Empirical Analysis of the Role of Digital Agriculture in Enabling Coordinated Development of Ecosystem Services and Human Well-Being: Evidence from Provincial Panel Data in China

Abstract

As digital transformation deepens, digital agriculture plays a crucial role in advancing the Coordinated Development Level of Ecosystem Services and Human Well-Being (CD-ESWB). However, aligning ecosystem service capacities with human development needs remains a challenge. This study investigates how digital agriculture influences CD-ESWB, using panel data collected from 30 provinces in China between 2014 and 2022. First, an index system, the Level of Digital Agriculture Development (LDAD), is discussed, followed by a quantitative analysis using the entropy-weight TOPSIS method. The CD-ESWB is then evaluated through the “Ecosystem Services–Human Well-Being” coupling coordination model. Empirical analysis, incorporating fixed effects, mediation, and moderation models, demonstrates that digital agriculture significantly enhances CD-ESWB, though its impact varies across regions. Robustness checks, including instrumental variable methods, sample interval adjustment, and variable substitution, confirm the reliability of the findings. Notably, the Technological Innovation Effect (TIE) mediates digital agriculture’s impact, while Industrial Structure Upgrade (ISU) acts as a positive moderator. This study emphasizes the importance of regional policies in promoting technological innovation and optimizing industrial structures, providing both theoretical and empirical insights into how digital agriculture shapes CD-ESWB.

1. Introduction

The rapid development of advanced technologies, including artificial intelligence (AI), blockchain, and cloud computing, has initiated a profound transformation in the global economy, ultimately giving rise to what is now recognized as the digital economy [1]. These technological breakthroughs have not only reshaped industries but also served as fundamental drivers propelling the expansion and maturation of the digital economy [2]. Within the framework of China’s 14th Five-Year Plan, the digital economy has progressed into a new stage characterized by extensive application, more systematic growth, and broader societal inclusion. This stage is distinguished by deep technological integration, expansive reach, high flexibility, and cost efficiency [3]. Moreover, the advancement of the digital economy has catalyzed the convergence of cutting-edge technologies, such as big data and the Internet of Things (IoT), with traditional industries, paving the way for innovative developments in sectors like digital agriculture [4]. As a vital component of the broader digital economy, digital agriculture is increasingly seen as a strategic pathway for fostering sustainable development in agricultural practices [5].

Ecosystems form the backbone of Earth’s life-supporting systems, directly affecting human survival and socioeconomic development [6]. Ecosystem services—defined as the beneficial contributions of ecosystems to human well-being—are essential for sustainability. Healthy ecosystems provide critical services, such as food supply, water purification, and climate regulation, which in turn support ecosystem stability and sustainable development [6]. Therefore, managing these services effectively is crucial to maintaining their long-term viability. The central goal of sustainability science is to balance ecosystem services with human well-being to achieve coordinated, long-term development [7]. However, growing human demands and the declining capacity of ecosystems are intensifying this challenge [8]. Ecosystem degradation poses a threat to both human well-being and economic development, making the effective management of ecosystem services essential for sustainable progress [6]. Thus, achieving CD-ESWB is critical for modern sustainability efforts [9]. CD-ESWB involves a balanced and sustainable integration of ecological functions with socioeconomic benefits, aiming to evaluate the extent to which ecological improvements support human well-being across regions and over time. Such coordination is crucial for maintaining long-term sustainability, ensuring that ecosystem health aligns with human socioeconomic requirements. Ecosystem services (ESs) significantly enhance human well-being by providing resources, regulating environmental processes, supporting social and cultural values, and increasing resilience against environmental stressors. However, the benefits derived from ESs are often shaped by social and institutional factors, such as governance policies and local power dynamics, which can influence access and distribution [10]. These factors may either enhance or limit ES availability for certain communities, potentially widening social disparities and limiting well-being gains. While ES development and conservation generally improve environmental quality and well-being, they can also place constraints on local economies by limiting land and water resources, thereby affecting employment and income in resource-dependent regions [10]. Balancing ecosystem protection with social and economic needs is therefore essential for achieving coordinated development between ESs and human well-being (CD-ESWB). Amid challenges posed by climate change and unsustainable resource utilization, securing food safety and advancing high-quality agricultural growth via digital transformation has become a necessary course of action [11]. China has implemented strong policy support for digital agriculture in recent years. The “Outline of the Digital Rural Development Strategy” emphasizes the crucial role of informatization in rural revitalization [12], while the “Digital Agriculture and Rural Development Plan (2019–2025)” calls for accelerating the digitalization of agriculture to improve agricultural production and optimize ecological resource allocation, thereby enhancing CD-ESWB [13]. However, advancing digital agriculture requires balancing ecological conservation and social equity to effectively achieve CD-ESWB. Against this backdrop, it is seen as a crucial means for fostering such coordinated development.

As a crucial part of the digital economy, digital agriculture now plays a leading role in advancing agricultural modernization and driving rural revitalization [4,14,15]. Through the integration of networks, cloud computing, and big data analytics across multiple disciplines and sectors, digital agriculture aims to establish a new agricultural model that enhances the efficient production of high-quality agricultural products [14]. However, despite its rapid growth, digital agriculture in China faces several challenges, including limited technological application, poor data integration, and a shortage of skilled personnel [4,14]. Nevertheless, by deeply integrating digital technologies with modern agricultural practices, digital agriculture plays a crucial role in advancing CD-ESWB.

At present, digital agriculture promotes CD-ESWB through three main approaches. The first involves driving technological innovation by integrating technologies such as data analytics, IoT, and AI into farming practices, boosting production efficiency and enhancing ecosystem services, ultimately achieving both ecological and economic gains [16]. Second, it integrates economic and environmental outcomes by focusing on balancing agricultural growth with environmental protection through digital technologies [17]. Third, it examines the social and policy impacts of digital agriculture, including the role of government policies in promoting sustainable rural development and advancing social equity [18,19]. Despite numerous studies, most have examined the separate impacts of digital agriculture on ecosystem services and human well-being, with insufficient exploration of its integrated role in CD-ESWB. This gap restricts a deeper understanding of the sustainable development potential of digital agriculture.

This study bridges the gap by utilizing panel data from 30 provinces in China, covering the period from 2014 to 2022, to examine how digital agriculture impacts CD-ESWB and to assess the development level of digital agriculture. The key contributions of this paper include the following: (1) developing an index system to evaluate LDAD, which provides both a theoretical and quantitative basis for analysis; (2) investigating the mechanisms through which digital agriculture affects CD-ESWB; (3) identifying the mediating role of TIE and the moderating effect of ISU in this relationship; and (4) presenting policy recommendations to fully harness digital agriculture’s potential in enhancing CD-ESWB.

2. Theoretical Analysis and Research Hypotheses

China has been a pioneer in embracing the concept of sustainable development, playing a key role in advancing the Sustainable Development Goals (SDGs) [20]. Its approach has evolved from environmental protection to the construction of an ecological civilization [21]. Within this context, digital agriculture has seen rapid growth. Since 2013, various provinces have introduced policies such as “Internet Plus” and “Agricultural and Rural Informatization”. The “Internet Plus” policy encourages the application of internet and digital technologies across traditional sectors, including agriculture, to foster innovation, increase efficiency, and drive sustainable practices. “Agricultural and Rural Informatization” focuses on integrating information and communication technologies (ICTs) with agricultural and rural development to improve productivity, optimize resource use, and enhance rural living standards [22]. These initiatives promote the integration of digital technologies with agriculture, thereby supporting CD-ESWB. By 2019, the digital agriculture and rural development levels across county regions reached 36% nationwide, and by 2020, over 210 national demonstration bases for agricultural and rural informatization had been established [23]. The rapid expansion of digital agriculture provides a crucial pathway for promoting CD-ESWB. Based on this, this study proposes a theoretical framework illustrating how digital agriculture contributes to CD-ESWB, as shown in Figure 1.

The theoretical framework illustrates how digital agriculture promotes CD-ESWB through TIE and ISU. Digital agriculture encompasses four key dimensions: infrastructure construction, macro-environment optimization, green production, and efficiency improvement. These dimensions work together to enhance agricultural sustainability. Infrastructure construction focuses on the development of digital tools and technologies such as broadband connectivity, mobile networks, and smart farming solutions, which are essential for modernizing agriculture. Macro-environment optimization involves creating a supportive policy and economic environment, encouraging the adoption of digital technologies and fostering innovation in the agricultural sector. Green production refers to the use of digital technologies to optimize resource use, reduce environmental impacts, and promote sustainable farming practices. Finally, efficiency improvement emphasizes enhancing agricultural productivity and operational efficiency through digital innovation. By deeply integrating digital technologies, agriculture is shifting towards intelligent, precise, and sustainable practices. This shift not only strengthens ecosystem services but also boosts the quality and output of agricultural products, promoting CD-ESWB. Hypothesis 1 (H1) posits that digital agriculture significantly impacts CD-ESWB.

In this context, TIE plays a crucial mediating role. By focusing on talent cultivation, research and development investment, and the application of innovations, digital agriculture improves resource utilization efficiency and optimizes ecosystem services, enabling sustainable development in production. Hypothesis 2 (H2) suggests that TIE is a key pathway through which digital agriculture indirectly promotes CD-ESWB.

Furthermore, ISU acts as a moderating variable, amplifying the positive impact of digital agriculture. By optimizing resource allocation and fostering high-value-added economic activities, ISU enhances the influence of digital agriculture on ecosystem services and human well-being. Hypothesis 3 (H3) indicates that ISU positively moderates digital agriculture’s effect on CD-ESWB.

The framework also clarifies the interrelationships among ecosystem services, human well-being, and human–ecosystem interactions. Ecosystems offer a range of services, including provisioning, regulation, support, and cultural contributions, which influence human material life, health, environmental conditions, and cultural well-being, both directly and indirectly. Digital agriculture enhances the capacity and quality of these ecosystem services, thereby improving human well-being and reflecting their coordinated development. The framework illustrates how digital agriculture advances CD-ESWB via both direct mechanisms and the indirect influence of technological innovation, with industrial structure upgrades serving as a moderating factor.

2.1. The Direct Impact of Digital Agriculture on the Coordinated Development of Ecosystem Services and Human Well-Being

Digital agriculture refers to the use of information as a key production factor, applying modern information technologies for visual representation, digital design, and data-driven management across agricultural objects, environments, and processes [24]. Any application of digital technology in agricultural research and production falls under this broad concept. Digital agriculture, as a crucial branch of the digital economy, serves as a vital contributor to the advancement of the CD-ESWB. The accelerated evolution of digital technologies has propelled the field of digital agriculture towards a trajectory of intelligent, precise, and environmentally sustainable development. This transformation directly promotes CD-ESWB. Technologies driven by innovation, such as IoT, precision farming, and data analysis, enable accurate monitoring and smart management across the agricultural production process. These advancements enhance resource efficiency, improve ecosystem service quality, and increase both the yield and quality of agricultural products [23]. AI and autonomous machinery are crucial in enhancing production decision-making and streamlining task execution, which in turn boosts agricultural efficiency [25]. The development of digital agriculture not only reduces costs and increases efficiency at the technical level but also minimizes resource waste and environmental pollution [26]. The deep integration of digital elements with labor, capital, and resources enables agricultural production to achieve both economic benefits and environmental protection, further advancing CD-ESWB [27,28]. Additionally, the widespread application of digital technologies allows remote regions to access market information, technological resources, and data support more effectively, helping unique agricultural products reach broader markets. This not only promotes balanced regional economic development but also significantly improves the well-being of local residents [29].

Although most studies support the positive effects of digital agriculture on CD-ESWB, alternative perspectives suggest potential challenges. One concern is that the promotion of digital agriculture may exacerbate regional income disparities. Particularly in areas where infrastructure is unevenly distributed, differences in the adoption of digital technologies may make it difficult for small-scale farmers to compete with large agricultural producers, leading to economic divisions within rural communities [18]. This economic disparity can negatively affect the well-being of rural populations and the sustainable supply of ecosystem services, potentially hindering overall well-being improvement. Furthermore, the development of digital agriculture faces structural challenges, such as the urban–rural divide, which are difficult to overcome in the short term, including weak digital infrastructure and difficulties in technology dissemination in rural areas [30]. Some research points to a “U-shaped” pattern, suggesting that during the initial phases of digital economy growth, the income disparity between urban and rural areas tends to decrease. However, as digitalization progresses, this gap may potentially widen once more. This imbalance could adversely affect CD-ESWB [31]. In addition, the penetration of digital technologies into traditional agriculture faces high barriers. Despite China’s early adoption of e-commerce, the digitalization of the production sector remains insufficient [32,33]. While the development of digital infrastructure and progress in digital agriculture has significantly improved production efficiency, the accompanying resource consumption and potential environmental impacts, such as higher electricity demand and the disposal of obsolete equipment, cannot be overlooked [34,35]. Moreover, the rise of digitalization has triggered a range of ethical and legal issues, including cybersecurity threats, data breaches, and privacy infringements [36]. Addressing these challenges is essential for fostering the long-term sustainability and overall health of both ecosystem services and human well-being. Proper management of these issues is crucial as digital agriculture continues to advance.

Drawing from the preceding analysis, Hypothesis H1 posits that the development of digital agriculture exerts a significant positive influence on the coordination between ecosystem services and human well-being.

2.2. The Indirect Impact of Digital Agriculture on the Coordinated Development of Ecosystem Services and Human Well-Being

The swift advancement of digital agriculture creates fresh opportunities for enhancing CD-ESWB and contributes to realizing sustainable development goals. The pervasive integration of digital technologies has facilitated the expansion and sophisticated advancement of agricultural production, thereby effectively advancing the CD-ESWB.

Mediating Effect: TIE serves as a key pathway through which digital agriculture indirectly influences CD-ESWB. Advanced digital technologies enable the scaling and intelligent management of agricultural production, indirectly enhancing the supply of ecosystem services and improving human well-being. The technological innovation in digital agriculture not only fosters balanced regional development but also helps optimize the relationship between population, resources, and the environment [37]. In light of the expansion of the digital economy, emerging technologies have provided agriculture businesses with digital advantages, leading to higher productivity and competitiveness [38].

Digital agriculture is not only a beneficiary of technological innovation but also a significant driver of it. Integrating technologies like big data, IoT, and AI drives innovation across agricultural production, particularly within digital agriculture. These advancements support precision farming, allowing real-time monitoring of crops, efficient use of water and fertilizers, and reduction in environmental impact [16]. The benefits of technological innovation are manifold: it shortens production cycles, reduces costs, creates safer working environments, and promotes environmentally friendly production processes while significantly increasing the yield and quality of agricultural products [39].

In addition, the introduction of innovative technologies, such as intelligent management systems, has provided farmers with more advanced and efficient tools for managing natural resources. This not only improves the efficiency of resource use but also helps to sustain and enhance critical ecosystem services. Studies have shown that TIE plays a crucial role in driving digital development and is fundamental to ensuring the long-term sustainability of both ecosystem services and human well-being [40]. The transformative impact of digital technologies extends beyond agriculture, driving significant changes in business models and operational strategies across a wide array of industries [41]. Digital platforms, in particular, facilitate more efficient allocation of resources within the physical economy, enabling sectors to identify opportunities for reducing emissions and thereby making substantial contributions toward the goals of “carbon peaking” and “carbon neutrality” [42,43]. Furthermore, the increasing use of big data analytics has catalyzed the emergence of new business models, which in turn boost the productivity of both farmers and enterprises [44,45]. Overall, the integration of digital technologies within agriculture fosters a balanced and coordinated approach to development, addressing social, environmental, and economic challenges in a synergistic manner [46].

However, different types of technological innovation may have varying impacts on ecosystem services: some innovations can significantly improve the quality of ecosystem services, while others may have no noticeable effect or even a negative impact [47,48].

In light of the analysis above, it is proposed that H2 be accepted as a working hypothesis. Integrating digital technologies within agricultural practices facilitates the interconnection between ecosystem services and human well-being, as mediated by the TIE.

Moderating Effect: ISU promotes the optimization of traditional industries through technological advancements, making industrial structures more coordinated and efficient [49]. The optimization and upgrading of industrial structures (ISU) is a double-edged sword in the context of CD-ESWB (coordinated development of ecosystem services and human well-being). While ISU can enhance resource efficiency, support sustainable practices, and amplify the benefits of digital agriculture, it may also introduce challenges such as environmental trade-offs, land use inefficiencies, and uneven regional development. Historically, China’s industrial structure has been centered on low-value sectors characterized by high energy usage and pollution, resulting in substantial resource and environmental costs, which place considerable strain on CD-ESWB [50].

A rational industrial structure can optimize the business environment and ensure the optimal allocation of resources and capital, thereby enhancing the positive effects of digital agriculture on CD-ESWB [51,52]. On the one hand, ISU reinforces the synergy between digital agriculture, ecosystem services, and human well-being. The innovative capabilities of digital agriculture, by integrating data, information, and platform resources, improve production and resource allocation efficiency, thus driving the improvement of ecosystem services [53]. Conversely, modernizing industrial structures can promote the growth of high-value economic activities, support the adoption of digital agricultural technologies, and foster sustainable, low-carbon farming practices, thereby improving CD-ESWB [54]. For instance, ISU is often accompanied by the transition to cleaner energy sources, which is crucial for maintaining ecosystem health and enhancing human well-being [55].

The positive moderating role of ISU in promoting the coordination between digital agriculture, ecosystem services, and human well-being is widely recognized. As industrial structures rationalize and advance, this typically leads to higher resource efficiency and lower environmental costs, making it easier to balance industrial development with ecological preservation. By encouraging the shift from resource-intensive industries to more knowledge-intensive and low-carbon industries, ISU can help reduce environmental pressures and improve the sustainability of ecosystem services. For example, the rationalization of industrial structure enhances the optimal allocation of resources, lowering overall energy consumption while fostering green total factor productivity (GTFP) growth, which directly benefits both human well-being and the environment [56,57]. Furthermore, the technological innovation resulting from ISU, by promoting the use of digital and clean technologies in agriculture, can reduce carbon emissions and facilitate the adoption of low-impact farming practices, significantly enhancing CD-ESWB in both urban and rural contexts [56].

However, some studies also point out potential adverse effects that need careful consideration. For instance, while ISU can lead to efficiency gains, it may also negatively impact land use efficiency, especially in regions where industrial expansion occurs at the expense of natural ecosystems. In particular, industries like heavy chemicals may cause long-term harm to ecosystem services such as soil fertility, water quality, and biodiversity [56]. Moreover, rapid industrial upgrading may temporarily suppress production efficiency and lead to increased ecological pressures as regions adjust to new technologies and regulations. For instance, the transition to more advanced industrial sectors may result in short-term inefficiencies, while technological upgrades often involve significant regulatory compliance costs that may further disrupt local economies [58]. Additionally, the pollution haven hypothesis suggests that ISU could potentially shift high-pollution industries to areas with less stringent environmental regulations, exacerbating environmental degradation in less developed regions [59].

A comprehensive approach, informed by empirical evidence, is necessary to ensure that ISU can simultaneously promote digital agriculture, ecosystem services, and human well-being while mitigating any negative externalities. Drawing from the discussion, H3 posits that ISU strengthens the beneficial effect of digital agriculture on fostering the coordination between ecosystem services and human well-being.

3. Research Design

3.1. Empirical Model Design

3.1.1. Entropy-Weighted TOPSIS Comprehensive Evaluation Method

• Data Standardization

To eliminate dimensional inconsistencies and ensure calculation accuracy, the range method is used to standardize the data, as the units of various indicators in the evaluation system differ.

For positive indicators, the following holds:

$$X_{ij}^{\ast }={\displaystyle \frac{X_{ij}^{\ast }-\min \left(X_{ij}\right)}{\max \left(X_{ij}\right)-\min \left(X_{ij}\right)}}\times 0.99+0.01$$

(1)

For negative indicators, the following holds:

$$X_{ij}^{\ast }={\displaystyle \frac{\max \left(X_{ij}\right)-X_{ij}}{\max \left(X_{ij}\right)-\min \left(X_{ij}\right)}}\times 0.99+0.01$$

(2)

where $i$ represents the province, $j$ represents the measurement indicator, and $X_{ij}$ and $X_{ij}^{\ast }$ represent the initial data and standardized data, respectively.

• Entropy Method for Determining Indicator Weights

The feature proportion $P_{ij}$ for the $i$th sample under the $j$th indicator is calculated as follows:

$$P_{ij}={\displaystyle \frac{X_{ij}^{\ast }}{{{\displaystyle \sum }}_{i=1}^n{X}_{ij}^{\ast }}}$$

(3)

Based on the entropy calculation formula, the information entropy $e_j$ for the jth indicator is calculated as follows:

$$e_j=-k{\displaystyle \sum }_{n=1}^n{P}_{ij}\ast ln{P}_{ij}$$

(4)

Based on $e_j$, calculate the divergence coefficient $d_j$ for the $j$th indicator:

$$d_j=1-e_j$$

(5)

Using $d_j$, calculate the weight $W_j$ for the $j$th indicator:

$$W_j={\displaystyle \frac{d_j}{{{\displaystyle \sum }}_{j=1}^n{d}_j}}$$

(6)

Calculate the comprehensive evaluation score $S_i$ using weighted summation:

$$S_i={{\displaystyle \sum }}_{j=1}^n\left(W_j{X}_{ij}^{\ast }\right)$$

(7)

where $n$ represents the number of indicators and $W_j$ represents the weight of the $j$th indicator.

• Construct the weighted decision matrix:

$$R={\left(r_{ij}\right)}_{mn}=\left(w_j^{\ast }{X}^{\ast }\right)$$

(8)

• Determine the positive and negative ideal solutions:

Positive ideal solution:

$$R^{+}=\max \left(r_{1j},r_{2j}\cdots r_{ij}\right)$$

(9)

Negative ideal solution:

$$R^{-}=\min \left(r_{1j},r_{2j}\cdots r_{ij}\right)$$

(10)

• Calculate the Euclidean distance $D_i$:

Distance to the positive ideal solution:

$$D_i^{+}=\sqrt{{{\displaystyle \sum }}_{j=1}^n{(r_{ij}-R_j^{+})}^2}$$

(11)

Distance to the negative ideal solution:

$$D_i^{-}=\sqrt{{{\displaystyle \sum }}_{j=1}^n{(r_{ij}-R_j^{-})}^2}$$

(12)

• Calculate the relative closeness coefficient $C_i$:

$$C_i={\displaystyle \frac{{\mathrm{D}}_i^{-}}{D_i^{+}+D_i^{-}}}$$

(13)

3.1.2. Coupling Evaluation Method

The coupling coordination degree is a metric used to assess how harmoniously different subsystems within a broader system develop together. It provides insight into the level of synchronization and mutual reinforcement among these subsystems. In the context of ecosystem services and human well-being, an inherent and complex coupling relationship exists, as the sustainability and health of ecosystems directly influence the quality of human life, while human activities, in turn, affect the state of the environment. Recognizing this interdependence, this study seeks to quantitatively evaluate their interconnectedness by constructing an “Ecosystem Services–Human Well-Being” coupling coordination model. This model is designed to measure the degree of integrated development between these two critical dimensions, offering a framework for understanding how improvements in one area can drive progress in the other.

• Coupling Coordination Degree Model

Since the subsystems in this study are two, the calculation formula is as follows:

$$C_{12}=2\times {\left[S_1{S}_2/{\left(S_1+S_2\right)}^2\right]}^{1/2}$$

(14)

where $C_{12}$ represents the coupling value between the development levels of ecosystem services and human well-being, ranging from 0 to 1.

The coupling coordination degree model is represented as follows:

$$D_{12}=\sqrt{{\mathrm{C}}_{12}\times T_{12}}$$

(15)

$$T_{12}=a_1{S}_1+a_2{S}_2$$

(16)

where $D_{12}$ represents the coupling coordination degree of the comprehensive index of ecosystem services and human well-being, ranging from 0 to 1; $T_{12}$ is the comprehensive evaluation index of ecosystem services and human well-being; $a_1$ and $a_2$ are the undetermined weight coefficients, with $a_1+a_2=1$. In this paper, it is assumed that ecosystem services and human well-being are equally important; thus, $a_1=0.5$ and $a_2=0.5$.

3.1.3. Baseline Model Construction

This study utilizes an econometric model to explore the impact of digital agriculture on CD-ESWB. The model is based on data from 30 Chinese provinces covering the period from 2014 to 2022, focusing on the development of digital agriculture and CD-ESWB. CD-ESWB serves as the dependent variable, with the digital agriculture development level as the primary explanatory variable, alongside several control variables, as outlined in Equation (17), to test Hypothesis H1.

$$CD-ESW{B}_{it}={\alpha }_i+{\beta }_{x1}LDA{D}_{it}+{\beta }_{k1}Control{s}_{it}+u_i+v_t+{\epsilon }_{it}$$

(17)

In Equation (17), $i$ and $t$ represent the individual and time, respectively; $u_i$ and $v_t$ represent the individual effects and time effects; $CD-ESWB$ represents the Coordinated Development Level of Ecosystem Services and Human Well-Being; $LDAD$ represents the Level of Digital Agriculture Development; $Controls$ represents the control variables; $\alpha$ represents the constant term; ${\epsilon }_{it}$ refers to the random disturbance term; ${\beta }_{k1}$ is the coefficient of the control variables, and ${\beta }_{x1}$ is the coefficient of the core explanatory variable, which is the key focus of this study. If ${\beta }_{x1}$ is significantly positive, it indicates that digital agriculture significantly promotes CD-ESWB; otherwise, the opposite is true.

3.1.4. Mediation Effect Model

This study refines the traditional three-step mediation effect method to further investigate the mediating role of digital agriculture on CD-ESWB—particularly focusing on its influence, whether digital agriculture promotes CD-ESWB through technological innovation. The traditional approach suffers from low statistical power and potential bias in effect estimation [60]. Therefore, following the recommendations of Jiang Ting et al. [61], this paper optimizes the traditional method and constructs the following mediation effect models to test Hypothesis H2.

First, Equation (17) is used to directly examine the enabling effect of digital agriculture on CD-ESWB.

Second, using technological innovation as the dependent variable and digital agriculture as the core explanatory variable, Equation (18) is constructed to investigate the impact of digital agriculture on technological innovation. In Equation (18), $TIE$ represents the mediating variable, technological innovation, and ${\beta }_{x3}$ represents the coefficient of the core explanatory variable.

$$TI{E}_{it}={\alpha }_i+{\beta }_{x3}LDA{D}_{it}+{\beta }_{k3}Control{s}_{it}+u_i+v_t+{\epsilon }_{it}$$

(18)

Third, based on a literature review, the impact of technological innovation on CD-ESWB is analyzed.

3.1.5. Moderation Effect Model

A moderation effect model is introduced to explore the moderating role of ISU in the impact of digital agriculture. By incorporating the moderating variable and its interaction term with digital agriculture into the baseline model, the following econometric model is constructed to test Hypothesis H3.

$$CD-ESW{B}_{it}={\alpha }_i+{\beta }_{x4}LDA{D}_{it}+{\beta }_{3}LDAD\times IS{U}_{it}+{\beta }_{k4}Control{s}_{it}+u_i+v_t+{\epsilon }_{it}$$

(19)

In Equation (19), $ISU$ represents the moderating variable, $LDAD\times ISU$ is the interaction term between the $LDAD$ and $ISU$. If the estimated coefficient ${\beta }_3$ of the interaction term is positive, it indicates that ISU has a positive moderating effect on the impact of digital agriculture on the CD-ESWB. Conversely, if ${\beta }_3$ is negative, it suggests that ISU negatively moderates the impact of digital agriculture on CD-ESWB.

3.2. Variable Selection and Data Description

3.2.1. Explanatory Variable

To scientifically evaluate the development level of digital agriculture, this study emphasizes objectivity, scientific rigor, and data availability. Drawing from the “Digital Agriculture and Rural Development Plan (2019–2025)” and relevant research findings [62,63,64], digital agriculture development is evaluated across four key dimensions: digital agriculture infrastructure, the macro-environment, green development, and production efficiency. Each dimension encompasses several specific indicators, forming a comprehensive evaluation framework (see

Table 1. Evaluation index system for digital agriculture development level.

Primary Indicator	Secondary Indicator	Indicator Direction
Digital Agriculture Infrastructure	Number of Taobao Villages (units)	+
	Length of Optical Cable (km)	+
	Number of Mobile Phone Base Stations (10,000 units)	+
	Rural Broadband Access Users (10,000 households)	+
	Proportion of Administrative Villages with Postal Services (%)	+
Macro-Environment of Digital Agriculture	Rural Electricity Consumption (100 million kWh)	+
	Number of Digital Agriculture Enterprises (units)	+
	Local Government Expenditure on Agriculture, Forestry, and Water Affairs (CNY billion)	+
	Number of Agricultural Meteorological Observation Stations (units)	+
	Degree of Agricultural Mechanization (kWh/hectare)	+
Green Development of Digital Agriculture	Multiple Cropping Index	+
	Effective Irrigation Rate (%)	+
	Cultivated Land Area (1000 hectares)	-
	Water Consumption per Unit Output (cubic meters/CNY)	-
	Intensity of Pesticide and Fertilizer Use (kg/1000 hectares)	-
Production Efficiency of Digital Agriculture	Per Capita Grain Production (kg)	+
	Agricultural Productivity (CNY 10,000/person)	+
	Rural Electricity Generation (10,000 kWh)	+
	Total Output Value of Agriculture, Forestry, Animal Husbandry, and Fishery (CNY billion)	+
	Total Wages of Employees in Agriculture, Forestry, Animal Husbandry, and Fishery (CNY billion)	+

Note: ‘+’ indicates a positive indicator, and ‘-’ indicates a negative indicator.

). Following data standardization, the entropy-weight TOPSIS method is used to comprehensively assess the level of digital agriculture development across provinces.

3.2.2. Dependent Variable

In this study, CD-ESWB is used as the dependent variable. Ecosystem services exert both direct and indirect influences on human well-being, functioning through various types of services—provisioning, regulating, cultural, and supporting—which are essential for the maintenance of broader ecological processes. These services not only sustain environmental balance but also underpin social and economic development. Nevertheless, the relationship between ecosystem services and human well-being is highly complex and non-linear [65]. While certain improvements in ecosystem services can enhance human well-being by promoting development and quality of life, other changes may have negative consequences, leading to adverse impacts on human health or economic stability. Conversely, human development also affects ecosystems, both directly—through land use, resource extraction, and pollution—and indirectly—through policies, technological advancements, and shifts in consumption patterns [66]. To capture these intricate dynamics, this research employs the coupling coordination degree model to assess the level of harmony between ecosystem services and human well-being. By using CD-ESWB as the dependent variable, the study aims to reveal the degree to which these two systems develop in a balanced and integrated manner.

Ecosystem services refer to the life-supporting goods and functions provided by ecosystems, which can impact human well-being either directly or indirectly. These services emerge from the complex structure, processes, and functions inherent to ecosystems, supporting life on Earth by offering essential resources and regulating environmental conditions. As the direct carrier and modifier of ecosystem services, Land Use/Cover Change (LUCC) plays a pivotal role in assessing Ecosystem Service Value (ESV), making it a crucial factor for evaluating how land alterations influence the overall functioning of ecosystems [9]. Drawing on the framework established by the Millennium Ecosystem Assessment, Xie Gaodi and colleagues refined the classification of ecosystem services into 11 secondary indicators, which are organized into four primary categories: provisioning services (such as food production, raw material supply, and water provision), regulating services (such as gas regulation, climate control, environmental purification, and water regulation), supporting services (such as soil conservation, nutrient cycling, and biodiversity), and cultural services (such as aesthetic value and recreational landscapes) [67,68]. This study utilizes the China Land Cover Dataset (CLCD) [69], integrating it with ecosystem service value coefficients derived from the Pipi Shrimp database [70]. By applying the equivalent factor evaluation method [67], the research assesses various land types—including farmland, forest land, grassland, water bodies, construction land, unused land, shrubland, wetlands, and snow cover—to calculate the ESV for each Chinese province over the period from 2014 to 2022. This comprehensive evaluation aims to provide a clearer understanding of how different land uses contribute to or detract from ecosystem service values across regions.

Human well-being generally encompasses aspects such as health, happiness, prosperity, and an overall positive state of living. It is achieved through various capabilities, opportunities, and conditions that together contribute to a high standard of life and personal fulfillment [71]. Drawing from extensive research and theoretical foundations, the concept of well-being has been explored and measured through multiple frameworks. In line with the relevant literature [65,72] and drawing inspiration from established indices such as the Human Development Index (HDI) [73], the World Health Organization (WHO) health guidelines [74], the Millennium Ecosystem Assessment (MEA) framework [8], and the Better Life Index developed by the OECD [75], this study constructs a comprehensive human well-being evaluation system. This evaluation framework encompasses six core dimensions: material living standards, health and well-being, education and knowledge, social governance and security, environmental quality and ecological sustainability, and cultural life. These dimensions were selected to capture the multifaceted nature of human well-being, ensuring a holistic approach to measuring quality of life across different regions. To maintain consistency and comparability across various regions and time periods, all indicators are quantified using relative indices. The entropy method is then applied to process the data, allowing for the derivation of the Human Well-being Development Index (HWDI), which provides a comprehensive measure of well-being over time and across geographic areas. The detailed evaluation system, including specific indicators for each dimension, is presented in

Table 2. Human well-being evaluation index system.

Primary Indicator	Secondary Indicator	Indicator Direction
Material Living Standards	Per Capita Household Consumption Expenditure (CNY)	+
	Per Capita Disposable Income (CNY)	+
	Per Capita Daily Water Consumption (liters)	+
	Per Capita Electricity Consumption (100 million kWh)	+
	Number of Household Cars per 100 Households (units)	+
Health and Well-Being	Health Checkup Coverage Rate	+
	Number of Health Technicians per 1000 People (persons)	+
	Number of Hospital Beds per 1000 People (units)	+
	Number of Medical Institutions per 1000 People (units)	+
	Incidence of Class A and B Infectious Diseases (per 100,000)	-
Education and Knowledge	Student–Teacher Ratio in Higher Education	-
	Per Capita Government Education Expenditure (CNY)	+
	Average Years of Education	+
	Proportion of Higher Education Graduates	+
	Illiteracy Rate of Population Aged 15 and Above (%)	-
Social Governance and Security	Urban–Rural Income Ratio	-
	Proportion of Social Security and Employment Expenditure in Government Budget (%)	+
	Minimum Living Security Coverage Rate	-
	Registered Unemployment Rate	-
	Employee Pension Insurance Coverage Rate	+
Environmental Quality and Ecology	Per Capita COD Emissions (tons)	-
	Average PM2.5 Concentration	-
	Per Capita Public Park Green Space (sqm/person)	+
	Green Coverage Rate in Urban Areas (%)	+
	Household Waste Harmless Disposal Rate (%)	+
Cultural Life	Per Capita Cultural Expenditure (CNY)	+
	The proportion of Cultural Expenditure in Government Budget (%)	+
	Cultural Facilities Area per 10,000 People (sqm)	+
	Per Capita Museum Visits (times)	+
	Per Capita Participation in Domestic Performances by Art Troupes (times)	+

Note: ‘+’ indicates a positive indicator, and ‘-’ indicates a negative indicator.

.

3.2.3. Mediating Variable

In most studies, the level of technological innovation is traditionally measured using patent data as a key indicator. However, relying on a single metric, such as the number of patents, may not provide a comprehensive view of the overall innovation landscape, as it fails to capture other critical aspects of innovation processes and capacities [76]. To offer a more complete and nuanced representation of technological innovation, this study employs a multidimensional approach by selecting three key indicators.

First, the full-time equivalent of R&D personnel is used to represent the reserves of technical talent, reflecting the human capital available for driving innovation. Second, the intensity of R&D expenditure is chosen to represent investment in research and development activities, highlighting the financial commitment to fostering technological advancements. Finally, the number of domestic patent applications serves as an indicator of innovation outcomes, providing a measure of the tangible results generated from research efforts.

Together, these three positive indicators capture different facets of technological innovation, from resource input to output. To ensure accuracy and comparability, the data for each indicator is standardized, after which the entropy method is applied for a comprehensive calculation. This approach allows the study to derive an overall metric that more accurately reflects the level of technological innovation across regions and time periods.

3.2.4. Moderating Variable

The process of upgrading industrial structures refers to the gradual transition from a lower-level, less complex industrial composition to a more advanced and diversified one. This shift typically involves the movement from reliance on primary industries, such as agriculture, toward secondary industries (manufacturing) and eventually to tertiary industries (services), reflecting an economy’s progression toward higher-value-added activities. As industries evolve, they contribute more significantly to economic growth and innovation.

To quantify the degree of industrial structure upgrading, this study adopts the methodology proposed by Pei Xiao et al. [70], which utilizes the industrial structure hierarchy coefficient. This coefficient serves as a measure of how advanced a region’s industrial structure has become. A higher coefficient indicates a greater degree of industrial sophistication and modernization, suggesting that the economy is moving toward sectors that contribute more to productivity and development.

The calculation of the industrial structure hierarchy coefficient is based on the relative contributions of the primary, secondary, and tertiary industries to the overall GDP. Specifically, the formula is expressed as follows:

$$ISU=L_1\times 1+L_2\times 2+L_3\times 3$$

(20)

where $L_1$, $L_2$, and $L_3$ represent the proportions of value added by the primary, secondary, and tertiary industries to GDP, respectively. This formula helps to capture the structural shifts within the economy, providing insight into the balance and growth of different sectors over time.

3.2.5. Control Variables

To prevent estimation bias from omitted variables and ensure an accurate assessment of the impact of digital agriculture on CD-ESWB, the following control variables are selected based on their theoretical relevance and empirical support: Urbanization Rate (UR): Calculated as the percentage of the urban population, this variable is included because urbanization is closely linked to economic development, infrastructure growth, and the adoption of digital technologies. As urbanization increases, there tends to be more access to technological resources and innovation, which can significantly influence the relationship between digital agriculture and the coordinated development of ecosystem services and human well-being [62]. Environmental Regulation Intensity (ERI): Measured by the occurrence of environmental regulation-related keywords in provincial government reports, this control variable captures the strength of government policies in promoting environmental sustainability. Since digital agriculture interacts with environmental policies (e.g., sustainable farming practices or resource conservation), incorporating ERI helps isolate the effect of digital agriculture from that of stringent environmental regulations that may either support or hinder the development of ecosystem services [77]. Human Capital Level (LHC): Measured by the average number of higher education students per 100,000 people, this variable is included to account for the role of education and workforce skills in adopting and implementing digital agricultural technologies. Regions with higher human capital are more likely to develop and apply digital technologies effectively, which could enhance the impact of digital agriculture on ecosystem service management and human well-being. Financial Development Level (LFD): Represented by the ratio of loans and deposits to regional GDP, LFD is included to control for the financial infrastructure and access to capital, which plays a crucial role in facilitating the adoption of new technologies, including digital agriculture. Higher financial development allows farmers and agricultural businesses to invest in digital tools, improve productivity, and thus promote CD-ESWB.

3.2.6. Data Sources and Descriptive Statistics

In developing the evaluation index system for this study, several key factors were carefully considered, including objectivity, scientific validity, and the availability of reliable data. To ensure that the analysis would be both comprehensive and robust, sample data were rigorously screened. Cases with missing or incomplete key indicators were excluded, ultimately yielding a balanced panel dataset from 30 Chinese provinces (excluding Tibet, Hong Kong, Macau, and Taiwan) for the period from 2014 to 2022. This selection ensured both the completeness and reliability of the data, providing a solid foundation for analysis.

For instances where data were missing for specific regions or particular years, linear interpolation methods were used to address missing data, ensuring consistency across the dataset. The Land Use/Cover Change (LUCC) data were obtained from the China 30 m Resolution Land Cover Dataset (CLCD) [69], which provides high-resolution spatial data. These LUCC data were further processed using ArcGIS 10.8 software to ensure accuracy in spatial analysis.

In addition, a wide array of statistical resources was utilized to gather data across various domains. These included several Statistical Yearbooks, such as the China Statistical Yearbook, China Agricultural Yearbook, China Rural Statistical Yearbook, China Information Industry Yearbook, China Environmental Statistical Yearbook, China Health Statistical Yearbook, and China Science and Technology Statistical Yearbook.

Beyond yearbooks, data were also sourced from various Data Platforms, including the EPS Data Platform and the Pipi Shrimp Database [78], which provided supplementary information for certain variables. Furthermore, relevant government statistics were gathered from the National Bureau of Statistics of China and the China Administrative Division Information Inquiry Platform, ensuring that official and authoritative data sources were incorporated into the analysis.

Logarithmic transformations were applied to some variables to improve the accuracy of model estimations and to address potential issues with heteroscedasticity in the data. All statistical analyses, including model estimations, were performed using Stata 17 software, a robust tool for conducting econometric analysis and handling complex panel data structures.

Table 3. Digital agriculture development levels in Chinese provinces from 2014 to 2022.

Province	2014	2015	2016	2017	2018	2019	2020	2021	2022
Beijing	0.082	0.091	0.090	0.100	0.109	0.126	0.118	0.114	0.119
Tianjin	0.091	0.093	0.080	0.080	0.069	0.078	0.079	0.078	0.079
Hebei	0.148	0.152	0.149	0.164	0.182	0.208	0.240	0.274	0.300
Shanxi	0.077	0.082	0.070	0.074	0.078	0.084	0.088	0.093	0.097
Inner Mongolia	0.118	0.128	0.130	0.134	0.124	0.101	0.107	0.115	0.118
Liaoning	0.100	0.107	0.110	0.114	0.101	0.096	0.094	0.100	0.105
Jilin	0.070	0.077	0.082	0.082	0.082	0.077	0.079	0.083	0.086
Heilongjiang	0.229	0.233	0.237	0.269	0.261	0.195	0.139	0.149	0.153
Shanghai	0.128	0.132	0.140	0.145	0.150	0.165	0.152	0.077	0.080
Jiangsu	0.244	0.263	0.278	0.290	0.319	0.350	0.366	0.322	0.329
Zhejiang	0.192	0.227	0.274	0.317	0.412	0.497	0.523	0.568	0.604
Anhui	0.087	0.097	0.106	0.115	0.131	0.140	0.155	0.170	0.180
Fujian	0.191	0.201	0.253	0.210	0.202	0.252	0.246	0.268	0.307
Jiangxi	0.095	0.101	0.117	0.114	0.114	0.130	0.135	0.142	0.155
Shandong	0.166	0.181	0.187	0.214	0.282	0.266	0.299	0.366	0.386
Henan	0.123	0.137	0.146	0.157	0.173	0.176	0.197	0.221	0.242
Hubei	0.111	0.116	0.129	0.141	0.139	0.143	0.170	0.170	0.189
Hunan	0.158	0.168	0.181	0.176	0.175	0.205	0.208	0.220	0.223
Guangdong	0.233	0.247	0.294	0.295	0.324	0.391	0.425	0.456	0.498
Guangxi	0.122	0.130	0.133	0.139	0.142	0.163	0.173	0.178	0.188
Hainan	0.062	0.060	0.060	0.064	0.065	0.070	0.077	0.084	0.092
Chongqing	0.072	0.075	0.082	0.088	0.092	0.105	0.144	0.158	0.141
Sichuan	0.267	0.269	0.281	0.292	0.302	0.315	0.321	0.304	0.334
Guizhou	0.095	0.104	0.108	0.113	0.126	0.144	0.147	0.144	0.144
Yunnan	0.227	0.243	0.258	0.269	0.273	0.258	0.263	0.266	0.287
Shaanxi	0.071	0.077	0.077	0.084	0.092	0.107	0.114	0.119	0.130
Gansu	0.087	0.092	0.092	0.099	0.110	0.115	0.119	0.117	0.124
Qinghai	0.066	0.069	0.071	0.074	0.076	0.079	0.080	0.071	0.079
Ningxia	0.056	0.057	0.050	0.050	0.051	0.053	0.056	0.059	0.060
Xinjiang	0.227	0.246	0.242	0.254	0.124	0.125	0.133	0.149	0.149

displays the development levels of digital agriculture across 30 Chinese provinces from 2014 to 2022. Overall, digital agriculture has shown a growing trend, particularly in the eastern regions where rapid growth is observed. The central regions follow with steady improvement, while the western regions, though relatively lagging, also exhibit positive growth. This indicates regional imbalances in the development of digital agriculture in China, warranting further attention and study.

Table 4. Descriptive statistics of variables.

Variable	Variable Name	Variable Code	N	Mean	p50	SD	Min	Max
Dependent Variable	Coordinated Development Level of Ecosystem Services and Human Well-Being	CD-ESWB	270	0.261	0.229	0.111	0.0830	0.596
Core Explanatory Variable	Digital Agriculture Development Level	LDAD	270	0.165	0.135	0.098	0.050	0.604
Control Variable	Urbanization Rate	UR	270	0.620	0.605	0.110	0.402	0.893
	Environmental Regulation Intensity	ERI	270	4.064	4.078	0.335	3.178	4.820
	Human Capital Level	LHC	270	7.929	7.900	0.275	7.107	8.599
	Financial Development Level	LFD	270	3.587	3.366	1.076	1.972	7.618
Mediating Variable	Technological Innovation Effect	TIE	270	0.158	0.115	0.165	0.002	0.885
Moderating Variable	Industrial Structure Upgrade	ISU	270	1.450	1.277	0.764	0.704	5.283

offers a detailed account of the descriptive statistics pertaining to the variables in question. The data indicate differences in the development levels of digital agriculture and the CD-ESWB across regions. The mean (0.261) and median (0.229) of CD-ESWB are close, indicating a relatively even data distribution without significant skewness. Similarly, the mean (0.165) and median (0.135) of the LDAD are also close, showing that the data distribution is fairly uniform. After the logarithmic transformation of certain variables, the value differences were reduced, effectively mitigating heteroscedasticity issues.

4. Research Results and Analysis

4.1. Analysis of the Coordinated Development Level of Ecosystem Services and Human Well-Being

This study utilizes the coupling coordination degree model to determine the extent of coordinated development between ecosystem services and human well-being across 30 Chinese provinces from 2014 to 2022. The findings are depicted in Figure 2. A higher degree of coupling coordination reflects greater harmony between the systems. In general, the level of coordinated development between ecosystem services and human well-being in China steadily increased, with the national average rising from 0.430 in 2014 to 0.485 in 2022. This reflects a sustained enhancement in the coordination between the two systems. Nevertheless, considerable regional disparities remain.

Specifically, the coordination level in the eastern region remained consistently below the national average, with slow growth, increasing from 0.352 in 2014 to 0.378 in 2022, with fluctuations during this period. This suggests that the more economically developed eastern regions face certain challenges in coordinating ecosystem services and human well-being, likely due to the pressure of rapid economic development on environmental protection. Research and assessments conducted on the ecological efficiencies of eastern China’s major metropolitan areas reveal how rapid economic development can indeed place significant stress on environmental resources, complicating the balance between ecosystem services and human well-being. For instance, in the economically vibrant urban clusters such as Beijing–Tianjin–Hebei, the Yangtze River Delta, and the Greater Bay Area, industrial and urban expansion have consistently elevated pollution levels and intensified resource consumption, even as these regions strive to implement ecological measures [79]. Furthermore, the evaluation of ecological efficiencies across these coastal clusters indicates that areas like the Pearl River Delta, while achieving considerable economic growth, have periodically faced declines in ecological efficiency due to high emissions and resource-intensive practices. This recurring impact on environmental health within economically advanced clusters like Beijing–Tianjin–Hebei underscores the trade-offs between economic ambitions and environmental protection efforts [80]. The central region showed a more noticeable improvement, with the coordination level rising from 0.431 in 2014 to 0.501 in 2022, gradually approaching the national average. The western region demonstrated the highest coordination level, consistently leading other regions. The coordination level increased from 0.506 in 2014 to 0.582 in 2022, exhibiting significant growth. This improvement can be attributed to several factors: Abundant Natural Resources: The western region is rich in natural resources, including vast forests, grasslands, and wetlands, which provide a strong foundation for ecological conservation efforts. Support from National Ecological Protection Policies: The Chinese government has implemented various policies aimed at protecting and restoring the ecological environment in the western region. For instance, initiatives like the Grain for Green Program and the establishment of national parks have significantly contributed to ecological preservation [81]. Lower Economic Development Pressure: Compared to the eastern regions, the western region experiences less industrialization and urbanization pressure, allowing for more sustainable development practices and less environmental degradation. Research conducted by Tsinghua University’s China Institute of New Urbanization further supports this view, highlighting that urbanization levels in China are significantly higher in the eastern regions compared to the west [82]. This disparity stems from differences in economic development and regional characteristics, aligning with the observation that the western region experiences less industrialization and urbanization pressure. Collectively, these factors have endowed the western region with distinct advantages for enhancing the coordination between ecosystem preservation and human well-being. This favorable environmental and policy context has promoted a notable improvement in achieving balanced development goals that align with sustainable ecological and social outcomes.

To analyze the specific situation of each province in more detail,

Table 5. Classification standards for the Coordinated Development Level of Ecosystem Services and Human Well-Being.

Coupling Coordination Degree	Level	2014	2022
0 < D ≤ 0.1	Extreme Imbalance		Shanghai
0.1 < D ≤ 0.2	Severe Imbalance	Tianjin, Shanghai	Tianjin
0.2 < D ≤ 0.3	Moderate Imbalance	Tianjin	Beijing
0.3 < D ≤ 0.4	Mild Imbalance	Hebei, Liaoning, Anhui, Shandong, Henan, Hainan, Chongqing, Ningxia	Shandong, Henan, Hainan, Ningxia
0.4 < D ≤ 0.5	Near Imbalance	Shanxi, Jilin, Jiangsu, Zhejiang, Fujian, Jiangxi, Hubei, Hunan, Guangdong, Guangxi, Guizhou, Shaanxi, Gansu	Hebei, Shanxi, Liaoning, Jilin, Jiangsu, Anhui, Chongqing, Guizhou
0.5 < D ≤ 0.6	Barely Coordinated	Heilongjiang, Sichuan, Yunnan	Zhejiang, Fujian, Jiangxi, Hubei, Hunan, Guangdong, Guangxi, Shaanxi, Gansu
0.6 < D ≤ 0.7	Primary Coordination	Qinghai, Xinjiang	Heilongjiang, Sichuan, Yunnan
0.7 < D ≤ 0.8	Intermediate Coordination	Inner Mongolia	Qinghai, Xinjiang
0.8 < D ≤ 0.9	Good Coordination		Inner Mongolia
0.9 < D ≤ 1	High Coordination

lists the coupling coordination levels of each province in 2014 and 2022, along with the changes over time.

Table 5 provides a classification and evolutionary analysis of the coordination levels of various provinces. Overall, from 2014 to 2022, the coupling coordination levels of the provinces improved, though significant regional disparities remain. Eastern provinces such as Shanghai and Tianjin were still at lower levels in 2022, classified as “Extreme Imbalance” and “Severe Imbalance,” respectively, indicating notable challenges in coordinated development in these economically developed regions. Beijing remained in the “Moderate Imbalance” category. In the central area, provinces such as Shandong, Henan, Hainan, and Ningxia were categorized as “Mild Imbalance” in 2022, showing relatively slow progress. The western region saw the greatest progress, with Inner Mongolia rising from “Intermediate Coordination” to “Good Coordination” and both Qinghai and Xinjiang shifting from “Primary Coordination” to “Intermediate Coordination”. This indicates substantial advancements in the coordination between ecosystem services and human well-being in these areas.

Significant progress was observed in the western region, where Inner Mongolia advanced from “Intermediate Coordination” to “Good Coordination”, while Qinghai and Xinjiang moved from “Primary Coordination” to “Intermediate Coordination”. This reflects notable improvements in the alignment of ecosystem services and human well-being within these regions.

Despite overall improvements, the coupling coordination levels reveal distinct regional challenges and opportunities across China’s eastern, central, and western regions. Shanghai and Tianjin, two of China’s most economically advanced cities, remain in the “Extreme Imbalance” and “Severe Imbalance” categories, respectively. This reflects significant challenges in aligning rapid economic growth with sustainable ecosystem services. High levels of industrialization and urbanization contribute to considerable ecological pressures, as observed in studies where rapid economic gains exacerbate environmental degradation and place immense stress on ecological balance [83]. Despite advances in digital and technological agriculture, these efforts are insufficient to offset the environmental demands of densely populated and highly industrialized urban centers [84]. Consequently, eastern provinces face complex dynamics in balancing economic output with ecosystem preservation, a task that requires innovative, resource-efficient policies targeting urban ecology and industrial emissions.

Provinces such as Shandong, Henan, Hainan, and Ningxia are classified as having a “Mild Imbalance” in 2022, indicating steady yet limited progress. These provinces are predominantly agricultural and possess a relatively robust ecological base, but they face structural challenges in fully integrating sustainable practices within their agricultural economies. The slower advancement in coordination levels suggests a need for enhanced investment in sustainable agricultural practices and infrastructure improvements, as well as supportive policies that facilitate the adoption of green technologies. Given their comparative advantage in agriculture, central provinces are well positioned to strengthen ecosystem services through resource-conserving agricultural modernization, which would also contribute to reducing interregional development gaps [83].

The western region demonstrates the most significant progress, with Inner Mongolia advancing from “Intermediate Coordination” to “Good Coordination” and Qinghai and Xinjiang moving from “Primary Coordination” to “Intermediate Coordination.” These advancements can be attributed to lower population densities and vast natural resources, which ease pressures on ecosystem services compared to the eastern and central regions. Policy support, including ecological conservation incentives, has facilitated better integration of ecosystem services and human well-being [65]. By continuing to leverage these advantages and expanding targeted technological support, western provinces can serve as models for sustainable growth that prioritizes both environmental and human well-being objectives. However, sustained progress will require addressing socioeconomic challenges, such as limited infrastructure and workforce skill gaps, to maintain and further enhance coordination levels [83].

Therefore, to achieve sustainable development nationwide, differentiated policies tailored to specific regional conditions are needed to further promote coordinated development across regions.

4.2. Baseline Regression Analysis

A regression analysis was performed to assess the effect of LDAD on CD-ESWB. The Hausman test was supported by the fixed effects model for the baseline regression. The Variance Inflation Factor (VIF) test revealed a maximum value of 2.71 and an average of 1.71, both under 10, indicating the absence of multicollinearity in the model. Given the Hausman test results and the higher R² value, the two-way fixed effects model was selected as the baseline.

As indicated in

Table 6. Baseline regression results.

Variable	(1)	(2)	(3)	(4)
	CD-ESWB
LDAD	0.364 **	0.358 ***	0.620 ***	0.466 ***
	(0.154)	(0.112)	(0.165)	(0.130)
Cons	0.171 ***	−1.474 ***	0.158 ***	−0.722 ***
	(0.021)	(0.387)	(0.027)	(0.167)
Control Variables	No	Yes	No	Yes
Individual/Time Fixed Effects	Yes	Yes	No	No
Robustness Adjustment	Yes	Yes	Yes	Yes
N	270	270	270	270
R2	0.536	0.651	0.604	0.480
RMSE	0.084	0.129	0.074	0.081

Note: Standard errors in parentheses; ** p < 0.05, *** p < 0.01.

, the LDAD coefficient in column (1) is 0.364, and it is statistically significant at the 5% level. When control variables are introduced in column (2), the LDAD coefficient shifts to 0.358, with its significance increasing to the 1% level. Adding control variables increases the R² from 0.536 to 0.651, indicating an improvement in the model’s explanatory power. The RMSE values in columns (1) and (2) are 0.084 and 0.129, respectively, with the lower RMSE indicating higher prediction accuracy in column (1) than in column (2). Thus, the results in column (2) are more indicative. Whether control variables are included or a random or fixed effects model is used, the positive influence of LDAD on CD-ESWB remains statistically significant, affirming the beneficial role of digital agriculture in fostering coordinated development.

The results in column (2) indicate that for every one-unit increase in LDAD, the CD-ESWB increases by 0.358 units. This result is significant both statistically and economically, suggesting that the development of digital agriculture significantly promotes the coordination between ecosystem services and human well-being. This may be closely related to the ability of digital agriculture to enhance ecosystem services through optimizing resource allocation, improving production efficiency, and promoting environmentally friendly agricultural practices.

Specifically, digital agriculture applies precision technologies, including those derived from big data and the IoT, enabling more refined management of agricultural production processes. This optimizes provisioning services (e.g., food production, water resource management) and regulating services (e.g., climate regulation, pollution control). Not only does this improve resource use efficiency, but it also reduces negative environmental impacts, thereby enhancing the contribution of ecosystem services to human well-being. The development of digital agriculture also promotes sustainability in agriculture, such as reducing dependence on fertilizers and pesticides, which improves environmental quality and boosts regulating services (e.g., air and water purification).

Furthermore, digital technologies enhance cultural services by promoting eco-tourism and leisure agriculture, not only stimulating local economies but also enriching residents’ cultural and life experiences, indirectly improving human well-being.

In summary, the baseline regression results confirm Hypothesis H1: digital agriculture significantly affects the CD-ESWB.

4.3. Endogeneity Issues

In addressing potential endogeneity issues, particularly the risk of reverse causality that could introduce bias into the analysis, this study adopts a methodological approach inspired by Zhang Xun and colleagues [85]. Specifically, the study utilizes the spherical distance between provincial capitals or municipalities and Hangzhou as an instrumental variable. This choice is grounded in two key considerations. First, Hangzhou is recognized as a significant hub for the digital economy in China, enjoying a pioneering advantage in the development of digital economic initiatives. The geographical distance from Hangzhou can serve as a proxy for the potential for digital agriculture development in various provinces. This relationship meets the relevance requirement for instrumental variables, as it is expected that closer proximity to a center of innovation like Hangzhou may correlate with greater advancements in digital agriculture. Second, it is crucial that the chosen instrumental variable does not have a direct effect on the coordinated development of ecosystem services and human well-being. In this case, geographical distance satisfies the homogeneity requirement, as it is unlikely to directly influence these broader development outcomes.

Given these parameters, the methodology employed in this study aligns with the framework established by Huang Qunhui et al. [86], which is specifically designed for balanced panel data in conjunction with cross-sectional geographical distance data. To further enhance the analysis, an interaction term is constructed by multiplying the spherical distance from each provincial capital or municipality to Hangzhou with the number of internet users nationwide from the previous year. This interaction term functions as the instrumental variable for LDAD (Land Use/Cover Change), effectively capturing regional disparities in digitalization while addressing potential endogeneity.

The results of the instrumental variable regression analysis are presented in

Table 7. Instrumental variable regression results.

Variable	(1)	(2)
	First	Second
	LDAD	CD-ESWB
LDAD		0.302 ***
		(3.46)
IV	−0.000 ***
	(−6.80)
Cons	−0.201	−1.820 ***
	(−0.59)	(−8.00)
Control Variables	Yes	Yes
Individual/Time Fixed Effects	Yes	Yes
Robustness Adjustment	Yes	Yes
Kleibergen–Paap rk LM statistic	25.630 ***
Cragg–Donald Wald F statistic	74.798
Kleibergen–Paap rk Wald F statistic	46.262
N	270	270
R2	0.248	0.954
RMSE	0.117	0.024

Note: Standard errors in parentheses; *** p < 0.01.

, providing insights into the robustness of the findings and the effectiveness of the chosen methodology in mitigating endogeneity concerns.

Table 7 presents the results from the two-stage instrumental variable (IV) regression. The first stage, shown in column (1), estimates the relationship between the instrumental variable (IV) and LDAD (Land Use/Cover Change), while the second stage, shown in column (2), explores the effect of LDAD on CD-ESWB (Coordinated Development of Ecosystem Services and Human Well-Being). The findings from both stages are significant and align with theoretical expectations.

In terms of instrumental variable validity, several key tests were performed. The Kleibergen–Paap rk LM statistic is 25.63, which is significant at the 1% level, allowing us to reject the null hypothesis of “under-identification.” This indicates that the instrumental variable is correctly identified. Furthermore, the Cragg–Donald Wald F statistic (74.798) and the Kleibergen–Paap rk Wald F statistic (46.262) are both significantly higher than the Stock–Yogo critical value of 16.38 at the 10% level, effectively rejecting the null hypothesis of “weak identification.” Thus, these tests demonstrate that the chosen instrumental variable is both relevant and strong, validating its use in this context.

In the first-stage regression, the coefficient of the instrumental variable on LDAD is −0.000, and it is highly significant (p < 0.01). This result confirms a robust negative relationship between the spherical distance from Hangzhou and the level of digital agriculture development (LDAD), reflecting the influence of geographic proximity to Hangzhou, a key digital economy hub, on regional digital agriculture potential.

In the second-stage regression, the coefficient of LDAD on CD-ESWB is 0.302, which is also significant at the 1% level. This result indicates that advancements in digital agriculture have a substantial positive effect on the coordinated development of ecosystem services and human well-being. In other words, regions with higher levels of digital agriculture tend to exhibit better alignment between environmental sustainability and human welfare.

In summary, the instrumental variable regression confirms that the endogeneity issue has been effectively addressed, with both stages of the analysis yielding statistically significant results. The use of the instrumental variable—spherical distance from Hangzhou—proves to be valid, and the results are robust, reinforcing the conclusion that digital agriculture plays a critical role in enhancing the synergy between ecosystem services and human well-being.

4.4. Robustness Checks

To ensure the reliability and validity of the findings, a series of robustness checks were conducted, which are outlined as follows:

4.4.1. Adjusting the Time Sample Interval

Recognizing that the COVID-19 pandemic in 2020 may have influenced the coordination between digital agriculture and ecosystem services as well as human well-being [87,88], the analysis was performed again by excluding data from 2020 to 2022. The focus shifted to the sample period from 2014 to 2019. The results presented in column (1) of

Table 8. Robustness test results.

Variable	Adjusting Time Sample	Adjusting Regional Sample	Replacing Explanatory Variable
	(1)	(2)	(3)
	CD-ESWB
LDAD	0.339 ***	0.293 ***	0.073 ***
	(0.099)	(0.093)	(0.009)
Cons	−0.560 **	−1.486 ***	−0.674 **
	(0.268)	(0.419)	(0.283)
Control Variables	Yes	Yes	Yes
Individual/Time Fixed Effects	Yes	Yes	Yes
Robustness Adjustment	Yes	Yes	Yes
N	180	234	270
R2	0.646	0.670	0.785
RMSE	0.087	0.088	0.076

Note: Standard errors in parentheses; ** p < 0.05, *** p < 0.01.

indicate that the coefficient for LDAD (Land Use/Cover Change) is 0.339, which remains statistically significant at the 1% level. This finding underscores that the positive impact of digital agriculture on the coordinated development of ecosystem services and human well-being persists even when pandemic-related data are omitted, reinforcing the reliability of the original conclusions.

4.4.2. Adjusting the Regional Sample Interval

To further strengthen the analysis, municipalities known for their significant economic scale, development level, and policy strength—namely Beijing, Shanghai, Tianjin, and Chongqing—were excluded from the sample. The analysis was then rerun, and the results shown in column (2) of Table 8 reveal an LDAD coefficient of 0.293, also significant at the 1% level. This consistency in results indicates that the initial findings are robust and not unduly influenced by the presence of these major urban centers, thus affirming the generalizability of the model.

4.4.3. Replacing the Explanatory Variable

To mitigate potential biases associated with a single measurement approach, principal component analysis (PCA) was employed to reassess the LDAD. The Kaiser–Meyer–Olkin (KMO) value was calculated at 0.747, and Bartlett’s test for sphericity yielded significant results, confirming the appropriateness of the data for factor analysis. Following the extraction of principal components, the LDAD was recalculated and reintegrated into the baseline model. As indicated in column (3) of Table 8, the LDAD coefficient stands at 0.073, which is significant at the 1% level. This result further validates the robustness of the model, affirming that the positive influence of digital agriculture remains intact regardless of the measurement method used.

In summary, all three robustness tests consistently demonstrate that the positive impact of digital agriculture development on the coordinated relationship between ecosystem services and human well-being remains statistically significant. Whether through adjustments to the time or regional samples or by employing different measurement methods for the explanatory variable, the findings are robust across various conditions. The high R² values—0.646, 0.670, and 0.785—indicate that the model possesses strong explanatory power, further validating the research conclusions. Overall, these robustness test results reaffirm the significant positive effect of digital agriculture on enhancing the synergy between ecosystem services and human well-being, solidifying the integrity of the research outcomes.

4.5. Further Discussion

4.5.1. Mediation Effect Analysis

This section extends the analysis by exploring how digital agriculture impacts the coordinated development of ecosystem services and human well-being through technological innovation. To investigate this indirect effect, TIE is introduced as a mediating variable in the baseline model. By incorporating TIE, this study aims to determine whether digital agriculture enhances technological innovation, which in turn contributes to the coordination between ecosystem services and human well-being. The mediation effect regression results are presented in

Table 9. Mediation effect regression results.

Variable	(1)	(2)
	TIE
LDAD	0.611 ***	0.569 ***
	(0.206)	(0.188)
Cons	0.033	0.264
	(0.033)	(0.336)
Control Variables	No	Yes
Individual/Time Fixed Effects	Yes	Yes
Robustness Adjustment	Yes	Yes
N	270	270
R2	0.642	0.704
RMSE	0.127	0.145

Note: Standard errors in parentheses; *** p < 0.01.

, providing insights into the direct and indirect relationships between these variables.

As shown in Table 9, the regression coefficients of LDAD on TIE are positive and statistically significant at the 1% level in both column (1) and column (2). In column (2), the LDAD coefficient is 0.569, indicating that digital agriculture development significantly fosters technological innovation. In particular, for each one-unit increase in LDAD, the TIE increases by approximately 0.569 units, which underscores the significant contribution of digital agriculture to technological innovation.

Technological innovation acts as a key driver of social, economic, and environmental benefits, directly impacting the integrated development of ecosystem services and human well-being [89]. Previous research suggests that by improving ecological efficiency, technological innovation not only reduces resource consumption and pollutant emissions but also effectively enhances environmental quality, laying the foundation for improving ecosystem services [48]. Increased ecological efficiency optimizes resource utilization while enhancing provisioning and regulating services. For example, more efficient resource management and the use of clean energy strengthen ecosystem functions, indirectly promoting human health, quality of life, and economic growth [90]. Particularly, the application of green technologies has significantly improved energy cleanliness and decarbonization, enhancing ecosystem services such as climate regulation and water resource management, thereby improving public health and well-being [91]. Furthermore, technological innovation has driven the widespread adoption of energy-saving and emission reduction technologies, improving ecosystem service quality and advancing the sustainability agenda [92].

On the economic and social fronts, technological innovation improves productivity and reduces costs, significantly increasing resource allocation efficiency, which directly improves living conditions and well-being [93]. In the realm of digital agriculture, technological innovation has made agricultural production more precise and efficient, reducing natural resource waste and further optimizing ecosystem services [90]. Moreover, with the widespread application of technologies, including big data and artificial intelligence, technological innovation facilitates the scientific management of land resources, thereby enhancing the efficiency of land resource allocation [94]. Additionally, technological innovation creates new job opportunities and industry models, broadening income channels, alleviating economic pressures on small and medium-sized enterprises, mitigating poverty, and advancing social equity and sustainable economic growth [95,96,97].

In conclusion, technological innovation is crucial in the relationship between digital agriculture, ecosystem services, and human well-being. Digital agriculture not only directly promotes the coordinated development of ecosystem services and human well-being but also indirectly enhances the overall social, economic, and environmental benefits through technological innovation, providing strong support for achieving sustainable development goals. This analysis further validates Hypothesis H2, which states that digital agriculture significantly contributes to the coordinated development of ecosystem services and human well-being through technological innovation.

4.5.2. Moderation Effect Analysis

This study incorporates ISU as a moderating variable to further investigate how digital agriculture influences CD-ESWB. An interaction term between LDAD and ISU is introduced for regression analysis, with results presented in

Table 10. Moderation effect regression results.

Variable	(1)	(2)
	CD-ESWB
LDAD	0.380 **	0.347 ***
	(0.139)	(0.096)
LDAD*ISU	0.538 *	0.595 ***
	(0.293)	(0.180)
Cons	0.169 ***	−1.431 ***
	(0.020)	(0.283)
Control Variables	No	Yes
Individual/Time Fixed Effects	Yes	Yes
Robustness Adjustment	Yes	Yes
N	270	270
R2	0.575	0.692
RMSE	0.086	0.140

Note: Standard errors in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01.

.

As shown in Table 10, the regression results demonstrate that the coefficient of the interaction term LDAD*ISU is positive in both columns (1) and (2) and significant at the 10% and 1% levels, respectively. This indicates that, regardless of the inclusion of control variables, ISU exerts a positive moderating effect on the relationship between digital agriculture development and CD-ESWB. In particular, the coefficient of the interaction term in column (2) is 0.595, suggesting that as the level of industrial structure upgrade increases, the positive impact of digital agriculture on the coordinated development of ecosystem services and human well-being is further strengthened.

This enhancement may occur because industrial structure upgrades help shift resources from less efficient sectors to more productive industries, enabling digital agriculture to utilize resources more effectively. As a result, the coordination between ecosystem services and human well-being is improved. For example, as digital technologies are more widely applied in agriculture, production processes are optimized, leading to increased efficiency in resource use and a reduction in environmental pollution. ISU supports this transition by fostering a more conducive environment for technological advancements in agriculture, thus reinforcing the beneficial effects of digital agriculture.

Moreover, industrial structure upgrades are often accompanied by policy incentives that attract both investment and skilled talent to the digital agriculture sector. These incentives not only accelerate the adoption of digital technologies but also amplify their positive impact on ecosystem services and human well-being by promoting sustainable and efficient agricultural practices.

In conclusion, the findings of this study confirm that digital agriculture positively influences the coordinated development of ecosystem services and human well-being. This supports Hypothesis 3 (H3), which posits that the integration of industrial structure upgrades with digital agriculture can enhance the relationship between ecosystem services and human well-being.

4.5.3. Heterogeneity Analysis

A heterogeneity analysis was conducted to explore the regional variations in the relationship between digital agriculture and CD-ESWB further. The results, presented in

Table 11. Heterogeneity analysis results.

Variable	Eastern Region	Central Region	Western Region
	(1)	(2)	(3)
	CD-ESWB
LDAD	0.494 ***	0.452 ***	0.142
	(0.153)	(0.107)	(0.184)
Cons	−1.944	−0.163	−0.808 **
	(1.138)	(0.574)	(0.281)
Control Variables	Yes	Yes	Yes
Individual/Time Fixed Effects	Yes	Yes	Yes
Robustness Adjustment	Yes	Yes	Yes
N	99	72	99
R2	0.685	0.880	0.668
RMSE	0.089	0.202	0.143

Note: Standard errors in parentheses; ** p < 0.05, *** p < 0.01.

, examine the effects of LDAD across three different regions of China: the eastern, central, and western regions.

The findings reveal significant regional differences. In the eastern region, the coefficient for LDAD is 0.494, and in the central region, it is 0.452, both of which are statistically significant at the 1% level. This indicates a strong positive relationship between digital agriculture and coordinated development in these regions. Specifically, for each one-unit increase in LDAD, the level of coordinated development between ecosystem services and human well-being rises by approximately 0.494 units in the eastern region and 0.452 units in the central region. These results suggest that digital agriculture has a substantial impact on improving coordination in these areas, likely due to more advanced technological infrastructure, stronger economic growth, and greater policy support.

In contrast, the results for the western region are notably different. The coefficient for LDAD is 0.142, and it is not statistically significant, indicating a weaker or negligible effect of digital agriculture on the coordinated development of ecosystem services and human well-being in this region. The lack of significance in the western region may be attributed to several factors, including slower economic development, less developed digital infrastructure, and fewer resources available for implementing and scaling digital agricultural technologies.

These heterogeneity results underscore the importance of regional context in understanding the impact of digital agriculture. The eastern and central regions, with their more mature digital economies and better access to technological resources, are able to leverage digital agriculture more effectively to promote coordination between environmental sustainability and human well-being. In contrast, the western region faces greater challenges in this regard, potentially due to geographical, economic, and infrastructural limitations that hinder the full realization of digital agriculture’s benefits.

The regional differences observed can be explained by multiple factors, including economic development, infrastructure quality, policy support, industry structure, and geographical conditions. Economic Development Levels: Eastern and central regions benefit from stronger economic foundations that allow for higher investments in digital agriculture, thus enhancing its positive impact on CD-ESWB. In these regions, the significant LDAD coefficients (0.494 and 0.452, respectively) reflect that economically robust areas can allocate more resources toward adopting advanced technologies, supporting precise resource management and sustainable agricultural practices [83,98]. Infrastructure and Technology Adoption: The eastern and central regions also enjoy superior digital infrastructure, with greater access to technologies like data collection, IoT, and precision management in agriculture. This digital foundation helps these regions leverage digital agriculture effectively for ecological benefits. By comparison, the western region’s digital infrastructure is less developed, contributing to a lower and non-significant LDAD coefficient (0.142), as limited infrastructure restricts the region’s ability to adopt digital practices in agriculture [84,99]. Policy Support and Resource Allocation: Policy and financial support are distributed unevenly across regions, with eastern areas receiving more targeted funding and clearer policy directives to support digital agriculture. Central regions also benefit from substantial policy support, though slightly less than the east. Western areas, however, receive limited policy resources for digital agriculture, weakening its role in improving CD-ESWB [83,98,100]. Industry Structure and Agricultural Practices: The eastern and central regions have more modernized, market-oriented agricultural systems, which allow digital tools to align more closely with ecological management needs. In contrast, the west predominantly relies on traditional farming, making it difficult for digital agriculture to penetrate small-scale, dispersed farms and thus diminishing its impact on ecosystem services [100,101]. Geographical and Environmental Constraints: Western China’s mountainous terrain and harsher climate conditions further hinder the integration of digital technologies with agricultural and ecological management. Such challenging geography complicates large-scale deployment of digital agriculture, limiting its contribution to CD-ESWB compared to the more favorable conditions in the east and central regions [83,99].

In summary, the heterogeneity analysis highlights the varying degrees of impact that digital agriculture has across different regions in China. The results suggest that policy interventions aimed at promoting digital agriculture may need to be tailored to regional needs, with a focus on improving digital infrastructure and support in the western region to enhance its development potential.

Figure 3 illustrates the mean distribution of CD-ESWB from 2014 to 2022 to better understand regional differences. Western provinces (such as Qinghai, Xinjiang, and Inner Mongolia) exhibit higher CD-ESWB levels, indicating relatively better ecological environments and lower human interference. On the other hand, eastern coastal provinces (such as Shanghai, Tianjin, and Beijing) show relatively lower CD-ESWB levels, which may be attributed to the significant ecological pressure caused by rapid economic development and intensive human activities in these regions.

In comparison, Figure 4 shows the mean distribution of LDAD during the same period. Eastern coastal regions (such as Zhejiang, Jiangsu, and Guangdong) exhibit higher LDAD levels, driven by strong economic power, robust policy support, and advanced technological capabilities. In contrast, LDAD levels are relatively lower in central and western regions, likely due to incomplete digital infrastructure and slower economic development, which limit the adoption and application of digital agriculture technologies.

By comparing the distribution patterns in Figure 3 and Figure 4, it becomes evident that the relationship between CD-ESWB and LDAD is not a simple positive correlation. The western region demonstrates higher CD-ESWB levels but lower LDAD levels, suggesting that digital agriculture development lags while the coordination between ecosystem services and human well-being is relatively good. In contrast, the eastern regions have higher LDAD levels but relatively lower CD-ESWB levels, indicating that digital agriculture should emphasize optimizing ecosystem services and mitigating the negative environmental impacts of economic activities in these areas.

5. Conclusions

5.1. Summary and Conclusions

This paper explores the dual role of digital agriculture in fostering the harmonious integration of ecosystem services and human well-being. First, an evaluation index system for LDAD was developed to comprehensively measure development across dimensions such as infrastructure, macro-environment, green growth, and production efficiency. Second, based on panel data from 30 Chinese provinces spanning from 2014 to 2022, empirical analyses were conducted using various models, including the coupling coordination model, fixed effects model, mediation effect model, and moderation effect model, to evaluate the influence of digital agriculture on CD-ESWB. In addition, robustness checks and heterogeneity analyses were carried out to ensure the reliability of the results.

The main findings are as follows: (1) Digital agriculture plays a crucial role in enhancing coordination between ecosystem services and human well-being. Baseline regression analysis indicates a significant positive impact of LDAD on CD-ESWB, with each unit increase in LDAD associated with approximately a 0.358-unit improvement in coordination. Robustness tests, which involved adjusting time intervals, regional samples, and alternative measurement methods, further validated the stability of this positive relationship. (2) Technological innovation acts as an important intermediary. Digital agriculture indirectly facilitates the alignment of ecosystem services and human well-being by driving advancements in technology. The regression results reveal that LDAD positively affects TIE, with a coefficient of 0.569, highlighting the pivotal role of digital agriculture in fostering innovation. This technological innovation, as a mediating variable, further strengthens coordinated development, indicating that it serves as a key channel through which digital agriculture exerts its effects. (3) Industrial structure upgrading has a moderating influence. ISU enhances the relationship between digital agriculture and coordinated development, as evidenced by the interaction term LDAD*ISU, which has a coefficient of 0.595. This suggests that industrial upgrades amplify the positive effects of digital agriculture on the coordination between ecosystem services and human well-being. (4) Regional heterogeneity was also observed. The analysis shows that digital agriculture has a more pronounced impact on coordinated development in the eastern and central regions, with LDAD coefficients of 0.494 and 0.452, both significant at the 1% level. In contrast, the effect in the western region is not statistically significant, indicating a weaker relationship in this area.

5.2. Policy Recommendations

Based on the research findings, the following policy recommendations are proposed to promote the development of digital agriculture and achieve the coordinated advancement of ecosystem services and human well-being at both the macro and micro levels, with a particular focus on addressing regional disparities across China.

First, strengthening digital agriculture infrastructure is crucial, especially for underdeveloped provinces. In eastern regions such as Jiangsu, Zhejiang, and Guangdong, where digital agriculture development is already advanced, the focus should be on further upgrading technological infrastructure. Policies should prioritize extending smart farming technologies, such as precision farming, IoT-enabled irrigation systems, and AI-driven crop monitoring, to enhance productivity and sustainability. Additionally, fostering digital market platforms and e-commerce models can help these provinces better integrate their agricultural sectors with national and international markets.

For central and western regions, including Shaanxi, Gansu, and Guizhou, the primary focus should be on boosting digital infrastructure and closing the digital divide. These provinces lag in terms of broadband connectivity and access to smart farming technologies. Governments should prioritize expanding 4G/5G networks, improving broadband coverage in rural areas, and providing subsidies for the purchase of digital farming equipment. Special attention should also be given to developing region-specific digital technologies that align with local agricultural practices. For example, Shaanxi could focus on integrating smart irrigation systems for its dominant crop—wheat—while Guizhou, with its mountainous terrain, could benefit from drone technology for crop monitoring and pest management.

In northern provinces like Heilongjiang and Inner Mongolia, where large-scale farming and livestock agriculture are prevalent, policies should focus on supporting digital solutions for large-scale farming management, such as automated machinery for planting, harvesting, and crop management. These provinces can also leverage precision farming technologies, such as satellite-guided tractors, to improve crop yields and reduce resource wastage. Additionally, government initiatives should encourage the establishment of data-sharing platforms that can provide farmers with real-time weather, market prices, and pest control information to enhance decision-making.

Second, optimizing the agricultural industrial structure is critical for promoting green transformation across regions. In the eastern provinces, such as Jiangsu, Zhejiang, and Shandong, where agricultural productivity is higher, the focus should be on expanding emerging industries such as agricultural e-commerce, eco-tourism, and agri-tourism. Zhejiang, with its strong e-commerce presence, could foster further growth in agricultural online sales platforms, promoting organic and eco-friendly products. Shandong, as a major producer of fruits and vegetables, could develop its agricultural value chain by incorporating agro-processing industries and creating value-added products for export.

For central and western provinces like Hunan, Sichuan, and Yunnan, the priority should be to shift towards more sustainable agricultural practices. Hunan, with its strong agricultural base in rice production, could promote digital tools that support environmentally friendly practices such as precision irrigation and smart fertilization techniques. Sichuan, known for its diverse crops and livestock, could focus on improving the efficiency of its agricultural systems by adopting digital tools that reduce energy consumption and enhance resource utilization. Similarly, Yunnan, with its rich biodiversity and eco-tourism potential, could integrate green agricultural practices with eco-tourism, offering organic agricultural products that cater to both local and international markets.

Third, enhancing policy frameworks and providing financial support tailored to the unique needs of different regions is essential. In eastern provinces, policies should focus on incentivizing innovation in digital agriculture through tax incentives for technology firms, subsidies for R&D, and fostering collaborations between universities and businesses to develop cutting-edge agricultural technologies. Guangdong and Beijing, with their thriving tech sectors, could spearhead initiatives to develop AI and blockchain technologies for the agricultural sector, while Shanghai could lead the creation of data platforms to streamline agricultural supply chains and improve market access.

In central and western provinces, where financial constraints limit the adoption of digital technologies, governments should offer more targeted financial support. Chongqing, Henan, and Anhui, which have significant agricultural potential, would benefit from low-interest loans, subsidies for digital farming equipment, and the introduction of agricultural insurance tailored to digital agriculture. Additionally, offering financial support for farmers to participate in training programs would be crucial for ensuring the effective adoption of digital tools.

Fourth, regional coordination and cross-regional cooperation should be strengthened. To address the existing regional imbalances in digital agriculture development, the government should establish mechanisms for cooperation between the more developed eastern regions and the less developed central and western regions. For example, Beijing, with its strong digital infrastructure, can collaborate with Gansu and Xinjiang by providing technical support and training programs, enabling these provinces to leapfrog in digital agriculture development. Moreover, Shandong and Guangdong, with their advanced agricultural sectors, could serve as models for other provinces to replicate successful digital agriculture practices. Cross-regional digital agriculture demonstration zones could be established to showcase successful models and foster knowledge exchange. These zones could offer a platform for research institutions, businesses, and farmers from different provinces to collaborate and share best practices in the use of digital technologies.

In addition, governments should support the creation of farmer cooperatives and industrial alliances, particularly in Sichuan and Hunan, where small-scale farming remains predominant. These cooperatives can facilitate resource sharing, provide digital tools to farmers, and create economies of scale to ensure the profitability and sustainability of digital agriculture initiatives. Such collaboration can help mitigate regional inequalities by promoting collective growth and shared access to advanced technologies.

By implementing these province-specific measures, China can promote the growth of digital agriculture, bridge the regional digital divide, and ensure the coordinated advancement of ecosystem services and human well-being across the country. Tailored policies that take into account the unique agricultural characteristics, infrastructure gaps, and technological capabilities of each province will be essential for achieving sustainable agricultural development in the digital age.

5.3. Limitations and Further Research

Despite the achievements of this study, several limitations remain that require further refinement in future research. First, limitations remain related to data availability and completeness. This study analyzed panel data from 30 Chinese provinces between 2014 and 2022. Although the data are relatively comprehensive, the absence of certain data for specific regions and years may have affected the accuracy of the models. Future research could consider incorporating higher spatial and temporal resolution data to fill gaps and improve data quality. Second, limitations remain in the construction of the digital agriculture development indicator system. While the indicator system developed in this study covers dimensions such as infrastructure, the macro-environment, green development, and production efficiency, these indicators may not fully capture all aspects of digital agriculture due to data availability constraints. Future research could further refine the indicator system to provide a more comprehensive assessment of digital agriculture development. Third, the depth of regional difference analysis requires enhancement. Although this study conducted a heterogeneity analysis across eastern, central, and western regions, it did not delve into the specific causes behind these differences. The primary reason is that this study focuses on the overall national macro-level impact rather than regional details, and obtaining detailed data on regional policies, technology adoption, and industrial structure was challenging. Additionally, an in-depth analysis of the causes of regional differences was not pursued for the sake of research focus. Future research could explore how digital agriculture affects the coordinated development of ecosystem services and human well-being across different regions by incorporating regional characteristics.