Ecological Functional Zoning and Conservation Strategies for Agricultural Heritage Sites Based on Ecosystem Service Bundles: A Case Study of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, China

Abstract

As agricultural heritage systems provide crucial ecosystem service functions, conducting functional zoning serves as a fundamental and essential approach to implementing the ecological civilization strategy and promoting targeted conservation and sustainable utilization. Taking the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, a site recognized as a China-Nationally Important Agricultural Heritage System, as a case study, this research integrates the equivalent factor method and the Self-Organizing Map neural network clustering method to evaluate ecosystem service values, identify ecosystem service clusters, and conduct ecological functional zoning. Protection and utilization strategies are subsequently proposed for each functional zone. The results show the following findings: (1) From 2005 to 2020, the total ecosystem service value of the system exhibited a fluctuating yet overall declining trend, decreasing by approximately 0.25%; (2) five ecosystem service clusters were identified, within which services generally showed synergistic relationships, while trade-offs were mainly concentrated between food provision and other ecosystem services; (3) based on these findings, the study area was divided into five functional zones—the Heritage Culture Core Zone, the Ecological Restoration and Conservation Priority Zone, the Industrial Integration and Development Zone, the Ecological–Industrial Transition and Optimization Zone, and the Multi-Value Protection and Exploration Zone. Specific protection and utilization strategies were proposed for each zone. This study provides a novel theoretical perspective and practical reference for rational ecological functional zoning, as well as the protection and sustainable use of agricultural heritage systems.

1. Introduction

As the culmination of human agricultural civilization accumulated over millennia, agricultural heritage systems embody unique production modes and ecological philosophies. These systems demonstrate significant contemporary value in promoting rural revitalization and balancing ecological conservation with economic development [1]. Consequently, their protection and utilization have become global focal points [2,3,4,5]. China pioneered the exploration of agricultural heritage protection, achieving remarkable results. In 2005, the designation of the Qingtian Rice–Fish Co-Culture System in Zhejiang Province as one of the first Globally Important Agricultural Heritage Systems (GIAHS) [6], marked the formal commencement of China’s heritage protection efforts. In 2015, the issuance of regulatory administrative measures established a solid institutional foundation for conservation. In 2025, the national strategy has been further emphasized, calling for strengthened identification, documentation, and sustainable utilization of agricultural heritage systems, thereby elevating their protection to the level of a national strategic priority. These precious heritage systems not only deliver provisioning services such as food production but also serve as strategic resources for advancing rural revitalization and achieving synergetic progress in ecological protection and economic development [7].

However, despite current conservation practices, many agricultural heritage sites are confronting a series of pressing challenges that require urgent solutions [8]. Among these, the gradual decline in traditional farming practices is particularly acute. With the acceleration of urbanization and the impact of modern agricultural technologies, the younger generation’s affinity for traditional farming methods has diminished, leading to the risk of losing vital indigenous agricultural techniques [9]. Simultaneously, public awareness of the ecological functions of agricultural heritage sites remains limited. Attention is often disproportionately focused on their economic or historical value, while their crucial roles in maintaining biodiversity, regulating climate, and conserving water resources are frequently neglected [10]. The persistence of these issues has significantly hindered the effective protection of agricultural heritage systems. Therefore, it is imperative to develop scientifically targeted conservation strategies from the perspective of ecological functions to address these current challenges in heritage protection effectively [11].

Scientific ecological function zoning serves as the cornerstone for achieving differentiated conservation of agricultural heritage sites. By identifying ecosystem services across different regions, researchers can formulate targeted strategies for conservation and sustainable utilization [12]. Current studies have adopted a variety of approaches, such as spatial overlay analysis, hierarchical zoning, empirical methods, and cluster analysis. For instance, Li and Liu applied a comprehensive index method, selecting four ecosystem services—biodiversity conservation, soil retention, water conservation, and nutrient retention—to conduct regional zoning of the Three Gorges Reservoir Area using RS and GIS technologies [13]. Similarly, Wang et al. employed a spatially constrained K-means clustering model to identify ecosystem service clusters and delineate ecosystem service management zones in the Yangtze River Delta region [14]. While these studies provide valuable methodological references for ecological function zoning, they exhibit certain limitations. Most existing research focuses on macro-scale administrative divisions—such as national, provincial, or municipal levels—where ecosystem types, climate, and topography are integrated to define large-scale functional regions with distinct ecological characteristics [15]. However, research on micro-scale ecological function zoning remains insufficient, leading to an incomplete understanding of the ecosystem services embedded within specific agricultural heritage sites. Consequently, conservation strategies often lack precision. Compounded by practical challenges such as rural depopulation and the decline of traditional farming practices [16,17,18], developing precise ecological function zoning has become a critical entry point. To address the aforementioned issues, this study aims to construct a micro-scale research framework for ecological functional zoning in agricultural heritage sites. Specifically, this study aims to achieve three main objectives: (1) to quantitatively assess the temporal evolutionary characteristics of ecosystem service value (ESV) using the equivalent factor method; (2) to identify the composition of ecosystem service bundles (ESBs) utilizing the Self-Organizing Map (SOM) clustering algorithm, thereby revealing the mechanisms of trade-offs and synergies among service functions across different regions; (3) to delineate functional zones based on the characteristics of these bundles and propose differentiated strategies for conservation and utilization. The Mountain Spring Water Fish Farming System in Kaihua, Zhejiang was selected as an empirical case study (Figure 1). This research intends to provide a theoretical basis and practical paradigms for the precise conservation and sustainable development of both the local site and similar agricultural heritage systems [19].

2. Study Area and Methods

2.1. Study Area

Kaihua County, located in the western part of Zhejiang Province and the northwestern area of Quzhou City, lies at the headwaters of the Qiantang River and borders Anhui and Jiangxi provinces. The county is situated between 28°54′30″–29°29′59″ N and 118°01′15″–118°37′50″ E, extending approximately 66 km from north to south and 59.2 km from east to west, with a total area of 2236.61 km². It administers 14 towns and townships, comprising 255 administrative villages. The Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, located in the mid-mountain terrain of Kaihua County (Figure 2), was inscribed in January 2020 on the Fifth List of China-NIAHS. It is also the only nationally recognized agricultural heritage system in Quzhou City, Zhejiang Province. The system features a unique composite ecological structure of “forest, terrace, tea garden, stream, pond, and villages,” characterized by high ecological, landscape, cultural, and economic values. In December 2024, this system was included in the tentative list of the fifth batch of GIAHS, marking an important milestone that enables Kaihua’s traditional fish-rearing practices to engage in broader international dialogue and gain access to a larger global platform.

2.2. Data Source

To ensure the scientific validity and historical representativeness of the analysis, this study constructed a temporal dataset spanning from 2005 to 2020 at five-year intervals. The selection of this observation period is based on three key considerations: (1) historical significance, as 2005 marks the inauguration of the GIAHS program in China; (2) data consistency, as these time points align with China’s 11th to 13th Five-Year Plan cycles; and (3) socio-economic context, as this timeframe captures the impact of rapid urbanization on traditional agricultural systems. Based on this temporal framework, the data used in this study primarily included national land-use datasets with a spatial resolution of 30 m for the years 2005, 2010, 2015, and 2020, obtained from the Resource and Environment Science and Data Center, Chinese Academy of Sciences [20,21] (https://www.resdc.cn/). In addition, data on the major grain cultivation areas, yields, costs, and profits in the study area for the same years were collected from the Quzhou Statistical Yearbook and the Compilation of National Agricultural Product Cost–Benefit Data.

2.3. Methods

2.3.1. Ecosystem Service Value Assessment Method

The equivalent factor method integrates the research findings of numerous scholars [22,23]. By introducing predefined equivalent factors, it quantifies ecosystem service functions in terms of their economic value [24,25]. This method offers several advantages, including a concise and transparent assessment process and the use of comprehensive and standardized variables. Given that agricultural heritage sites possess unique land-use characteristics [26,27], a trait shared by the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, the equivalent factor method is particularly suitable. It can comprehensively capture the system’s diverse ecosystem services, such as water conservation, climate regulation, and agricultural production, encompassing land-use types like forest land, water bodies, and cultivated land [28,29,30]. Although some ecosystem services lack direct market transaction prices, this method effectively addresses the challenge of unifying the quantification of multiple services by constructing a reference system based on standard farmland output as a general equivalent. This approach ingeniously transforms non-market services into comparable value metrics. Crucially, the ESV calculated in this study is primarily intended to reveal the spatial differentiation characteristics and evolutionary trends of ecosystem services within the system. It serves as a relative quantitative indicator for functional zoning rather than pursuing absolute monetary accounting. Consequently, after comprehensive consideration, the equivalent factor method was selected as the primary approach for ESV assessment in this study.

In this method, the economic value of one standard equivalent factor unit (E_a) per hectare (1 hm²) is defined as one-seventh of the market value of the average grain yield per unit area. To mitigate the impact of price fluctuations, the average market prices of rice, wheat, maize, and soybean in 2005, 2010, 2015, and 2020 were adopted from the Compilation of National Agricultural Product Cost–Benefit Data. Based on calculations, the market value of grain crops per unit area within the system was determined to be 14,427.02 yuan/hm², and thus the value of one equivalent factor, representing one-seventh of this amount, is 2061 yuan/hm². The specific formula, derived from the method proposed by Xie Gaodi [31], is expressed as follows:

$$E_a={\displaystyle \frac{1}{7}}{\sum }_n^i{\displaystyle \frac{M_i{Q}_i{P}_i}{M}}$$

(1)

where $M_i$ represents the planting area of crop type i (hm²); $Q_i$ denotes the yield per unit area of crop type i (t/hm²); $P_i$ is the national average market price of crop type i in the corresponding year (yuan/t); and M is the total planting area of all crops.

Based on the land-use characteristics of the study area and local realities, adjustments to land-use types were made based on the equivalent factor table developed by Xie Gaodi. The equivalent factors for paddy fields and dryland correspond directly to the respective categories in Xie Gaodi’s table. For forest areas, the equivalent factor for coniferous and broad-leaved mixed forests was adopted, informed by field surveys and literature review. Shrubland was assigned the equivalent factor for shrubs. Depending on actual vegetation conditions, high-coverage grasslands were assigned the equivalent factor for shrub-grassland mixtures. For medium-coverage and low-coverage grasslands, the equivalent factors derived for high-coverage grasslands were adjusted by applying coefficients of 0.8 and 0.6, respectively. The equivalent factors for reservoirs, ponds, and water systems were applied directly. For built-up land, the equivalent factor was adapted from the research findings of Chen [32], while bare land was assigned the equivalent factor for barren land. Through these adjustments, a customized equivalent factor table specific to the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang was compiled (

Table 1. Equivalent factor table of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang.

Land Use Type	Provisioning Services			Regulating Services				Supporting Services			Cultural Services
	Food Production	Raw Material Production	Water Supply	Gas Regulation	Climate Regulation	Environmental Purification	Hydrological Regulation	Soil Conservation	Nutrient Cycling	Biodiversity	Aesthetic Landscape
Paddy field	1.36	0.09	−2.63	1.11	0.57	0.17	2.72	0.01	0.19	0.21	0.09
Dryland	0.85	0.4	0.02	0.67	0.36	0.1	0.27	1.03	0.12	0.13	0.06
Forest land	0.31	0.71	0.37	2.35	7.03	1.99	3.51	2.86	0.22	2.6	1.14
Shrub-land	0.19	0.43	0.22	1.41	4.23	1.28	3.35	1.72	0.13	1.57	0.69
Open forest land	0.248	0.568	0.296	1.88	5.624	1.592	2.808	2.288	0.176	2.08	0.912
High-coverage grassland	0.38	0.56	0.31	1.97	5.21	1.72	3.82	2.4	0.18	2.18	0.96
Medium-coverage grassland	0.304	0.448	0.248	1.576	4.168	1.376	3.056	1.92	0.144	1.744	0.768
Low-coverage grassland	0.228	0.336	0.186	1.182	3.126	1.032	2.292	1.44	0.108	1.308	0.576
Reservoirs and ponds	0.8	0.23	8.29	0.77	2.29	5.55	102.24	0.93	0.07	2.55	1.89
Built-up land	0	0	0	0	−3.87	−2.87	−4.53	1.65	0	0	2.08
Bare land	0	0	0	0.02	0	0.1	0.03	0.02	0	0.02	0.01

).

2.3.2. Analysis Method of Ecosystem Service Trade-Offs and Synergies

Trade-offs and synergies among ecosystem services refer to the relationships of mutual constraints or mutual enhancement between two or more ecosystem service functions within an ecosystem. To explore these relationships, Spearman non-parametric correlation analysis was applied to the identified ecosystem service clusters [33,34,35]. The analysis was conducted using the corrplot package in R to visualize the correlation patterns among the eleven ecosystem services. Each cluster contained 55 pairwise correlations. Quantitatively, correlation coefficients range from −1 to 1: values below zero indicate trade-offs, whereas those above zero indicate synergies. Coefficients exceeding 0.5, and particularly those above 0.7, were interpreted as strong correlations. Furthermore, the trade-offs and synergies were analyzed independently for each of the five ecological functional zones. This approach helps to validate the identification of ecosystem service clusters and provides a scientific basis for zonal management of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang [36,37]. The specific formula is as follows:

$$r={\displaystyle \frac{\sum i(x_{ij}-\overline{x})(y_{ij}-\overline{y})}{\sqrt{\sum i{(x_{ij}-\overline{x})}^2\sum {(y_{ij}-\overline{y})}^2}}}$$

(2)

where r is the correlation coefficient ranging from −1 to 1; a positive value of r indicates a synergistic relationship, while a negative value represents a trade-off relationship; $x_{ij}$ and $y_{ij}$ denote the values of different types of ecosystem services.

2.3.3. Ecological Function Zoning Method

For the identification of ecosystem service bundles, the SOM neural network model, specifically the SOM clustering algorithm, was adopted. In this study, model construction and computation were conducted on the RStudio 4.4.3 platform utilizing the kohonen package [38]. The specific workflow is as follows: First, the values of the eleven ecosystem services assessed previously were selected as input variables. To eliminate dimensional discrepancies, a Z-score standardization was performed on the data. To ensure temporal consistency in classification criteria across the four study periods (2005, 2010, 2015, and 2020), the data were integrated into a unified dataset comprising 14,044 grid units, with 3511 cells for each year. Second, to ensure the scientific rigor of the zoning, the Within-Cluster Sum of Squares (WCSS) method was employed to optimize the number of clusters. Based on the “elbow” method, where the WCSS curve flattens significantly after a certain point, the optimal number of clusters was determined to be five. Subsequently, the SOM model parameters were configured as follows: the network topology was set to rectangular, and the grid dimension was defined as 5 × 1 to correspond to the five identified service bundles. Additionally, the neighborhood radius was set to 1, while the number of iterations remained at the default value to ensure network convergence. Through a competitive learning algorithm, the model mapped high-dimensional ecosystem service data onto a low-dimensional output layer, thereby achieving automatic clustering and functional identification of the sample units (see Supplementary Materials, File S1). The SOM clustering algorithm effectively handles complex high-dimensional data without strictly requiring a predefined number of clusters. Moreover, it intuitively reveals the distribution characteristics and intrinsic structures of the data, facilitating a deeper understanding of the spatial and functional heterogeneity of ecosystem services [39].

3. Results

3.1. Spatiotemporal Variation in Ecosystem Service Values

3.1.1. Temporal Variation

Using the Reclassify tool in ArcGIS software 10.8, the land use types within the system conservation area were reclassified into six primary categories: cultivated land, forest land, grassland, water bodies, construction land, and unused land. Based on the changes in the area of these land use types, a Sankey diagram was generated to visualize the transition flows (Figure 3). Additionally, the calculated area variations were plotted as line charts (Figure 4). Figure 4 illustrates the temporal variation in the ESV of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang from 2005 to 2020. As shown, the overall ESV exhibited a gradual declining trend during this period. Specifically, from 2005 to 2010, the ESV decreased by 9.2564 million yuan, representing the largest decline among the three intervals. Between 2010 and 2015, the ESV showed a slight rebound, increasing by 2.3854 million yuan. However, from 2015 to 2020, the ESV declined again by 2.5624 million yuan, although the rate of decline was smaller than that observed from 2005 to 2010.

In terms of land-use types, the ESV of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, from 2005 to 2020 was highest for forest land, followed by open forest land, shrubland, high-coverage grassland, paddy fields, reservoirs and ponds, dryland, low-coverage grassland, medium-coverage grassland, and bare land, while built-up land exhibited the lowest ESV. The ESV of paddy fields, dryland, shrubland, open forest land, high-coverage grassland, and medium-coverage grassland exhibited similar temporal patterns: increasing during 2005–2010 and 2015–2020 but declined during 2010–2015. For forest land, the ESV showed a declining trend during 2005–2010 and 2015–2020, with a slight rebound observed during 2010–2015. The ESV of low-coverage grassland decreased continuously from 2005 to 2015, followed by an increase during 2015–2020. For reservoirs and ponds, the ESV remained relatively stable from 2005 to 2010, increased between 2010 and 2015, and then slightly declined from 2015 to 2020. The ESV of built-up land exhibited a consistent upward trend from 2005 to 2020, with the most pronounced increase occurring during 2010–2015, whereas the ESV of bare land remained unchanged throughout the study period (

Table 2. ESV for different land-use types from 2005 to 2020 (million yuan).

Land Use Type	2005	Variation	2010	Variation	2015	Variation	2020
Paddy field	5389.67	13.71	5403.38	−20.06	5383.32	22.30	5405.62
Dryland	1927.45	23.21	1950.66	−15.92	1934.74	14.13	1948.87
Forest land	328,050.21	−4025.13	324,025.08	337.93	324,363.00	−330.65	324,032.36
Shrubland	11,178.31	85.54	11,263.85	−23.15	11,240.70	18.35	11,259.05
Open forest land	12,049.54	2660.92	14,710.46	−18.85	14,691.61	14.39	14,706.00
High-coverage grassland	9937.55	304.97	10,242.52	−29.58	10,212.93	18.63	10,231.56
Medium-coverage grassland	773.12	32.14	805.26	−5.26	800.00	3.80	803.80
Low-coverage grassland	1291.60	−21.69	1269.91	−11.83	1258.07	12.27	1270.34
Reservoirs and ponds	2066.66	0.00	2066.66	23.30	2089.96	−30.29	2059.67
Built-up land	−333.71	0.70	−333.01	1.96	−331.05	0.84	−330.21
Bare land	0.37	0.00	0.37	0.00	0.37	0.00	0.37

).

In terms of ecosystem service types, regulating services accounted for the largest proportion, exceeding 65% of the total ESV throughout the study period. Supporting services ranked second, contributing approximately 24%, followed by provisioning services, while cultural services represented the smallest proportion. From 2005 to 2020, the overall variation in ESV across the four service categories was relatively small, with the most pronounced changes occurring during 2005–2010. Specifically, within provisioning services, the value of food production continuously declined from 2005 to 2015, followed by an upward trend during 2015–2020. The values of raw material production and water supply decreased from 2005 to 2010, increased during 2010–2015, and declined again between 2015 and 2020. For regulating services, the values of gas regulation, climate regulation, environmental purification, and hydrological regulation all decreased during 2005–2010, with climate regulation showing the largest reduction of 3.0366 million yuan. During 2010–2015, all four regulating services increased, with climate regulation exhibiting the greatest rise of 0.7542 million yuan. From 2015 to 2020, the values of all four regulating services declined again. Regarding supporting services, the values of soil conservation, nutrient cycling, and biodiversity all decreased during 2005–2010, with soil conservation experiencing the greatest decline of 1.1502 million yuan. Between 2010 and 2015, all three services exhibited similar upward trends, led by biodiversity, which increased by 0.2725 million yuan. However, during 2015–2020, all three services declined again. For cultural services, the value of aesthetic landscape decreased during 2005–2010 and 2015–2020, while showing an increase between 2010 and 2015 (

Table 3. Individual ESV from 2005 to 2020 (million yuan).

Ecosystem Service Type		2005	Variation	2010	Variation	2015	Variation	2020
Provisioning services	Food production	7243.29	−1.45	7241.85	−7.14	7234.70	7.25	7241.95
	Raw material production	11,435.72	−27.93	11,407.79	5.82	11,413.61	−6.35	11,407.26
	Water supply	2302.48	−24.81	2277.67	19.06	2296.73	−21.26	2275.47
Regulating services	Gas regulation	38,722.96	−91.57	38,631.39	17.42	38,648.81	−18.48	38,630.33
	Climate regulation	110,658.76	−303.66	110,355.09	75.42	110,430.52	−77.59	110,352.93
	Environmental purification	31,547.95	−81.38	31,466.57	21.98	31,488.54	−22.38	31,466.17
	Hydrological regulation	61,868.91	−115.79	61,753.13	39.40	61,792.52	−44.91	61,747.61
Supporting services	Soil conservation	45,449.32	−115.02	45,334.30	26.82	45,361.12	−29.59	45,331.53
	Nutrient cycling	3767.73	−8.02	3759.71	0.97	3760.68	−1.04	3759.64
	Biodiversity	41,173.57	−108.38	41,065.19	27.25	41,092.44	−28.83	41,063.61
Cultural services	Aesthetic landscape	18,160.07	−47.63	18,112.44	11.54	18,123.98	−13.05	18,110.93

).

3.1.2. Spatial Variation

Using the Create Fishnet tool in ArcGIS 10.8, the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang Protection Area was divided into 3511 square grid cells. The Tabulate Area tool was then employed to calculate the area of each land-use type within every grid cell, based on which the ESV of each grid cell was derived. This procedure enabled the spatial visualization of ESV for different years within the system. For the classification of ESV, the Natural Breaks (Jenks) method was applied to divide the total ESV into five levels: low value areas (−0.29309089 to 0.37059418 million yuan), relatively low value areas (0.37059419 to 0.72503663 million yuan), medium value areas (0.72503664 to 0.98749126 million yuan), relatively high value areas (0.98749127 to 1.18728810 million yuan), and high value areas (1.18728811 to 3.68521145 million yuan).

As shown in Figure 5, the overall ESV of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, corresponds closely with the local land-use patterns. Among all categories, high value areas accounted for the largest proportion, and were primarily distributed in the central and western parts of the protection area, forming large and continuous patches. Field verification indicates that these regions mainly consist of forest land and open forest land. The relatively high value and medium value areas largely overlapped with paddy fields and dryland, while relatively low value areas were mostly distributed across the northwestern and eastern parts of the system. Low value areas were concentrated mainly in the eastern region, with some scattered patches along the periphery; these areas were predominantly built-up land where local residents are concentrated. Regarding spatial evolution, high value areas occupied the largest proportion in 2005. From 2005 to 2010, the relatively low value areas, medium value areas, and relatively high value areas increased in scattered patches, while high value areas decreased correspondingly in both area and proportion. During 2010–2015, the spatial distribution of relatively low value areas, medium value areas, and relatively high value areas showed minor positional adjustments with limited overall change. From 2015 to 2020, both medium value areas and relatively high value areas expanded, while high value areas further declined in extent. Overall, the medium value areas, relatively high value areas, relatively low value areas, and low value areas of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, exhibited a slight westward expansion trend, although the magnitude of change remained small. In contrast, high value areas demonstrated a gradual decline in their overall proportion.

3.1.3. Sensitivity Analysis

To verify the potential impact of value coefficient selection on subsequent spatial analysis and functional zoning, the Coefficient of Sensitivity (CS) was introduced in this study. Although the primary aim of this research is to reveal spatial relative differentiation through ecosystem service values rather than to pursue absolute monetary accounting, the stability of the coefficients is directly related to the reliability of the inputs for zoning. By adjusting the ecological value coefficients for each land-use type by an increase and decrease of 50%, the calculation results (Figure 6) indicate that the CS values for all land-use types were consistently far less than 1, with the highest value being 0.87 for forest land. This result suggests that the spatial distribution pattern of ESV in the study area exhibits strong robustness. Its core driving force stems from the spatial structure of land use rather than the numerical settings of individual equivalent factors. Consequently, even if the coefficients fluctuate within a reasonable range, the ecological rank order among grid cells remains stable. Therefore, the ecosystem service clusters identified based on this spatial pattern of ESV, as well as the finally delineated functional zoning boundaries, possess high robustness and are unlikely to undergo significant shifts or deviations due to the subjective selection of parameters.

3.2. Ecological Function Zoning Based on Ecosystem Service Clusters

3.2.1. Identification Results of Ecosystem Service Clusters

The determination of the optimal number of clusters is presented in Figure 7. As indicated by the overall trend, the WCSS values decreased significantly as the number of clusters increased. The curve exhibits a steep slope as the number of clusters ranges from 1 to 4; however, after the number of clusters reaches 5, the downward trend flattens. This suggests that a five-cluster classification effectively explains data heterogeneity while avoiding the redundancy associated with over-classification. Consequently, the optimal number of clusters for this study was determined to be five. Subsequently, the classification results were visualized to provide an intuitive and clear representation of the spatial distribution patterns and temporal dynamics of ecosystem service clusters across different years [40].

In the counts plot (Figure 8), the number of circles represents the distinct categories of ecosystem service clusters, while the color intensity of each circle denotes the sample size within that cluster. The results showed that the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, was categorized into five distinct ecosystem service clusters. Specifically, three clusters contain fewer than 2000 samples, one cluster comprises between 2000 and 4000 samples, and one cluster encompasses more than 6000 samples.

The codes plot illustrates the specific classification of ecosystem service clusters within the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang (Figure 9). It is important to note that to eliminate dimensional differences among various ecosystem services, the data presented are relative weight values processed by Z-score standardization. The origin represents the regional average level of the service (Z = 0), and the size of the fan area (or bar length) reflects the relative dominance of the service within a specific cluster: positive values indicate that the supply capacity is above the average, and larger values (typically > 0.5 standard deviations) signify significant dominant characteristics, whereas negative values indicate limiting services. Through this proportional transformation, the trade-off and synergy structures of services within each functional zone are clearly revealed. The results showed that Cluster A shows no dominant ecosystem service function, with generally low levels across all services. Cluster B maintains moderate levels across all eleven ecosystem services, characterized by water supply, environmental purification, hydrological regulation, and aesthetic landscape services. Cluster C is characterized by dominant functions in food production, nutrient cycling, and gas regulation, while water supply and hydrological regulation remain relatively low. Cluster D is dominated by raw material production, gas regulation, and nutrient cycling, whereas hydrological regulation is relatively weak. Cluster E exhibits dominant functions in raw material production, gas regulation, climate regulation, soil conservation, nutrient cycling, biodiversity, and aesthetic landscape services, with hydrological regulation remaining at a lower level.

3.2.2. Spatiotemporal Variation in Ecosystem Service Clusters

From 2005 to 2020 (Figure 10), the ESV of Clusters A and E showed an overall declining trend, whereas that of Cluster C exhibited a steady increase. In Cluster B, all ESV indicators increased except for food production, water supply, and hydrological regulation. In Cluster D, only water supply displayed a downward trend. The spatial visualization of the ecosystem service cluster identification results from 2005 to 2020 was conducted using ArcGIS 10.8. Overall, the spatial distribution of ecosystem service clusters in the Zhejiang Kaihua Mountain Spring Water Fish Farming System showed only minor spatial changes throughout the study period. Cluster A covered a relatively small area, mainly distributed in the eastern region, with several scattered patches across the protection area; its area proportion fluctuated slightly, showing a small increase in 2020 compared with 2005. Cluster B had the smallest area, primarily distributed across reservoirs and ponds. Cluster C was mainly concentrated in the eastern region, with several strip-shaped patches extending into the central and western parts; its area proportion increased from 2005 to 2010, remained stable from 2010 to 2015, and decreased from 2015 to 2020. Cluster D occupied the second-largest area proportion, located adjacent to Cluster C and serving as a transitional zone between Clusters C and E. Its area proportion increased during 2005–2010, decreased during 2010–2015, and increased again during 2015–2020, showing an overall upward trend during the study period. Cluster E, with the largest area proportion, was mainly distributed in the central and western regions, forming continuous patches. Its area proportion decreased from 2005 to 2010, increased from 2010 to 2015, and decreased again from 2015 to 2020, showing an overall downward trend during the study period.

3.2.3. Trade-Offs and Synergies Among Ecosystem Service Clusters

The Spearman correlation analysis revealed distinct interaction patterns among the eleven ecosystem services within the five identified bundles. The results showed that in Cluster A (Figure 11a), except for a trade-off between food production and water supply, most ecosystem services exhibited synergistic relationships. This finding reflects the interdependence and mutual reinforcement among different ecosystem functions, which enhances ecosystem stability and functional performance. In Cluster B (Figure 11b), most ecosystem services also displayed synergistic interactions, with the dominant services—water supply, environmental purification, hydrological regulation, and aesthetic landscape—exhibiting strong synergies. Trade-offs mainly occurred between food production, raw material production, and the remaining services, suggesting that the relationships between provisioning services and other ecosystem services within this cluster require careful management.

In Cluster C (Figure 11c), the overall ecosystem services exhibited synergistic relationships. The dominant services—food production, nutrient cycling, and gas regulation—were mutually synergistic, with strong synergies observed between nutrient cycling and gas regulation. The synergy between food production and the other two services was relatively weak. Trade-offs were mainly concentrated between food production and other ecosystem services, suggesting that while promoting industrial integration, attention should be paid to maintaining the balance among ecosystem functions. In Cluster D (Figure 11d), synergistic relationships were also predominant. The synergies between food production, water supply, and other ecosystem services were relatively weak, with some trade-offs observed. The dominant services—raw material production, gas regulation, and nutrient cycling—exhibited strong mutual synergies, indicating a stable internal functional structure within this cluster.

In Cluster E (Figure 11e), the dominant services—raw material production, gas regulation, climate regulation, soil conservation, nutrient cycling, biodiversity, and aesthetic landscape—exhibited strong synergistic relationships with one another. The synergy between hydrological regulation and other ecosystem services was relatively weak. Except for nutrient cycling, food production showed trade-off relationships with the other ecosystem services, with a particularly strong trade-off observed between food production and water supply.

3.2.4. Process and Results of Ecological Function Zoning

To achieve the transformation from the attribute classification of “ecosystem service clusters” to the spatial management of “ecological functional zoning,” this study adopted an “attribute clustering-spatial mapping” zoning strategy. Specifically, based on the five ecosystem service clusters identified by the SOM model, the cluster identifiers (Cluster IDs) from the clustering results were mapped back onto the geospatial grid according to the principle of functional similarity, thereby forming the final ecological functional zoning pattern. Unlike traditional macro-zoning based on administrative boundaries, this study adhered to the principle of refined management and retained the spatial heterogeneity details at the grid scale. Although this results in zoning patterns characterized by certain fragmented patch mosaics, it more authentically reflects the micro-scale ecological functional differentiation of the agricultural heritage site under complex terrain conditions. This provides a scientific basis for high-resolution classified protection and precision utilization. Based on this, the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang was divided into the following five zones with distinct functions: the Heritage Culture Core Zone, the Ecological Restoration and Conservation Priority Zone, the Industrial Integration and Development Zone, the Ecological–Industrial Transition and Optimization Zone, and the Multi-Value Protection and Exploration Zone (

Table 4. Ecological Functional Zoning.

Ecological Function Zoning	Ecosystem Service Clusters	Description
The Heritage Culture Core Zone	Cluster A	The area is predominantly designated as built-up land, characterized primarily by village settlements. Serving as a core repository of the cultural heritage of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, it preserves a rich array of traditional folk customs, historical relics, and heritage craft techniques. This area plays a pivotal role in sustaining and promoting the region’s distinctive agricultural culture and in maintaining the continuity of local farming traditions.
The Ecological Restoration and Conservation Priority Zone	Cluster B	The area is primarily characterized by reservoirs and ponds, which serve as distinctive and indispensable components of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang. Given their critical ecological roles, this area should be accorded priority protection to strengthen the ecological foundation that underpins the sustainable development of the entire system.
The Industrial Integration and Development Zone	Cluster C	The area comprises various land-use types, including paddy fields and dryland, which correspond to the terraced fields and tea plantations within the system. As the core zone for food production, it possesses distinct agricultural advantages that should be fully leveraged to promote the integrated development of agriculture and related industries.
The Ecological–Industrial Transition and Optimization Zone	Cluster D	The area comprises land-use types such as dryland and grassland and is located within the transitional zone between Cluster C and Cluster E. It faces dual demands for ecological conservation and industrial development. On the one hand, it is essential to strictly uphold the ecological red line, strengthen the protection of dryland and grassland ecosystems, and prevent ecological degradation resulting from overexploitation. On the other hand, the area’s ecological functions can be appropriately harnessed to regulate the local environment and support sustainable industrial development.
The Multi-Value Protection and Exploration Zone	Cluster E	The area consists primarily of forest land, which exhibits the highest ecosystem service value and covers the largest proportion of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang, thereby forming the ecological foundation of the entire system. Accordingly, the area’s diverse ecological, cultural, and economic values should be systematically identified and sustainably harnessed, under the strict prerequisite of safeguarding the natural ecosystem from any form of degradation.

, Figure 12).

4. Discussion

4.1. Spatiotemporal Variations and Driving Factors of Ecosystem Service Value in the Zhejiang Kaihua Mountain Spring Water Fish Farming System

The spatiotemporal evolution characteristics of ESV serve as a crucial basis for identifying the operational status of agricultural heritage systems. The research results indicate that from 2005 to 2020, the total ESV of the study area exhibited a fluctuating downward trend. In terms of service composition, regulating and supporting services contributed more than 60% to the total value, whereas provisioning services, which directly generate economic benefits, accounted for a relatively low proportion. This finding suggests that the system currently faces a risk of ecological functional degradation.

Regarding the fluctuating downward trend in the ESV of the study area, although this study did not construct an econometric model for quantitative verification, it is possible to infer, based on field investigations and cutting-edge literature, that socio-economic transformation is a key inducing factor. Previous studies have confirmed that labor shortages caused by rural population aging pose a serious threat to the sustainability of smallholder farming in China. As farmers age, their capacity for refined management declines, directly leading to farmland abandonment and reduced utilization efficiency [41]. This conclusion is highly consistent with the phenomenon of fish pond abandonment due to lack of maintenance observed during our field surveys. Furthermore, compounded by the transformation of livelihood strategies under the impact of the market economy, farmers’ willingness to invest in low-return traditional agriculture has diminished. The combined effect of these factors may have led to the functional decline of the agricultural heritage system, which relies heavily on refined management.

4.2. Protection and Utilization Strategies for the Zhejiang Kaihua Mountain Spring Water Fish Farming System Based on Ecological Functional Zoning

The spatial heterogeneity of ecosystem services dictates that the protection of agricultural heritage cannot rely on a single, uniform administrative management model. Based on the SOM clustering results, this study identified five distinct ecosystem service clusters with varying characteristics, thereby confirming the necessity of implementing differentiated management at the micro-scale. Given the differences in the dominant functions of each service cluster, this study proposes the following targeted zoning management strategies:

The Heritage Culture Core Zone, centered on traditional villages such as Chaijia Village and Lulian Village in Hetian Township, serves as the origin and central area of the Mountain Spring Water Fish Farming Technique. For this region, a strategy for the protection of agricultural cultural authenticity should be implemented [42]. On one hand, relying on traditional village protection funds, the restoration of ancient fish ponds, ancestral halls, and traditional dwellings within the region must strictly follow the principle of “repairing the old as the old.” The volume and style of new constructions must be rigorously controlled to maintain the integrity of the heritage site’s historical landscape. On the other hand, focus should be placed on the living utilization and empowerment of culture. Relying on the Qingshui Fish Museum and the landscape of the core villages, an agricultural heritage systems field study base should be established [43]. Furthermore, the value of intangible cultural heritage should be deeply excavated. Drawing upon the ecological protection concepts inherent in traditional celebrations, such as the “Xianghuo Caolong” in Suzhuang Town, the “Baomiao Festival” in Pingkeng Village, and the “Gufu Festival” in Tangtou Village, events like the “Qingshui Fish Cultural Festival” should be planned and organized. Through community participation mechanisms, indigenous residents should be guided to engage in living displays, thereby enhancing cultural cohesion and establishing the core zone as a window for showcasing the unique value of the heritage [44].

The Ecological Restoration and Conservation Priority Zone is a critical area for maintaining the system’s water cycle. The control focus for this region lies in constructing a comprehensive water environment security barrier. First, a dynamic water quality monitoring system should be established [45,46]. Vegetation buffer zones should be constructed along stream banks to reduce non-point source pollution. Furthermore, ecological red lines must be strictly delineated to prohibit high-intensity development and modern feed-based aquaculture, ensuring the purity of the traditional ancient farming environment. Second, under the premise of ensuring ecological security, an ecological experience model with extremely low environmental impact should be explored. For example, utilizing the unique high-altitude “terrace–rapeseed flower–fish pond” composite landscape of Gaotiankeng Ancient Village in Changhong Township, activities such as agricultural landscape photography and nature science popularization education should be developed. This transforms the well-preserved traditional farming landscape into ecological value, avoiding the development of high-load consumptive tourism and ensuring the sustainable development of the core water source conservation area [47].

The Industrial Integration and Development Zone, represented by areas with convenient transportation and abundant arable land resources, serves as the system’s primary base for grain and raw material production. This region should fully leverage its locational advantage of being adjacent to Majin Industrial Park and highway interchanges to promote agricultural industrialization and the deep integration of secondary and tertiary industries [48]. On one hand, moderate-scale ecological farming should be developed, focusing on extending the industrial chains of characteristic agricultural products such as soybeans and Qingshui Fish, and developing high-value-added processed products. On the other hand, the authorization and franchising system for the “Qianjiangyuan” regional public brand should be implemented. Products meeting the standards should be sold under a unified label to achieve agricultural product premiums. At the same time, leveraging its advantages as a commercial and logistics hub at the gateway to the national park, a regional agricultural product distribution center should be established to promote the transformation and upgrading of industrial forms from low-value-added raw material output to high premium product processing and service provision [49].

The Ecological–Industrial Transition and Optimization Zone, typically represented by Suzhuang Town, is situated in the ecotone between production and ecological areas. Characterized by a high degree of landscape fragmentation, it undertakes critical ecological buffering functions. The control strategy should aim to construct a composite system of ecological buffering and understory economy. Regarding landscape fragmentation restoration and the construction of ecological corridors, priority should be given to utilizing native tree species to establish landscape ecological corridors. These corridors will connect fragmented habitats and reinforce the zone’s function as an ecological barrier on the periphery of core ecological areas such as the Gutianshan National Nature Reserve. Secondly, utilizing the unique high-altitude climate and abundant woodland resources, the understory economy and characteristic agriculture in high mountains should be vigorously developed. For example, the planting of high-value-added Chinese herbal medicines under the forest, such as Polygonatum and Dendrobium officinale, should be promoted. Additionally, high-mountain vegetables should be moderately developed. This approach aims to increase farmers’ income without compromising forest cover, thereby establishing a sustainable livelihood model [50,51].

The Multi-Value Protection and Exploration Zone is dominated by large areas of contiguous forest land. Centered on Qixi Town and covering the core area of the Qianjiangyuan National Park, it constitutes the foundation with the largest area and the highest ecological function grade in the entire agricultural heritage system. The control strategy should promote the advanced transformation of ecological product values. First, focus should be placed on regulating services and ecological compensation mechanisms. Given the zone’s critical role in carbon sequestration and climate regulation as the source of the Qiantang River, priority should be given to establishing an ecological compensation mechanism based on forestry carbon sinks. Models for carbon sink trading and green finance should be explored to effectively realize the internalization of external ecological values. Second, the high-end forest wellness industry should be prioritized for development. Relying on the unique headwater ecological environment of Qixi Town, and under the premise of strictly protecting the forest structure, immersive experience projects such as forest healing and summer wellness retreats should be developed. These projects should utilize the high concentration of negative oxygen ions and cool climate resources. This approach aims to achieve an industrial upgrade from the supply of primary forest products to the supply of high-quality ecological services, realizing a deep synergy between ecological protection and green development.

4.3. Dynamic Utilization and Integrated Industrial Development of the Zhejiang Kaihua Mountain Spring Water Fish Farming System

As a typical coupled socio-ecological system, the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang embodies a high degree of integration among ecological, economic, and cultural dimensions. This study recognizes that maximizing its overall value depends on fully tapping the potential of its ecosystem services and establishing a multi-dimensional, multi-level value transformation framework.

At the level of ecological value transformation, the regulating and supporting services of the Mountain Spring Water Fish Farming System constitute the largest proportion of its total ESV. By establishing an ecological asset accounting system and exploring mechanisms such as ecological compensation and carbon sequestration trading, ecological benefits can be converted into sustainable economic returns, thereby promoting the long-term accumulation of ecological capital. At the level of industrial integration and economic development, the Industrial Integration and Development Zone and the Ecological–Industrial Transition and Optimization Zone should serve as the core areas. Priority should be given to developing characteristic industries such as organic tea plantations and ecological aquaculture to enhance land-use efficiency and ecological productivity per unit area. Under the premise of safeguarding ecological carrying capacity, new business models—such as ecotourism and educational experiential programs—should be introduced to facilitate a transition from primary production to ecological experiences and cultural consumption. At the cultural level, the Mountain Spring Water Fish Farming System embodies unique traditional fish-rearing techniques. Its cultural value should be revitalized through cultural presentation and educational dissemination. The establishment of facilities such as a “Cultural Exhibition Hall” within the Heritage Culture Core Zone, together with the use of digital technologies, can effectively showcase and transmit the heritage cultural system.

4.4. Limitations and Future Research

By introducing ecosystem service clusters and the SOM clustering method, this study analyzed the protection strategies of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang from quantitative and spatial perspectives, forming a beneficial dialogue and supplement to existing international research. Compared with existing GIAHS studies, which mostly focus on the qualitative description of single services or macro-scale evaluation, this study achieved refined zoning at the micro-scale, aligning well with the trends advocated by the international landscape ecology community. However, benchmarked against international frontier research standards, this study still has certain limitations. First, there is a lack of cross-regional comparative studies. This research focused solely on a typical case in Zhejiang, China. Compared to the increasing number of multi-case analyses internationally, single-case studies fall short in terms of the global generalizability of their conclusions, which to some extent limits the external validity of the findings. Second, the localization of value assessment parameters needs improvement. In the ESV assessment, there is currently no unified international standard for the equivalent factor values of special land types such as built-up land. The correction coefficients adopted in this study may not fully capture the unique socio-ecological characteristics of the locality, indicating a gap relative to the requirements for parameter calibration based on local scenarios.

Based on the aforementioned limitations, future research should focus on the following aspects: First, the breadth of international comparative studies should be expanded. It is recommended to establish transnational cooperation networks to conduct parallel comparative studies between China’s Mountain Spring Water Fish Farming System and similar global systems, such as Japan’s Satoyama landscapes and the Philippines’ Ifugao Rice Terraces. The aim is to verify the universality of the “zoning management model based on service clusters” proposed in this study across different geographical contexts and to summarize the universal experiences of GIAHS in balancing the conflict between protection and development, thereby compensating for the deficiencies of single case studies. Second, dynamic adjustment and adaptive management mechanisms should be optimized. Agricultural heritage is a dynamically evolving system. Future research should integrate time-series data to continuously monitor the implementation effects of zoning management measures. Accordingly, management boundaries and strategies should be dynamically adjusted to enhance the adaptability of planning schemes to socio-economic and environmental changes.

5. Conclusions

From 2005 to 2020, the total ESV of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang decreased from 3723.3077 million yuan to 3713.8743 million yuan, showing a fluctuating but overall declining trend. Among the various categories of ecosystem services, regulating services accounted for more than 65%, dominating functions such as climate regulation and flood control; supporting services represented approximately 24%; while provisioning and cultural services were relatively lower, reflecting that the system primarily functions as an ecological regulation-oriented system. Although the value of individual services declined slightly, the reduction was minor, highlighting the system’s strong ecological resilience and stability. Spatial visualization of ESV divided the protected area of the system into five value categories. High value areas were mainly concentrated in the central and western parts of the region, forming the core zones for climate regulation and biodiversity. In contrast, low value areas were located in the southeast, overlapping with zones of intensive human activity. Over the 15-year period, the area of high value zones slightly decreased, while the other four categories expanded correspondingly. However, the overall spatial pattern remained stable, with only minor local changes occurring in peripheral areas affected by agricultural adjustments.

Based on the identification of ecosystem service bundles, the system was divided into five functional zones: the Heritage Culture Core Zone, focusing on preserving traditional villages and agricultural culture; the Ecological Restoration and Conservation Priority Zone, emphasizing ecological restoration and environmental protection; the Industrial Integration and Development Zone, promoting the integration of agriculture, culture, and tourism; the Ecological–Industrial Transition and Optimization Zone, serving as a buffer between core conservation and production areas with controlled development intensity; and the Multi-value Protection and Exploration Zone, concentrating on forested regions and exploring pathways for transforming ecological products into economic value. This zoning scheme effectively clarifies the dominant functions of different spatial units, providing a scientific basis for resolving spatial conflicts between protection and development. This study confirms that by identifying the spatial heterogeneity of ecosystem service clusters, the organic compatibility of production, living, and ecological functions can be achieved at the micro-scale. This research pathway not only provides decision support for the sustainable management of the Mountain Spring Water Fish Farming System in Kaihua, Zhejiang but also offers a Chinese case and experience for similar small-scale agricultural heritage sites globally to transition from “qualitative description” to “quantitative spatial governance.”