Introduction

Ecosystems provide critically important provisioning, regulating, supporting, and cultural services, which are indispensable for human survival and development1,2. However, the United Nations’ Millennium Ecosystem Assessment points out that human activities are continuously altering ecosystems3. With the rapid advancement of urbanization fueling an intensification of human activities, substantial alterations in ecosystem services have ensued, exerting profound impacts on human well-being, health, and socio-economic development4,5. This situation has drawn the attention of researchers and policymakers worldwide, who stress the necessity of exploring the dynamic relationship between human activities and ecosystem services in the context of rapid urbanization. Such exploration is of paramount importance for laying the foundation for regional sustainable development6.

Currently, research on the impact of human activities on ESV primarily focuses on scales such as watersheds7economic belts8urban agglomerations9provinces10and ecological barrier regions11. At the urban scale, the exploration of the impact on ESV remains a relatively underserved area of research. Nonetheless, a few studies have provided valuable insights. For example, Kang et al. used random forest to reveal that anthropogenic factors such as human activities, population, and economy have gradually replaced natural factors as the dominant influences on ecosystem services12. Zhu et al. found through Spearman’s correlation coefficient that changes in landscape patterns have significantly different impacts on the ESV of different land types in Nanchang City13. Let et al. demonstrated through rank correlation analysis that there is a negative correlation between human impact and ESV in Old Malda Municipality14. Liu et al. found that the proportion of water area and the output value of the secondary industry in Nanjing have significant independent effects on ESV based on a deep learning model15. Song et al. used convergent cross mapping and global sensitivity and uncertainty analysis to find that GDP, NDVI, and population have significant impacts on the ESV of Sanya City16. Degefu et al. employed bivariate Moran’s I analysis and found significant positive spatial correlations between the ESV of four cities in Ethiopia and comprehensive land-use dynamic degree, land-use intensity, and land-use diversity17.

However, most studies on the impact of ESV at the urban scale primarily focus on identifying key driving factors and lack quantitative characterization of driving effects and their thresholds. With the in-depth study of the mechanisms influencing ESV, some scholars have recognized the existence of effect relationships between ecological and human activity thresholds, which have significant implications for ecological environmental management and decision-making. For example, Peng et al. found through piecewise linear regression that when the population density in Beijing exceeds 229 people per square kilometer and economic urbanization surpasses 1.0715 billion Chinese yuan per square kilometer, ESV declines rapidly18. Liu et al. utilized machine learning methods to reveal that when the population density in the Guangdong-Hong Kong-Macao Greater Bay Area is below 669 people per square kilometer, social economies and ecosystems may develop together19. Li et al. discovered that when the Human Activity Intensity (HAI) is below 0.83, the grasslands on the Tibetan Plateau still possess considerable adaptive capacity to climate change20. Li et al., through restricted cubic splines regression analysis, determined the thresholds for human factors such as the Human Impact Index, Land Use Intensity, and population density affecting ESV in the Beijing-Tianjin-Hebei urban agglomeration to be 29.8, 61.9, and 312 people per square kilometer, respectively. Before these thresholds, ESV declines rapidly, and the rate of decline slows after exceeding them21. Guo et al., using a conditional inference tree model, found that within specific ranges in the upper and middle Yellow River (with HAI between 0.11 and 0.29 and LUI between 184.32 and 210.6), enhanced human activities can significantly improve ESV, but beyond these boundaries, they have negative impacts22. Zhao et al., employing a threshold regression model, analyzed that the impacts of population urbanization and land urbanization on carbon storage in mega-urban agglomerations both exhibit inverted U-shaped curves, with thresholds of 272 people per square kilometer and 1.063%, respectively. The impact of land urbanization on food production, however, shows a U-shaped curve with a threshold of 0.704%23. Yan et al.‘s quadratic term fitting analysis revealed that the response threshold of ecosystem services to urban-rural transformation in the Beijing-Tianjin-Hebei region gradually increases over time, indicating that ecosystem services exhibit a certain degree of stability and resilience when facing urban-rural transformation24. In summary, the methodologies employed in ESV threshold studies have evolved from simple linear models to more sophisticated machine learning and nonlinear regression techniques. These advances have enabled researchers to better capture the complex, nonlinear relationships between ecological and human activity thresholds, providing valuable insights for ecological environmental management and decision-making.

Although current research on the impact mechanisms of ecosystem services has thoroughly demonstrated the positive and negative effects of human factors on ecosystem services and clearly confirmed the existence of threshold effects among the influencing factors, there are still two core questions that urgently need to be answered: (1) How do human activities affect ESV of various functions within the unique context of urban environments? (2) Under what specific conditions do these threshold effects play a critical role? Cities, as areas with the most concentrated and frequent human activities, have profound and complex impacts on ESV. However, regrettably, this key research area lacks in-depth exploration, especially in the identification of local anomalies when exploring the spatiotemporal correlation between human activities and ESV.

Hefei, the capital city of Anhui Province in China, also serves as a sub-central city within the Yangtze River Delta urban agglomeration and a strategic dual-node city in the Yangtze River Economic Belt. Its unique geographical location has made it one of the top ten high-speed rail hubs in the country. The convenient transportation and strategic location have facilitated the rapid development of the city over the past 20 years. As of the end of 2023, Hefei has a population of 9.853 million and a total GDP of RMB 1,267.38 billion, with the urbanization rate increasing from 32.64% in 2000 to 85.55%. The urbanization process in Hefei not only emphasizes the dual goals of rapid development and high quality but also integrates technological innovation, ecological conservation, regional harmony, and policy guidance as key pillars of growth. This holistic approach positions Hefei as a model city within the global trend of rapid urbanization, where sustainable urbanization practices are implemented to minimize ecological impacts while promoting economic prosperity. However, despite these efforts, the ecosystem in Hefei has experienced significant transformations due to human activities and economic development, necessitating a closer examination of ecosystem dynamics25. Therefore, analyzing the process of ecosystem changes under the background of rapid urbanization and their response to human activities is of great significance for the harmonious development of urban ecology and economy at a high level.

Therefore, Hefei provides a highly valuable case for addressing the aforementioned key issues. This study selects Hefei as the research area, employing the ESV flow direction loss-gain matrix to examine the driving factors behind changes in urban ESV. It utilizes the GeoDetector to determine the thresholds of HAI’s impact on ecosystem service values across different functions and accurately identifies local abnormal influences. The aim is to delve into the impact mechanisms of human activities on urban ecosystem services. The research objectives are defined as follows: to reveal the temporal and spatial changes in urban ESV; to explore the determinants of urban ESV changes; to determine the thresholds of HAI on ESV of different functions, and to precisely identify localized anomalous impacts, with the aim of providing scientific evidence for urban ecological environment management and decision-making.

Materials and methods

Study area

Hefei City (116°41ʹ–117°58ʹE, 30°57ʹ–32°32ʹN) is situated in the central part of Anhui Province and spans an expansive 11,500 square kilometers. The terrain in this region is dominated by hills and mounds, with the Jianghuai Watershed running east - west across the entire territory. It is situated in a subtropical monsoon humid climate zone, characterized by distinct seasons, pronounced monsoons, mild weather, and moderate rainfall. The administrative area of Hefei City includes Shushan, Yaohai, Luyang, and Baohe Districts, as well as Chaohu City, Changfeng County, Feidong County, Feixi County, and Lujiang County (Fig. 1). Hefei City has undergone rapid urbanization in recent decades, particularly in its main urban districts. The main urban area, comprising Shushan, Yaohai, Luyang, and Baohe Districts, forms the core economic and administrative hub of the city. These districts are highly urbanized, with dense populations, extensive commercial and industrial activities, and well - developed transportation networks. Beyond the urban core, Hefei City boasts a significant amount of forestlands. These forestlands are scattered across the city’s landscape, particularly in the hilly regions and along the margins of the Jianghuai Watershed. Water bodies are an integral part of Hefei’s geographical and ecological makeup. The most prominent water feature is Chaohu Lake, one of the largest freshwater lakes in China, located to the south of the main urban area. Besides urban and agricultural land, Hefei City has also designated ecological spaces such as nature reserves, parks, and greenbelts to protect the environment and enhance residents’ quality of life.

Fig. 1
figure 1

Study area and LULC in 2020. Note: This map was created by ArcGIS 10.8 (http://www.esri.com/). The administrative division data of China in the map come from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences. (https://www.resdc.cn/).

Full size image

Data sources

Land Use and Land Cover (LULC) data were sourced from the Chinese GlobeLand30 Global Land Cover Dataset provided by the National Earth System Science Data Center (https://www.geodata.cn), which provides a spatial resolution of 30 m and an overall accuracy rated at 83.50% by third-party evaluations. The types of land cover encompass cropland, forestland, grassland, shrubland, wetland, water area, tundra, built-up land, bare land, glaciers, and permanent snow. In Hefei City, the covered types include cropland, forestland, grassland, wetland, water area, and built-up land. The past two decades have witnessed rapid urbanization in Hefei City, thus, land cover data from 2000, 2010, and 2020 were selected for this study. Agricultural data from 2000 to 2020 were obtained from the “Anhui Statistical Yearbook,” which included statistics on the sown area, yield, and value of crops.

Methods

ESV accounting

ESV serves as a core indicator in assessing ecological civilization’s effectiveness and characterizing ecosystem services’ functions. This includes provisioning service value (PSV), regulating service value (RSV), supporting service value (SSV), and cultural service value (CSV)26. This study adopts the ESV equivalent factor methodology developed by Xie Gaodi et al.27and to eliminate the impact of market price fluctuations, a standard equivalent factor, equal to one-seventh of the economic value of grain produced by per hectare of farmland (based on the average market price of grain yield in 2020) was used as the price standard28. The biomass factor in Hefei is significantly correlated with ecological characteristics such as climate, vegetation, and land use. Its biomass distribution pattern is consistent with the overall trend in Anhui Province. Consequently, adjustments were made using the biomass factor coefficient of 1.17 for Anhui Province, as determined by the research of Xie Gaodi et al.29culminating in a table that presents the value equivalent factors for various ecosystem service functions across different LULC types in Hefei (Table 1). The formula for calculating ESV used is as follows:

$$\:ESV=\sum\:_{k=1}^{n}{S}_{k}\times\:V{C}_{k}$$
(1)
$$\:V{C}_{k}=\sum\:_{k=1}^{n}V{C}_{fk}$$
(2)

Where Sk is the area of the k-th land type (hectares), VCk is the ESV coefficient of the k-th land type (yuan/hectare), and VCfk is the ESV coefficient of the f-th item for the k-th land type (yuan/hectare).

Table 1 Equivalent factor table of ESV per unit area in Hefei (Yuan/hm2).
Full size table

ESV flow direction loss-gain analysis

This study constructs an ESV flow direction loss-gain matrix from the land use transition matrix30revealing changes in ESV’s spatial movement. The flow direction of ESV reflects the transfer pathways of service functions among different land use types. The conversion of cropland to built-up land is accompanied by the loss of ESV related to soil conservation, carbon sequestration, etc., while the expansion of water area and wetland will enhance the gains in ESV associated with hydrological regulation, biodiversity maintenance, etc. By quantifying the losses and gains along this flow direction, we can reveal the spatial reconfiguration process of regional ecosystems and its impacts on human well-being. The formula is defined as follows:

$$\:{P}_{ij}=\left(V{C}_{j}-V{C}_{i}\right)\times\:{A}_{ij}$$
(3)

Where Pij is ESV loss-gain from type i land transitioning to type j land, VCi and VCj are the ESV coefficients for type i and type j land respectively, and Aij is the area of type i land transitioning to type j land.

HAI calculation

HAI is defined as the degree of natural cover use, transformation and exploitation of land surface by human in a certain region. This degree can be reflected by land use/cover types. Based on the requirements for analyzing the scale of the study area and the impact mechanism of ESV, this paper calculates the HAI for 1 km grid cells using LULC data with a spatial resolution of 30 m. The formula used is:

$$\:HAI=\sum\:_{i=1}^{n}\frac{{A}_{i}\times\:{P}_{i}}{TA}.$$
(4)

Where HAI is human activity intensity; Ai is the total area of land use/cover type i; Pi is the parameter reflecting the intensity of human impact for the land use/cover type i. Based on previous research results and considering the actual situation of land use type conversions in the study area, this paper assigns values to the Pi parameter using the average of the Lohani Inventory Method, Leopold Matrix Method, and Delphi Method (Table 2). The Lohani Inventory Method, Leopold Matrix Method, and Delphi Method are used to assign values to the human impact intensity coefficients reflected by different landscapes from the dimensions of resource dynamics, environmental impacts, and expert judgments, respectively. Adopting the average value derived from these three methods aims to balance the strengths of each method and reduce potential biases31; TA is the total land area; n is the numbers of land use/cover types. The range of HAI values is from 0 to 1, with higher values indicating greater human impact on land use and lower values indicating less impact.

Table 2 Strength indices of human impact for different landscape elements.
Full size table

Geographic detection analysis

The GeoDetector, a spatial statistical model, employs the principle of variance to identify driving factors behind spatial heterogeneity32. This study utilizes both factor and risk detectors to analyze the spatiotemporal associations between HAI and changes in ESV. The factor detector, measuring a value “q” from 0 to 1, explains the driving force of HAI behind ESV’s spatial heterogeneity, where higher values indicate greater explanatory power33. The formula is expressed as:

$$\:\text{q}=\text{1}-\frac{\text{1}}{\text{N}{{\sigma}}^{\text{2}}}\sum\:_{\text{h=1}}^{\text{L}}{\text{N}}_{\text{h}}{\sigma\:}_{\text{h}}^{\text{2}}$$
(5)

.

Where h is 1,…, L; Nh is the sample size for stratum h, N is the total sample size; \(\:{\sigma\:}_{\text{h}}^{\text{2}}\)is the variance within stratum h, \(\:{\text{σ}}^{\text{2}}\) is the total variance.

The risk detector is used to assess whether there are significant differences in the mean values of specific attributes across different regions, and it has been widely applied in the field of public health34. Given the advantages of the risk detector in identifying differences, this study utilizes it to assess significant differences in ESV mean values within HAI graded regions, determine the thresholds of HAI’s impact on ecosystem service values of different functions, identify local abnormal influences, and provide a more nuanced perspective for understanding the complex relationship between human activities and ecosystem services. Regions with significant differences are marked “Y” (Yes); those without are marked “N” (No), indicating no ESV change due to HAI. This detector employs statistical t-tests to identify areas where ESV does not change despite unusual HAI variations, with the significance level (α) set at 0.05.

$$\:t=\frac{{\bar{Y}}_{h=1}-{\stackrel{-}{Y}}_{h=2}}{{\left[\frac{Var\left({\stackrel{-}{Y}}_{h=1}\right)}{{n}_{h}=1}+\frac{Var\left({\stackrel{-}{Y}}_{h=2}\right)}{{n}_{h}=2}\right]}^{1/2}}$$
(6)

.

\(\:{\bar{Y}}_{h}\) is the mean attribute value within sub-region;nh is the sample count within sub-region; Var is the variance.

Results

Analysis of changes in ESV

In Hefei, the spatial distribution of ESV shows high concentrations around Chaohu Lake and Huangpi Lake, whereas areas undergoing urban construction display notably lower ESV (Fig. 2). Over the last twenty years, zones of high ESV have seen gradual spatial and temporal growth, mainly in the northern regions, while zones of low ESV around Hefei’s urban district have expanded rapidly due to urbanization. ESV declined from 47.012 billion yuan in 2000 to 46.652 billion yuan in 2010, and decreased further to 45.534 billion yuan in 2020, with an average annual decline rate of 3.14%.

Fig. 2
figure 2

Spatial distribution of ESV in Hefei (created by ArcGIS 10.8, http://www.esri.com/).

Full size image

Using a 1 km grid scale and hotspot analysis tools35,36this study analyzed the spatial distribution of ESV changes for the periods 2000–2010, 2010–2020, and 2000–2020 with a confidence level of ≥ 95% (Fig. 3). From 2000 to 2010, significant ESV loss coldspots occurred in Hefei’s urban district, Chaohu City, and Lujiang County, particularly along the southwest shores of Chaohu Lake. Conversely, ESV gain hotspots were more widespread across the northern and southwestern parts of the study area, especially in the northern regions. From 2010 to 2020, ESV loss coldspots within Hefei’s urban district, Chaohu City, and Lujiang County expanded, whereas those along the Chaohu shoreline receded. ESV gain hotspots were primarily concentrated in the northern, western, and southern parts of the study area, illustrating a significant shift in ESV distribution over the past two decades.

Fig. 3
figure 3

Spatial distribution of cold spots or hot spots of ESV changes in Hefei (created by ArcGIS 10.8, http://www.esri.com/).

Full size image

According to the statistics (Fig. 4), regulating services represented 29.121 billion yuan or 63% of the total ecosystem service functions, followed by supporting services at 12.590 billion yuan or 27%. Provisioning and cultural services were less significant, comprising only 10%. Over the past twenty years, regulating services initially increased but subsequently declined. Supporting and provisioning services exhibited a continuous decline, whereas cultural services initially improved but subsequently decreased. Among secondary services, waste treatment recorded the highest value, while raw material production registered the lowest. Hydrological regulation, waste treatment, and provide aesthetic landscape services initially increased but later decreased, mirroring declining trends in other individual service values. Between 2000 and 2020, regulating services decreased by 47.85% and supporting services by 41.78%, collectively accounting for 89.63% of the total change.

Fig. 4
figure 4

The variation of ESV in Hefei.

Full size image

Spatial and temporal analysis of ESV flow changes

Between 2000 and 2010, ESV loss predominantly occurred in Baohe District, Yaohai District, and Chaohai City, with ESV gains more widely dispersed. From 2010 to 2020, ESV loss areas expanded outward from Baohe, Yaohai, and Shushan Districts, accompanied by a broader dispersion of ESV gains. From 2000 to 2020, ESV losses in Hefei concentrated in urbanized regions, whereas gains were broadly dispersed across various counties. The total ESV loss area significantly exceeded that of ESV gain, with losses clustering and gains dispersing (Fig. 5). Between 2000 and 2010, the total ESV changes included a loss of 2.394 billion yuan and a gain of 2.034 billion yuan, resulting in a net loss of 0.360 billion yuan. From 2010 to 2020, ESV gains of 1.758 billion yuan were offset by losses of 2.876 billion yuan, leading to a net loss of 1.118 billion yuan and a total two-decade loss of 1.478 billion yuan.

Fig. 5
figure 5

Spatial distribution of ESV gains and losses in Hefei (created by ArcGIS 10.8, http://www.esri.com/).

Full size image

Analysis of ESV directional changes (Fig. 6) between 2000 and 2010 reveals significant losses primarily from the conversion of cropland to built-up land (1.525 billion yuan, 64%), water area to cropland (0.430 billion yuan, 18%), and forestland to cropland (0.221 billion yuan, 9%). Notable increases were observed from the conversion of cropland to water area (0.956 billion yuan, 47%), built-up land to cropland (0.671 billion yuan, 33%), and cropland to forestland (0.176 billion yuan, 9%). From 2010 to 2020, substantial losses occurred with cropland being converted to built-up land (1.881 billion yuan, 65%), water area to cropland (0.313 billion yuan, 11%), and cropland to grassland (0.265 billion yuan, 9%). Primary increases were derived from the reversion of built-up land to cropland (1.204 billion yuan, 68%) and cropland to water area (0.306 billion yuan, 17%). It is clear that the main contributors to ESV increases in Hefei were the reversions of built-up land to cropland and cropland to water area. However, considerable conversion of cropland to built-up land and some water area to cropland led to significant ESV losses. Despite some compensation from low-value ESV land categories converting to higher-value wetland and water area, the total output was inadequate to offset the losses, resulting in significant ESV depletion in Hefei.

Fig. 6
figure 6

Flow matrix of ESV gains and losses in Hefei (billion yuan).

Full size image

Impact of HAI on ESV

Utilizing 30 m LULC data within a 1 km grid scale, the HAI for Hefei was assessed for the years 2000, 2010, and 2020. The average HAI exhibited minor fluctuations around a value of 0.567, indicating minimal overall change. Spatial analysis revealed that the most significant changes occurred in urban areas such as Hefei and Chaohu, where the HAI and ESV demonstrated an inverse relationship. Regions most impacted by human activities were those experiencing the most substantial degradation of ecosystem services.

Fig. 7
figure 7

Spatial distribution of HAI in Hefei (created by ArcGIS 10.8, http://www.esri.com/).

Full size image

To investigate the relationship between HAI and ESV more thoroughly, this study utilized GeoDetector methods, specifically factor and risk detectors, to ascertain the extent of HAI’s impact on ESV and to identify the thresholds of its anomalous impacts. By applying an equal interval classification approach, the HAI was discretized into four distinct categories of 6, 8, 10, and 12 levels in ascending order. Through rigorous experimental analysis, it was determined that dividing the HAI into 10 levels provided the most optimal results for geographic detection, as depicted in Fig. 7. The factor detector revealed that HAI impacts ESV, PSV, RSV, SSV, and CSV with impact values of 0.590, 0.227, 0.567, 0.275, and 0.681 respectively; CSV exhibited the greatest explanatory power, as depicted in Fig. 8. Over various periods, the influence of HAI on ESV changed within the context of rapid urbanization; this was evidenced by a gradual decrease in impact on RSV and CSV, an increase in SSV, fluctuating levels in PSV, and initially rising then declining ESV impacts.

Fig. 8
figure 8

Q value of HAI detection for spatial differentiation of ecosystem service value.

Full size image

In the risk detector analysis with a confidence level of 95%, as the HAI increased, the ESV, RSV, and CSV generally exhibited a decreasing trend, becoming negatively correlated. Conversely, PSV and SSV initially increased before declining, indicating a transition from positive to negative correlation, as detailed in Fig. 9. At HAI levels between 1 and 2 (0.090 to 0.262), rapid fluctuations in ecosystem service values were noted, with pronounced decreases in ESV, RSV, and CSV, alongside marked increases in PSV and SSV. At HAI levels between 7 and 9 (0.606 to 0.864), significant reductions in ESV, PSV, and SSV were observed, suggesting that intense human activities rapidly erode ecosystem service values. At HAI levels above 7 (greater than 0.606), CSV declined almost to zero. For other HAI ranges, reductions in ecosystem service values were less pronounced, exhibiting varied declines as HAI increased.

Fig. 9
figure 9

Mean ecosystem service value of HAI graded zones.

Full size image

The study identified three critical ranges of anomalous HAI impacts with a confidence level of 95% (Fig. 10): At levels 2–3 (0.176–0.348), human activities can maintain equilibrium in ESV, RSV, and CSV. During human activities with a relatively low HAI, it is essential to manage land type proportions effectively, limiting cropland to a maximum of 30% and built-up land to a maximum of 5%, so that various land - use types can maintain relatively stable proportions. Natural ecosystems such as forestland, grassland, wetland, and water area can maintain their own ecological processes and functions within a relatively small spatial area, thus enabling the ecosystem service value to remain in balance. This strategy facilitates potential land use changes, particularly in urban areas where the development of parks and green spaces during urban sprawl can expand the areas of forestland, grassland, wetland, and water area, enhance the service functions of ecosystems, thereby supporting ecosystem services and mitigating losses in their values. At levels 4–6 (0.348–0.606), human activity intensity increases but remains within the ecosystem’s tolerable range, allowing for the maintenance of the SSV balance. In this context, cropland should not exceed 80% of the total land area, while forestland, grassland, wetland, and water area collectively account for 15%, with built-up land making up 5%. Rational utilization of cropland can meet human basic needs without causing excessive damage to the natural ecosystem. Keeping the proportion of cropland within 80% can reserve some space for natural ecosystems such as forestland, grassland, wetland, and water area. These lands play a significant role in preventing soil erosion and, at the same time, provide habitats and migration corridors for wild animals and plants, contributing to biodiversity. Notably, the limited extent of built-up land (5%) helps to minimize its direct impact on these natural processes and habitats. At levels 5–7 (0.434–0.692), where human activity intensity is relatively high, maintaining an equilibrium in PSV requires increasing the proportion of cropland to 90% to meet the substantial human demand for food and raw materials. Meanwhile, it is essential to effectively manage other land types and reasonably plan the distribution and utilization of forestland, grassland, wetland, and water area. For example, the rational utilization of grassland resources can yield livestock products, and the protection of wetlands can ensure the sustainable utilization of fishery resources. Through this land-use approach, while meeting the high-intensity material demands of humans, the supply service capacity of ecosystems is preserved to the fullest extent possible.

Fig. 10
figure 10

Proportion of land area in HAI anomaly range.

Full size image

Discussion

Changes and causes of urban ESV

Over the past two decades, amid rapid urbanization, Hefei’s ESV has exhibited a consistent decline, with an annual decline rate of 3.14%. This trend mirrors those observed in Shanghai37 and Wuhan38among other rapidly urbanizing cities, where the annual decline rates are 3.39% and 2.12%, respectively. Compared with previous studies, the uniqueness of this research lay in revealing a certain spatial regularity in the changes of urban ESV. The cold spot areas of ESV loss were primarily concentrated in urban zones, and as built-up areas rapidly expanded outwards, a significant amount of land resources were encroached upon, leading to a continuous expansion of the cold spot areas of ESV loss. Amidst the wave of rapid urbanization, the government had actively responded to the call for ecological civilization construction by implementing a series of policies such as “returning housing back to cropland” and “converting cropland to wetland,” which had led to a slight increase in the total area of hot spot regions where ESV was on the rise. Notably, the ESV changes along the shoreline of Chaohu Lake exhibited distinct migration characteristics, transitioning from original cold spot areas of loss to hot spot areas of gain.

Declines in ESV in rapidly urbanizing cities have consistently been linked to changes in land use. As urban expansion rapidly encroaches on cropland, forestland, and water areas, it triggers significant regional ESV reductions39,40,41,42. This study establishes an ESV flow direction loss-gain matrix to elucidate the dynamics of ESV alterations and the root causes behind its decline. Specifically, in Hefei, urban expansion has converted substantial cropland areas into built-up land, accounting for the majority of ESV losses—a trend that has intensified with ongoing urbanization. Despite this, enhanced ecological and environmental protection measures, coupled with proactive policy implementations, have contributed to ESV gains. While these land-use conversions have yielded some regional benefits, the magnitude of ESV loss far outweighs the gains, resulting in a net decline in regional ESV. Therefore, strengthening the scientific management of land resources in ESV loss cold spots and guiding the transformation of low-value land types into high-value alternatives are effective strategies for enhancing urban ESV.

Threshold of HAI’s impact on different functional ESV and identification of local anomalies

This study used risk detector to determine the differential characteristics of mean ESV within regions classified by HAI grade, and fitted the numerical relationship between HAI grades and mean ESV, effectively identifying the threshold of human activity’s impact on the spatiotemporal changes of ESV and clarifying the relationship between HAI and the value of ecosystem services with different functions. The study found that there were significant differences in the driving effects of HAI on ESV with different functions, with ESV, RSV, and CSV all showing a decreasing trend as HAI increased. However, the changes in PSV and SSV were relatively complex, initially showing an increasing trend followed by a gradual decrease, with the threshold for this transition ranging between HAI values of 0.176 and 0.262. This is consistent with the research findings of Guo et al. on the upper and middle Yellow River of China. In their study, it was pointed out that intensifying human activities and land use within a certain range (0.11< HAI<0.29 and 184.32< LUI<210.6) could greatly promote ESV22. The difference lies in that they only defined the threshold range of the combined influence of HAI and LUI on PSV, whereas this study further clarified the thresholds of HAI’s impact on the value of ecosystem services with different functions.

Furthermore, this study precisely identified three interval ranges (0.176–0.348, 0.348–0.606, 0.434–0.692) where HAI has an abnormal impact on ecosystem service values. By overlaying LULC data for analysis, we found that under relatively low and high levels of HAI, effective planning and development of land resources, coupled with strict control over the proportion of various land areas, can effectively maintain the equilibrium of ecosystem service values.

This study not only fills the gap in identifying the thresholds of HAI’s impact on different functional types of ecosystem service values and recognizing local anomaly information, but also provides a scientific basis for achieving harmonious and coordinated development of human-land relationships in Hefei. At the same time, this research outcome can also serve as a useful reference and guidance for other cities that are undergoing rapid urbanization in terms of rational land use and ecological protection.

Policy implications

In fact, policies are important factors influencing urban LULC changes, and changes in LULC in turn lead to variations in the economic and ecological values of various ecosystem services across different regions43. Intensifying the implementation of ecological protection policies is an effective strategy to enhance urban ESV on a regional level44,45. Considering the significant impact of HAI on ESV in urban and rural areas within a macro - regional context, the following policy recommendations are proposed:

  1. 1.

    Urban ecological enhancement policies

    At the regional level, cities should focus on enhancing urban parks and green spaces. This can be achieved by developing regional green infrastructure networks that connect different urban areas. Diversifying urban green spaces with features like pedestrian greenways, shaded parking lots, and rooftop gardens not only enriches urban ecological diversity but also contributes to the overall ESV of the region. For example, regional planning authorities can set standards for the proportion of green spaces in new urban developments and encourage the integration of green elements into existing urban areas through renovation projects.

  2. 2.

    Rural land protection and intensive use policies

    In rural areas of the macro - region, stringent enforcement of cropland protection policies is crucial. Regional governments should establish a comprehensive monitoring system to prevent illegal land conversion and ensure the sustainable use of cropland. Expanding pilot projects for land conservation and intensive use can serve as models for other regions. These projects can include innovative agricultural practices such as precision farming and crop rotation, which can improve agricultural productivity while minimizing environmental impact. Additionally, advocating for the conversion of non - agricultural land back to cropland, where feasible, can enhance agricultural and rural development on a regional scale. For instance, regional agricultural development plans can allocate funds for land reclamation and restoration projects in areas with degraded cropland.

  3. 3.

    Wetland and water area protection policies

    In regions with significant water bodies and wetlands, such as the Chaohu Lake basin and similar areas, continuing comprehensive management and wetland restoration projects are essential. Regional environmental protection agencies should coordinate with local governments to intensify efforts to restore cropland, aquaculture, and housing areas back to wetland. This can be achieved through policy incentives such as subsidies for wetland restoration projects and penalties for illegal land use. Safeguarding water areas and wetlands on a regional level contributes to the overall ESV of the area. Developing leading ecological civilization demonstration zones that integrate urban and aquatic environments, like the proposed concept in the Chaohu Lake basin, can create spaces conducive to living and business. Regional planning should consider the ecological carrying capacity of wetlands and water bodies when allocating land for development, ensuring a balance between economic growth and ecological protection.

Limitations and future work

Land use change is a defining feature of rapid urbanization. Consequently, this study investigates the impact of HAI on ESV, using the land type change method to quantify HAI. However, there are two limitations. Firstly, the HAI indicator is based solely on land types and does not account for non-land-based human activities such as pollutant emissions and energy consumption, potentially leading to an underestimation of the overall impact of human activities on ecosystems. Secondly, the study does not consider the potential effects of climate change on ESV. Changes in precipitation patterns can alter hydrological regulation services, while temperature variations may affect carbon sequestration and biodiversity maintenance. Future research should systematically compare the differences in ESV assessments resulting from various HAI measurement methods46 (e.g., net primary productivity human appropriation, composite indices, human footprint index) and incorporate climate variables to construct a comprehensive HAI assessment framework that includes multi-dimensional indicators such as land use, emissions, energy, and socioeconomic factors. This will enable a more comprehensive analysis of the complex interactions between human activities and ecosystem services.

In this study, we investigated the impact of HAI on ESV at 1 km grid scale, based on considerations of the spatial extent of the study area. It is noteworthy that both HAI and ESV exhibit scale effects, presenting different spatial distribution patterns and spatial correlation characteristics at various scales47,48. Therefore, in future research, we aim to explore how scale effects influence the coupling process between human activities and ecosystem services, in order to gain a deeper understanding of the dynamic changes and complexity of the interaction mechanisms between humans and ecosystems at different scales.

Conclusion

This study calculated ESV and HAI in Hefei from 2000 to 2020 based on LULC data. By analyzing the changes in ESV and the impact of human activities on the value of different ecosystem services, the study determined the threshold of HAI’s influence on ESV and identified local abnormal impact information. These findings contribute to understanding how human activities and ecosystem services interact in rapidly developing cities.

  1. 1.

    Over the past two decades, Hefei’s ESV had experienced a marked decline, from 47.012 billion yuan in 2000 to 45.534 billion yuan in 2020, particularly in areas of rapid urbanization.

  2. 2.

    The decline in ESV primarily stemmed from significant losses due to the transformation of cropland into built-up land and water area into cropland, propelled by urban development and agricultural intensification. Conversely, notable gains arose from reconverting built-up land to cropland and transforming cropland into aquatic environments, guided by policies such as “returning housing to cropland” and “converting cropland to wetland.”

  3. 3.

    The impact of HAI on the value of ecosystem services across different functions was ranked from high to low as CSV, ESV, RSV, SSV, and PSV. HAI exhibited a negative correlation with ESV, RSV, and CSV, and shifted from a positive to a negative correlation with PSV and SSV, with the transition threshold range locked between 0.176 and 0.262. This analysis identified three critical HAI thresholds (0.176–0.348 for ESV, RSV, SSV, and CSV; 0.348–0.606 for SSV; 0.434–0.692 for PSV) within which the value of ecosystem services could be maintained in equilibrium.