.

(3)

Each factor exhibits a synergistic effect on ACE and AGWF, meaning that the influence of each factor on ACE and AGWF is in the same direction across Zhejiang. Agricultural labor productivity and per capita GDP are the main drivers for the increase in ACE and AGWF, accounting for 98.01% of the ACE growth and 98.03% of the AGWF growth, respectively. Conversely, the technological level and the labor-R&D ratio significantly contributed to the reduction in ACE and AGWF, accounting for 78.38% and 79.41% of the total reductions, respectively. In addition, the impact of agricultural R&D expenditure intensity on ACE and AGWF exhibits spatiotemporal heterogeneity and sectoral differences.

(4)

During approximately half of the period from 2010 to 2020, ACE and AGWF exhibited synergistic changes, with water pollution reduction measures in the agricultural sector having a more significant impact than carbon reduction strategies.

6.2. Policy Implications

Based on the above findings, this study posits that implementing policies aimed at enhancing technological levels and reducing the labor-R&D ratio (either by increasing R&D investment or reducing the labor force) will significantly reduce ACE and AGWF. Although the effectiveness of these policies may vary due to differences in socio-economic conditions among regions. However, while reducing the labor force may increase mechanization and agricultural labor productivity, it could also lead to a rise in ACE and AGWF. Therefore, it is essential to weigh the relationship between mechanization and environmental impact according to the specific agricultural conditions in different regions. The core of collaborative emission reduction strategies lies in advancing the research, development, and introduction of green farming technologies, as well as increasing investment in R&D for these technologies, particularly in regions where green agricultural technologies are not yet fully developed or widely applied. In the planting industry, priority should be given to promoting technologies aimed at reducing the use of pesticides and chemical fertilizers while improving their application efficiency, such as precision fertilization, pest prediction, and monitoring technologies. These technologies should be adjusted according to the agricultural practices and environmental conditions of different regions. In livestock husbandry, emphasis should be placed on optimizing manure treatment technologies, strengthening comprehensive manure management, and improving the efficiency of collection, storage, treatment, and resource utilization. These measures should be adjusted according to the needs and capacities of different regions to maximize their environmental and economic benefits.