Spatiotemporal evolution and driving factors of ecosystem service value in the Upper Minjiang River of China

Abstract

The stability of ecosystems in high mountain canyon areas is poor, and the interaction between humans and the land is complex, making these ecosystems more vulnerable to destruction. Quantitatively assessing the ecosystem service value (ESV) in high mountain canyon areas and revealing its spatiotemporal evolution patterns and driving factors play a crucial role in the construction of regional ecological barriers and the assurance of ecological security. This study focuses on the Upper Minjiang River as the research area, using the InVEST model and the Equivalent Factor Method to estimate ESV. This combination aims to address the inadequacy of the Equivalent Factor Method in reflecting the variability of ESV across different regions, and the sensitivity of the InVEST model to data changes that results in insufficient accuracy of ESV assessments. By harnessing spatial au-tocorrelation and the geodetector method, we unravel the spatiotemporal evolution characteristics and driving factors of ESV. The results show that: (1) From 2000 to 2020, the ESVs estimated by the two estimations increased by 31.28% and 22.47%, respectively, both indicating that the eco-environment quality of the upper Minjiang River has been continuously improved. (2) When Moran’s I was greater than 0.5 (p < 0.05), the spatial clustering of “High-High” and “Low-Low” ESV was obvious. It is clear that the ESV varies geographically. High values are primarily found in the study area’s center and southern regions, as well as on both banks of the Minjiang River, whereas low values are more common in the region’s northern region. (3) Slope and human activity intensity (HAI) are the principal contributors to the spatial differentiation of the ESV, more than 60% of the interaction types between the two factors were classified as dual-factor enhancement. The synergistic reinforcing effects of HAI, slope, elevation, and temperature collectively shape the shifts in ESV spatial distribution. This study offers a novel evaluative lens on the ESV of the Upper Minjiang River area, supplying a sturdy data support for crafting specific ecological preservation and rejuvenation strategies in the coming years.

Introduction

Ecosystem services reflect the holistic contributions of ecosystems to humanity, encompassing supporting, regulating, provisioning, and cultural services¹. We have access to energy, food, clean water, medication, and chances to engage with nature because to these services^1–4. However, due to global climate change and human activities, over 60% of the world’s ecosystems are either degrading or being exploited unsustainably (Millennium Ecosystem Assessment). IPBES (Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services) highlights that while ecosystem services offer essential elements for global economic development, these systems are changing dramatically due to human activities. Consequently, 75% of the Earth’s terrestrial surface has undergone significant transformation, and more than 85% of wetlands have been lost. The degradation or loss of ecosystem services poses direct threats to human health, ecological security, and sustainable development.

In order to maintain the stability of ecosystems and enhance the capacity of ecosystem services, many countries and regions have adopted a series of ecological environmental protection measures^5,6. For example, landscape engineering, ecological restoration project, returning farmland to forest and grassland, soil improvement, and runoff storm water^5–11. These projects have played a great role in ecological conservation and protection, and have become an important way to realize the construction of ecological civilization¹². At the same time, it is important to quantify the capacity of ecosystem services in order to clarify the ecosystem service problems and to regulate and maintain regional ecological security^13,14. Therefore, the study of ecosystem service assessment and the factors affecting the change of ecosystem services has gradually become a research hotspot.

ESV is a way of valuing the capacity of ecosystem services by translating them into monetary units, it become an increasingly emphasized research topic in geography and ecology over the past few decades^15–18. Various assessment methodologies for ESV exist, including ecological footprint analysis, ecological value coefficient method, and market value approach^19–24. The market value approach presents difficulties for ecological services without obvious market values but is appropriate for those with clear market values^25,26. Notably, the Equivalent Factor Method, represented by Costanza¹and Xie²., focuses on converting various ecological services into a unified economic value using equivalent coefficients^27–30. Although these coefficients need to be reevaluated and adjusted, as shown in references^23,24, the approach is widely used in China. The ecological footprint method aims at accounting for human consumption of ecological resources against ecosystem production capacity^33,34, but it may not adequately represent the regional ESV and has certain assessment limitations^35,36. The InVEST model applies the supply-service-value framework, connecting ecological production functions with economic value, and can spatially display ecological and environmental issues³⁷. Its accuracy in ecosystem service assessment represents a qualitative leap compared to Costanza and others^38,39. However, the InVEST model requires a variety of spatial base data and is sensitive to data changes, leading to significant uncertainty in assessment results³⁹. Therefore, more research is needed to determine how to accurately and quickly evaluate ESV in particular places.

In studies on the spatiotemporal evolution of ESV, spatial autocorrelation, Geary’s C, Getis-Ord Gi*, among others, are statistical methods reflecting the spatial relationships of ESV. Global autocorrelation offers a statistical model for the overall spatial distribution pattern of ESV⁴⁰. In contrast, Geary’s C focuses on differences between adjacent observations⁴¹, and Getis-Ord Gi* identifies high or low value clusters⁴². The semivariance function describes spatial continuity and variability⁴³. Spatial autocorrelation has been used more frequently in some ESV spatiotemporal evolution research because of its ease of use and comprehensive descriptive capacity⁴⁴. Current hotspots of research also include drivers and driving forces of ESV spatial differentiation. The drawbacks of factor analysis, principal component analysis, and grey relational analysis for driving factor research include missing important variables and poor presentation of variable interactions⁴⁵. Given these limitations, geodetectors and geographically weighted methods have been adopted for ESV evolution research in arid regions⁴⁶, grasslands⁴⁷, and riverbanks⁴⁸.

High mountain canyon areas are complex spaces where natural processes and human processes intertwine. Due to their unique energy gradients, they are high-incidence areas for mountain disasters and ecologically fragile areas. Species and community changes are more sensitive and unstable under the influence of global changes, and changes in ecosystem service functions are characterized by significant regional, spatial and temporal heterogeneity and uncertainty^5,12,17. Dynamically tracking changes in ecosystem service functions in high mountain canyon areas is of great significance for regional ecological environment protection and ecological barrier construction. This paper takes the upper Minjiang River as a typical case study area. This area is characterized by high mountains, steep slopes, deep valleys, a complex geological environment, loose and fragmented rock masses, frequent seismic geological disasters, poor ecological environment stability, and prominent human contradictions. The aims of this study were therefore to: (1) Use the InVEST model and the Equivalent Factor Method to evaluate the ESV of the Upper Minjiang River, verifying the similarities and differences in the estimation results of the two methods; (2) Reveal the spatiotemporal variation patterns of ESV in the study area, enhancing our understanding of changes in ecosystem service functions in ecologically fragile areas; (3) Analyze the driving factors of ESV spatiotemporal changes to provide data support for regional ecological environment protection and land resource management.

Materials and methods

Study area overview

The upper reaches of the Minjiang River are located in the southwest of China, spanning the coordinates 102°35′E to 103°57′E and 30°45′N to 33°10′N. This region is characterized as a transitional physical space between mountainous regions and ecologically fragile zones. It acts as the transition zone between the Qinghai-Tibet Plateau and the Sichuan Basin. Dominated by plateau and mountainous terrains, the region has altitudes ranging from 743 m to 5927 m. There is a pronounced vertical distribution of climate and natural vegetation, with a rich biodiversity⁴⁹. The geological structure is complex due to tectonic extensions and compressions, leading to the frequent occurrence of natural disasters such as flash floods, mudslides, and landslides. The region experiences considerable vertical temperature fluctuations due to its subtropical monsoon climate, which features distinct seasons and concentrated rainfall. The soil is mostly composed of layers of diverse brown and brown-yellow soil found in mountainous areas. Vegetation primarily consists of broadleaf forests, mixed conifer-broadleaf forests, coniferous forests, shrubs, and meadows. This area has historically been associated with concentrated poverty, which has led to a heavy reliance on natural resources and noticeable environmental disruption from human activity (Fig. 1).

Fig. 1.

Location map of study area. The map of Upper Minjiang River is created in the support of ArcGIS 10.2 (ESRI). The China map were collected from Standard Map Service System (http://bzdt.ch.mnr.gov.cn/). The Tibetan Plateau boundary data were collected from the Global Change Research Data Publishing & Repository (http://www.geodoi.ac.cn/).

Data sources

The data used in this study include elevation, research boundaries, land-use data, evapotranspiration, temperature, annual rainfall, annual average temperature, NDVI, and socio-economic data (Table 1). The land-use data are sourced from the Resource Environment Science and Data Center of the Chinese Academy of Sciences. Land types are classified into ten categories: dry land, paddy fields, shrubs, forests, grasslands, wetlands, barren land, water bodies, glaciers and snow, and built-up areas.

Table 1.

Main data sources and types.

The Data Source	Type	Resolution	Year	Data Source
DEM	Raster	30 m	-	http://www.geodata.cn
administrative boundaries	Vector	--	2020	http://www.geodata.cn
land use	Raster	30 m	2000, 2005, 2010, 2015, 2020	http://www.resdc.cn
potential evapotranspiration	Raster	1 km	2020	http://www.geodata.cn/
soil dataset	Raster	1 km	2014	https://data.tpdc.ac.cn/
temperature	Raster	1 km	2020	http://www.geodata.cn/
annual average rainfall	Raster	1 km	2020	http://www.geodata.cn/
annual average temperature	Raster	1 km	2020	http://www.geodata.cn/
NDVI	Raster		2020	http://www.resdc.cn/
per kilometer of GDP spatial distribution(GDP)	Raster	1 km	2020	http://www.gis5g.com
per kilometer of population spatial distribution (PD)	Raster	1 km	2020	https://landscan.ornl.gov
other socio-economic data	Text	-	2020	“Sichuan Statistical Yearbook” “Aba Prefecture Statistical Yearbook” and “National Economic and Social Development Bulletin”

Methods

This study, leveraging land-use, socio-economic, and hydro-meteorological data, assesses the ecological service value (ESV) of the upper Minjiang River and analyzes its spatiotemporal evolution and driving factors. The research approach is as follows: (1) Based on the InVEST model, we selected three modules: carbon storage value, water conservation value, and soil conservation value. By integrating the opportunity cost method, market value method, and shadow engineering, we calculated the region’s ESV. Using the equivalent factor method and incorporating NPP, rainfall, and population data, we derived an equivalence factor table suitable for the study area to further calculate the ESV. (2) We employed a spatial autocorrelation analysis to reveal the spatiotemporal evolution of the ESV in the study area. (3) Using the geographical detector, we identified the primary drivers of spatial differentiation in the ESV of the upper Minjiang River. The technical approach is detailed in Fig. 2.

Fig. 2.

Technology Roadmap.

ESV calculation based on the invest model

The InVEST model is a technique used to evaluate ecosystem services, which include soil, water, carbon storage, and biodiversity conservation. This study selected these three elements for ESV computation because they are crucial for reducing climate change, maintaining soil health as a major indicator of ecological balance, and impacting water supply and quality. To convert ES to ESV, the shadow engineering technique was applied. (1) Carbon Storage Value Calculation: This study focuses on four pivotal carbon reservoirs: above-ground carbon, underground carbon, soil carbon, and dead organic matter carbon. Based on the region’s geography, climate, and ecological data, the carbon storage was computed. Considering the continuous improvement in China’s carbon trading market over the past decades and the release of the “First Compliance Period Report on China’s Carbon Emission Trading Market” in December 2022. Therefore, the carbon storage value in this paper is based on the average transaction price data published in the report, which is 42.85 yuan/t. Using the conversion factor of 0.2727 between C and CO₂, we calculated the per-unit carbon storage value³⁸.

• The carbon storage value calculation formula is as follows:

1

Here, ‘’ stands for carbon storage value, ‘’ stands for the above-ground carbon pool, ‘’ for the underground carbon pool, ‘’ for the soil carbon pool, ‘’ for the organic matter carbon pool, and ‘’ is the carbon trading price.

• Calculation of Water Conservation Value

When assessing water conservation value, the InVEST model primarily considers factors such as land cover, climate, soil properties, and basin topography⁵⁰. The formula for calculating water conservation value is as follows:

2

Here, ‘’ represents the water conservation value; ‘’ is the annual water yield of the region (m³); ‘’ stands for the annual average rainfall (mm) of grid unit ‘x’; ‘’ signifies the annual average potential evapotranspiration on land use type ‘j’ of grid unit ‘x’; ‘’ points to the dryness index on land use type ‘j’ of grid unit ‘x’; ‘’ adjusts the ratio of available annual water amount to precipitation in vegetation. ‘A’ is the unit reservoir cost value, and in this study, we use the unit reservoir cost value at the time of the construction of the Zipingpu Hydropower Station within the research area, which is 6.34 yuan/m³.

• Calculation of Soil Retention Value

Using the SDR module in the InVEST model, the potential soil erosion quantity (RKLS) under bare ground conditions is calculated. Subsequently, the actual soil erosion amount (USLE) under the influence of vegetation cover and soil and water conservation is computed. The difference between the two represents the soil conservation quantity (SD). The formula is as follows⁵¹:

3

R, the rainfall erosivity factor (MJ·mm(ha·hr)⁻¹); K, the soil erodibility factor (t·ha·hr(MJ·ha·mm)⁻¹); LS, the slope length-gradient factor; C, the cover and management factor; and P, the conservation practice factor.

The value of soil conservation is mainly reflected in terms of sediment retention value, nutrient retention value, and reduced silt deposition value⁵¹. Opportunity cost, market value, and shadow project methods are used respectively to evaluate these three values.

4

In the equations: , ,represents the sediment retention value, the nutrient retention value, and the reduced sediment deposition value. measured in tons; is the soil bulk density, measured in t/m³; h is the average soil thickness in China, approximately 0.6 m. Average annual forestry income, in yuan/ha. The soil and water conservation measures implemented in the study area mainly include converting farmland back to forest and afforestation. At the same time, considering the feasibility of obtaining value conversion data and the significant contribution of forestry to soil conservation in the study area, this study refers to the findings of Li et al. (2010) and uses average annual forestry income as an economic value⁵². The variable ‘a’ is the nutrient content in the soil, ‘b’ is the nutrient content in fertilizers, and ‘p’ is the cost of the fertilizer.C is the cost of sediment removal, yuan/m³.

Based on prior studies of the upper reaches of the Minjiang River^52,53, the is estimated at 1.195 t/m³. According to the Sichuan Province Statistical Yearbook for 2020, the is 158324.30 yuan/ha In this study, the nutrient retention value is derived from diammonium phosphate and potassium chloride. Based on current market conditions, the nitrogen content of diammonium phosphate fertilizer is 18% and its phosphorus content is 46%, with a price point of approximately 3200 yuan/t. On the other hand, potassium chloride fertilizer contains 60% nitrogen and is priced at about 3800 yuan/t. Drawing from the literature on soil studies in the upper reaches of the Minjiang River¹⁸, the soil content is found to be 0.29% for nitrogen (N), 0.02% for phosphorus (P), and 0.29% for potassium (K). Based on the same study¹⁸, the cost for sediment removal in this research is established at 6.5 yuan/m³.

• Calculation of ESV

5

The carbon storage value, soil retention value, and water source conservation value are added up to determine the ESV for the study area. In the equation, , , and correspond to the carbon storage value, water source conservation value, and soil retention value for region i, respectively.

Calculation of ESV based on the equivalence factor method

Based on China’s ESV equivalence factor table, ESV

The ESV equivalent factor in Table 3is based on data for all of China and cannot be directly applied to the study area. Ecosystem services are closely related to biomass, population, and rainfall^54,55. The greater the biomass, the stronger the ecosystem services; the higher the population density, the greater the demand for resources; and rainfall affects supply services and regulating services. Therefore, this study uses Net Primary Productivity (NPP) as a measure of spatial heterogeneity, rainfall to reflect climatic spatial differences, and population density to indicate resource scarcity. The results of the correction factor calculations are presented in Table 2.

Table 3.

Upper Minjiang River ESV Equivalence Factor Table yuan/ha.

Land Use	Provisioning services	Regulating services	Supporting services	Cultural services	Total
Dryland	1846.40	2035.40	1860.94	87.23	5829.98
Paddy field	-1715.55	6629.60	596.08	130.85	5640.98
Shrubland	1221.24	14931.14	3518.34	1003.16	20673.89
Forest	5364.74	60800.43	21866.05	4390.66	92421.88
Grassland	465.24	4754.12	1788.25	363.47	7371.07
Wetland	5233.90	48457.15	15061.99	6876.76	75629.80
Barren land	87.23	1279.40	436.16	87.23	1890.02
Water body	13549.98	161160.38	5161.20	2747.80	182619.35
Glacier and snow	3140.34	11892.57	14.54	130.85	15178.30
Settlement	0.00	0.00	0.00	0.00	0.00

Table 2.

Upper Minjiang River ESV equivalence factor Adjustment table.

Spatial heterogeneity coefficient	Rainfall spatial variability coefficient	Resource scarcity coefficient
0.87	1.4	1.07

Based on these adjustment factors, we derived the ESV equivalence factor table for the upper Minjiang River, showcased in Table 3.

The calculation formula is as follows: In this formula:

6

In this formula:“"represents the area of the i-th land-use type. A stands for the ESV equivalence factor per standard unit in the study area. signifies the ESV equivalence factor per national standard unit. corresponds to the national grain production value, while represents the grain production value of the study area. N is the NPP adjustment factor. R is the precipitation adjustment factor. P is the population adjustment factor. denotes the net primary productivity (NPP) of vegetation in the study area, and is the NPP of vegetation at the national level. is the average annual precipitation in the study area, whereas indicates the national average annual precipitation. reflects the population density of the study area, and stands for the national population density.

Spatial autocorrelation

Global spatial autocorrelation analysis is employed to detect the overall clustering or dispersion trend of ESV spatial data. By utilizing the Moran’s I statistical indicator, the degree of spatial autocorrelation is quantified. Moran’s I values range from − 1 to 1. I value greater than 0 indicates a positive spatial autocorrelation. The larger the value, the more pronounced the spatial relationship of ESV is. Conversely, I value less than 0 suggests negative correlation. The smaller the value, the greater the ESV spatial disparity. I value equal to 0 denotes random spatial distribution.

7

In the formula, refers to the deviation of attribute i from its average value , while signifies the spatial weight between elements i and j. n is the total number of elements, and is the aggregate of all spatial weights.

Local Spatial Autocorrelation is commonly quantified using the Local Indicators of Spatial Association (LISA) statistics. One of the most frequently used LISA statistics is Local Moran’s . The formula for Local Moran’s for an observation at location i is:

8

n is the number of spatial units (locations). is the deviation of the attribute value at location i from its mean. is the spatial weight between location ii and location j. is the sum of all spatial weights.

Geodetector model

Geodetector, a non-parametric spatial statistical technique, was created specifically to determine and measure the driving force of geographical factors on a particular phenomenon⁵⁶. It comprises factor detection, risk detection, interaction detection, and ecological detection. The factor detector is primarily used to assess the driving force of impact factors on the change in ESV, revealing the strength of each factor’s influence on ESV spatial differentiation. The formula is as follows:

9

10

11

In the formula, h = 1,…, L stands for the stratification of variables or factors. and N respectively represent the number of units in layer h and the entire region. and are variances of the dependent variable for layer h and the entire region. SSW and SST stand for the sum of intra-layer variances and the total variance of the entire region.

The interaction detector is designed to probe the complex interrelationships between multiple factors. It can not only detect the combined effects of two or more factors but also reveals how these factors interact with each other to either amplify or diminish their impact on the target phenomenon. The specific relationship between two factors can be categorized into the following five types⁵⁶, as shown in Table 4.

Table 4.

Dual-factor Interaction Outcome types.

Judgment Basis	Interaction Type
q(X1X2) < Min(q(X1),q(X2))	Non-linear weakening
Min(q(X1),q(X2)) < q(X1X2) < Max(q(X1),q(X2))	Single-factor non-linear weakening
q(X1X2) > Max(q(X1),q(X2))	Dual-factor enhancement
q(X 1 X 2 ) = q(X 1 ) + q(X 2 )	Independent
q(X1 X2) > q(X1) + q(X2)	Non-linear strengthening

The reasons for the spatial differentiation characteristics of Ecosystem Service Value (ESV) include natural factors such as topography and climate conditions, as well as socio-economic factors such as population and economic level. Through field surveys and literature analysis, it was found that the upper Min River region features typical high mountain canyon topography, where land use types exhibit a significant elevation gradient effect, and slope has an important impact on land cover. At the same time, there are substantial spatial differences in temperature and rainfall, which affect ecosystem services. Moreover, the higher the intensity of human activity (HAI), the greater the degree of stress on the ecology; population distribution and economic conditions not only influence land use patterns but also affect soil and water conservation and carbon storage^14,46. Therefore, this study selects elevation, slope, NDVI, rainfall, temperature, human activity intensity (HAI), GDP, and population as the driving factors for the spatial differentiation of Ecosystem Service Value (ESV) (Fig. 3).

Fig. 3.

Driving Factors. Map generated with ArcGIS 10.2 (ESRI).

Results and analysis

Seasonal variation characteristics of ESV

According to the ESV data Fig. 4a calculated with the InVEST model, the upstream of Minjiang River had ESV values of 937.22 × 10⁸ yuan, 2,053.60 × 10⁸ yuan, 1,103.59 × 10⁸ yuan, 1,044.62 × 10⁸ yuan, and 1,230.41 × 10⁸ yuan for the years 2000, 2005, 2010, 2015, and 2020 respectively. Over the past 20 years, there was an increase of 31.28%, showcasing a fluctuating growth trend. Specifically, the ESV growth was the most pronounced between 2015 and 2020, with a 5 year increase of 185.79 × 10⁸ yuan, and it was the least between 2010 and 2015, with a decrease of 58.97 × 10⁸ yuan over the 5 year period.

In contrast, evaluations based on the equivalence factor method (Fig. 4b) revealed ESV values for the upstream of Minjiang region of 652.15 × 10⁸ yuan, 747.12 × 10⁸ yuan, 795.57 × 10⁸ yuan, 795.20 × 10⁸ yuan, and 798.71 × 10⁸ yuan for the same five years. This also exhibits a fluctuating growth trajectory. Over the past 20 years, there was an increase of 22.48%, the most substantial ESV growth was seen between 2000 and 2005, with an increase of 94.96 × 10⁸ yuan. Meanwhile, the least growth occurred between 2010 and 2015, decreasing slightly by 0.37 × 10⁸ yuan over 5 years. When analyzing the variations in the individual ESV, every category presents its own value trajectory while maintaining a steady growth pattern. This reflects an ongoing enhancement of the ecological system functions in the upstream Minjiang region. Thus, even though the two assessment methods differ in calculation techniques and data types, both highlight the overall ESV growth in the upstream of Minjiang region.

Fig. 4.

Statistical Results of ESV from 2000 to 2020. aInVEST (Unit: 108 yuan). bEFM (Unit: 108 yuan).

Spatial variation of ESV in the upstream of minjiang region

As depicted in Fig. 5a, the ESV in the upstream Minjiang region presents a spatial distribution characteristic of “low in the north and high in the south.” This spatial pattern gets more and more noticeable with time. The ESV is comparatively low in the northern areas, including the upper reaches of Zhenjiang Pass and the Xiao Xing Gully watershed. In contrast, the southern areas encompassing the Heishui River, Zagunan River, and Yuzi Creek watersheds exhibit higher ESV values. Over the past 20 years, areas with low ESV have continuously shrunk, especially evident within the territory of Songpan County in the north. Concurrently, the ESV along the Minjiang River and its main tributaries has consistently grown, contributing to a significant increase in the ESV throughout the region.

Fig. 5.

Spatial distribution of ESV in the study area during 2000–2020. Map generated with ArcGIS 10.2 (ESRI). a The ESV calculated through the In VEST model. b The ESV calculated through the Equivalence Factor Method.

There is a noticeable contrast between Fig. 5b and a. The stark north-south disparity in ESV spatial distribution is diminished in Fig. 5b. The northern part of Wenchuan County, the central area of Li County, and the western region of Heishui County emerge as high ESV zones. Meanwhile, flat terrains along the Minjiang Riverbanks, densely populated areas, and high-altitude regions predominantly appear as areas with lower ESV values. Nonetheless, the overarching trend of ESV growth remains unaltered, with 16.22% of the region showcasing an increase. These areas exhibiting growth primarily lie in the northern Songpan County territory and nearby areas of the Minjiang watershed, paralleling the trends seen in Fig. 5a. A comparison of the results reveals that regions such as the upper reaches of Zhenjiang Pass, Heishui River, and Zagunan River consistently stand out as significant high-value ESV zones in both models, playing a pivotal role in the ecosystem service functions of the upstream of Minjiang region.

From Fig. 6, based on the Equivalence Factor Method, the Moran’s I values for ESV in the upstream Minjiang over five intervals are respectively 0.635, 0.635, 0.612 0.612, and 0.612, with p < 0.05. Over the 20-year period, these Moran’s I values have fluctuated, yet they depict an overall ascending trend. With a continuously expanding spatial agglomeration effect of ESV, the growth suggests a constantly better ecological environment and a continuing increase in ecological service capability. With reference to Fig. 7, the central and western regions of the study area are predominantly related to the high-high clusters based on this method.

Fig. 6.

Calculation Results for Global Autocorrelation.

Fig. 7.

Calculation Results for Local Autocorrelation. Map generated with ArcGIS 10.2 (ESRI).

In contrast, from Fig. 6, as per the VEST model’s calculation, the Moran’s I values for the same region across the five time frames are 0.556, 0.552, 0.578, 0.565, and 0.557, with p < 0.05. Over these two decades, the Moran’s I values present a fluctuating pattern, initially rising and later declining. The spatial agglomeration effect of ESV first amplifies, but subsequently, anthropogenic activities and various ecological risks exert pressure on the regional ecological environment, leading to a reduced spatial clustering effect. As seen in Fig. 7, the high-high regions using this model predominantly gather in the central and southern sections of the studied region.

Comparing the data from Figs. 6 and 7, both the Equivalence Factor Method and the VEST model exhibit fluctuations in the Moran’s I values for the upstream of Minjiang’s ESV over the past 20 years. The two methodologies yield different Moran’s I trends. While the former model highlights high-high clusters primarily in the central and western sectors, the VEST model indicates their prevalence mainly in the central and southern parts. Despite the minor differences in the results of the two models, both consistently reveal a pronounced clustering trend in the spatial distribution of ESV in the upstream Minjiang from 2000 to 2020¹⁶. Solid empirical support is provided by the concurrence for future efforts in ecological conservation and spatial planning.

Analysis of driving forces using the geodetector

Spatial autocorrelation

Each single factor was examined in relation to the spatial differentiation of ESV estimated by the two methods (Table 5). Based on the results from the In VEST model, slope emerged as the key driving factor causing ESV spatial differentiation, contributing 30.9%. Its influence greatly outweighs other elements, indicating the slope’s critical function in processes such as surface runoff, soil erosion, and carbon storage in the upstream of Minjiang River. The contributions of HAI, elevation, and temperature were 19.7%, 13.0%, and 11.0%, respectively, indicating that besides natural factors, human activity intensity also has a significant impact on the ecosystem. GDP, NDVI, total population, and annual precipitation have relatively minor influences on ESV spatial differentiation.

Table 5.

Driving force analysis of spatial differentiation in ESV in the Upstream Minjiang River.

Methond	value	elevation	slope	NDVI	rain	temperature	HAI	GDP	population
In VEST	q statistic	0.130	0.309	0.058	0.040	0.110	0.197	0.093	0.052
	p value	0	0	0	0	0	0	0	0
EFM	q statistic	0.205	0.077	0.147	0.078	0.185	0.382	0.038	0.028
	p value	0	0	0	0	0	0	0	0

Using the Equivalence Factor Method, HAI contributed the most to ESV spatial differentiation at 38.2%, followed by elevation (20.5%), temperature (18.5%), and NDVI (14.7%), with other factors having lesser impacts. Comparatively, the driving factors for ESV spatial differentiation estimated by both methods showed some differences, and the driving power for ESV spatial differentiation calculated by the Equivalence Factor Method was noticeably lower. However, the q-values of the HAI, elevation, and temperature from both methods were relatively high, suggesting that the spatial distribution variations of ESV in the upstream of Minjiang River are influenced by both natural and socio-economic factors.

Interaction detection analysis

Based on the interaction detection results of various driving factors (Fig. 8), it was found that pairwise factor interactions consistently enhanced the driving force for the spatial differentiation of ESV in the upper Minjiang River, indicating a high degree of coupling between the ESV and its drivers. According to estimations of InVEST model estimations, most interaction types were driven by two factors simultaneously, accounting for 82.14%, while nonlinear enhancements constituted 18.86%. The strongest interaction was observed between slope and elevation with a q-value of 0.387. Other pairwise interactions explaining more than 0.35 of the ESV spatial differentiation include slope∩temperature (0.371), slope∩rain (0.359), and slope∩HAI (0.356). The results based on the equivalence factor method slightly differed from the InVEST model. While dual-factor enhancement still dominated, its proportion decreased to 60.71% and nonlinear enhancement rose to 39.29%. The strongest interaction here was between HAI∩GDP with a q-value of 0.453, followed by HAI∩population (0.451), HAI∩temperature (0.436), and HAI∩elevation (0.430), all other interactions exhibited driving forces below 0.400.

Fig. 8.

Heatmap of Interaction Factor Detection Results for ESV.

This indicates that interactions between socioeconomic and natural factors manifested greater driving forces on the ESV spatial differentiation, the synergistic reinforcing effects of slope, HAI, elevation, and rainfall collectively shape the shifts in ESV spatial distribution. The driving forces of spatial differentiation in ESV were more pronounced when the factors interacted with slope and HAI.

Discussion

Changes in the ESV in the upstream of minjiang river

Both model assessment results indicate that the ESV in the upstream of Minjiang River has gradually increased from 2000 to 2020. However, ESV declined slightly from 2010 to 2015. Comparing the two models, their evaluations of ESV trends are similar, offering consistent evidence from two independent methods. This consistency corroborates the conclusion that the ESV in the upstream of Minjiang River has steadily grown over these 20 years, The results are consistent with previous studies^55,56. Concerning the temporal and spatial changes in ESV, the spatial distribution features evaluated by the two methods have certain differences. However, high ESV was mainly located in the central and southern parts of the study area and on both sides of the Minjiang River and its tributaries. The main reason is that the central and southern parts are dominated by forests and grasslands^55,58, which have a stronger capacity for water conservation, soil and water conservation, supply services, and regulating service^58–60, and the two sides of the rivers have sound hydrothermal conditions and rich biodiversity, so ESV is also higher.

Overall, the ecosystem service function of the upper Minjiang River has been improving, which is related to the government’s attention and the residents’ increased awareness of environmental protection^49,58. According to data from the official website of Aba Prefecture, the implementation of projects such as natural forests conservation project, soil and water conservation projects has increased financial investment and strengthened ecological environment management, and the region has made significant progress in returning farmland to forests, returning pasture to grassland, sealing off mountains for forestation, sandy management. Meanwhile, the awareness of environmental protection has been enhanced, vegetation cover has increased, and soil erosion as a whole has been effectively curbed, greatly improving the regional ecosystem service function⁶¹⁻⁶³. the ESV decreased slightly from 2010 to 2015, mainly due to the occurrence of several geologic disasters during the perio⁵⁸. Coupled with the rapid development of the regional socio-economy after 2010, the impermeable layer continued to increase, which affected the ecosystem service function to a certain extent. It indicates that the stability of the ecosystem in the region is not strong enough, and relevant measures need to be implemented continuously⁶⁴.

Driving forces of ESV spatial differentiation

In the ESV calculated by the two evaluation models, the factor detection results vary slightly. The key driving factors for regional ESV spatial differentiation are slope and HAI, respectively. The driving force of each single factor on the ESV calculated by the Equivalence Factor Method is notably weaker, primarily due to the method’s reliance on land-use data, thus highlighting areas of high human activity intensity. However, the interaction between pairs of factors enhancing the driving force on ESV spatial differentiation is consistent, and these factors and their interactions decisively affect the spatial distribution of ESVs. Slope and HAI play a determining role in that slope has an important and significant role in hydrological regulation, climate regulation, soil conservation, and biodiversity⁶⁵, while HAI influences land use types, resource utilization, and ecosystem quality^66,67. Slope affects the way of human activities, which in turn affects ecosystem services, human-land interactions are strongest in areas with smaller slopes, steep-slope areas are inherently ecologically fragile, and human activities exacerbate ecological damage, and human activities change the slope, which indirectly affects ecosystem services^68,69. This indicates that the ESV in the upstream of Minjiang River is influenced by both natural factors and socio-economic factors. Natural factors affect the spatial differentiation of ESV in the study area, while socio-economic factors further exacerbate changes in ecosystem service functions^13,14. By means of comparison, disparities in the drivers and driving forces of ESV spatial differentiation¹⁹, can be shown as a result of the various ESV evaluation models’ varying emphasis on ecosystem service functions⁴⁹.

Insights for upgrading ESV

In order to enhance the ESV of the upper reaches of the Minjiang River, realize ecological environmental protection and sustainable development, and ensure the ecological security of the Chengdu Plain, it is imperative to ceaseless carry out integrated protection and restoration of mountains, water, forests, fields, lakes, grasses, sand and ice in the region. First, establish an ecological value realization mechanism oriented to ESV enhancement. Insist on good zoning prevention and control, strengthen water and soil prevention and control, and take the construction of the Giant Panda National Park as an opportunity to build an ecological security barrier and improve the ecological protection compensation system^63,70. Second, improve the pattern of land space protection and development. The red line of permanent basic farmland protection, the red line of ecological protection, and the urban development boundary as the insurmountable alart for adjusting economic structure, planning industrial development, and promoting urbanization. Combined with the Minjiang River National Wetland Park, Dagu Glacier Geological Park and other nature reserves⁷¹, the ecological corridor formed along the Minjiang River, to build a “one corridor, two belts, three axes, multiple points” of the land space development and protection pattern (Aba Prefecture Spatial Planning, 2021–2035). Refine the main function zoning, coordinating agricultural function, pastoral function and urban function. Third, to build a comprehensive disaster prevention system. Taking the Minjiang River and its tributaries as the core area, build a “sky-space-ground” integrated natural disaster hazard identification and monitoring and early warning system⁷², and grasp the dynamic changes of natural disasters. Taking the improvement of the human environment as an entry, villagers will be motivated to participate in disaster prevention and mitigation.

Limitations and future research

According to the study’s findings, the Equivalent Factor Method is based on land use data, which has a low data volume, is simple to understand and apply, and can be used to assess regional ESV. However, it can be challenging to determine the actual SEV in various regions, and the dynamics are insufficient. The InVEST model realizes the spatialization and dynamization of the quantitative assessment of the ESV of the Upper Minjiang River, but it is very sensitive to the change of the data, and has poor accessibility to data, which has a certain impact on the assessment accuracy during the transformation of the amount of physical quantity into the amount of value. The combined use of the InVEST model and the Xie Gaodi Equivalent Factor Method compensates to some extent for the shortcomings when using a single method⁷³, thus providing a more accurate assessment of ESV variations in the study area. The present investigation employed the InVEST model to achieve the spatialization and dynamization of the quantitative evaluation of ESV in the upper Minjiang River. This study does, however, have many shortcomings. The ESV estimated by the two approaches is the sole method compared and analyzed in this work, and the overall ESV change trend is comparatively similar. The overall ESV change trend is reasonably close, but there is a significant geographical divergence, necessitating more research into the two techniques’ integration methods. Based on the available study, the geographical and temporal evolution mechanism of ESV has not been analyzed, therefore in order to identify the ecosystem’s development trend for certain goals, attention must be paid to the evolution process of ESV. The pressure of human activities on ecosystems has increased, and people are paying more attention to the social value of ecosystems, but the current ESV assessment lacks an assessment of the social value aspect.

Conclusions

Over the past two decades, the ESV in the upper Minjiang River has shown a consistent growth trend, reflecting the continual strengthening of regional ecosystem services. Areas such as the upper reaches of Zhenjiang, Heishui River, Zagunao River, and Yuzixi Watershed exhibit overall high ESV values, playing a crucial role in the upper Minjiang River’s ecosystem services and hence warrant special protection. Significant spatial autocorrelation features are seen in ESV, with “high-high clustering and low-low clustering” being particularly noticeable. Slope and HAI dominate the spatial differentiation of ESV in the study area, with the combined interactive effects of slope, HAI, elevation, and rainfall driving the spatial distribution changes of ESV. To promote the stable enhancement of ecosystem services in the upper Minjiang River and safeguard regional ecological security, it is advised that distinct regional ecological compensation policies be put into place, together with robust protection measures, optimized land use plans, and dynamic monitoring and evaluation mechanisms.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by L, S, M, W, and J. The first draft of the manuscript was written by L and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data availability

All data, models, or codes generated or used during this study are available from the corresponding author upon reasonable request. E-mail: xiangmingshun19@cdut.edu.cn.

Competing interests

The authors declare no competing interests.