Introduction

Ecosystem services refer to the material basis and environmental conditions that sustain and meet the various needs of human survival, development, and well-being21,60. They not only maintain the stability of regional ecological security patterns, but also play a key role in providing for the material and spiritual needs of human society, as well as in ensuring sustainable development. Ecosystem services are primarily divided into two aspects: the provision of tangible material products and the provision of intangible services. They mainly include provisioning services, regulating services, and cultural services, as well as supporting services necessary to maintain other types of services11. Against the background of rapid urbanization, China’s economic growth and population increase have put pressure on the ecological environment. Among these pressures, the impact of land use/land cover changes as a consequence of human activities has a significant influence on ecosystem services55. Degradation of regional ecosystem services caused by land use/land cover changes directly threatens ecological security, sustainable economic and social development, and human well-being12,13,19,23. Therefore, scientific assessment of the value of ecosystem services is helpful in the rational formulation of policies for ecological land conservation, planning, and compensation55.

Among the existing approaches for evaluating ESV, the equivalence factor method proposed by11 has gained widespread attention across disciplines such as ecology, geography, and economics due to its broad applicability and ease of use40. Subsequently, domestic and foreign scholars further studied the evaluation method proposed by Constanza et al. (1997). Building on this foundation, they have explored and improved theoretical approaches for valuing ecosystem services46,47,65. Research by scholars revealed that the equivalence factor assessment method proposed by11 is based on a global scale and is not applicable to the actual ecological conditions in China. Chinese scholars56,58, after conducting a questionnaire survey among Chinese ecologists, proposed an equivalence factor table for ecosystem services that is consistent with the actual ecological conditions in China. This table is currently one of the more systematic and scientifically referenced equivalence factor frameworks in the field of domestic ecosystem services research, which essentially reflects the average levels of different types of ecosystem services across the country.

Research indicates that different ecological processes and environmental conditions influence the functional intensity of ecosystem services31. For example, functions such as organic matter production, gas regulation, and nutrient cycling7 are closely linked to Net Primary Productivity (NPP). In contrast, water supply and regulation, soil retention, biodiversity conservation, and cultural services are more influenced by factors such as rainfall, soil erosion, habitat quality, and recreational accessibility15,42,50. Therefore, spatial adjustments are necessary when applying the equivalence factor method to evaluate regional ESV37,62. The impact of land use on ecosystem services is a complex, multi-factorial process involving various ecological responses59. Most existing studies on the relationship between regional land use and ecosystem services rely on a single variable, such as NPP or biomass30, for spatial correction. Others use regional food crops as the standard economic equivalent for ESV10,32,28,56,57,58. However, these approaches often fail to reflect real-world ecological complexity.

Many studies on the interaction between land use and ecosystem services focus on specific ecosystem types, such as cultivated land20, forests14, grasslands67, and water areas59. These studies often overlook regions characterized by complex ecological conditions. Moreover, static cross-sectional analyses static cross-sectional analyses neglect the temporal dynamics of land use–ecosystem service interactions and often lack quantitative tools for intuitive exploration. Furthermore, ecosystem services in inter-provincial border regions—often characterized by complex ecological contexts—remain largely underexplored. Grid-scale analysis offers a finer resolution for evaluating regional ecosystem services.

Based on these considerations, this paper focuses on the HEZ, a key eco-economic region in China’s north-south climate transition zone. The study applies the correction methods proposed by Xie et al.55,56 to spatially adjust the ESV per unit area, assess dynamic changes in ESV at the grid scale, and explore its temporal and spatial evolution. The main research ideas are as follows: (1) Analyzing the structure and intensity of land use/cover changes in the HEZ from 1995 to 2020; (2) Using modified ecosystem service equivalent factors, calculating the ESV of various ecosystems in the HEZ at the grid scale; and (3) Utilizing spatial data analysis methods to investigate the spatiotemporal differentiation characteristics of the ESV in the HEZ. The marginal contribution of this study is reflected in the following aspects: (a) It captures the dynamic relationship between land use and ESV through the construction of an ESV profit and loss statement; (b) The spatially corrected grid-scale analysis improves the accuracy of ESV estimation; (c) By focusing on inter-provincial boundary zones and climatic transition areas, the study provides regionally representative insights and offers valuable references for similar regions.The obtained research results can provide scientific references and theoretical basis for ecological environmental protection and sustainable use of land resources in the HEZ.

Study area and data

Study area

The HEZ is one of the earlier established regional economic cooperation organizations in China and occupies an important position in the overall context of national economic and social development29. This region is located between 114°45′–120°54′E longitude and 33°12′–36°13′N latitude, at the southern edge of the Huang-Huai-Hai Plain. It lies north of the middle and lower reaches of the Yellow River and south of the Yangtze River Delta urban agglomeration. The region is characterized by flat terrain and a warm, semi-humid monsoon climate in the northern temperate zone with favorable hydrothermal conditions for ecological development.

Fig. 1
figure 1

Location of the HEZ.

Full size image

In 2018, with the State Council’s approval of the “Development Plan for the Huaihe River Ecological Economic Belt,” the promotion of coordinated development in the HEZ, and the exploration of a new model for building a basin ecological civilization, officially became a national strategy (www.gov.cn/zhengce/content/2018-10/18). The HEZ includes 10 cities in 4 provinces: Shandong Province (Zaozhuang, Jining, Linyi and Heze), Jiangsu Province (Xuzhou, Lianyungang and Suqian), Henan Province (Shangqiu), and Anhui Province (Suzhou and Huaibei)(Table 1). The total land area of this region is 96,543 km2. The HEZ is not only the junction area of the four provinces a`nd the marginal area of eastern coastal economic development, but also an important ecological and economic zone located in the transitional climate zone between the north and south of China (Fig. 1, plotted under the GCS_WGS_1984 of the geographic coordinate system.).

Table 1 Full names and abbreviations of 10 cities in the HEZ.
Full size table

Data

Land use and land cover (LULC) data at a spatial resolution of 30 m were obtained from the Data Center for Resources and Environmental Sciences (www.resdc.cn). Based on random survey sampling and field research, the overall classification accuracy of the dataset is greater than 90%52. Considering the actual situation of the HEZ and research needs, the land use types in the study area are reclassified into seven categories: farmland, woodland, grassland, water area, wetland, built-up land, and bare land. The administrative map data are based on the 2015 spatial vector data of the provincial administrative boundaries, which were obtained from the Data Center for Resources and Environmental Sciences (www.resdc.cn)(Figure1). During the study period, no changes occurred in the administrative boundaries of the region; therefore, using the 2015 administrative division map did not affect the consistency of the analysis across the entire study period.

Net Primary Productivity (NPP) data (2000–2010), rainfall data (2000–2015), soil erosion data (2005) and road density data (2010–2015) are used for spatial adjustments of equivalent factors. All data are freely available from the Data Center for Resources and Environmental Sciences (www.resdc.cn) and OpenStreetMap (https://www.openstreetmap.org/).

The data for calculating the ESV per unit area, including grain yield, planting area, unit price of crops (wheat, corn, soybeans), are from the “Statistical Yearbooks of Shandong,” “Statistical Yearbooks of Jiangsu,” “Statistical Yearbooks of Henan,” “Statistical Yearbooks of Anhui,” and “China Agricultural Product Price Survey Yearbook” for the years 1995–2020 (Table 2).

Table 2 Data sources.
Full size table

Methods

This study estimated the total ESV and individual ESV components in the HEZ based on land use changes and spatially adjusted ESV coefficients of unit area. Second, it introduced an improved cross-sensitivity coefficient for assessing the impact of net land use conversion on changes in ESV. Finally, it used exploratory spatial data analysis to examine the spatiotemporal patterns of ESV in the HEZ (Fig. 2).

Fig. 2
figure 2

Flowchart of spatiotemporal characteristics analysis of ESV in the HEZ.

Full size image

Methods of evaluation land use change

Degree of land use dynamism (K)

The index of \({\text{K}}\) quantitatively describes the range and pace of land use changes17:

$${\text{K=}}\frac{{{U_b} - {U_a}}}{{{U_a}}} \times \frac{1}{T}$$
(1)

where \({\text{K}}\) indicates the degree of land use dynamics for a specific land use type; \({U_a}\) and \({U_b}\) are, respectively, the areas that are annually under specific land use types; and T indicates the time in years.

Land use transfer matrix

The land use transfer matrix (\({K_{ij}}\)) is used to represent the dynamic transformation of each land use type in the study period25 as follows:

$${K_{ij}}=\left[ {\begin{array}{*{20}{c}} {{K_{11}}}&{{K_{12}}}&{...}&{{K_{1n}}} \\ {{K_{21}}}&{{K_{22}}}&{...}&{{K_{2n}}} \\ {...}&{...}&{...}&{...} \\ {{K_{n1}}}&{{K_{n2}}}&{...}&{{K_{nn}}} \end{array}} \right]$$
(2)

where K is the LULC area, n is the number of LULC types, and i and j are the LULC types at the start and at the end of study period, respectively.

Assessment of ecosystem services

Model of spatio-temporal correction of equivalence factors

In this study, ecosystem services were classified into four primary and eleven secondary categories based on the equivalent factor table proposed by Xie et al.58. As discussed above, neglecting regional differences in ecological conditions and socio-economic development may lead to biased estimates of ESV32. Therefore, spatial adjustment of equivalence factors is essential prior to ESV calculation. Drawing on existing research and accounting for the natural and socio-economic characteristics of the HEZ, four spatial adjustment factors—NPP, rainfall, soil erosion, and accessibility—were selected to correct the regional ecosystem service equivalence coefficients. The study area lies at the intersection of provincial boundaries and a climatic transition zone, and includes several major transportation hub cities. Therefore, this study assumes that NPP, rainfall, soil erosion, and accessibility exert equally important influences on the regional ecosystem, and assigns them equal weights in the analysis38.

$$V_{{if}}^{a}={V_f} \times {E_{if}}$$
(3)

where \(V_{{if}}^{a}\) refers to the equivalent factor of the f ecosystem service in the region i; \({V_f}\) refers to the average equivalent factor of the f ecosystem service at the national level; and \({E_{if}}\) refers to the spatial regulation factor of the f ecosystem service in the region i.

The NPP regulation factor corresponds to food production, raw material production, gas regulation, climate regulation, nutrient cycling, landscape aesthetics, and environmental purification. The equation for calculating the NPP spatio-temporal regulation factor is as follows:

$${P_i}=\frac{{{B_i}}}{{\overline {B} }}$$
(4)

where \({B_i}\) refers to the annual average NPP of the ecosystem in the region i; and \(\bar {B}\) refers to the annual average NPP at the national level of the same ecosystem.

The rainfall regulation factor corresponds to water supply and hydrological regulation. The equation for calculating the spatio-temporal rainfall regulation factor is as follows:

$${R_i}=\frac{{{W_i}}}{{\overline {W} }}$$
(5)

where \({W_i}\) refers to the average unit area rainfall in the region i; and \(\bar {W}\)refers to the annual average of rainfall per unit area at the national level.

The soil erosion regulation factor corresponds to soil conservation and biodiversity. Soil erosion is an important influencing factor affecting soil conservation in the region55. Based on the soil erosion classification of the Institute of Geographic Sciences and Natural Resources Research of the Chinese Academy of Sciences, the intensity of soil erosion is assigned values from 1 to 6, ranging from slight to severe. The average erosion intensity in the HEZ was calculated using ArcGIS. The equation for the soil erosion regulation factor is as follows:

$${S_i}=1/\left( {\frac{{{e_i}}}{{\overline {{{e_i}}} }}} \right)$$
(6)

where \({e_i}\) refers to the average soil erosion intensity in the region i; and \(\overline {{{e_i}}}\)refers to the national average intensity of soil erosion.

The accessibility regulation factor corresponds to culture and recreation. Road density was used as an indicator of regional transportation accessibility. According to the recreation opportunity spectrum theory43, the supply of recreational opportunities is generally positively correlated with transportation accessibility. The equation for calculating the accessibility spatio-temporal regulation factor is as follows:

$${E_i}=\frac{{R{D_i}}}{{\overline {{R{D_i}}} }}$$
(7)

Where \(R{D_i}\) refers to the average road density in the region i; and \(\overline {{R{D_i}}}\) refers to the national average raod density.

Value of ecosystem services for one standard equivalent factor

One standard unit of ESV equivalent factor refers to the economic value of grain production on one hectare of farmland, which is the average annual economic value of grain production in farmland in the entire country. It is used to measure the average contribution capacity of different types of ecosystem services58. The ESV of one standard equivalent factor is calculated as 1/7 of the crop production value of grain crops in farmland ecosystems. Based on the actual cultivation of crops in the HEZ, three grain crops, namely wheat, corn, and soybeans, were selected for the calculation. The equation is as follows:

$${E_a}=\frac{1}{7} \times \sum\limits_{{i=1}}^{n} {\frac{{{m_i}{p_i}{q_i}}}{M}} \left( {i=1,2,...,n} \right)$$
(8)

where \({E_a}\) represents the economic value of the food production service provided per unit area of farmland (CNY/hm2); i is the type of crop; \({p_i}\)is the average price of the crop (CNY/t); \({q_i}\) refers to the crop yield per unit area of i (t/hm2); \({m_i}\) refers to the area of i (hm2); and M refers to the total area of grain crops (hm2). In this study area, \({E_a}\) is 1434.90 CNY/hm2.

Calculation of ESV

The equation for total ESV is as follows45:

$$ESV={E_a} \times {A_k} \times V_{{kf}}^{a}$$
(9)

where ESV is the total value of ecosystem services; \({A_k}\) is the area (hm2) of k land use types; and \(V_{{kf}}^{a}\) is the adjusted equivalent factor of land use types.

Coefficient of improved cross sensitivity

Many existing studies have used sensitivity coefficients (SC) to calculate the sensitivity and stability of ESV to changes in the value coefficients of different land use types by varying the ESV coefficients for each land use type by ± 50% (Aschonitis et al.2,; Gao et al.18,. However, some research has noticed that the sensitivity coefficients used in current studies do not consider the effect of prices on product demand, resulting in their values always falling between 0 and 116,24. Therefore, this method can only be used to estimate the contribution of each land type to total ecosystem services, but cannot accurately estimate the stability and sensitivity of ESV coefficients (Aschonitis et al.3,. Referring to the research done by22,53, this study uses CICS to examine the impact of conversion between different land classes in the HEZ on changes in the value of ecosystem services63.

The calculation of CICS is as follows:

$$CICS=\frac{{\left( {{V_{cj}} - {V_{ci}}} \right) \times \Delta {S_{ji}}}}{{\Delta ESV}}$$
(10)

where \(CICS\) refers to the Coefficient of Improved Cross Sensitivity for the transition between the i−th and j−th land use types; \({V_{ci}}\) and \({V_{cj}}\) represent the ecosystem value coefficient of the i−th and j−th land use types; \(\Delta {S_{ji}}\) represents the net conversion area of land use types i and j between the k−th year and the (k-1)−th year; and \(\Delta ESV\) represents the change in ESV between the k−th year and the (k-1)−th year.

Spatial heterogeneity of ESV

Spatial autocorrelation is an effective tool for analyzing spatial relationships and degrees of association between variables. Given the complex land use types and high spatial heterogeneity of ecosystems in the HEZ, 3 km grid scale was chosen as the optimal observation scale for estimating ESV. This scale better captures both small-scale ecosystem differences and large-scale ecosystem service characteristics compared to the 1 km and 5 km grid scales21,41. The global spatial autocorrelation provides a general description of the spatial characteristics of correlated variables across regions, and the global Moran’s I reflects the spatial autocorrelation of ESV at the regional level, with values from − 1 to 1. The calculation is as follows:

$$I=\frac{{\sum\nolimits_{{i=1}}^{n} {\sum\nolimits_{{j=1}}^{n} {{W_{ij}}\left( {{X_i} - \overline {X} } \right)\left( {{X_j} - \overline {X} } \right)} } }}{{\left( {\sum\nolimits_{{i=1}}^{n} {\sum\nolimits_{{j=1}}^{n} {{W_{ij}}} } } \right)\sum\nolimits_{{i=1}}^{n} {\left( {{X_i} - \overline {X} } \right)\frac{2}{n}} }}(i \ne j)$$
(11)

where n is the total number of spatial locations; \({X_i}\) and \({X_j}\) are the values of x at the pairwise adjacent spatial locations, respectively; \({W_{ij}}\)denotes the spatial weight matrix of the neighborhood or distance; and \(\bar {X}\) is the average value of the variables.

The local spatial autocorrelation shows the degree of correlation between a local small area variable and the same variable in surrounding cells. The index of local spatial autocorrelation is calculated as follows:

$$LISA=\frac{{\left( {{X_i} - \overline {X} } \right)}}{{\sum\nolimits_{i} {\left( {{X_i} - \overline {X} } \right)2/n} }}\sum\limits_{j} {{W_{ij}}\left( {{X_j} - \overline {X} } \right)} (i \ne j)$$
(12)

where \({X_i}\), \({X_j}\), \({W_{ij}}\), and \(\bar {X}\) are of equal significance. A value of LISA less than 0 indicates a spatial aggregation of similar values, and a value greater than 0 indicates a spatial aggregation of high or low values around this regional unit.

Results

Dynamic changes in LULC

Based on the land use in the HEZ (Fig. 3), it can be noted that farmland has consistently been the dominant land use type in the study area from 1995 to 2020, accounting for over 68% of the total area. The second largest category of land use is built-up land, which represents a share of over 15% of the total area. Other land categories, ranked in descending order of area share, include water area, grassland, woodland, wetland, and bare land, none of which accounted for more than 5% of the total area during the study period.

Fig. 3
figure 3

Distribution of land use/land cover types in HEZ, 1995–2020.

Full size image

(Based on the standard map GS(2023)2676 from the Ministry of Natural Resources, China, with unaltered base map boundaries. This map was created using ArcGIS 10.8.)

From the perspective of changes in the area of different land types in the HEZ (Table 3), it can be observed that the areas of farmland, woodland, grassland, wetland, and bare land have all decreased from 1995 to 2020, while the water area and built-up land increased year by year. Among them, the most dramatic change is registered for the built-up land area, which increased by 4,272.34 km2 from 1995 to 2020. The water area shows a growing trend, increasing by 1,267.66 km2. Farmland also changed significantly, decreasing by 2,571.33 km2 during the study period. Grassland is the next most affected, decreasing by 1,561.43 km2. Woodland and wetland areas changed relatively little, decreasing by 644.86 km2 and 623.55 km2, respectively. Bare land changed the least, decreasing by 42.54 km2 during the study period.

However, from the perspective of land use dynamics (Table 3), wetlands exhibited the most significant changes between 1995 and 2020, with a dynamic degree of 0.033%, followed by water areas and grasslands, with values of 0.024% and 0.022%, respectively. Farmland showed the least change, at only 0.003%. Notably, although construction land underwent substantial area changes, its dynamic degree was only 0.020%, indicating that the extent of land use change cannot be evaluated solely based on the absolute area change.

Table 3 Dynamic change of land use types in the HEZ, 1995–2020.
Full size table

Farmland is the primary land use type in the HEZ. Built-up land is sporadically distributed within the study area, roughly along the Heze to Suzhou axis. In the northeastern part of the axis, built-up land is more densely concentrated, and these dense areas are often located in the surrounding regions with well-developed water systems, such as around Weishan Lake, the northern part of Hongze Lake, along the Grand Canal, in the southern and northern parts of Luoma Lake, and in the coastal areas around Lianyungang. On the southwestern side of the axis, the built-up land is more scattered. As urbanization progresses, the built-up land distribution characteristics become more apparent, with a focus on county town centers and major transportation routes. Woodland and grassland are primarily located in the northern hilly regions of the study area, while bare land is generally located in some relatively difficult-to-develop areas within hilly regions.

Patterns of transformation of land use types

Different types of land use in the study area showed varying transformation directions during different phases of the study period (Fig. 4). From 1995 to 2020, built-up land showed the most dramatic changes. 6,140.92 km2 of farmland were converted into built-up land, accounting for 91.27% of the total conversion to built-up land. Grassland and woodland follow, with 263.25 km2 and 170.57 km2 converted to built-up land, respectively. Wetlands, water area, and bare land accounted for a smaller share.

The largest area reduction was recorded in farmland, which is primarily converted into built-up land and water area, accounting for 85.08% and 9.82% of the converted farmland area, respectively. A smaller part of farmland has been converted into woodland and grassland, with bare land having the smallest share.

Fig. 4
figure 4

Chord diagrams of land use and land cover conversion during 1995–2020.

Full size image

During the study period, woodland was primarily converted to farmland, built-up land, and grassland, with only 4.36% of woodland being transformed into water area, wetlands, and bare land. Grassland have mostly been converted into farmland and built-up land, while the remaining parts have been transformed into woodland, wetlands, bare land, and other land categories. Water area and wetlands had similar directions of transformation, primarily converting into farmland, and the trend of transformation of wetlands into a water area is also noticeable. Bareland has undergone relatively slow changes, with the main transformation directions being reclamation into farmland or conversion into built-up land.

Change in ESV

Change in total ESV

The ESV in the HEZ showed a fluctuating downward trend (Table 4) in the period 1995–2020. From 1995 to 2005, ESV showed a moderate growth trend and reached a peak of 264.704 billion CNY during the study period. From 2005 to 2015, ESV experienced a significant decline, reaching the lowest point of 246.8 billion CNY in the study period. The reason for this decline is a substantial reduction in the area of farmland, woodland, grassland, and wetlands during the period from 2005 to 2015, which had a significant impact on the total ESV. From 2015 to 2020, the total ESV started to show a slight recovery.

From 1995 to 2020, the ESV of farmland showed a slight and continuous downward trend. It decreased from 111.685 billion CNY in 1995 to 107.452 billion CNY in 2020, which is a decrease of 4.232 billion CNY. Its contribution to the total ESV remained stable, ranging between 42% and 44%. The ESV of woodland showed a slight decline from 51.561 billion CNY in 1995 to 41.765 billion CNY in 2020, which is a reduction of 9.795 billion CNY. Its contribution to the total ESV remained stable, ranging between 16% and 20%. The ESV of grassland exhibited a slight decline from 1995 to 2005, followed by a significant decline from 2005 to 2020. It decreased by 10.855 billion CNY in the period from 2005 to 2010. Its contribution to the total ESV ranged between 8% and 12%. The ESV of wetlands consistently and significantly declined during the study period. It decreased from 10.940 billion CNY in 1995 to 5.556 billion CNY in 2020, which is a decrease of 49.22%. The ESV of water area showed a continuous upward trend, increasing from 57.275 billion CNY in 1995 to 78.018 billion CNY in 2020, which is an increase of 36.22%. As shown in Table 3; Fig. 4, the areas of land types with high ESV coefficients—such as farmland, woodland, and grassland—have declined. Although water areas possess the highest ESV coefficient, the increase in their area is insufficient to compensate for the losses caused by the reduction in other land types, leading to an overall decline in total ESV. Finally, the ESV of bareland showed a pattern of first decline and then increase from 1995 to 2020, with little overall change. Its contribution to the total ESV was relatively small (Fig. 5).

Table 4 ESV in the HEZ, 1995–2020.
Full size table
Fig. 5
figure 5

ESV in the HEZ, 1995–2020.

Full size image
Table 5 ESV flow profit and loss in the HEZ, 1995–2020.
Full size table

Based on the land use transition matrix for the HEZ, a profit and loss table for ESV in the HEZ from 1995 to 2020 was calculated (Table 5). From 1995 to 2020, ESV in the study area showed a decreasing trend of 3.60%. ESV profits primarily came from the conversion of farmland into water area, wetlands into water area, and built-up land into water area. Loss in ESV is mainly caused by the conversion of farmland into built-up land, grassland into farmland, and woodland into farmland. Changes in land types such as farmland, built-up land, grassland, woodland, and water area have had a significant impact on ESV. From 1995 to 2020, the conversion of farmland into water area had the largest impact on ESV in the study area as it increased ESV by 10.435 billion CNY. The second biggest impact was the conversion of farmland into built-up land, resulting in a decrease of 10.108 billion CNY in ESV. Conversion of built-up land to other land types has led to an increase in ESV. Among these conversions, the transformation from built-up land to farmland and water area had the most significant impact on ESV, increasing it by 3.521 billion CNY and 4.909 billion CNY, respectively. Conversion between grassland and built-up land or farmland had a noticeable effect on ESV. From 1995 to 2020, the conversion of grassland to farmland led to a reduction in ESV of 8.238 billion CNY, and the conversion of grassland to built-up land led to a reduction in ESV of 1.829 billion CNY. When woodland was converted to farmland, the ESV decreased by 6.778 billion CNY. During the process of converting woodland into built-up land, the ESV decreased by 2.591 billion CNY. Conversion of water area to other land use types had a negative impact on the regional ESV. When the water area was converted to farmland, the ESV decreased by 2.903 billion CNY. Conversion of water area into built-up land led to a decrease in ESV of 1.640 billion CNY. When the water area was converted to wetlands, the ESV decreased by 430 million CNY. Mutual conversions between bareland and other land use types had relatively little impact on the regional ESV.

Change in individual ESV

In the HEZ, among individual ecosystem services, four ecosystem services have a relatively high contribution to total ESV, and they are ranked from highest to lowest as follows: hydrological regulation, climate regulation, gas regulation, and food production. From 1995 to 2020, within the provisioning services, food production and raw materials production showed a downward trend, decreasing by 1.515 billion CNY and 987 million CNY, respectively. On the other hand, water supply showed a slight increase of 1.646 billion CNY. Within the regulation services, hydrological regulation showed a significant growth trend, increasing by 10.181 billion CNY. Gas regulation, climate regulation, and environmental purification, on the other hand, showed a downward trend, decreasing by 3.457 billion CNY, 7.066 billion CNY, and 1.160 billion CNY, respectively. Recreation and culture services showed a continuous decline from 1995 to 2020, decreasing from 9.843 billion CNY in 1995 to 8.471 billion CNY in 2020, which resulted in a decrease of 1.372 billion CNY (Table 6).

Hydrological regulation, climate regulation, gas regulation, and food production are the four ecosystem services with the highest percentage contribution. Among them, hydrological regulation has the highest contribution, consistently above 30% on average from 1995 to 2020, and shows a continuous growth trend from 30.17% in 1995 to 35.30% in 2020. Climate regulation follows, with an average contribution of over 15% during the study period but shows a declining trend from 16.90% in 1995 to 14.75% in 2020. Gas regulation maintains an average contribution of over 14%, with a slightly fluctuating decline from 14.38% in 1995 to 13.56% in 2020. Food production has the lowest contribution to the total ESV, with an average contribution of over 13%, and shows a small fluctuating decline from 13.64% in 1995 to 13.55% in 2020, with relatively minor changes.

Table 6 Change of individual ESV in HEZ, 1995–2020.
Full size table

Hydrological regulation, climate regulation, gas regulation, and food production together account for over 70% of the total ESV, while the other seven ecosystem services contribute less than 30%. Most of these seven services show a decreasing trend during the study period (Table 6). Among them, biodiversity protection and soil formation and retention show the most significant declining trends, with a decrease of over 20%, amounting to 3.211 billion CNY and 2.171 billion CNY, respectively, between 1995 and 2020. Next are raw materials production, environmental purification, and recreation and culture, each with a decline of over 10%, amounting to 0.987 billion CNY, 1.160 billion CNY, and 1.372 billion CNY, respectively. Maintain nutrient cycling shows the smallest decline, with only 0.407 billion CNY. Water supply is the only ecosystem service that shows a growth trend, increasing by 1.646 billion CNY from 1995 to 2020.

Cross sensitivity coefficient of ESV

Fig. 6
figure 6

Cross-sensitivity coefficient of ESV in HEZ, 1995–2020.

Full size image

CICS for 21 different conversion processes among seven land use types in the HEZ from 1995 to 2020 was calculated using Eq. (9) (Fig. 6). The absolute value of CICS indicates the sensitivity of regional ecosystem services to the mutual conversion of two types of land use. A higher absolute value indicates a higher sensitivity, while a lower absolute value indicates a lower sensitivity. The sensitivity of mutual conversion between different land use types in the study area is as follows.

The conversion between farmland and grassland, farmland and water area, farmland and built-up land has relatively high sensitivity, with absolute CICS values of 0.5807, 0.7913, and 0.6920, respectively. The conversion between woodland and built-up land shows a higher sensitivity, with an absolute CICS value of 0.2510. The conversion between water area and wetland as well as water area and built-up land is relatively sensitive, with absolute CICS values of 0.4086 and 0.3435, respectively. To further illustrate the calculation process, the land conversion with the highest absolute value of the CICS—namely, the transition between farmland and water areas—is used as an example. From 1995 to 2020, the net change in area between farmland and water areas was 51,178.21 hectares. According to Eq. (3), the ESV coefficient for farmland was 16,460.07 yuan/ha, while that for water bodies was 163,628.82 CNY/ha. As shown in Table 3, the net change in total ESV amounted to 9,518.63 million yuan, and the result calculated using Eq. (10) was 0.7913.

Regardless of how land use types change, conversion to built-up land always results in a reduction in the total ESV in the region. Regulation services have the largest contribution to the total ESV, which is consistent with the findings of other researchers35,49,66.

Analysis of Spatial heterogeneity of ESV

Distribution of ESV

This study referred to research by scholars such as26,41, as well as existing studies on ecosystem services in the HEZ32. Based on a 3 km × 3 km grid scale, we assessed the ESV in the HEZ and analyzed their spatial heterogeneity. Considering the different ranges of ESV values at different stages and grid scales, a natural break classification method was used in ArcGIS 10.8 software in order to classify the ESV grid values for 2005 and 2010. Then, the average of these classifications was taken as the classification standard for ESV in the HEZ from 1995 to 2020. The ESV values were classified into different categories as follows. ESV values from 0 to 9.2885 million CNY are categorized as extremely low ESV. ESV values from 9.2886 to 25.0336 million CNY are categorized as low ESV. ESV values from 25.0337 to 52.5614 million CNY are categorized as medium ESV. ESV values from 52.5615 to 101.7790 million CNY are categorized as high ESV. ESV values from 101.7791 to 148.7155 million CNY are categorized as extremely high ESV. The spatial distribution map of the ESV in the HEZ from 1995 to 2020 is presented in Fig. 7.

Fig. 7
figure 7

Spatial distribution of ESV in HEZ, 1995–2020.

Full size image

(Based on the standard map GS(2023)2676 from the Ministry of Natural Resources, China, with unaltered base map boundaries. This map was created using ArcGIS 10.8.)

The spatial distribution of ESV in the HEZ remained relatively stable from 1995 to 2020 (Fig. 7). The study area is mainly dominated by regions with low and extremely low ESV, the latter being sporadically distributed without obvious spatial layout characteristics and increasing in area year by year. Medium and high ESV regions are primarily located in low hills and hillock regions in the central and northeastern parts of the study area. In the central part of the study area, the medium ESV regions form a significant east-west axis (Shangqiu—Xuzhou—Lianyungang), where the ESV to the north of the axis is substantially higher than in the regions to the south of the axis. During the study period, regions with medium and high ESV continuously decreased. The changes are more pronounced in the western part of Linyi City and the border areas between Linyi City and Zaozhuang City. This is mainly due to the reduction in land use types such as grassland and woodland in that region. The spatial characteristics of regions with extremely high ESV remained stable during the study period. Their spatial distribution is largely consistent with the spatial layout of water bodies within the study area. Regions with extremely high ESV are primarily located around Lake Weishan, Lake Luoma, the northern part of Lake Hongze, and the coastal tidal flats. Moreover, the area covered by these extremely high ESV increased over time, primarily due to the expansion of water bodies within the study area.

In terms of changes in the areas of the five ESV categories, regions with extremely low ESV experienced the most dramatic changes. From 1995 to 2020, the area of extremely low ESV increased by 86.62%, which is equivalent to an increase of 4545 km². Most of this increase is the result of conversions from low ESV regions. Considering the LUCC distribution of the HEZ (Fig. 3), this is primarily due to the continuous expansion of urban development, which has encroached upon a significant amount of farmland, further contributing to the decline in the regional ESV. Regions with extremely high ESV increased by 513 km², with their share increasing by 24.05%. This increase is primarily due to the gradual expansion of water area and other land use types in the study area, leading to an overall increase in the regional ESV. Both the middle ESV and high ESV regions experienced a decrease in area, with decreases of 14.69% and 13.27%, respectively, compared to 1995. This decline is closely related to the reduction in land use types such as grassland, woodland, and wetland within the study area. Regions with low ESV experienced the smallest change, decreasing by only 2.46%. However, it still accounts for over 60% of the ESV.

Spatial autocorrelation of ESV

The spatial clustering characteristics of ESV in the HEZ were analyzed using GeoDa software. The Global Moran’s I and LISA values for ESV in the study area were calculated on a 3 km×3 km grid scale (Table 7; Fig. 8).

Table 7 Global moran’s I of ESV in HEZ, 1995–2020.
Full size table

From 1995 to 2020, the global Moran’s I of ESV in the HEZ was greater than 0 and the significance level was below 0.001, which indicates a significant positive spatial autocorrelation. The spatial distribution of ESV in the study area showed a strong positive correlation, with neighboring grid cells showing a high level of spatial clustering and similarity. This trend showed an initial increase followed by a decrease over time.

Fig. 8
figure 8

Moran’s I scatter plots of the value of ecosystem services in HEZ, 1995–2020.

Full size image

The local Moran’s I can reflect the clustering locations and spatial correlation of ESV in the grid space of the study area. By visualizing the calculated LISA values using GeoDa software, this study obtained a spatial clustering graph of ESV in the HEZ (Fig. 9).

Fig. 9
figure 9

LISA cluster graph of ESV in HEZ, 1995–2020.

Full size image

(Based on the standard map GS(2023)2676 from the Ministry of Natural Resources, China, with unaltered base map boundaries.)

From 1995 to 2020, the spatial clustering of ESV in the HEZ has remained relatively stable (Fig. 9). In 1995, the “High-High”(H-H) aggregation regions were mainly distributed in the low hilly areas of the central and northeastern parts of the study area, around Weishan Lake, Luoma Lake, and the northern part of Hongze Lake, accounting for 12.40% of the total area. The “Low-Low”(L-L) aggregation regions were sporadically distributed, mainly in the southwestern and southeastern regions, accounting for the largest proportion (26.94%) of the total area. The “High-Low”(H-L) and “Low-High”(L-H) aggregation regions had relatively small proportions, accounting for 0.20% and 1.15% of the study area, respectively. The spatial clustering characteristics of ESV within the study area remained relatively stable from 2000 to 2005, with little change compared to 1995.

By 2010, noticeable changes in the spatial clustering characteristics of ESV within the study area began to appear. The proportion of the H-H region decreased to 10.91%, but its spatial distribution characteristics remained relatively stable. The proportion of the L-L region decreased to 19.00%, and its spatial characteristics also did not change significantly. The H-L and L-H regions showed little change, with their proportions remaining below 1.5%.

The spatial clustering characteristics of ESV continued to decline in the study area by 2020. The proportion of the H-H region was 10.79%, which is a decrease of 1.44% compared to 1995. The L-L region accounted for 21.87%, which is a decrease of 5.07% compared to 1995. The proportions of the H-L and L-H regions remained relatively small, with a slight increase of 0.08% and a decrease of 0.03%, respectively, compared to 1995.

From the LUCC distribution graph of the HEZ (Fig. 3) and the spatial clustering graph of ESV in the HEZ (Fig. 9), it can be noticed that the distribution and evolution of the H-H region of ESV are closely related to the spatial characteristics of water area, woodland, grassland, wetlands, and other land use types. The H-H region is primarily located in areas covered by water, woodland and grassland. The L-L region is mainly influenced by the spatial distribution of built-up land. From 1995 to 2020, the urbanization in the HEZ accelerated, leading to the continuous expansion of built-up land at the expense of farmland, woodland, grassland, and other land use types. Overall, the spatial clustering of ESV in the HEZ is gradually weakening year by year.

Discussion and conclusion

Discussion

This study selected key ecological and socio-economic factors from a land use perspective to spatially correct the ecosystem service equivalent factors proposed in previous research. The study emphasizes spatial heterogeneity in ESV estimation, enhancing the accuracy of regional ecosystem service evaluations. To balance small-scale ecosystem heterogeneity with large-scale ecosystem integrity, the ESV of the HEZ was assessed at an optimal grid scale. Additionally, a profit and loss table of ESV was introduced to analyze the dynamic response of ecosystem services to land type changes. The improved cross-sensitivity coefficient (CICS) quantifies the impact of net transformations between land types on ecosystem services. This method overcomes the limitations of traditional sensitivity coefficients, which fail to accurately assess the stability and sensitivity of ESV coefficients, providing a more effective tool for analyzing the relationship between land use and ecosystem services. To expand the research perspective and concepts in this study, the research methods and key conclusions will be further discussed:

LUCC has a wide-ranging impact on ecosystem services. Human activities and environmental pressures lead to changes in land use types, which result in corresponding changes in ecosystem services. This is a rather complex non-linear process54. Previous studies have primarily focused on unidirectional change in land use types26,34,36,41. In this study, we take a dynamic perspective and investigate ESV changes resulting from dynamic processes of land class conversions. Between 1995 and 2020, the ESV profit of the HEZ mainly came from the conversion of farmland to water area, wetland to water area, and built-up land to water area. The regional ESV loss is primarily from the result of processes such as conversion of farmland to built-up land, grassland to farmland, and woodland to farmland. The conversion of different land use types to built-up land consistently leads to a decrease in regional ESV, a finding consistent with the results of other researchers35; Munthali et al.39; Su et al.49,. In general, regardless of the estimation method used, the expansion of woodland and natural vegetation tends to have a positive impact on ecosystem services, while the expansion of farmland and urban development can have positive or negative effects, which depends on the value coefficients used in the study period48. found that the regional total ESV decreased after the increase in farmland area in the Afar region of Ethiopia. In contrast4, found an increasing trend in regional ESV when desert areas in Pakistan were converted to farmland51. observed that during the same period, the increase in woodland, farmland and urban areas, along with the decrease in natural vegetation, led to a decrease in total ESV. Research by1,5 also indicated a crucial role of ecosystems like woodlands and water in enhancing regional total ESV, which is largely consistent with the results of this study. Based on the results of this study, the total ESV in the HEZ exhibited a declining trend, primarily due to the reduction in woodland, grassland, farmland, and other land types with high ESV coefficients. However, the substantial increase in water areas—characterized by the highest ESV coefficient—contributed to a continuous rise in their individual ESV, partially mitigating the overall decline. In light of these findings, future land use planning in the HEZ should refer to the ESV profit and loss table to (1) appropriately limit the conversion of high-ESV land types to those with lower ESV, and (2) adjust the regional ecosystem service supply structure by considering both the ESV coefficients and the net area changes of different land use types.

This study introduced CICS to assess the degree of influence of net conversions between different land use types on ESV changes. Compared with previous studies that used the sensitivity coefficient to measure the impact of value coefficient changes on ESV32,64, the use of CICS in this study allows for a better representation of the sensitivity of ecosystem services to the conversion of two different land use types. Essentially, it serves as an indicator of how ESV responds to land use changes per unit area63. In this study, Water area, woodland, and farmland contribute significantly to ESV, while bare land and built-up land contribute relatively less to ESV. Therefore, when conversion occurs between these two ecosystem types, it has a crucial impact on the overall ecosystem services in the region. This impact is particularly evident in the changes between woodland and built-up land and between farmland and built-up land, which is consistent with the findings of8,34. Similar findings were observed in studies of ecosystem services in the Pearl River Delta region34, the Shule River Basin41, and major watersheds in Morocco27, where urban development and indiscriminate use of high-quality cropland led to a significant decrease in the total ESV. The CICS can partially reflect the impact of land use conversions on changes in ESV. In future land use planning, CICS can be applied to guide the management of specific land type transitions, while the ESV profit and loss table can serve as an effective tool for monitoring the sustainable use of regional land resources and evaluating the quality of ecosystem services9,22.

Building on the research of Xie et al.55,56, this study selected four ecological-economic indicators: NPP regulation factor, rainfall regulation factor, soil erosion regulation factor and accessibility regulation factor, in order to spatially adjust the equivalent factors of ESV in the HEZ. The research results accurately reflect the real situation in the HEZ. In previous research, differences in regional biomass, rainfall, soil erosion, vegetation density, and other factors were often neglected, leading to inaccuracies in ESV assessments33,44,61. Selecting appropriate regulatory factors for spatial adjustment can improve the accuracy of research results. In their study on ESV in the Pearl River Delta region34, proposed that the ESV coefficients suggested by Costanza et al.(1997) are primarily applicable at the global rather than the regional level. In another study focusing on ESV in the Guangdong-Hong Kong-Macao Greater Bay Area33, used vegetation normalization indices for spatial adjustment and found significant differences in ESV obtained in 2000 compared to those obtained by60, who did not perform spatial adjustments. In a previous study on the assessment of ecosystem services in the HEZ32, researchers directly used the nationwide ecosystem service equivalent factor table proposed by Xie et al.58 without spatial adjustments. They obtained an equivalent factor of 2217.12 CNY/hm2, which is slightly higher than the equivalent factor results in this study.

As shown in Sect. the total ESV in the study area began to recover gradually after a marked decline, starting in 2015. Based on the analysis, this recovery appears to be relatively stable and is primarily attributed to changes in land use patterns with limited influence from other external factors. According to the above analysis of land use change, the expansion of built-up land slowed markedly after 2015, indicating a reduced rate of encroachment on farmland, woodland, grassland, and other high-value land types. Meanwhile, the increase in grassland and water areas—both associated with high ESV coefficients—contributed to the gradual recovery of overall ESV in the study area. From the perspective of regional policy, the national government officially approved the Development Plan for the Huaihe River Ecological Economic Belt at the strategic level in 2015 (www.gov.cn/zhengce/content/2018-10/18). This policy marked a shift toward green developmentby placing constraints on extensive and unregulated economic expansion and construction activities. It also enhanced public awareness of ecological protectionand promoted a transition toward more sustainable land use practices thereby contributing to the gradual restoration of regional ESV. Although multi-period factor data were employed for spatial correction in this study, the temporal variation in key correction factors—such as NPP, precipitation, and soil erosion—was relatively minor over the study period, with the exception of road network data. Moreover, the use of average values further smoothed temporal fluctuations. As a result, the observed temporal changes in regional ESV were only marginally influenced by the correction factors.

In addition to the above discussions, this study still faces several sources of uncertainty. For example, the 30-meter resolution land use data employed were derived from a multi-period database constructed primarily from Landsat remote sensing imagery. After preprocessing steps such as clipping and image fusion, these data were directly applied to land use analysis within the study area. However, limitations remain, including relatively low spatial resolution, insufficient classification accuracy, and a coarse land use classification system. Regarding the modified equivalent factor table used in this study, its applicability may be limited due to regional differences in ecological conditions and socio-economic development at larger spatial scales. Nevertheless, the approach of selecting appropriate correction factors for spatial or temporal adjustment provides a valuable reference for studies in other, non-similar regions. In this study, NPP, rainfall, soil erosion, and accessibility were selected as parameters for the correction of ESV coefficients. However, due to limitations in data availability and manuscript length, the sensitivity of these parameters and their relative weight distribution were not fully considered, which may have introduced some degree of uncertainty into the ESV estimation results.

It is important to consider regional variations when estimating the value of regional ESV. The level of the same type of ESV can vary due to factors such as soil and rainfall6. Therefore, selecting appropriate regulation factors for spatial adjustment is crucial to overcome calculation biases caused by regional differences and is an important prerequisite for objectively and accurately assessing regional ecosystem services.

Conclusion

Based on land use data covering six periods from 1995 to 2020, and using the revised standard equivalence factors, this study selected the HEZ as the study area and estimated the spatio-temporal characteristics of ESV in the study area. The key findings and main conclusions are as follows:

(1) From 1995 to 2020, there have been changes in all land use types in the HEZ. Water area and built-up land showed an upward trend, increasing by 1267.66 km² and 4272.34 km², respectively, while the others decreased.

(2) From 1995 to 2020, the total ESV in the HEZ showed a fluctuating downward trend. The regional ESV profit is mainly derived from the conversions of farmland to water area, wetlands to water area, and built-up land to water area. Conversely, the conversions of farmland to built-up land, grassland to farmland, and woodland to farmland are the main reasons for ESV losses in the HEZ.

(3) From 1995 to 2020, regional ESV sensitivity to land use transformations in the HEZ followed this order: farmland to water area, farmland to grassland, farmland to built-up land, farmland to woodland, water area to wetland, water area to built-up land, and finally, woodland to built-up land.

(4) The spatial distribution of ESV in the HEZ remained relatively stable from 1995 to 2020. A significant portion of the total area is characterized by low and extremely low ESV. Throughout the study period, Moran’s I statistic remained consistently greater than 0, initially increasing and then decreasing, indicating significant positive spatial autocorrelation in the ESV distribution.

This study adopts a dynamic approach to examine the impact of land use type conversions on regional ESV. It introduces an improved cross-sensitivity coefficient to quantify the impact of net conversions between land use types on ESV. Additionally, it reveals the spatial differentiation patterns of regional ESV at the grid scale. The findings of this study provide a reference for exploring the interaction between land use and ecosystem services, offering theoretical insights for the construction of ecological civilization and sustainable development in the HEZ.