Introduction

Ecosystem services serve as a crucial bridge between ecosystems and socio-economic systems, providing essential services and products to humanity. They play a pivotal role in ecological conservation and sustainable development, serving as the foundation of human existence and progress1. The structure and dynamics of ecosystems profoundly influence the functioning of modern society and the economy, ensuring human well-being, improving quality of life, and serving as an integral component of constructing ecological civilization2. As urbanization accelerates, human activities rapidly alter urban land use, resulting in short-term changes in ecosystem structure and function. These changes, in turn, impact ecosystem dynamics and the quantification of urban ecosystem service values3. To better manage the relationship between human activities and the provision of ecosystem services, it is essential to study the variation in ecosystem service values from a land-use perspective. Such research illuminates the effects of human activities on ecosystems and is crucial for fostering harmonious human-environment relationships and promoting regional sustainability4.Urban ecosystems in arid and semi-arid regions are facing significant challenges due to the uneven spatial and temporal distribution of water resources. This has a direct impact on the stability of the local ecological environment and hinders human socio-economic development. Therefore, it is crucial to study ecosystem services in arid regions and identify their driving elements in order to formulate sustainable development strategies and protect the ecological environment.

Nearly two-thirds of the world’s natural resources have been depleted in recent years5. The assessment of ecosystem service value can play a crucial role in maintaining urban ecological security, achieving a better balance between development and utilization, and environmental protection, as well as promoting sustainable development.Land use and land cover changes are critical factors influencing Ecosystem Service Value, playing a pivotal role in preserving ecosystem service functions. They hold significant importance amidst global environmental changes and efforts towards sustainable development. However, With the increase in productivity in human society, human utilization and transformation of land has become more and more frequent, and human activities have accelerated the destruction of the surface environment, leading to large-scale land-use changes, seriously affecting ecosystem service functions, and the ecological environment has suffered serious challenges leading to a series of ecological and environmental issues6,7. Consequently, both domestic and international scholars have conducted extensive research on land use, land cover changes, and ecosystem services8,9. In the realm of ecosystem service research, Costanza and his colleagues, drawing on a global species dataset, have investigated how ecosystem services respond to changes in land use and land cover10. As research progresses, scholars worldwide have shifted their focus from a global scale to specific regions, such as Lake Ebi11, the Ganges Plain12, coastal areas of Tanzania13, and regions within China like the Qinghai-Tibet Plateau and Karst regions14,15,16. Regarding research methodology, common practices include using approaches such as the equivalent factor method17, the functional value method, and the energy value method to quantify ecosystem service values18,19. Additionally, studies frequently employ tools like InVEST20,21, spatial autocorrelation22, and geographically weighted regression models20 to explore how ecosystem service values respond to changes in land use23. Research tends to focus on aspects such as value assessment24,25,26, the evolution and mechanistic analysis of ESV26,27,28, the relationship between ecosystem services and land use28,29, and future multi-scenario predictions of land use and ESV30,31. In conclusion, there is a trend towards more detailed assessments at smaller scales, combining multiple methods for quantitative evaluations, and achieving a balance and synergy of ecosystem services in terms of content. However, it is crucial to elucidate the spatiotemporal patterns and driving mechanisms of ecosystem service values to better regulate regional ecological restoration32, especially in the context of research on eco-functional cities in China, which is of utmost urgency. Therefore, this study, based in Lanzhou City—a climate-sensitive area with ecological functional zones—integrates fine-scale grid-based spatial mapping of land use changes with sensitivity analysis of ecosystem service values to analyze the spatiotemporal aspects of ecosystem service values in Lanzhou City.The contribution of this study is beneficial to enrich methodological and theoretical research and bridge the research gap to some extent. Secondly, it comprehensively analyzes the spatial pattern of ecosystem service value in Lanzhou from the perspectives of spatial distribution and clustering, reveals the evolution trend of spatial pattern, and provides empirical basis for policies based on local conditions.

Lanzhou City is located in the upper reaches of the Yellow River, a critical region for water source conservation, characterized by arid conditions and low rainfall typical of Northwestern urban centers. The rapid pace of urbanization, combined with its unique geography nestled between mountains and flanked by a river, has intensified ecological carrying capacity pressures31. Therefore, investigating the impact of land use on Lanzhou City’s ecosystem is essential for current development32. In conclusion, this study utilizes three periods of land-use data spanning 2000 to 2020 from Lanzhou City. By analyzing patterns of land use changes, it employs methods such as equivalent factor coefficient correction, spatial autocorrelation analysis, and clustering analysis. The study conducts spatiotemporal analysis on the evolution of ESV and utilizes geographic detectors to explore dominant factors influencing spatial heterogeneity in ESV and their interactions. The objective is to uncover the driving forces behind spatial variations in ESV, ultimately providing scientific support for decision-making in land use planning and ecological conservation efforts in Lanzhou City.

Materials and methods

Overview of the study area

Lanzhou City (92°13′-108°46′ E,32°11′-42°57′ N) is strategically positioned as a crucial transportation hub and logistics center in Northwest China, serving as a core node city within the Silk Road Economic Belt. The Yellow River flows through the city from east to west, making it the only provincial capital intersected by this river (Fig. 1). Situated at an average elevation ranging from 1530 to 1580 m, Lanzhou experiences a temperate continental climate with an annual mean temperature of 10.3 °C, around 2446 h of sunshine annually, and an average precipitation of 327 millimeters, concentrated mainly from June to September. Within its administrative boundaries lie two notable national-level nature reserves: the Gansu Xinglongshan National Nature Reserve and the Gansu Liancheng National Nature Reserve. It also borders the eastern segment of the Qilian Mountains Nature Reserve, collectively enhancing Lanzhou City’s ecological security. Established as a heavy industrial center in the 1950s, Lanzhou City underwent rapid urban expansion, significantly contributing to China’s rapid socio-economic development. However, it also faced challenges such as severe thermal inversion and environmental pollution issues that garnered national attention33. Presently, with the implementation of national ecological compensation policies, Lanzhou City has strengthened its ecological security measures to some extent.

Fig. 1
figure 1

Location of the study area.Map created using ArcGIS Desktop version 10.8 downloaded from a third-party source. URL: https://pan.baidu.com/s/1vVJXMNeRVSqbPFcpNA-xBg Extractioncode: tvxj).

Full size image

Data sources

This study utilized three phases of TM remote sensing images for the years 2000, 2010, and 2020, obtained from the China National Geomatics Center. Each phase comprised four scenes with four bands, and the spatial resolution for each TM image band was 30 m. Land use, topography, population, and NDVI data were sourced from the Chinese Academy of Sciences Resource and Environmental Science and Data Center. Socio-economic data primarily originated from the National Bureau of Statistics and the Global Roads Open Access Database, while historical climate data were provided by the National Meteorological Information Center (Table 1).

Table 1 Data sources.
Full size table

Research methods

Estimation of ecosystem service value

In recent years, Chinese scholars such as Xie Gaodi have researched the equivalence value table of ecosystem services in China. Their work has identified equivalent factors for ecosystem services related to cultivated land, forest land, grassland, water areas, and unused land. They have determined that the economic value of one ecosystem service equivalent factor equals 1/7 of the annual per-hectare food value. Based on this research, the economic value of one ecosystem equivalent in China is established at 3406.50 CNY per hectare17. According to data from the Statistical Yearbook, the national average grain yield per unit area was 4974 CNY per hectare from 2000 to 2020, compared to 2515.08 CNY per hectare in Lanzhou. This suggests that the correction factor for ecosystem service equivalence in Lanzhou is 0.5056, setting the value of one standard unit of ecosystem services in Lanzhou at 1722.32 CNY per hectare. This determination enables the assessment of ecosystem service values per unit area for different land use and cover types in Lanzhou34 (Table 2).

Table 2 Ecosystem service value per unit area of land use/coverage types (RMB.hm-2).
Full size table

Finally, the ecosystem service value for each land use/cover category in Lanzhou City was calculated by multiplying the area of each category by its respective ecosystem service value. The calculation formula is as follows:

$$ESV=\sum U_i \times V_{ij}$$
(1)
$$V_i=\sum\nolimits_{\text j=1}^{\text n} ESV_{ij}$$
(2)

In the formula, Ui represents the ecosystem service value of land use/cover type i, Vij denotes the area of land use/cover type i, j represents the item f ecosystem service function, and ESVij signifies the ecosystem service value associated with land use/cover type i.

Sensitivity calculations

Using a sensitivity analysis approach, this study adjusted the ecosystem service value coefficients per hectare for cultivated land, forest land, grassland, water bodies, unused land, and construction land by ± 50%. This adjustment aimed to calculate the ESV sensitivity index, showing how different land use types in Lanzhou City depend on ecosystem service values over time33. The calculation formula is as follows:

$$\text{CS}=\frac{(\text{ESV}_{\text j}-\text{ESV}_{\text i})/\text{ESV}_{\text i}}{(\text{VC}_{\text {jk}}-\text{VC}_{\text {ik}})/\text{VC}_{\text {ik}}}$$
(3)

Spatial autocorrelation analysis

Spatial autocorrelation assesses spatial dependence and variability of a variable within a specific region. Moran’s I was utilized to characterize the spatial autocorrelation of ESV34,35.

$$\text{Moran}^\prime\ \text{I}=\frac{\text{n}\sum\limits_{\text{i=l}}^{\text n}\times \sum\limits_{\text{i=l}}^{\text n}\times\ \text{W}_{\text{ij}}(\text{X}_{\text i}-{\overline X})(\text{X}_{\text j}-{\overline X})}{\text{n}\sum_{\text{i=1}}^{\text{n}}\times(\text{X}_{\text i}-{\overline X})^2(\sum \text{i}\sum\text j\ \text{W}_\text{ij})}$$
(4)

Where Xi and Xj represent the attribute values of grids i and j, respectively, \(\overline X\) is the average value of the attribute, Wij is the spatial weight matrix, and n is the number of grids. Moran’s I exceeding 0 indicates positive spatial autocorrelation in ESV, with higher values indicating stronger positive autocorrelation. Conversely, Moran’s I below 0 indicates negative spatial autocorrelation, with more negative values indicating stronger negative autocorrelation. A Moran’s I value of 0 suggests no autocorrelation, indicating a random distribution of ESV.

Clustering and outlier analysis methods can evaluate the spatial differentiation patterns of changes in ecosystem service values influenced by the spatiotemporal patterns of land use in the study area36, as shown in the following formula:

$${{\text{LISA}}_{\text i}=\frac{(X-\overline X)}{m_0}}\sum_{\text i}^{}\;W_{ij}\left(X-\overline X\right)$$
(5)
$$m_0\sum_{\text i}^{}\left(\text{Xi}-\overline{\chi}\right)/n$$
(6)

In the formula, Wij represents the spatial weight matrix between units, i denotes the attribute value of each unit, \(\overline X\)  is the mean of all attribute values, and n signifies the total number of regional units. A LISA value greater than 0 indicates spatial clustering of ecosystem services in units with high-high or low-low values; conversely, a LISA value less than 0 suggests spatial clustering of ecosystem services in units with high-low or low-high values.

The geodetector

The Geodetector is an innovative statistical methodology designed to explore spatial heterogeneity and reveal its underlying driving factors. It consists of four distinct detectors - the Factor Detector, Interaction Detector, Risk Area Detector, and Ecological Detector. This method finds widespread application in various fields including land use, public health, regional planning, environmental ecology, remote sensing, and computer networking37. In this study, our emphasis is on the quantitative exposition and analysis of the Factor and Interaction Detectors.

$$\text q=1-\frac{\text 1}{\text{N}\delta^2}\sum_{h=1}^L\;N\ h\delta h^{2}$$
(7)

Here, q represents the degree to which a specific factor explains the spatial distribution of ESV in Lanzhou City, ranging from 0 to 1. The symbol h = 1, 2, …, L denotes the classification of natural factors, where Nh and N denote the number of units in layer h and across the entire study area, respectively.\(\text {N}\delta^2\) and \(\delta^2\) indicate the variances of ESV within layer h and across the entire area, respectively. The interaction analysis evaluates how the combined effects of multiple factors influence changes in ESV, contrasting with the impact of individual factors.

Results

Spatial and temporal evolution of land-use types

Using land use/cover change data spanning from 2000 to 2020, this study delineates the trends in land use and cover changes within Lanzhou City, as illustrated in Fig. 2.

Fig. 2
figure 2

Change diagram of land use/cover in Lanzhou from 2000 to 2020.Map created using ArcGIS Desktop version 10.8 downloaded from a third-party source. URL: https://pan.baidu.com/s/1vVJXMNeRVSqbPFcpNA-xBg Extractioncode: tvxj).

Full size image

Based on Figs. 2 and 3, it is clear that from 2000 to 2020 in Lanzhou City, grasslands consistently covered over 60% of the total land area, while cultivated lands consistently represented more than 18%, together accounting for nearly 80% of the total area. This highlights grasslands and cultivated lands as the predominant land types in the city. Following these are forested areas at 5.8%, and then construction lands, unused lands, and water bodies at 2.58%, 1.33%, and 0.5%, respectively. The primary trends in land use changes during this period included a noticeable decrease in cultivated and grassland areas, accompanied by a significant increase in construction lands. Specifically, forested areas and water bodies experienced some reduction between 2000 and 2010, followed by an upward trend from 2011 to 2020. The expansion of construction land largely occurred at the expense of cultivated land, a trend amplified by Lanzhou’s role as a provincial capital and the development of the Lanzhou New District. The dynamics of land use and land transfer matrices indicate that the decrease in grassland areas resulted from the conversion of cultivated lands between 2010 and 2020. Spatially, cultivated lands are predominantly located in the eastern regions of Yuzhong and Yongdeng counties; forests are mainly situated in the western parts of Yongdeng County and southern Yuzhong County; construction lands are concentrated along the Yellow River in central Lanzhou and the Lanzhou New District, with dense distribution in the eastern areas of Yuzhong County. The spatial distribution of water bodies and unused lands remained relatively stable.

Fig. 3
figure 3

Degree of land use change in Lanzhou from 2000 to 2020.

Full size image

Spatio-temporal change of ESV

Temporal changes in the value of ecosystem services

Calculations based on Formulas 1 and 2, along with Fig. 4, reveal that Lanzhou City’s total ESV experienced significant growth from 17.937 billion yuan in 2000 to 19.386 billion yuan in 2020, marking an increase of 1.45 billion yuan and a growth rate of 7.74%. Specifically, the period from 2000 to 2010 saw a modest rise of 0.035 billion yuan, equating to a growth rate of 0.20%. In contrast, the decade from 2010 to 2020 experienced a more pronounced increase of 14.49 billion yuan, translating to a growth rate of 8.07%. The accelerated growth of ESV in the latter decade, compared to the former, can be primarily attributed to the substantial baseline of grassland coverage and the expansion of forest and water areas observed during the study period.

Fig. 4
figure 4

ESV of different land use types in Lanzhou city from 2000 to 2020.

Full size image

From 2000 to 2020 in Lanzhou City, grasslands dominated the ESV attributed to different land use types, comprising over 52% of the total ESV. Following were forest lands, cultivated lands, and water bodies, contributing 17.99%, 12.64%, and 3.23% of the total ESV, respectively, while unused lands constituted the smallest proportion. The most significant increase in ESV occurred in forest lands, rising by 2.817 billion yuan, equating to an increase rate of 86.88%. This was followed by water bodies, which saw an increase of 0.086 billion yuan, corresponding to a growth rate of 14.57%. In contrast, the ESV of cultivated lands decreased by 1.488 billion yuan, a decline rate of 37.79%, and both unused lands and grasslands experienced reductions at rates of 10.11% and 0.18%, respectively. Therefore, during the period of accelerated urban development from 2010 to 2020, Lanzhou City’s ESV expanded rapidly, primarily driven by increased areas of forest lands and water bodies.

Analyzing the ecosystem service values (Table 3) from 2000 to 2020 reveals that soil conservation held the highest value among different types of ecosystem services. Hydrological regulation, biodiversity conservation, climate regulation, water resource provision, gas regulation, food production, raw material generation, and aesthetic landscapes followed in descending order of significance. These findings indicate that Lanzhou City relies significantly on its natural resources to sustain its ecological environment. During the same period, the ecosystem service value of food production declined due to a reduction in cultivated land area by 1741.83 km², thereby decreasing its service value and functionality. In contrast, the ESV of other ecosystem service types consistently increased annually. The most significant changes were observed in raw material generation and aesthetic landscapes, followed by food production, gas regulation, water resource provision, hydrological regulation, biodiversity conservation, climate regulation, and soil conservation. This trend highlights Lanzhou City’s successful utilization of natural ecosystems, supported by artificial ecosystems.

Table 3 ESV of ecological service types from 2000 to 2020 (RMB.108).
Full size table

Spatial variation in the value of ecosystem services

Using ArcGIS 10.8 software and a 1 km×1 km grid as the unit of analysis, this study spatially visualized four types of ecosystem services within Lanzhou City. The natural break method categorized these services into five distinct intervals. Figure 5 illustrates that the spatial distribution of Ecosystem Service Value in Lanzhou City correlates closely with land use patterns. Areas with high ESV are predominantly located in the western region of Yongdeng County’s Tulu Valley National Forest Park, the southern part of Yuzhong County’s Xinglong Mountain National Nature Reserve, and the basins of the Yellow River and its tributary, the Huangshui River. These regions benefit from intersecting rivers, extensive vegetation cover, well-preserved water functional zones, and minimal human activity impact. In contrast, lower ESV areas are concentrated in Lanzhou City’s urban and Lanzhou New Districts. These areas feature flatter terrains, intense human activity, extensive development, and widespread construction land, resulting in lower ESV values. In conclusion, the ESV distribution in Lanzhou City exhibits distinct spatial patterns, with higher values in the western and southern regions and lower values in the central and eastern areas. This pattern indicates a clustering of high-value zones and a radial expansion of low-value zones, leading to noticeable spatial polarization between areas with high and low ESV.

Fig. 5
figure 5

Spatial distribution of ESV in Lanzhou city.Map created using ArcGIS Desktop version 10.8 downloaded from a third-party source. URL: https://pan.baidu.com/s/1vVJXMNeRVSqbPFcpNA-xBg Extractioncode: tvxj).

Full size image

Global trend analysis

The trend surface analysis depicted the ecosystem service value of Lanzhou City through three-dimensional curves, illustrating the spatial distribution trends of ESV across the city. The spatial trend line generally maintains a pattern of higher values in the west and south, and lower values in the east and north. Specifically, the trend surface slopes more steeply in the northwest-southeast direction (Fig. 6), indicating a pronounced spatial distribution along these axes. Given that the forested areas of Lanzhou City are predominantly located in the west and south, the ESV size correlates significantly with these regions; denser vegetation and better ecosystem conditions contribute to larger ESV values.

Fig. 6
figure 6

Surface Changes in Spatial Trends of ESV in Lanzhou City.

Full size image

Sensitivity analysis of the value of ecosystem services

Using Formula 3, we computed the ESV sensitivity index (Table 4). Among the six land use categories, grasslands exhibited the highest sensitivity index at approximately 0.48%. This suggests that a 1% increase in the value coefficient of grasslands results in a 0.48% increment in ESV. Water bodies followed with the next highest sensitivity index, while forest land, cultivated land, and unused land showed the least sensitivity indices, with construction land registering a zero index. Throughout the study period, the ESV sensitivity index showed minimal variation between consecutive years. Over the entire research span, sensitivity indices for water bodies and forest lands generally indicated an increasing trend, whereas those for other land types displayed a decreasing tendency. Notably, from 2000 to 2020, the ESV sensitivity index consistently remained below 1, indicating a lack of ESV elasticity and affirming the reliability of the study’s estimations.

Table 4 Sensitivity index of ESV for different land types in 2000–2020.
Full size table

Spatial autocorrelation analysis of the value of ecosystem services

Most scatter points representing ESV in Lanzhou City across the three time periods from 2000 to 2020 are concentrated in the first quadrant (Fig. 7). The global Moran’s I values were 0.708, 0.718, and 0.775, respectively, with p-values of 0.05. Moran’s I value exhibited a gradual upward trend from 2000 to 2020, indicating a positive spatial correlation in the ecosystem service value across the study area. The spatial clustering effect of ESV in Lanzhou City strengthened over time, suggesting a viable approach for studying the distribution pattern of spatial correlation.

Fig. 7
figure 7

Moran scatter plot of ESV in the study area in 2000–2020.

Full size image

By employing the fishnet tool with a 1 km×1 km grid as the analytical unit and utilizing cluster and outlier analysis within ArcGIS 10.8, this study investigated the spatial distribution pattern of ESV changes in Lanzhou City from 2000 to 2020 (Fig. 8). The classifications HH, HL, LH, and LL correspond to four spatial clustering types: high-high, high-low, low-high, and low-low values, respectively. Over the 2000–2020 period, the spatial agglomeration effect of ESV in the study area showed a progressive intensification. HH clustering zones were primarily concentrated in the western and southern regions of Lanzhou City, particularly in Yongdeng County and Yuzhong County, with minimal occurrences elsewhere. LL clusters were predominantly observed in central Lanzhou City and around the Lanzhou New District, including the main urban area along the Yellow River and the northern region of Yuzhong County. HL clusters were mainly situated at the peripheries of HH clustering areas, presenting in block-like formations, while LH clusters were concentrated in the central urban area, with scattered occurrences around the HH clusters. The widespread yet spatially confined distribution of both HH and LL clusters suggests that the ecological balance of Lanzhou City has been partially disrupted, despite many areas remaining unaffected. Analysis across different phases revealed minimal changes in clustering distribution between 2000 and 2010 across various counties and districts within the study area. However, during the 2010–2020 period, the high-high (HH) clustering of ESV in Lanzhou City showed a declining trend, while the low-low (LL) clustering demonstrated a strengthening tendency. This shift could be attributed to the expansion of cultivated land in Lanzhou City, leading to a transition in ecosystem service values from moderate to lower-value zones.

Fig. 8
figure 8

Distribution pattern of variable spatial correlation in Lanzhou city.Map created using ArcGIS Desktop version 10.8 downloaded from a third-party source. URL: https://pan.baidu.com/s/1vVJXMNeRVSqbPFcpNA-xBg Extractioncode: tvxj).

Full size image

Analysis of the drivers of spatial differentiation in the ESV

Geodetector-based factor detection analysis

Vector points generated in ArcGIS 10.8 were used to extract reclassified results, which served as X data for GeoDetector analysis. The average ESV from 2000 to 2020, categorized by different service types, constituted the Y data. These X and Y datasets were then input into GeoDetector for analyses on differentiation, factor detection, and interaction detection. The spatial distribution map of the reclassified data is depicted in Fig. 9.

Fig. 9
figure 9

Spatial layout of ESV spatial differentiation driver.Map created using ArcGIS Desktop version 10.8 downloaded from a third-party source. URL: https://pan.baidu.com/s/1vVJXMNeRVSqbPFcpNA-xBg Extractioncode: tvxj).

Full size image

Using a 1 km×1 km grid for analysis in Lanzhou City, and considering ESV as the dependent variable, along with independent variables such as elevation, slope, temperature, precipitation, NDVI, GDP, population, and distances to roads, railways, and rivers, GeoDetector was employed to determine the q value of single driving factors (Table 5). The results of the interaction and factorial probes showed that all variables passed the significance test of 0.005. NDVI emerged as the most significant contributor to ESV variations, accounting for 42.9%. Population followed with a 25% contribution, while precipitation and distance to roads contributed 24%. GDP and DEM contributed 22%, and distance to railways contributed 21%, with other factors contributing less than 20%. The distribution of low and high-value zones of ESV is closely linked to land use types, with low-value areas predominantly occupied by construction and unused lands, which have relatively low ecosystem service values of 0 yuan and 0.087 × 107 yuan, respectively. In summary, the spatial disparities in ESV within Lanzhou City are influenced by the interplay of natural and societal factors, natural factors also had more significant effects on ESV in the study area.

Table 5 Factor detection for spatial heterogeneity of ESV.
Full size table

Geodetector-based factor interaction detection analysis

In this study, interaction detection was utilized to quantitatively expose and analyze the driving factors of ecosystem service values in Lanzhou City. The interaction detector determines whether two influencing factors operate independently or in conjunction. By employing the interaction detection module, this research identified the interactive effects of ten influencing factors on ESV. The results depicted in Fig. 10 from the interaction detection indicate that all driving forces amplify the spatial distribution and differentiation characteristics of ESV when interacting with each other, rather than acting in isolation. Consequently, ESV in Lanzhou City is shaped by the interplay of multiple factors. The interaction between NDVI and precipitation emerges as the most potent, with a q value of 0.57. NDVI’s interaction with other factors significantly surpasses the impacts among other factors, with all q values exceeding 0.4. Natural factors such as DEM, temperature, and slope exhibit a close association with land use. The interaction between GDP and population, indicated by a q value of 0.44, also stands out significantly above others. The combined impact manifests in both natural elements and socio-cultural factors, particularly with human factors like GDP, population, and proximity to roads and railways, which intensify human activities. The pronounced interaction among other natural factors suggests that the driving forces of ESV in Lanzhou City result from a culmination of interactions among multiple factors.Therefore, in the process of urban development and the construction of ecological civilization, it is essential to not only consider the impact of natural and economic factors on urban development and the ecological environment but also to integrate the reinforcing effect produced by the interaction of various factors.

Fig. 10
figure 10

GeoDetector interaction detection and its type results.

Full size image

Discussions

Changes in LULC and ESV

This research conducted a quantitative assessment and spatial analysis of land use changes in Lanzhou City between 2000 and 2020, along with their implications for ecosystem service value. The findings indicate pronounced shifts in land use during this period, marked by a reduction in cultivated and grassland areas, and a notable expansion in construction land38. Since the 21st century, rapid urbanization in Lanzhou City has intensified development, human activities, and land use. This trend contrasts with Gu’s findings. Overexploitation of land resources has increased the total value of ecosystem services and strengthened them, attributed to extensive occupation of grassland, and expansion of forested land and watersheds39. ESV in Lanzhou City gradually increased during this period, driven primarily by gains in grasslands, forests, and water bodies. These results suggest a correlation between changes in land use and ESV, consistent with other scholarly studies40,41. Guo used an enhanced hybrid frog-leaping algorithm to develop a land optimization model integrating ecosystem service value as the fitness function. Focusing on land use trends in Lanzhou City and aiming to enhance ESV, this study confirms a strong link between the city’s land use changes and ESV42. It’s notable that expanding forest and water areas significantly impact the total value of ecosystem services in Lanzhou City. Grasslands have an ESV per unit area of 12831.3 yuan/hm2, ranking third among ecosystem types after water bodies and forests. Therefore, increasing forest and water areas, with their higher economic value within ecosystems, helps mitigate the overall reduction in ecosystem service value.

Analysis of ESV drivers

Understanding the drivers of ESV is crucial for formulating effective land-use strategies. Driver factor analysis reveals that NDVI effectively delineates spatial variations in ESV. Elevation differences influence human activities, with the low-lying plains of Lanzhou City being more favorable for human habitation and production. ESV is significantly affected by cultivated and constructed lands resulting from human activities. Factors such as precipitation, temperature, and other climatic and locational variables also play important roles. This highlights the considerable influence of Lanzhou City’s climate on human activities, where extensive construction reduces ESV. Moreover, the impact of road networks and transportation on ESV distribution underscores human activities’ significant role. Interaction detection shows a pervasive dual-factor enhancement effect in explaining ESV’s spatial variability. This emphasizes the substantial impact of interactions between multiple factors on ESV distribution, linking natural and socio-economic factors. Natural factors provide stronger explanatory power than socio-economic factors, underscoring the natural environment’s significant influence on ESV. Therefore, local governments, in formulating land-use policies, should consider integrating various factors and employing diverse regulatory strategies to prevent regional declines in ESV. Addressing the current decline in ESV and strained human-environment relations in the study area necessitates implementing scientifically-based land-use conservation practices. This includes establishing rational land-use scales, enhancing ecological environment protection, and prudently minimizing the impact of human production and activities on the natural environment.

Uncertainties and implications

The methodology used to calculate ESV in this research has inherent limitations. It assumes uniformity in ESV across entire biological community types over different years, leading the study to generalize the unit value of specific commodities to average values over diverse years, Instead of a specific value for a particular year. This approach may introduce deviations in ESV outcomes.In terms of research scale, only a 1 km×1 km grid was utilized as the unit of analysis. The effects of different grid scale changes on the degree of data aggregation and spatial and temporal variations in ecosystem service value were not further examined. Due to the complexity of the aggregation process, the driving factors influencing spatial heterogeneity of ESV will differ significantly at various grid scales. Therefore, when enhancing ecosystems, it is essential to select an appropriate analysis scale and consider not only the differences in influencing factors at different grid scales but also their interactions.Drawing from terrestrial ecosystem service equivalence values proposed by Xie and adapting them to suit Lanzhou City’s dry urban context, this study adjusted ESV evaluation coefficients. However, due to constraints in remote sensing imagery and a lack of appropriate equivalence factors, ESV for construction land in Lanzhou City was not included. Given the complexity and diversity of ecosystem services, and the nonlinear relationships among different services characterized by mutations, thresholds, and substitutions, future studies could benefit from using high-resolution remote sensing data to delineate more nuanced land use categories. By employing additional secondary classifications, a more accurate assessment of ESV can be achieved. This approach would allow for an exploration of how refined land use types in Lanzhou City influence the spatio-temporal evolution of ESV.

Conclusions

Building on the analysis of land use changes in Lanzhou City, this research calculated ESV for the city from 2000 to 2020. The spatiotemporal evolution and driving forces of ESV in the study area were thoroughly examined, revealing that spatial heterogeneity of ESV was predominantly influenced by natural factors. Therefore, there is a recommendation to prioritize understanding the impact of natural factors on ecosystem services in Lanzhou City. The main findings of the study are summarized as follows:

  1. (1)

    Between 2000 and 2020, Lanzhou City underwent significant changes in land use, characterized mainly by a notable decrease in cultivated and grassland areas, and a considerable increase in construction and unused lands. The area covered by forests and water bodies initially decreased from 2000 to 2010 but showed an upward trend from 2010 to 2020. The decline in grassland areas resulted primarily from the development and utilization of existing cultivated land between 2010 and 2020. Moreover, there was a dramatic 138% increase in construction land, highlighting Lanzhou City’s rapid development during this period.

  2. (2)

    The ESV of Lanzhou City increased from 17.937 billion yuan in 2000 to 19.386 billion yuan in 2020, reflecting a growth rate of 7.74%. Grasslands dominate in terms of overall magnitude, accounting for over 52% of the total ESV, followed by forest lands, cultivated lands, and water bodies, contributing 17.99%, 12.64%, and 3.23% respectively, while unused lands make the smallest contribution. In terms of ecosystem service types, soil conservation emerges as the most valuable, followed by hydrological regulation, biodiversity, climate regulation, water resource supply, gas regulation, food production, raw material production, and aesthetic landscapes. These findings indicate that Lanzhou City heavily relies on its natural resources to sustain its ecological environment.

  3. (3)

    In Lanzhou City, the spatial distribution of ESV predominantly shows higher values in the western and southern regions and lower values in the central and eastern areas. Throughout the study period, sensitivity indices for water bodies and forest lands generally trended upward, while those for other land types decreased. Spatially, high-high (HH) clusters are mainly found in Tulu Valley National Forest Park, Xinglong Mountain National Nature Reserve, the Yellow River basin, and the Huangshui River basin. Low-low (LL) clusters are primarily located in the central urban area and Lanzhou New District, expanding radially. High-low (HL) and low-high (LH) clusters form block patterns, highlighting significant spatial polarization between high and low ESV areas.

  4. (4)

    In the context of urbanization in Lanzhou, NDVI emerges as the most significant contributor to ESV among single-factor drivers, accounting for 43%, followed by GDP, precipitation, and proximity to roads. ESV in Lanzhou City is shaped by a complex interaction of multiple factors. The interactions involving NDVI with other factors exceed the impacts of individual factors, each contributing over 40%. These interactions exhibit a pattern of nonlinear enhancement. NDVI reflects the vegetative state in Lanzhou City, where grasslands dominate as the most extensive vegetation type. Despite high vegetation coverage, the ecosystem service value per unit area for grasslands is relatively low, resulting in a scenario of abundant vegetation cover but reduced ESV. Socioeconomic factors like GDP and population indicate the level and dynamics of economic development. Regions with higher economic development experience increased human activities, which significantly affect the ecological environment. These findings underscore the profound influence of climate change on ecosystem service values in Lanzhou City.