Is Land Fragmentation Undermining Collective Action in Rural Areas? An Empirical Study Based on Irrigation Systems in China’s Frontier Areas

Abstract

A comprehensive understanding of the impact of land fragmentation on collective action is essential for rural governance in developing countries. Prior publications have argued that land fragmentation impedes the sustainable development of agricultural economies and rural societies, while the connection between humans and nature has not been considered comprehensively. Therefore, the conclusion that the impact of land fragmentation on collective action is purely negative may be one-sided. To examine this conclusion, this paper re-evaluates the relationship between land fragmentation and rural collective action from a multidisciplinary perspective. Based on a rural field survey using stratified random sampling, Oprobit regression was employed to conduct an econometric analysis on data from 798 rural households across 14 cities in the border region of Guangxi, China. The following research findings were obtained: (1) When the human–nature connection is considered, the relationship between land fragmentation and collective action follows an inverted U-shaped curve. Specifically, moderate initial increases in land fragmentation can lead to improvements in collective action; however, when the degree of land fragmentation exceeds a certain threshold, further increases in land fragmentation will decrease the collective action capacity. (2) This nonlinear relationship between land fragmentation and collective action may be realized through three pathways: agricultural production, land use patterns, and the ecological environment. Under the context of collective action, this study shows that a moderate level of land fragmentation objectively exists in reality. This insight provides a new impetus for developing countries to shift policy perspectives to increase their land use efficiency. Additionally, this paper integrates relevant findings from both social sciences and natural sciences. Thereby, it not only expands the existing understanding of key factors influencing rural household collective action but also emphasizes the potential for cross-disciplinary integration between social sciences and natural sciences.

1. Introduction

Rural decline is a substantial challenge for the urbanization process of developing countries worldwide [1]. The general decline of public affairs governance is considered an essential cause of rural decline in developing countries, as it involves a variety of issues such as education, infrastructure, healthcare, and ecology [2]. At the same time, a lack of effective governance mechanisms for all things public has led to injustice and corruption. Collective action is the core issue of public affairs governance [3], and the lack of resources and collective action mechanisms has led to a decline in rural commons affairs governance in developing countries [4]. For example, the lack of mechanisms for public affairs governance has led to a shortage of infrastructure and public services in rural areas and effective regulatory mechanisms that would otherwise ensure the proper allocation and use of rural public resources. Therefore, it is essential to summarize the characteristics and rules of collective action for the rural areas of developing countries to improve the prevailing failure of public affairs governance.

Land fragmentation (LF) is a form of land management with the following characteristics resulting from the effects of natural, social, and economic factors on land: a large number of land parcels, small and varying individual plot sizes, inconsistent fertility level, and varying household proximity to the parcel [5]. In less developed agricultural systems, LF is the prevalent form of land use and has been recognized by mainstream research as one of the causes of the general decline of rural collective action capacity [6]. Specifically, Zang et al. (2019) [7] found that LF negatively affects farmers’ participation in collective action on irrigation; this result holds even when controlling for theoretical variables such as social and institutional rules, physical attributes of resources, community attributes, and household characteristics. Wang et al. (2020) [6] specifically examined the mechanisms by which LF influences collective action, showing that LF exerts a pronounced negative influence on collective action. This negative influence can be primarily attributed to factors such as agricultural dependency, formulation of irrigation systems, economic pressures, and land transfer.

Although current mainstream research concluded that the impact of LF on the economy and society is adverse, with the development of interdisciplinary research and the growing understanding of social–ecological systems, more and more scholars are beginning to reexamine the systemic value of LF. Such reexaminations include the ecological value of land, which has previously been ignored by social science research. A growing number of studies have shown that moderate LF actually contributes to agricultural diversification, food security, and sustainable land development [8]. Also, research on land ecology has suggested that smaller plots may be more suitable for diversified agricultural systems, such as agroforestry, horticulture, and organic farming, as they contribute to the overall productivity and sustainability of agricultural systems [9].

As a result, natural sciences research has led to the recognition that LF has both beneficial and detrimental effects on economic and social development. Further, the existing mainstream conclusion that the impact of LF on collective action is purely negative may be biased. This paper argues that the main reason for this biased conclusion is that prior research has not fully integrated knowledge from the social sciences with that from the natural sciences. This fragmentation between both disciplines has limited discussions of the role of LF on collective action and the mechanisms of its impact. To overcome this limitation, this paper addresses the following research questions: (1) Does LF decrease collective action? (2) Is the relationship between LF and collective action necessarily linear? In light of the aforementioned content, this paper examines the relationship between LF and collective action in a systematic manner, building on research data obtained from 798 farmers across 14 cities in Guangxi, China. Considering the connection between humans and nature, this paper further explores the potential mechanisms through which LF affects collective action. This exploration is based on crucial factors that influence collective action, including land systems, irrigation facilities, agricultural production, general institutional rules, and village economy.

This paper offers the following two contributions: First, this examination starts from human intervention in the ecological environment. By adding an analysis of relevant knowledge in natural sciences, both the relationship and influence mechanism between LF and collective action are then reexamined from multiple disciplines (i.e., agricultural economics, land management, sociology, geography, and resource environment). The key contribution is that the relationship between LF and collective action presents an “inverted U-shape”, thus further extending and expanding relevant research in this field. Second, based on data from Chinese rural households, the views and perceptions of residents are fully considered. Moreover, a detailed analysis of how rural residents and other stakeholders maintain the sustainability of land ecosystems through collective action is also presented. These considerations further bridge the gap in the research on how rural residents contribute to the diversification of socio-ecological conditions in the field of natural sciences. Third, the impact of LF on collective action is studied against different social and cultural contexts. For example, analyses of village topography, location, land transfer, village harmony, and heterogeneous experiences of natural disasters are employed to explain the causes and differences of the relationship between LF and collective action impacts. These contributions provide scientific references and suggestions for policymakers.

2. Literature Review

2.1. The Causes of Land Fragmentation

The causes of LF can be endogenous and exogenous. Fundamentally endogenous causes mainly cause the fragmentation of agricultural land in China by the effects of complex topography, landforms, geology, hydrology, and other natural conditions on the formation of agricultural land. Land use is the result of the mutual selection between people and nature over an extended history, and different climates, altitudes, and slopes all impact land use patterns. Regarding exogenous causes, the four main factors that lead to LF in China are institutional factors, market transaction mechanisms, land scarcity, and population conflicts [10]. In general, LF at the present stage can be understood as the result of a balance between the conflicts of “human–land–power” and the causes and influencing factors of natural, social, economic, institutional, and historical contexts [11]. Evidently, LF is a manifestation of both the physical attributes of land parcels and the dynamic interplay between human agents and the land. In the historical context of development, land use patterns are the result of multiparty games, and both “fractionalization” and “anti-fractionalization” are reasonable outcomes [12]. Although the determinant of land use patterns is productivity, land use patterns are also influenced by traditional institutions, economic factors, human–land relations, and other factors.

2.2. Research on Collective Action

Collective action aims to remedy the inconsistency between individual rationality and collective rationality, which often manifests as collective action dilemmas. By focusing on the provision of public goods and the pursuit of collective interests, Olson and Mancur identified several cases of inconsistency between individual and collective rationality [13]. Over the past 50 years, scholars have widely applied diverse disciplinary perspectives, theoretical tools, and research methods to study target objects, research domains, solutions, and influencing factors of public goods. These scholarly efforts have gradually shaped first- and second-generation theories of collective action with collective action at their core. This field continues to advance towards the development of the third-generation theory of collective action.

The first-generation collective action theory represents an initial observation and exploration of the dilemmas associated with collective action. A systematic exploration of the collective action dilemma originated from Hardin’s (1968) [14] proposition of a phenomenon Hardin termed the “tragedy of the commons”. Subsequently, this phenomenon continually evolved and ultimately constituted the main component of the first-generation collective action theory. It includes the “tragedy of the commons” arising from cost–benefit dynamics, the “prisoner’s dilemma” arising from information asymmetry, and the collective action logic arising from free-riding. The second-generation collective action theory aims to explore the conditions for the realization of collective action. Notable contributors to this theoretical framework, such as Ostrom, have advocated a mindset that avoids relying on a single solution to overcome the “tragedy of the commons”; instead, the adoption of user self-organizing governance has been proposed as a third alternative. Scholars have presented diverse perspectives through in-depth analyses of classic cases, theoretical studies, and empirical research. For instance, Ostrom (1990) [15] extensively categorized and summarized multiple classic cases. Fueled by these efforts, common characteristics leading to the failure of collective action institutions have been distilled, and eight institutional design principles necessary for successful collective action have been identified [16]. In summary, the second-generation collective action theory emphasizes that appropriate institutional arrangements can achieve effective governance of common pool resources; its goal is to identify the conditions for the successful management of shared resources.

2.3. The Impacts of Land Fragmentation on Collective Action: A Social Science Perspective

First, scholars have focused on the relationship between the scale of agricultural production and collective action. Research has shown that if the scale of agricultural production is more extensive, farmers’ willingness to participate in collective action is higher [17]. Research on the scale of agricultural production has shown that the larger the farm size, the more motivated farmers will be to participate in the construction and maintenance of irrigation facilities [18]. However, it has also been argued that rural collective action declines with the increase in the area under cultivation [19]. As farmers cultivate larger areas, conflicts between individual and collective interests will deepen, and farmers who cultivate large farmland areas tend to focus on individual interests rather than on collective interests [20]. In addition, the coordination cost between farmers will increase with the increase in the area under cultivation, leading to more complications during collaboration among farmers, thus leading to the emergence of collective action dilemmas [21].

Second, the relationship between farmers and social organizations can be affected by LF. For rural collective action, social organization is crucial to determine whether collective action can be implemented. On the one hand, research has shown that LF reduces trust between smallholders and social organizations [22]. Specifically, the greater the distance between agricultural land, the lesser the communication and cooperation between farmers. In turn, reduced communication and cooperation among farmers weaken rural social networks and loosens any existing ties between farmers and rural social organizations. This loosening may lead to a decrease in the credibility of village social organizations and the disorganization of collective action [23]. At the same time, LF limits the amount of land available to each farm household and may also lead to increased competition among farmers. In such a situation, farmers’ first consideration will be their own interests rather than those of the collective, and cooperation between farmers and social organizations will become more difficult because of increased coordination costs caused by fragmentation. On the other hand, LF has been shown to generate a smallholder economy because smallholders have limited economic resources and capacity. At the same time, because of their limited production capacity, small farmers may rely more heavily on the power of social organizations and collective action and may thus participate more actively in social organizations.

Again, many scholars have explored the specific mechanisms by which LF affects collective action linked to the issue of collective action in irrigation. Specifically, certain scholars argued that the presence of LF affects collective action in four ways: first, by influencing farmers’ livelihood decisions and thus their irrigation behavior and intentions [24]; second, by influencing irrigation rulemaking and thus farmers’ participation in irrigation collective action [2]; third, LF reduces farmers’ willingness to participate in collective action by affecting their income costs; fourth, as an essential mechanism that influences farmers to engage in land transfer, LF can promote the intensive use of farmland and optimize the allocation of household labor, capital, and time [25]. Contrary to the negative mechanisms regarding the impact of LF on collective action, as discussed above, research has shown that moderate levels of LF can enhance farmers’ income. This is achieved by increasing farmers’ dependence on agriculture and raising their household income, thereby boosting their willingness to participate in collective irrigation action [26]. Moderate levels of LF imply greater participation of farmers in collective irrigation actions. In irrigation schemes that are dominated by large-scale landholding farmers, those operating at a moderate scale of farmland can leverage the collective construction of irrigation facilities to reduce operating costs. This is one of the land utilization mechanisms that enhance farmers’ willingness to engage in collective irrigation action [27].

2.4. The Impacts of Land Fragmentation on Social–Ecological Systems: A Natural Science Perspective

In the field of natural science, most existing studies on LF have based their analysis on the influencing factors of LF, and the research factors involved mainly focused on the following four aspects: land use and cover (LUCC), ecological and environmental changes, and climate change.

Firstly, in terms of LUCC, the way farmers use their agricultural land is influenced by the resource endowment of the agricultural land, such as its distance from residential areas and transportation hubs [28]. The utilization of land primarily impacts LUCC by exacerbating the unevenness of farmland coverage and reducing land coverage rates. On the one hand, increasing unevenness of LUCC may lead to over-cultivation or abandonment of land, both of which negatively affect the sustainable use of land. On the other hand, farmers may add considerable amounts of chemical fertilizers and pesticides to the land to obtain a high production income, which eventually leads to land degradation, pollution, and soil erosion. Sklenicka (2016) [29] showed that LF is the root cause of LUCC and the direct cause of various kinds of land degradation. However, it has also been shown that in long-term human–land interaction, farmers possess extensive production experience to “tailor” their agricultural activities according to soil quality, slope, and precipitation to promote food security and land ecological sustainability [30]. In certain cases, LF may be better able to promote ecological sustainability. The reason is that LF can be better adapted to local topographic, geological, and hydrological conditions in specific areas, especially in steep mountainous and arid areas, and can further reduce environmental problems such as soil erosion [31]. In other words, there may not be a standard response to the impact of LUCC by LF, and potential responses need to be discussed and analyzed on a “site-specific” basis.

Secondly, in terms of changes of the ecological environment, recently, more and more scholars have considered the sustainability of land ecosystems, market risk offset, and sustainable countryside development. Many studies have argued that LF enables farmers to diversify their agricultural practice by planting according to the resource endowment of the agricultural land while considering the local economic development and geographical location [32]. This improves the heterogeneity and stability of the ecosystem. In addition, research has shown that the diversification of farming by farmers on more fragmented land helps to disperse market risks and reduce risks for farmers, thus increasing their income [11].

Thirdly, in terms of climate change, the utilization of LF—with its diversified cropping practices and crop rotation farming system—further improves the stability of the climate of land ecosystems [30]. On the one hand, land utilization practices associated with farm LF can help to mitigate the risk of complete crop loss caused by natural disasters and promote diversification in soil and growing conditions [5]. In this context, LF can both reduce and control the spread of pests and other crop diseases throughout the farm area [33]. On the other hand, the climatic impact of LF is also influenced by carbon storage and carbon emissions. In fragmented land use, farmers often apply excessive amounts of fertilizer and pesticides to a small piece of land to increase production, thus reducing the carbon storage capacity of the land and increasing its carbon emissions [34]. Soil erosion caused by LF further decreases the organic matter content of the soil, thus reducing the land’s carbon storage capacity and contributing to global warming and climate change [35].

2.5. Summary of the Literature

Research on LF is complex and involves knowledge from various disciplines, and a singular determination of whether LF is positive or negative may overlook the scientific mechanisms underlying LF. Employing the research findings and ideas from the field of natural sciences for the study of the impact of LF on collective action may potentially generate new insights and results in the field of social sciences. Scholars have conducted in-depth examinations of the impact of LF on collective action, and there have been several valuable explorations of the relationship between LF and rural governance.

However, existing studies are still limited by the following three shortcomings: first, in the research on the impact of LF on collective action, the two fields of social science and natural science are compartmentalized, and there is little communication and cooperation between both fields; therefore, their division of labor is strongly limited. Second, in the research on the impact of LF on collective action, scholars have focused on in-depth discussions on factors such as land size and village organization. However, factors such as the ecological environment and environmental values have often been neglected, making it necessary to further incorporate these factors into research on the impact of LF on collective action. Third, existing studies have already shown that LF has both benefits and drawbacks, and therefore, whether the relationship between LF and collective action is purely negative and linear remains to be examined. In summary, existing studies have yet to consider the impact of LF on collective action from a relatively comprehensive and systematic perspective. Based on this analysis of the literature, the present paper examines the impact and mechanism of collective action from a multidisciplinary perspective based on existing research findings.

3. Theoretical Analysis and Research Hypotheses

The construction of irrigation facilities is a typical rural public affair that is closely related to land system change and rural collective action; irrigation systems have been the focus of rural collective action research [36]. The construction, maintenance, design, and implementation of irrigation systems have improved residents’ ability to cope with climate change while also ensuring sustainable economic and social development [37]. Therefore, in this paper, the impact of the LF on the local population is discussed. In this exploration of the relationship between LF and collective action, this study primarily focuses on the irrigation governance system as the research object. By specifically examining the potential relationship between LF and collective irrigation actions, this study uncovers the potential correlations between LF and collective action.

3.1. Connections between Humans and Nature

The connection between humans and nature refers to the interdependence and mutual influence between humanity and the surrounding natural environment. This connection encompasses various aspects, such as human utilization of natural resources, conservation and maintenance of ecosystems, responses to climate change, and human health [38]. In other words, the relationship between humans and nature can yield both material and non-material benefits, thus providing the foundation for livelihoods, emotional support, and health protection. Studying the connection between humans and nature contributes to a better understanding of how human activities impact the natural environment and how, in turn, the natural environment affects human society. In this paper, the explored connection between humans and nature specifically refers to how both the production and life of rural residents are linked to the land, and how, in turn, the land is associated with humans through ecosystem services and other relevant factors.

3.2. The Impact of Land Fragmentation on Collective Action in the Context of Connection between Humans and Nature

Human adaptation to socio-ecological system change has been identified as a social process in which people within the system develop adaptive capacity with those they trust in the public sphere and form collective action [23]. As a result, the degree to which people adapt to changes in the social-ecological system associated with resource use depends on their skills and organizational capacity [39]. Specifically, if resource users have the ability to organize collective actions in response to changes, changes in the social–ecological system resulting from resource use can promote the formation of collective action. Conversely, if changes in the social–ecological system require adaptation that exceeds the adaptive capacity of resource users, changes caused by resource use can disrupt collective action.

The central carrier of rural socio-ecological systems is land, and interactions between human society, the natural environment, and economic activities contained in rural socio-ecological systems are carried out based on land [40]. Therefore, the socio-ecological system changes with the degree of LF (from low to high). Thus, based on the discussion above, this paper argues that the process of human adaptation to changes in the socio-ecological system (triggered by changes in the degree of LF) will also be characterized by stages. On the one hand, in the process of changing the degree of LF from low to high, the degree of LF will gradually reach the appropriate level. This change process will induce ecosystem improvements such as increased biodiversity and improved habitat conditions [5,29]. These ecosystem improvements reduce the ecological risks for farmers as users of land resources and increase the ecosystem services farmers can obtain. As a result, people living in this socio-ecological system are more likely to take proactive measures to adapt to such improvements, i.e., they may organize practical collective actions to maintain improved ecosystem outcomes and thus increase the resilience of the socio-ecological system. On the other hand, when the degree of LF exceeds the appropriate level, inefficient land production will not be able to meet the demand for food and resources. This will further lead to a haphazard and uncontrolled expansion of agricultural production [41], which may eventually have irreversible consequences such as land degradation and soil erosion. When such irreversible ecological degradation occurs, people living in social–ecological systems lose access to certain ecosystem services [32] and can no longer reverse the collapse of such ecosystems through collective action. As a result, more and more people will leave such socio-ecological systems, and eventually, collective action will decline as a result. Figure 1 summarizes the two pathways through which changes in LF affect collective action.

4. Materials and Methods

4.1. Research Area

The researched area examined in this paper is located in Guangxi, China. Situated on the southwestern border of China and adjacent to Vietnam, the province of Guangxi serves as a crucial gateway for China’s external openness, engagement with ASEAN, and global engagement. The topography of Guangxi is diverse: its western and northern regions are surrounded by plateaus, while the southern and eastern regions consist of a mosaic of hills and plains. According to scholarly research, the level of LF in Guangxi in 2022 was 1652, indicating a high degree of LF [42]. This high degree of fragmentation implies that farmland in the Guangxi region is sub-divided into many small and scattered plots, leading to problems such as low agricultural efficiency, increased production costs, and difficulties associated with resource management. Amid continuous developmental transformations, Guangxi has advanced to the forefront of China’s ecological environmental quality. Its ecological advantages have become one of the driving forces for the promotion of high-quality development. From 2018 to 2023, Guangxi’s Ecological Quality Index ranked second in China1. Moreover, Guangxi is home to various ethnic minority groups, with a total resident population of 18.9 million in this region in 2021. Ethnic minority groups account for 37.5% of the total population in Guangxi and 4.7% of the national minority population, based on a long-standing tradition of ethnic autonomy2. Guangxi is a predominantly agricultural region that mainly focuses on cultivating crops such as rice and sugarcane. In 2022, Guangxi’s total early rice production was 4.8 million tons, ranking fourth among China’s provinces3; the sugarcane production of Guangxi during the crushing season was 41.2 million tons, ranking first among China’s provinces4. The mountainous terrain of Guangxi imposes certain limitations on mechanized farming operations, necessitating high levels of labor-intensive cultivation for these crops and requiring favorable conditions for collective irrigation. Consequently, agriculture in Guangxi relies heavily on both collective action and irrigation systems. In summary, Guangxi Province provides an excellent subject for observing the interactions among LF, the connections between humans and nature, and resulting collective actions in this dynamic context.

4.2. Research Data

The data utilized in this paper are derived from the “Hundred Villages Thousand Households” survey conducted by Guangxi University in Guangxi’s border regions between 2023 and 2024. The survey questionnaire, administered at both the village and household levels, sampled various aspects, including the basic conditions of villages and households over the past three years, agricultural water management, land utilization, infrastructure, digital technology application, rural tourism, natural disasters, social capital, rural governance, rural education, and environmental management.

From March to June 2023, following expert consultations and incorporating recommendations and experiences from social survey teams at Tsinghua University, Peking University, and Renmin University of China, the research team developed an initial survey questionnaire. From July to August 2023, the research team conducted a pre-survey in the Babu District of Hezhou, Guangxi, where the effectiveness and practicality of the questionnaire were tested. Based on the outcome of this pre-survey, the questionnaire contents were adjusted and the final version of the questionnaire was obtained. By the end of August 2023, the research team had recruited 79 questionnaire surveyors from the School of Public Administration at Guangxi University. In early September 2023, 10 graduate and doctoral students experienced in conducting field surveys were invited to train newly recruited surveyors. Concurrently, to select the research sample, the research team employed a stratified random sampling method based on the economic and social development levels of Guangxi. Specifically, 14 counties were randomly selected from the 14 cities in Guangxi. Within each of these counties, 6–8 villages were randomly chosen, and within each village, 7–11 households were randomly selected for interviews. In January 2024, the research team successfully completed the survey for this project. Focusing on the specific issues addressed in this paper, a total of 798 household questionnaires from 114 villages in the 14 cities of Guangxi Province were selected as research samples. Figure 2 displays the distribution of our samples, plotted using ArcGIS Pro 3.1.5.

4.3. Variable Selection

4.3.1. Dependent Variable

The irrigation system is a socio-ecological system in which people adapt by organizing collective action. When studying irrigation collective action, there are two main ways to measure the capacity of collective action: the output approach and the process approach [43]. Of these approaches, the output approach refers to the ability of farmers to participate in collective action, measured by the outcome of collective action, for example, the discussion of farmers’ participation in the maintenance of irrigation facilities. In contrast, the process approach measures farmers’ participation through actions taken, such as the collective resolution of village water disputes. According to Zang et al. (2019) [7] and Wang et al. (2020) [6], the process approach is applied based on the availability of data and the practicality of the study; therefore, “whether the household regularly participates in irrigation facility repair activities” (ICA) (on a scale of 1 = never participate to 5 = frequently participate) was selected as the dependent variable of irrigation collective action.

The maintenance of irrigation infrastructure encompasses a large number of collective action problems. First, because of limited conditions, irrigation canals in many rural areas of China are constructed out of kilometers of earthen canals. To ensure that these canals can function properly, the villages where these canals are located need to organize regular canal dredging, which is conducted by the villagers, and which includes the removal of weeds, debris, and silt; this dredging is a labor-intensive process and thus requires a high level of collective action in the villages [44]. Second, even if the irrigation canals in the villages are made of better-quality cement masonry, the valves, weirs, small dams, and other units that are present in large numbers in irrigation canals still need to be operated through collective action [45]. Third, in terms of the management of irrigation canals, several issues need to be addressed, such as determining who is responsible for managing and overseeing the operation of infrastructure. Further, the amount of maintenance costs required needs to be decided, water needs to be allocated to farmers along the canal, and water disputes must be resolved. The resolution of these issues requires mutual consultation and cooperation among villagers, necessitating significant collective action [46]. Therefore, the maintenance of irrigation infrastructure is an indicator that is commonly used to measure the capacity for collective irrigation action [6].

The continuing impact of external economic and social development has eliminated the need for traditional rural collective actions. However, the implementation of the household joint production contract responsibility system has allowed traditional collective irrigation actions to remain widely used in rural areas of China. The long-term implementation of the household contract responsibility system has resulted in the small and scattered nature of rural land in China [6]. This has always necessitated the coordination of many small farmers and their land in the process of carrying out agricultural production. This coordination process is usually carried out from door to door [7]. For example, in irrigation management, irrigation water must flow through each household’s land. Ensuring this requires extensive collective action to coordinate the use of irrigation water and the maintenance of the entire irrigation system. Even in areas where advanced irrigation facilities have been installed, ownership of the land used by irrigation facilities is dispersed among households.

Therefore, when irrigation equipment requires maintenance, there is a high risk of free-riding behavior. In this case, to meet the maintenance needs of irrigation facilities, it is necessary to conduct door-to-door coordination, thus eliminating the free-rider phenomenon by promoting mutual commitment. Although the impact of urbanization on China’s rural society, the unique land system of rural China has not weakened the need for traditional collective irrigation action.

4.3.2. Core Independent Variables

LF refers to the condition of the block structure within land, which is typically assessed through indicators such as land area, block structure, inter-block distances, and land tenure relationships [5]. Firstly, the land area reflects the scale and continuity of land use, which is crucial for comprehending the degree of LF. Smaller cultivated land areas typically indicate that the land has been divided into multiple smaller plots, potentially leading to inconveniences and inefficiencies associated with agricultural production. Therefore, by measuring the land area, the extent of LF can be understood intuitively, thereby providing fundamental data and references for agricultural development and land management. Secondly, the analysis of the block structure involves a quantitative assessment of different block sizes, shapes, and spatial distributions, which discloses the complexity of the internal structure of the land. Inter-block distances examine the relative positions of specific blocks within the land, which directly influences land use efficiency. The clarity of land tenure relationships, which is an assessment indicator for fragmentation, reflects the impact of land systems on fragmentation through in-depth examinations of ownership and land use relationships.

The selection of these measurement indicators is based on their ability to comprehensively reflect both the internal structure and management characteristics of land; thereby, they provide a scientific basis for a deeper understanding of land fragmentation. China’s per capita land area is less than 1000 m², and therefore it ranks 118th among 195 countries in terms of per capita land area [47]. In studies addressing issues associated with LF in China, many scholars directly use the area of land managed by households to represent China’s LF level. This paper follows this tradition for measuring LF. In reference to Wang (2020) [6], the “reciprocal of the area of land managed by households” is selected as a variable for measuring LF.

4.3.3. Control Variables

Firstly, based on existing research and the hypothesis tested in this paper, the control variables consist of five categories of factors that are known to influence collective action. These five categories include natural geographical features, general institutional rules, socio-economic attributes, household characteristics, and variables representing interactions between humans and nature [4,48,49].

Drawing on the research of Wang et al. [6] and Zang et al. [7], in terms of family characteristics, the following three variables were selected: “Do the family members have religious beliefs?” (FAITH), “What was the family income in 2022?” (INCOME), and “How many people are there in the family?” (POPULATION). For natural geographical features, the following five variables were selected: “Is the village located in a coastal area?” (COASTAL), “Is the village located in a mountainous area?” (MOUNTAIN), “Has there been any degradation in the quality of family farmland in recent years?” (DEGENERAT), “How is the soil fertility level of the family’s land?” (Soil Health), and “How far is the family residence from the town center?” (DISTANCE). In terms of socio-economic attributes, the following nine variables were selected: “Does the family purchase elderly insurance?” (ENDOWMENT), “Is the family aware or informed about the ‘Environmental Protection Law’?” (LAW), “Has there been any improvement in the village’s farmland infrastructure conditions in recent years?” (FACILITIES), “Did the family join a rural cooperative in 2022?” (COOPERATIVE), “How familiar is the family with other villagers in the village?” (RELATION), “What percentage of the grains planted by the family is used for external sales?” (SELL), “Does the family seek agricultural technology assistance through the Internet?” (TECHNOLOGY), “Have natural disasters caused substantial damage to the house and garden properties in recent years?” (DISASTER1), and “Have natural disasters caused substantial harm to family agricultural production in recent years?” (DISASTER2). For general institutional rules, the following two variables were selected: “How well do other people in the village comply with the village regulations?” (REGULATIONS) and “Do you think the decision-making on various village affairs is truly fair, just, and transparent?” (EQUITY).

Secondly, human impacts on ecology have been examined through human pro-environmental behaviors [50]. This paper draws on existing research in this specific area. The following two variables were selected: “Has the family reduced pesticide use in actual cultivation?” (PESTICIDE) and “Has the family reduced fertilizer application in actual cultivation?” (FERTILIZER). These variables reflect the connection between humans and nature. Additionally, regional dummy variables were included in the estimation process to control for the impact of regional differences on the estimated results.

Table 1. Selected variables and descriptive statistics.

Variable	Description	Mean	Std. Dev.	Min.	Max.
Dependent variable
ICA	Does the household regularly participate in irrigation facility repair activities? (ICA) (1 = Never participate to 5 = Frequently participate)	4.286	0.879	1	5
Core independent variables
LF	Reciprocal of the area of land managed by households	1.089	0.651	0.143	2.915
LF2	Reciprocal of the area of land managed by households (square)	1.610	1.890	0.020	8.500
The connection between humans and nature
PESTICIDE	Did your family reduce pesticide application in actual cultivation? (0 = No; 1 = Yes)	1.519	0.500	1	2
FERTILIZER	Did your family reduce fertilizer application in actual cultivation? (0 = No; 1 = Yes)	0.436	0.496	0	1
Family characteristics
FAITH	Do your family members hold religious beliefs? (0 = No; 1 = Yes)	0.039	0.193	0	1
INCOME	What was the family income in 2022? (Yuan)	90,594.160	218,947.800	0	3,750,000
POPULATION	How many people are there in your family?	5.149	2.308	1	35
Natural geographical characteristics
COASTAL	Is the village located in a coastal area? (0 = No; 1 = Yes)	0.033	0.179	0	1
MOUNTAIN	Is the village located in a mountainous area? (0 = No; 1 = Yes)	0.471	0.499	0	1
DEGENERAT	Has there been any degradation in the quality of family farmland in recent years? (0 = No; 1 = Yes)	0.368	0.483	0	1
Soil Health	How is the fertility level of the family land? (1 = Very barren; 5 = Very fertile)	3.428	0.868	1	5
DISTANCE	How far is the family residence from the town center? (km)	7.540	6.563	0.01	50
Social and economic attributes
ENDOWMENT	Does the family purchase elderly insurance? (0 = No; 1 = Yes)	0.779	0.415	0	1
LAW	Have you heard or are you familiar with the “Environmental Protection Law”? (1 = Never heard of it; 5 = Very familiar)	2.808	1.145	1	5
FACILITIES	Has there been any improvement in the village’s farmland infrastructure conditions in recent years? (0 = No; 1 = Yes)	0.762	0.426	0	1
COOPERATIVE	Did your family join a rural cooperative in 2022? (0 = No; 1 = Yes)	0.145	0.353	0	1
RELATION	How familiar are you and your family with other villagers in the village? (1 = Not familiar at all; 5 = Very familiar)	4.489	0.729	1	5
SELL	What percentage of the grains planted by the family is used for external sales? (%)	25.852	37.696	0	100
TECHNOLOGY	Do you seek agricultural technology assistance through the Internet? (1 = Strongly disagree; 5 = Strongly agree)	1.998	1.212	1	5
DISASTER1	In recent years, natural disasters have caused substantial damage to properties such as houses and gardens (1 = strongly disagree; 5 = strongly agree)	2.354	1.318	1	5
DISASTER2	In recent years, have natural disasters caused substantial harm to household agricultural production? (1 = strongly disagree; 5 = strongly agree)	2.874	1.437	1	5
General institutional rules
REGULATIONS	How well do other villagers adhere to village rules and agreements? (1 = Never adhere; 5 = always adhere)	4.059	0.774	1	5
EQUITY	Do you believe that decision-making on various village affairs is genuinely fair, just, and transparent? (1 = strongly disagree; 5 = strongly agree)	4.064	1.135	1	5

presents the situation and descriptive statistical analysis of selected variables.

4.4. Research Methodology

4.4.1. Ordered Probability Regression Model

In this study, “Whether the household regularly participates in irrigation facility repair activities” (ICA) was chosen as the indicator to measure collective irrigation actions. This indicator was divided into five levels, ranging from one to five, corresponding to never participate, occasionally participate, sometimes participate, relatively frequently participate, and frequently participate, respectively. Therefore, the selected indicator for the characterization of collective action is an ordered categorical variable, and an ordered Probit regression model needs to be used to estimate the results. The model for this research is formulated as follows:

$$y=\{\begin{array}{l}0,y^{\ast }\le r_0\\ 1,r_0\le y^{\ast }\le r_1\\ 2,r_1\le y^{\ast }\le r_2\\ \dots \dots \\ J,r_{J-1}\le y^{\ast }\end{array}$$

(1)

where r₀ < r₁ < r₂ < … < r_J−1 is the parameter to be estimated, referred to as the “threshold”. Assuming that $\mathsf{\epsilon }\sim \mathrm{N}$(0,1) (normalizing the variance of the perturbation term to 1), the following assumptions can be made:

$$\begin{array}{ll}\mathrm{P}\left(y=0|x\right)& =\mathrm{P}\left(y^{*}\le r_0|x\right)=\mathrm{P}\left(x^{\prime }\beta +\epsilon \le r_0|x\right)\\ & =\mathrm{P}\left(\epsilon \le r_0-x^{\prime }\beta |x\right)=\mathrm{\Phi }\left(r_0-x^{\prime }\beta \right)\\ \mathrm{P}\left(y=1|x\right)& =\mathrm{P}\left(r_0\le y^{*}\le r_1|x\right)=\mathrm{P}\left(y^{*}\le r_1|x\right)-\mathrm{P}\left(y^{*}<r_0|x\right)\\ & =\mathrm{P}\left(x^{\prime }\beta +\epsilon \le r_1|x\right)-\mathrm{\Phi }\left(r_0-x^{\prime }\beta \right)\\ & =\mathrm{P}\left(\epsilon \le r_1-x^{\prime }\beta |x\right)-\mathrm{\Phi }\left(r_0-x^{\prime }\beta \right)\\ & =\mathrm{\Phi }\left(r_1-x^{\prime }\beta \right)-\mathrm{\Phi }\left(r_0-x^{\prime }\beta \right)\\ \mathrm{P}\left(y=2|x\right)& =\mathrm{\Phi }\left(r_2-x^{\prime }\beta \right)-\mathrm{\Phi }\left(r_1-x^{\prime }\beta \right)\\ & \cdots \cdots \cdots \cdots \\ \mathrm{P}\left(y=J|x\right)& =1-\mathrm{\Phi }\left(r_{J-1}-x^{\prime }\beta \right)\end{array}$$

(2)

By extrapolating the above equation, the sample likelihood function and the maximum likelihood estimator (i.e., the ordered probability model) were obtained.

4.4.2. Instrumental Variables Test

Simple multiple regression estimation results are susceptible to interference from endogeneity issues. Specifically, LF and collective action may be linked through an inverse causal relationship. For instance, related research indicated that one way to overcome LF is through collective action by multiple stakeholders for large-scale land operations. Therefore, the collective action capacity can actually impose a reverse impact on LF [44]. Factors such as climate change are essential in influencing collective action, but as the variables used in this paper do not consider these issues, relevant variables may have been omitted. The endogeneity problem described above can be controlled by introducing instrumental variables into the regression. A valid instrumental variable must satisfy two necessary conditions: The first necessary condition is correlation, i.e., the selected instrumental variable should be highly correlated with the core independent variable of LF in Table 1; the second necessary condition is independence, i.e., the selected instrumental variable needs to be uncorrelated with the random error term in the regression equation.

Based on the aforementioned criteria for an instrumental variable, two variables were selected as instrumental variables for LF. These are “Distance from the village to the nearest county hospital” (CXD) and “Whether the village of the household is located in Beihai City” (BEIHAI). The rationale is as follows: on the one hand, CXD is a geographical variable that is naturally exogenous to the natural conditions of the village and changes according to the economic and social environment. This distance reflects the geographical location of the village and the level of difficulty associated with agricultural activities in the village. The closer the village is to the nearest county hospital, the better the economic development of the area in which the village is located. Farmland may become more fragmented because of the economic development of the county town [6]. Therefore, CXD can influence the degree of LF for households.

On the other hand, BEIHAI is correlated with the independent variable. Because distance is a geographical variable, it is influenced neither by economic and social development nor by environmental changes. Simultaneously, in China, the residence of households in a particular area is highly correlated with their status as villagers. The acquisition of a villager identity by households is related to the historical development process of modern Chinese society [51]. Therefore, BEIHAI reflects a decision made by the ancestors of the household based on historical circumstances, and it is not meaningfully related to the individual decisions of the surveyed households. In summary, the selected instrumental variables are unrelated to other unobservable variables.

5. Estimated Results

5.1. Benchmark Regression

Table 2. Overall effect of land fragmentation on collective action according to regression analysis.

Variable	Model 1 (Oprobit)	Model 2 (2SLS)	Model 3 (IV-Probit)
	ICA	ICA	ICA
LF	0.679 **	2.474 **	0.599
	(2.39)	(2.14)	(1.49)
LF2	−0.261 ***	−0.887 **	−0.263 ***
	(−2.70)	(−2.17)	(−2.72)
PESTICIDE	−0.0391	−0.0493	−0.0294
	(−0.36)	(−0.63)	(−0.26)
FERTILIZER	0.0262	0.0284	0.0148
	(0.24)	(0.36)	(0.12)
FAITH	−0.174	−0.189	−0.164
	(−0.79)	(−1.15)	(−0.73)
INCOME	−0.022	−0.013	−0.022
	(−1.27)	(−0.98)	(−1.26)
POPULATION	−0.00555	0.00462	−0.00729
	(−0.29)	(0.33)	(−0.37)
COASTAL	0.0183	0.123	0.0312
	(0.08)	(0.70)	(0.13)
MOUNTAIN	0.170 *	0.170 **	0.148
	(1.82)	(2.25)	(1.22)
DEGENERAT	0.0392	0.0386	0.0490
	(0.42)	(0.57)	(0.49)
FERTILITY	0.176 ***	0.0988 **	0.177 ***
	(3.22)	(2.56)	(3.24)
DISTANCE	0.00119	0.00240	0.00243
	(0.17)	(0.48)	(0.30)
ENDOWMENT	0.0937	0.0601	0.0944
	(0.91)	(0.79)	(0.92)
LAW	0.106 ***	0.0715 **	0.106 ***
	(2.64)	(2.44)	(2.65)
FACILITIES	0.0322	0.0554	0.0309
	(0.32)	(0.73)	(0.30)
COOPERATIVE	−0.0415	−0.0234	−0.0434
	(−0.34)	(−0.27)	(−0.35)
RELATION	0.356 ***	0.272 ***	0.353 ***
	(6.16)	(6.09)	(6.00)
SELL	−0.00103	−0.00150	−0.000794
	(−0.86)	(−1.60)	(−0.54)
TECHNOLOGY	0.0481	0.0328	0.0480
	(1.30)	(1.26)	(1.29)
DISASTER1	−0.126 ***	−0.0788 ***	−0.127 ***
	(−3.13)	(−2.77)	(−3.15)
DISASTER2	0.0803 **	0.0431	0.0808 **
	(2.11)	(1.60)	(2.13)
REGULATIONS	0.314 ***	0.171 ***	0.310 ***
	(5.41)	(3.93)	(5.11)
EQUITY	0.0307	0.0122	0.0331
	(0.77)	(0.44)	(0.82)
Regional variables	Controlled	Controlled	Controlled
Observed value	798	798	798
Wald chi-squared	183.73	11.613	183.73
p > chi-squared	0.000	0.000	0.000
Pseudo-R-squared	0.103	0.071	0.103

t statistics are shown in parentheses, * p < 0.1, ** p < 0.05, *** p < 0.01.

presents the results of the estimates of LF affecting collective action obtained with Stata 17.0. As shown in Model 1, the coefficient of LF2 is negative and statistically significant after considering connections between humans and nature, suggesting that LF has a significant inverted U-shaped relationship with collective action. At the same time, this paper argues that the above empirical results may be biased because of endogeneity problems such as omitted variables. To address the impact of potential endogeneity issues on the empirical results, instrumental variables are introduced into Model 2 to control endogeneity problems. The result of weak instrumental variable identification shows that the Cragg–Donald Wald F-statistic is 11.61. This result indicates that the instrumental variables are well correlated with core independent variables and there is no weak instrumental variable problem. Moreover, the p-value of the Sargan statistic test, which is used to test the correlation between the instrumental variables and the random error term, is 0.437. Consequently, the null hypothesis that the instrumental variables are uncorrelated with the random error term is not rejected. In summary, it can be reasonably concluded that the instrumental variables selected in this paper are reasonably valid. Further, the estimation results of Model 3 show that the coefficient estimate of LF2 is still negative and significant. This result further confirms the hypothesis that LF has an inverted U-shaped relationship with collective action.

Figure 3 reveals whether a significant inverted U-shaped relationship exists between LF and the dependent variable in Model 1. Specifically, in reference to Haans (2016) [52], the marginal effect of LF is verified by testing the steepness of the slopes at the ends of the inverted U-shape. Specifically, on the one hand, at an LF level of 0.143 (the minimum value of LF), the slope of the curve is 0.604 and significant (p = 0.009), indicating that for every 1% increase in the LF level, collective action increases by 0.604 units. On the other hand, at an LF level of 2.915 (LF maximum), the slope of the curve is −0.844 and statistically significant (p = 0.002), indicating that for every 1% increase in the LF level, collective action decreases by 0.844 units. Finally, the level of LF at the inflection point is 1.299 (which actually corresponds to the average size of a single plot for a farm household 1/1.299 = 0.770 mu5, that is 513.3 m²). This value is located between the confidence intervals [0.143, 2.915], and conclusively proves the existence of an inverted U-shaped relationship between the core independent variable, LF, and the dependent variable.

5.2. Robustness Analysis

Rural public affairs in China encompass a variety of aspects, one of which is irrigation collective action. Here, the robustness of the baseline regression is examined by substituting the dependent variable. Therefore, in this study, the dependent variable is replaced with “disaster collective action” (DCA), “Sewage Canal Cleaning” (SCA), “Public Toilet Cleaning” (TCA), and “Participating in Public Affairs Discussions via Mobile Phone” (PCA). These substitutions not only reduce the sensitivity of the model and dependence on specific variables but also better reflect uncertainty. Moreover, they comprehensively consider the robustness of the research results, thereby enhancing the explanatory power of the findings.

Specifically, disaster collective action, cleaning sewage ditches, cleaning public toilets, and participating in public affairs discussions with mobile phones are classic rural public affairs. Their implementation is usually accomplished through collaborative mechanisms and collective participation. Driven by common goals, farmers have established close cooperative networks among themselves, forming entities such as villagers’ self-governing organizations or cooperatives to coordinate and organize collective actions [4]. These organizations or cooperatives are usually led by village leaders or heads of cooperatives and are elected by villagers. For example, in disaster collective action, farmers usually divide their workload according to their respective skills and resources. They collectively participate in rescue and relief work, and the employed division of labor and cooperation enables them to respond to disasters in a more orderly and efficient manner. Moreover, sanitation work such as cleaning sewage ditches and public toilets is usually accomplished through regular collective cleaning activities. In these activities, villagers determine the cleanup plan and division of responsibilities through joint discussion and negotiation, in which each household holds the responsibility to participate. Such concerted efforts both increase the efficiency of sanitation work and strengthen village cohesion. In contrast, mobile phone participation in public affairs discussions is usually accomplished through the establishment of community communication platforms or the use of tools such as social media platforms. Farmers can obtain up-to-date information, participate in discussions, and make suggestions through mobile phones, which realizes the instantaneous transmission of information and the full expression of public opinion. The introduction of this approach has strengthened the sense of participation and democratic involvement of rural residents in relevant affairs.

Table 3. Robustness analysis results.

Variables	Model 4	Model 5	Model 6	Model 7
	DCA	SCA	TCA	PCA
LF	0.688 **	−0.539 **	−0.407	1.034 ***
	(2.49)	(−2.08)	(−1.38)	(3.14)
LF2	−0.252 ***	0.182 **	0.228 **	−0.358 ***
	(−2.69)	(2.04)	(2.23)	(−3.19)
Connection between humans and nature	Controlled	Controlled	Controlled	Controlled
Controlled variables	Controlled	Controlled	Controlled	Controlled
Regional variables	Controlled	Controlled	Controlled	Controlled
Observations	798	798	798	798
Wald chi-squared	181.61	120.87	102.83	124.74
Chi-squared	0.000	0.000	0.000	0.000

t statistics are shown in parentheses, ** p < 0.05, *** p < 0.01.

presents the results of the robustness test. The estimation results of Models 4–7 show that the coefficients of LF2 are negative and significant, which further verifies the robustness of the model selected in this paper.

6. Discussion

6.1. The Impact of an LF Threshold of 513.3 m² on Collective Action and Its Policy Implications

In this study, we discovered a significant inverted U-shaped relationship between LF and collective action, with a threshold of 513.3 m². This threshold indicates that maintaining LF levels around 513.3 m² in the Guangxi can achieve optimal levels of collective action. When land areas approach 513.3 m², the frequency of interactions among farmers increases, trust is built, and social capital is significantly enhanced. However, when LF exceeds this threshold, excessive land fragmentation leads to a sharp increase in coordination and transaction costs, significantly reducing the community’s willingness and ability to cooperate. This result is consistent with social capital theory and transaction cost economics theory, suggesting that moderate LF can enhance social networks and cooperative capacity, while excessive LF disrupts this mechanism. Although we obtained precise LF threshold data based on a large-scale social survey, it is important to emphasize that the relationship between people and land varies locally, and the optimal land size needs to be considered based on local conditions in different regions.

From a policy perspective, the 513.3 m² threshold provides an important reference for land management and rural development. To promote rural collective action and diversified agricultural development, policymakers could consider limiting excessive land fragmentation to ensure plot sizes do not fall below this threshold. This can be achieved through land consolidation programs or cooperative agricultural incentive measures. Additionally, when formulating rural development plans, measures can be taken to enhance community cohesion and cooperative capacity in line with this threshold, such as supporting the establishment and development of cooperatives and providing economic incentives to encourage moderate land use and management. Through such policy measures, the benefits of collective action can be maximized while maintaining moderate LF, promoting sustainable development in rural areas.

Currently, many underdeveloped areas focus their policies on simplified traditional agriculture rather than transitioning to diversified agriculture. Diversified agriculture, which intentionally increases the cultivation of both agricultural and non-agricultural crops, is a solution for achieving sustainable environmental development [48]. The moderate LF studied in this paper may provide an opportunity to achieve diversified agriculture. Specifically, diversified agriculture can enhance the ecological stability of farms and their ability to withstand risks, while also providing farmers with more sources of income. This paper may offer a reference for the level at which diversified agriculture can be achieved.

6.2. Possible Mechanisms through Which Land Fragmentation Affects Collective Action

In the long-term interaction between rural residents and the land, numerous rural stakeholders spontaneously form collective actions. Often, these stakeholders demonstrate superior self-organizational abilities to form a set of practical rules and institutions in the face of public resource management. This further corroborates Ostrom’s research observation that the context for collective action is complex and diverse. Synthesizing existing studies and considering knowledge at the natural science level, LF will impact collective action through the three intermediate mechanisms of agricultural production, land use, and the ecological environment. The detailed diagram of the above-mentioned mediation mechanism is shown in Figure 4.

Firstly, in terms of agricultural production, compared to high levels of LF, moderate levels of LF can benefit agricultural production [53]. In areas with moderate LF, farmers often adopt multi-crop planting strategies, thus enabling them to adapt to heterogeneous land ecological conditions, population development, and volatile market systems [54]. That is, the utilization of moderate levels of LF is actually a flexible agricultural practice and an effective risk management strategy that can promote agricultural production. For example, certain lands may be more suitable for water-loving crops, while others are more suitable for drought-resistant crops. By planting according to the natural characteristics of the land, if one plot fails, other plots can still continue production; thereby, the use of land resources is maximized and the overall yield of agricultural production increases. This example reflects a concrete manifestation of farmers’ affinity with nature. Furthermore, good agricultural production can promote farmers’ participation in collective actions, as it ensures farmers’ household income and access to ample production resources. At the same time, farmers seeking sustainable and stable agricultural production will cooperate with other farmers to improve agricultural public facilities and services. In this process, good agricultural production provides farmers with more time and funds to participate in community activities. These additional resources enhance their sense of belonging to the community and can motivate them to engage in collective actions [55]. Moreover, during the negotiation and cooperation phase in agricultural production, trust and social capital among farmers are further strengthened. Consequently, the transaction costs of collective actions are reduced and farmers’ participation in collective actions is promoted [56].

Secondly, in terms of land use, moderate levels of LF have multifaceted impacts and can promote the development of village collective actions. Moderate LF levels enable cultivated land to better meet diverse utilization needs, thus allowing the full utilization of traditional land use, thereby improving land use conditions [57]. Moderate LF levels also reduce land monopolization, allowing more small farmers to participate in agricultural production, thus further enhancing the fairness and efficiency of land use. Additionally, moderate LF levels can promote diversified land use, thus allowing different plots to adapt to different planting needs or agricultural models [58]. Furthermore, a suitable land use model can encourage farmers to participate in collective actions. Specifically, reasonable and effective land use can bring substantial economic benefits to farmers, thereby increasing the common benefits for community members and further enhancing farmers’ willingness to engage in collective actions. Suitable land use models usually require coordinated management to ensure sustainability. In this context, farmers need to collaborate in the management of agricultural resources and the sharing of resources to ensure the sustainable development of rural social–ecological systems [59]. In certain villages in China, villagers entrust their land to large-scale households through the village collective. During this process, farmers need to engage in extensive negotiations, including determining land ownership and changing land use types [60]. As shown by this discussion, it is evident that farmers’ exploration of optimal land use models not only enhances their economic benefits but also improves interaction patterns among farmers. Suitable models can effectively promote collective actions.

Thirdly, in terms of the ecological environment, under moderate levels of LF, farmers tend to reduce the use of agricultural fertilizers and pesticides based on their farming experience. This reduction not only increases biodiversity and protects the ecological environment but also enhances the standard of living of farmers and promotes harmony between humans and nature [30]. In China, farmers managing moderately scaled land often have multiple occupations, but land management remains their most important source of income. Because of the substantial social security land resources provide, these farmers are unlikely to damage the land lightly. Moreover, these lands are often inherited through generations, creating a strong attachment among farmers to their land. As a result, farmers are careful not to overuse pesticides and fertilizers, thereby inadvertently protecting the ecology of their lands. A well-maintained land ecology ensures an environment that can further promote collective actions. The reason is that a good land ecology provides a common foundation for survival and development within rural communities, thus enhancing farmers’ sense of belonging and attachment to their villages. Additionally, good land ecology provides stable livelihood support for farmers, thus strengthening their reliance on the land and the rural community [61]. In other words, good land ecology can increase farmers’ sense of identity, attachment, and dependency on the place, further boosting their willingness to participate in collective actions.

6.3. Promoting an Extended Understanding of the Relationship between Land Fragmentation and Collective Action: A Dynamic Perspective Discussion

The passage of time needs to be addressed in many sociological studies, but a lack of relevant research conducted through the lens of time and human social mobility hinders the understanding of corresponding social issues [62]. Indeed, economic technologies and institutions change over time, and once a temporal dimension is introduced, a change in the conclusions of static-based social research can be observed [63]. Thus, introducing a temporal dimension can provide insights into the complexity of social systems and disclose the dynamic evolution of social phenomena and problems.

Based on a static analysis of cross-sectional data and according to a static perspective, the results of this paper indicate that LF and collective action follow an inverted “U”-shaped relationship that considers human impacts on the ecological environment. However, the maintenance of the resilience of social–ecological systems is a dynamic process that involves adaptation to change and uncertainty [64]. Therefore, introducing a further temporal dimension to the static results of the analysis enables a better understanding of how the relationship between LF and collective action changes over time; moreover, this approach shows how resilience can be maintained through change and adaptation when social–ecological systems are challenged. Specifically, Wang and Xu (2022) [65] analyzed the spatial and temporal characteristics of long-term land use data and concluded that the trend of LF in China between 2000 and 2020 showed a “continuous increase—slow growth—fluctuating change” dynamic. Further, they pointed out that there are two time-nodes when the trend of LF in China changes. These two time-nodes correspond to the beginning of the abolition of agricultural tax and the implementation of the land management rights transfer policy in China. Therefore, both high and low degrees of LF can be found in China, which fluctuate at a scientific level under human adaptation and change. This conclusion can be further expressed as shown in Figure 5a. When the different effects of various levels of LF on collective action are considered, the fluctuation of the level of LF in China at a scientific level corresponds to the maintenance of a high level of LF to preserve the resilience of the social–ecological system. In Figure 5, section AB represents the best effect range of rural irrigation collective action. Thus, a suitable level of fluctuation in LF means that a high level of collective action can always be maintained under the continuous adaptation and change of the socio-ecological system. The analysis of how the relationship between LF and collective action changes over time actually points to three keywords: adaptability change, system transformation, and overcoming the “resilience trap6”. Both adaptive and transformative changes can contribute to the resilience of social ecological systems and help to overcome the resilience trap [66]. In the case of rural land systems, over time, while innovation and economic development can generate a balance and harmonization of conflicting old land use patterns, economic and social development can also generate new conflicts [67] As a result, new policies, engineering technologies, and other external dynamics will reconcile land conflicts and promote the complementarity of implicit landforms. Thus, a new round of land use transformation is prompted and ultimately, the “resilience trap” of economic and social development is overcome [68] (Figure 5b). The existing discourse on the transformation of rural land systems provides a systemic approach to change, based on exogenous dynamics to achieve conflict coordination.

However, the findings of this paper suggest that endogenous dynamics provided by collective action-based approaches also play an essential role in driving regional land use patterns from conflict to coordination. On the one hand, interdependence between people as maintainers of the socio-ecological system and the ecosystem helps to achieve the resilience of the socio-ecological system [69]. For example, under the pressure–state–response mechanism, in response to external pressures, rural systems will initiate internal responses when dealing with risks and contingencies. People will begin to construct close social relationship structures, maintain capital, co-build infrastructure, improve ecological conditions, and provide means to resolve resource conflicts. On the other hand, when the resilience of the rural social–ecological system is maintained, based on their previous experiences, people within the system will continue to learn and change to resolve system conflicts and elevate the social–ecological system to a new level of operation [70]. In this study, the process of “continuous increase—slowdown—fluctuation” in the time series of LF is accompanied by systemic change. When the rural social–ecological system faces conflicts and shocks, people start to organize collective actions to restore, renew, and reorganize the elements of the system. Thus, land trust, land transfer, and other farmland management methods emerge so that the level of LF constantly fluctuates around a suitable level [71]. The high level of collective action that is formed by this process gradually mitigates the land use pattern of conflicts. Ultimately, the rural socio-ecological system is resilient and sustainable in a dynamically changing environment (Figure 5b).

7. Conclusions, Implications, and Limitations

7.1. Conclusions

Based on existing research, this study uses collective action theory and focuses on irrigation governance as the research object. How LF affects rural collective action is explored from various dimensions, such as the general social system, the economic situation, and agroecology. The impacts of LF on agricultural production, irrigation system maintenance, the ecological environment, and village social organization are analyzed in a comprehensive manner from both natural science and social science perspectives. The multidimensional impacts of LF on agricultural production, the maintenance of the irrigation system, ecology, and village social organization are examined. The main findings are summarized as follows: (1) When the trend of LF continues to increase without human intervention, LF exerts a negative impact on collective action. (2) The resilience of the socio-ecological system will guide people to take collective actions to adapt to and resist changes in the socio-ecological system when the fragmentation of agricultural land use poses a threat to humans. Thus, the fragmentation of agricultural land will not have a purely negative impact on collective actions. (3) In the case of human–land interaction, the resilience of the social–ecological system will keep LF at a moderate level and the collective action at the most appropriate level; once the level of LF becomes either too high or too low, the socio-ecological system will face collapse, and collective action will cease.

7.2. Implications

As ecological values are increasingly appreciated by society, the ecological, economic, and social values embodied in the promotion of moderate LF make it possible to maintain a moderate level of LF as an effective strategy for achieving sustainable social and ecological systems, combating climate change, and ensuring food security. Based on this, the present paper proposes the following three policy recommendations.

First, in underdeveloped areas, when promoting large-scale land management, it is essential to ensure appropriate levels of LF. Because of the influence of endowments with natural and social resources of a region, specific indicators of LF levels need to be determined based on the actual development situation of the region. Ensuring an appropriate level of LF in regional development can prevent excessive rural class differentiation while enhancing farmers’ risk resilience. Second, governments or rural collectives should support rural communities in establishing irrigation systems. By providing training and financial support, governments or rural collectives can help rural areas manage their irrigation systems collectively; moreover, appropriate resource allocation mechanisms can be established, thus further promoting rural resource sharing and cooperation. Third, when conducting agricultural land reforms, it is crucial to fully understand the opinions and opinions of local farmers. Agricultural land is essential for farmers’ livelihoods; therefore, before reforms related to agricultural land policies can be implemented, it is necessary to conduct assessments based on local conditions. Additionally, governments are encouraged to devolve power to the village collective level during agricultural land reforms, and conflicts should be resolved through negotiation within villages. These measures can effectively address contradictions associated with the agricultural land reform.

7.3. Limitations

As with any research, this study is subject to limitations, and these will provide avenues for future research. First, this paper only considers a single province of China, namely Guangxi; therefore, the generalizability of the results may be limited to a certain extent because of limitations imposed by the study area. In future studies, an expanded sample should be used to test the universality of the conclusions of this study. Second, because of the heterogeneity of local development, this paper does not provide explicit methods or criteria for the determination of the most appropriate LF level. This issue can be further explored in future research. Third, significant differences in topography, climate, and the ecological environment exist between the south and north of China, which may result in different effects and outcomes of LF on collective action. Therefore, follow-up studies should consider further regional differences to broaden the generalizability of the findings. Again, contingencies such as natural disasters are essential factors that affect local production and socioeconomics, and research has demonstrated that contingencies (such as natural disasters and human conflict) may present valuable opportunities for agricultural change [66]. However, specific knowledge on how contingencies such as natural disasters stimulate farmers to participate in collective action and increase the resilience of social–ecological systems still needs to be uncovered, and future research needs to be expanded accordingly.