Introduction

The accelerated processes of urbanization and industrialization has resulted in considerable damage to land resources1, generating an obvious alteration in the pattern and structure of the Land Use/Land Cover (LULC) on the surface2. The rapid transformation of LULC has altered the original, stable landscape structure and functions, and considerably elevated the landscape ecological risk (LER). As a consequence, the discrepancy between the exploitation of resources and the preservation of the natural environment, along with the growth of urban areas, has become more pronounced. This presents a significant challenge to the long-term sustainability of the human-land relationship and the quality of human well-being3,4,5. Accordingly, an objective assessment of LER could facilitate regional sustainable growth, inform rational LULC planning, ensure the security of the ecosystem6, promote global ecological and environmental collaboration, advance harmonious development of the global human ecosystem and safeguard the wellbeing of human civilization.

In 1990s, the concept of Landscape Ecological Risk Assessment (LERA) first appeared. It is a crucial subdiscipline of ecological risk assessment, utilizing landscape ecology to ascertain the probability of the regional-scale ecosystems encountering adversaries from exterior influences7. LERA has emphasized the coupling of landscape spatial pattern and landscape ecological process8, quantified the structure and function of ecosystems, reflected the adverse ecological effects of external disruptions, and offered indicators for maintaining the health and function of ecosystems9. Among the aforementioned approaches, LERA which is based on LULC, establishes a connection between landscape structure and ecological process10, thereby enabling the reflection of spatial heterogeneity of LER on a defined geographical scale and the more precise revelation of the influence of LULC on the relationship between ecological process and function. The majority of LULC-based LERA studies have focused on ecologically sensitive and vulnerable areas, as well as areas where human activities are concentrated11, such as river basins12,13, ecological barrier areas1, nature reserves14, mountainous regions15,16 and developed economic belt17. Additionally, some scholars have conducted studies on LER in mining areas2,18,19, but few studies have been conducted on resource coupling areas in inhabited mineral belts. The resource coupling areas in the inhabited mineral belt are characterized by a high level of human-mineral interaction. The exploitation of natural resources, social development and ecological restoration are intertwined in these areas, giving rise to complex interactions that shape the region, thereby endangering the ecological security and increasing the landscape ecological risk. Meanwhile, the mineral belt’s resource coupling zone has undergone a series of persistent and intricate transformations as a result of prolonged, high-intensity development and unsustainable alterations to land use and land cover (LULC). The aforementioned changes are reflected in the following ways: firstly, the mode of exploitation has shifted from unplanned and haphazard development to a more planned and intelligent approach. Secondly, the field of ecological restoration has progressed from initial reconstruction of form to subsequent improvement of function, and finally to reconstruction of function enhancement. Therefore, this has resulted in a spatial heterogeneity of LER under different time periods. Accordingly, a precise assessment of the LER and an investigation into its temporal and spatial evolution characteristics represents crucial prerequisites for the sustainable development management strategy of ecosystems in resource coupling areas in inhabited mineral belts20. This assessment could provide reference for the differentiated management of LER in analogous special regions, and furthermore offer constructive suggestions for the resolution of analogous complex ecological issues in special regions worldwide in the future. Furthermore, this has important practical significance for the development of an effective mitigation strategy for LER and the efficient allocation of scarce resources.

The regional LER is subject to a multitude of influencing factors. Static factors, including geomorphology and climate, represent the primary contributors to the spatial distribution of LER21,22,23. Human activities, including resource exploitation, population growth and industrialization24, have an apparent impact on the spatial heterogeneity of LER over different time periods8. The application of geographic detectors could not only facilitate the identification of spatial heterogeneity among variables3, but also elucidate the driving factors and processes that shape ecosystem stability21. Furthermore, these techniques could provide a quantitative assessment of the influence of driving factors on spatial distribution of LER25, and can well explain the interaction between driving factors23. In comparison with other techniques used to identify driving mechanisms, this method could offer a more efficient means of interpretation26. Consequently, the accurate grasp of the driving mechanism of LER changes in long time series through geographical detectors is not merely related to on an efficacious reflection of regional ecological environment conditions. Furthermore, it facilitates the subsequent optimization of LULC structure and the establishment of an early warning mechanism for ecological risk27.

The Beijing-Tianjin-Hebei region functions as the political, economic, and cultural epicenter of China, exhibiting rapid economic growth and an elevated degree of urbanization. Furthermore, it is the most significant represent of urban agglomeration in northern China28. The Beijing-Tianjin-Hebei major mineral belt (JMMB) represents a typical resource coupling area within an inhabited mineral belt. The region boasts a wealth of mineral resources, primarily comprising metallic and iron ore29, with open-pit mining representing the predominant mining technique employed. Concurrently, it facilitates the advancement of the smelting and processing industry within the region30. However, in the context of accelerated economic expansion, the discrepancy between ecological sustainability and human settlement development is becoming increasingly prominent. The continuous increase of LER is an important issue to be solved urgently in the current Beijing-Tianjin-Hebei coordinated development strategy. Consequently, the effective implementation of LERA could provide a crucial reference point and decision-making support for the identification and early warning of LER in the area, as well as the optimization of resource allocation. This study took JMMB as the research object, constructed a LERA model based on LULC and landscape index from 1985 to 2022, analyzed the spatial evolution of LER, and discussed the contribution rate of LER driving factors in different periods to reveal its driving mechanism. The findings could optimize the coupling relationship between LULC and LER in JMMB, maintain the ecological security and stability of JMMB, and have great significance in promoting the coordinated development of ecological environment and social economy in JMMB, and provide practical basis for improving the ecological protection management and planning decision-making of cities in mineral belt, which also offer a scientific foundation and technical means for the implementation of differentiated ecological restoration, management and land reuse in mineral belts, as well as offering a reference for reasonable landscape allocation and regional human-land relationship optimization in similar mineral belts around the world16. Furthermore, the findings may offer a theoretical basis for assessing the impact of ecological restoration initiatives in global mineral belts and for the stewardship of the global ecological environment in the subsequent phase31.

Study area

The Beijing-Tianjin-Hebei area is a typical area where mineral resources gather in China, with obvious regional characteristics, which mineral resources are mainly distributed in the vicinity of Taihang Mountain and Yanshan Mountain (Fig. 1) .The Beijing-Tianjin-Hebei area is located between 113°28’ -119 °50’ east longitude and 36°05’ -42 °40’ north latitude, surrounded by Taihang Mountain to the west, Bohai Bay to the east, North China Plain to the south, and Yanshan Mountain Range to the north, which including Beijing, Tianjin and Hebei Province, with an area of about 218000 km[2 [31. There are four distinct seasons and same period of hot and water, with the temperate continental monsoon climate in Beijing-Tianjin-Hebei region, that is suitable for plant growth31. JMMB is rich in mineral resources, with the majority of metal minerals and energy minerals. There are more than 1300 types of minerals listed above, including energy minerals, metallic and non-metallic minerals, as well as many coal and iron mines32 Mineral resources ensure the basic needs of infrastructure construction and people’s livelihood in Beijing-Tianjin-Hebei Provence area, and promote the coordinated development of regional economy and society. In the process of economic development, the destruction of LULC structure by mining and urbanization leads to the decline of ecological environment quality and hinders the process of sustainable development. The contamination and deterioration of the natural environment are concerns, and the implementation of measures to prevent and control LER is imperative. Therefore, an analysis of the evolution of LER in JMMB is of great importance.

Fig. 1
figure 1

Map of the JMMB.

Full size image

Data sources and methods

Sources of dataset

The LULC data in 1985 and 2022 (spatial resolution is 30 m×30 m) was came from the CLCD land cover classification dataset (https://doi.org/10.5281/zenodo.5816591), the LULC data in 1990–2020 (spatial resolution is 1000 m×1000 m) was came from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences(Data Platform for Resources and Environmental Sciences (resdc.cn)). According to the “Land Use Status Classification” standard33, the LULC data was reclassified into six LULC types: cropland, woodland, grassland, waters, construction land and unused land. The webpage of the Loess Plateau Branch Center of the National Earth System Science Data Center (http://loess.geodata.cn) provided the temperature and precipitation data; The slope data were extracted from DEM data obtained from geospatial data cloud platform (http://www.gscloud.cn/) ; The NDVI data were obtained from the University of Maryland platforms, and the MODIS sensor synthetic products next month MOD13A2 dataset was released by the United States Geological Survey (https://earthexplorer.usgs.gov/); The human disturbance degree data was calculated based on landscape index and human disturbance degree calculation model; The population density data was came from Worldpop (https://www.worldpop.org/) ; The data on night light was obtained from the DMSP/OLS and NPP/VIIRS databases at the Resources and Environmental Sciences and Data Center, Chinese Academy of Sciences (https://www.resdc.cn/). The secondary industry’s worth was obtained from the statistical yearbook of Beijing-Tianjin-Hebei Province. The driving force analysis section of this study covered the period of 1990–2022 due to the difficulty in obtaining economic data in 1985. In order to ensure ease of subsequent research, all raster images were resampled and reprojected into LULC data with the same pixel size, data resolution, and projection mode (1000 m, UTM-WGS_1984_Zone_50N).

Method

Division of LERA units

In order to facilitate subsequent research and analysis, better accurately depict the temporal and geographical change features of the landscape ecological risk index (ERI) during different periods in JMMB (1985, 1990, 1995, 2000, 2005, 2010, 2015, 2020, 2022), the reference was made to the national grid GIS standard “Geographic Grid” (GB12409-2009)34 and elevation relief, area size, natural environment characteristics and other factors in JMMB35, the study area and landscape type patch size were taken into consideration, following the principle that the average area of landscape patches is 2–5 times36, considering the amount of sampling work, the study using fishing net analysis and statistical tools divide into the grid units through repeated experiments, which resolution was 30 km×30 km, resulting in 239 ecological risk zones (Fig. 2). The ERI of each grid was computed based on the central point’s ERI of the ecological risk zone. Interpolation calculations were used to determine the distribution of the LER pattern throughout the entire research region.

Fig. 2
figure 2

Ecological risk evaluation cells on a simple map of the JMMB.

Full size image

Construction of the LERA Mode

Landscape index condenses landscape pattern information, it can employ to examine diverse ecological processes at varying scales, thereby elucidating the structural attributes and spatial and temporal patterns of ecological landscapes37. Referring to previous research results38, in light of the landscape characteristics of JMMB and the interconnections between its ecosystem landscape pattern and LER, the landscape disturbance index, landscape vulnerability index, landscape separation index, and landscape sub-dimension index were selected to establish the ERI. The formula is as follows:

$$\:\text{E}\text{R}\text{I}=\sum\:_{\text{i}=1}^{\text{N}}{\text{A}}_{\text{k}\text{i}}/{\text{A}}_{\text{k}}\text{*}{\text{R}}_{\text{i}}$$
(1)

Where, Ak represents the total area of the kth evaluation unit, Aki represents the area of the ith land use type in the kth evaluation unit, Ri represents the landscape loss index corresponds to the land use type. Higher values indicate higher risk, while lower values indicate reduced risk. The ERI value represents the degree of the LER in the studied area. The formula for calculating each ERI38 is shown in Table 1.

Table 1 Landscape index introduction.
Full size table

Exploratory spatial data analysis

  1. (1)

    Spatial semi-variance function. Geostatisticians utilize the spatial semi-variance function to study the spatial relationship and distribution pattern of geographical variables. It helps achieve optimal model fitting and judge whether geographical variables have spatial correlation in the cell grid, thus revealing their spatial heterogeneity28. The semi-variance function’s theoretical model is calculated as:

    $$\:{\upgamma\:}\left(\text{h}\right)=\frac{1}{2\text{n}\left(\text{h}\right)}\sum\:_{\text{i}=1}^{\text{n}\left(\text{h}\right)}{\left[\text{Z}\right({\text{x}}_{\text{i}}+\text{h}-\text{Z}\left({\text{x}}_{\text{i}}\right)\left)\right]}^{2}$$
    (2)

    Where: γ(h) represents the variance function value, h is the sample spacing, n(h) is the total number of sample pairs when the sampling spacing is h, Z is a random variable for a particular landscape attribute, xi is the spatial location, Z (xi + h) and Z (xi) is the calculated value of ERI on the spatial location of xi + h and xi.

  2. (2)

    Spatial autocorrelation. Spatial autocorrelation relevant to the aggregation of unit attribute values in space and whether the spatial distribution of nearby variables effect each other. This is typically measured using the global Moran Index (Moran’s I) and local Moran index (LISA). Moran’s I is a statistical indicator that analyzes the correlation between attribute values and their neighboring items. When Moran’s I > 0, it shows a positive correlation, and the ERI exhibits a cluster distribution. When Moran’s I = 0, it means that there is no correlation and the ERI is randomly distributed. When Moran’s I < 0, it indicates that there is a negative correlation and the ERI shows a discrete distribution. The LISA value measures the correlation between attribute values in a small local unit and those in the adjacent area, reflecting the local characteristics of spatial correlation of the LER. When LISA > 0, it manifests a high similarity in ERI between sampling unit and adjacent local unit demonstrating a ‘high-high’ or ‘low-low’ aggregation of local space. When LISA < 0, it indicates a low similarity of the ERI between sampling units and adjacent local units, resulting in a ‘high-low’ or ‘low-high’ aggregation in local space. When LISA = 0, it indicates that the sample area is not significantly related to adjacent areas12,15.

Geographical detectors

The Geographical Detector, which a spatial analysis model designed to discover regional heterogeneity and indicate probable driving forces was presented by Wang Jinfeng et al.39. It consists of four main parts: ecology, factor, risk and interactive detector, each with different functions and applications. This study employed a factor detector to examine the spatial heterogeneity of the LER in the dependent variable, as well as the degree to which the independent variable X was able to account for this difference. The paper discusses the degree to which driver X interpretations the spatial change of the LER, using q value to convey the driver factors’ explanatory power for the independent variable. According to the interaction detection to examine whether the 2 factors interact and to evaluate whether the driving factors enhance or lessen the explanatory power of the LER spatial differentiation40. The formula is as follows:

$$\:\text{q}=1-\frac{\sum\:_{\text{h}=1}^{\text{L}}{\text{N}}_{\text{h}}{{\upsigma\:}}_{\text{h}}^{2}}{\text{N}{{\upsigma\:}}^{2}}$$
(3)

Where: The variable q represents the explanatory power of a given factor on the LER, with a value ranging from 0 to 1. The q value is higher, the explanatory power is more greater. The h is the partition sequence number of the independent variable, L represents the stratification of the independent or dependent variable, Nh and N indicate the total number of samples across the entire research area and each partition, respectively. \(\:{{\upsigma\:}}_{\text{h}}^{2}\)and σ2 represents the sample variance throughout the entire research area and each partition, respectively.

Results

Analysis of the LULC changes in the JMMB

Temporal variation of the LULC

The proportions of land use and cover (LULC) in the JMMB from 1985 to 2022 were shown in Figs. 3 and 4. The JMMB’s primary land types include grassland, woodland and cropland. The cropland areas covered more than 38.50% of the entire study area, and its area has decreased by 39.91km2 over the past 40 years. The proportion of woodland area has exhibited a gradual increase, reaching a current total of over 22.50%. This has represent a 6.19% growth from the figures recorded in 1985. In contrast, the proportion of grassland area has been decreasing and currently accounts for over 19.50%. From 1985 to 2022, the proportion of grassland area decreased from 24.25 to 19.67%, with a total decrease of 20.98km2. The remaining three types of LULC were relatively small in area and were primarily located within or adjacent to cropland, woodland, and grassland. The percentage of LULC for construction has increased from 4.93% in 1985 to 11.64% in 2022, showing an upward trend. It accounted for over 4.50% of the entire study area. From 1985 to 2022, the percentage of water areas remained comparatively stable, ranging from 0.70 to 0.80%. Meanwhile, the amount of unused land decreased significantly from 0.16km2 in 1985 to 0.03km2 in 2022, representing a total decrease of 0.13km2 over the study period.

Fig. 3
figure 3

Spatial distribution of the LULC from 1985 to 2022.

Full size image
Fig. 4
figure 4

Proportion of the LULC areas from 1985 to 2022.

Full size image

Spatial variations of the LULC

Obvious changes have occurred with LULC types from 1985 to 2022(Fig. 5). Overall, the LULC types of the JMMB have undergone obvious changes from 1985 to 2022. The main type of land transfer was cropland, which covered a total area of 19145.38 km2. Approximately 10136.12 km2 of cropland area was converted to construction land, which showed 52.94% of the total proportion of cropland converted. This transfer phenomenon primarily occurred in the southeast and central-eastern parts. Part of the cropland has been converted into woodland and grassland in JMMB, which can alleviate the vulnerability of the ecosystem. Approximately 32.49% (about 6220.71 km2) of cropland converted to grassland, and 12.85% (about 2464.84 km2) of cropland converted to woodland, which was mainly distributed in the northeast and southwest parts. The second largest area of land conversion was from grassland, covering a total of 14320.10km2. The grassland was transformed into woodland (69.79%, approximately 9994.93km2) and cropland (26.11%, approximately 3739.30km2), the northern, northeast and southwest parts was its mainly distributed areas. Other LULC types have transferred less area, covering a total of 3532.94km2, which accounts for 9.55% of the total area transferred out. These types were primarily distributed in the cropland, woodland, and grassland areas surrounding the region. The main woodland transfer source was grassland, and the transferred areas were 12485.98km2, which primarily distributed in the northwest, northeast and southwest parts. The primary source of grassland transfer was cropland, with a total transferred area of 7547.62km2, which primarily distributed in the north and southwest parts. The area transferred from other LULC types were relatively small, with a total transferred area of 5826.16km2, which accounts for 15.74% of the total transferred area. In general, for almost 40 years, the change of cropland was the most obvious, with a change of 13836.81 km2. The mainly types of the LULC transfers were from cropland to construction land, as well as from cropland to woodland and grassland, which indicating that with the quickly development of economic and the implementation of ecological restoration policies in JMMB, LULC types were transformed into each other.

Fig. 5
figure 5

Change map of the LULC in JMMB from 1985 to 2022. (Note: 12 is cropland-woodland, 13 is cropland-grassland, 14 is cropland-waters, 15 is cropland-unused land, 16 is cropland-construction land, 21 is woodland-cropland, 23 is woodland-grassland, 24 is woodland-waters, 25 is woodland-unused land, 26 is woodland-construction land, 31 is grassland-cropland, 32 is grassland-woodland, 34 is grassland-waters, 35 is grassland-unused land, 36 is grassland-construction land, 41 is waters-cropland, 42 is waters-woodland, 43 is waters-grassland, 45 is waters-unused land, 46 is waters-construction land, 51 is unused-cropland, 52 is unused-woodland, 53 is unused-grassland, 54 is unused-waters, 56 is unused-construction land, 61 is construction land-cropland, 62 is construction land-woodland, 63 is construction land-grassland, 64 is construction land-waters, 65 is construction land-unused land.).

Full size image

Spatial and temporal evolution of LER

Temporal characteristics of the ERI

The average LER of the JMMB has undergone a clear change during the past 40 years, which the pattern of continuous increased followed by a declined (Table 2). The ERI was between 0.1 ~ 0.2, the average value of ERI was above 0.15, and the lowest value was mostly above 0.10. From 1985 to 2015, the average ERI exhibited a consistent rising trend. However, from 2015 to 2022, the implementation of the national mine restoration policy led to a decline in the average ERI of the JMMB.

Table 2 The result of ERI.
Full size table

Spatiotemporal variations of LER

The semi-variance function was utilized in geostatistics to investigate the spatial differentiation of the LER and assess the characteristics of this process in the JMMB. The results of fitting the LER from 1985 to 2022 were shown in Table 3. The Gaussian model was found to have the most ideal effect compared to other models, with an R2 higher than 0.87, indicating the best fitting result. In 1985, the ratio of nugget value to abutment value was over 75%, indicating weak spatial correlation of variables. In 1990, 1995, 2000, 2005, 2010, 2015, 2020 and 2022, the ratio of nugget value to abutment value ranged from 25 to 75%, indicating moderate spatial correlation28. Furthermore, from 1990 to 2022, the value of nuggets and abutments decreased gradually in JMMB. This suggests that the spatial correlation of the LER was increasing in JMMB.

Table 3 Semi-variogram of the LER from 1985 to 2022.
Full size table

Based on the fitting results presented above, a kriging interpolation has been performed on the LER value of the JMMB in order to obtain its spatial distribution. The distribution was then divided into five different levels of LER regions using the natural break point method (Fig. 6): low risk area (0.116080–0.132777), lower risk area (0.132777–0.143831), medium risk area (0.143831–0.154414), higher risk area (0.154414–0.164761) and high risk area (0.164761–0.176285), the proportion of the LER levels areas in different phases was statistically analyzed (Fig. 7). From 1985 to 2022, the risk level zones have undergone continuous development in JMMB. However, the spatial distribution of the LER’s different levels was not immediately apparent. Generally, in the central-northern and central-western parts, LER values were high; in the northeast, northwest and southwest parts, LER values were higher. The medium risk areas were distributed in a scattered manner, with the majority situated in proximity to the higher-risk areas. The low-risk and lower-risk areas were predominantly situated in the central-eastern and southeastern parts. Changes in landscape ecological risk in the study area, using 2015 as a dividing line. Before 2015, the LER in the JMMB comparatively high, with over 50.00% of the area classified as having higher and high LER. The low-risk and higher-risk areas decreased by 8254 km2 and 7115 km2, while the proportion of medium-risk and high-risk areas increased by 2.58% and 10.54%, respectively. After 2015, there is a significant change in the percentage of the area of high-risk zones in 2022, with an amount of change of 5.52%, the amount of change in the area share of the higher-risk area is 4,153km2, and the amount of change in the area share of the low, lower, and medium-risk areas is 0.5%, 1.33%, and 1.07%, respectively. In general, under the guidance of the scientific outlook on development and the new development concept, regional development is more focused on green and efficient, and the stability within the ecosystem of the JMMB is gradually increasing.

Fig. 6
figure 6

Spatial distribution of the LER levels from 1985 to 2022.

Full size image
Fig. 7
figure 7

Areas of the LER grades in the JMMB from 1985 to 2022.

Full size image

To further explore the temporal and spatial transformation properties of the various LER levels in more detail, the transfer matrix of different levels of the LER in the JMMB was obtained by means of analysis tools. The changes of the LER areas in the JMMB at different levels from 1985 to 2022 were shown in Fig. 8. From 1985 to 1995, a total of 14,413 km2 were transferred from predominantly low-risk areas to medium- and high-risk areas within the JMMB region. From 1995 to 2005, the transfer changes were mainly manifested as the transfer from medium risk area to other risk areas, and the medium-risk areas were transformed into higher-risk and lower-risk areas, with the transformation area of 12,193 km2 and 510 km2, respectively. From 2005 to 2015, the most significant expansion in the size of high-risk areas was observed. Transfers to other risk areas encompassed up to 15,987 km2. From 2015 to 2022, the risk grade transfer within the study area resulted in the high-risk region being converted into the intermediate and low-risk areas. Specifically, the high-risk areas were moved to a higher-risk area, while the medium-risk area remained unchanged. The total transfer area covered 10,042 km2. The higher risk area was transferred to the medium risk area, covering an area of 2736 km2. From 1985 to 2015, the LER in the JMMB shifted from other risk levels to medium and high risk levels. From 2015 to 2022, the high-risk areas of the ecosystem began to shift to other LULC types in JMMB, indicating an increase in the internal stability of the ecosystem after 2015.

Fig. 8
figure 8

Sankey diagram depicting the LER transformation, 1985–2022 (Unit: km2).

Full size image

Analysis of spatial autocorrelation

In JMMB, the ERI global spatial autocorrelation analysis was carried out in Table 4, which showed the Moran’s I values for the LER in each of the JMMB’s phases. The Moran’s I value of the LER in different phases were 0.7171, 0.6887, 0.6343, 0.6088, 0.6008, 0.5604, 0.5153, 0.4870 and 0.3253, respectively. All P-values in the study area were 0.001, and all of the Z-scores above 2.58, indicating that Moran’s I passed the significance test. The LER showed the significant positive spatial correlation, with all of the Moran’s I values being more than 0, with a clustered spatial distribution. During the study period, there may have been a progressive weakening of the spatial autocorrelation of the regional LER, as indicated by the different stage Moran’s I gradually declining, as well as a weakening of the aggregation and convergence of its spatial distribution.

Local autocorrelation analysis was used to obtain a local LISA cluster diagram (Fig. 9), as Moran’s I was unable to demonstrate spatial similarity between the attribute and adjacent units in JMMB. Based on the LISA map, the study region contained mostly ‘high-high’ and ‘low-low’ spatial clusters of the LER, with a small number of sporadically distributed ‘high-low’ and ‘low-high’ clusters. Overall, there was a steady increase in the number of “low-high” and “high-low” aggregations, while there was a drop in the number of “high-high” and “low-low” aggregations. The ‘high-high’ aggregations were primarily located in the northern, as well as the central and western parts of the JMMB. This suggests that the ecological environment in these areas was particularly vulnerable, with a higher level of LER. ‘Low-low’ aggregation zones were mainly located in the southeastern study region, and with scattered distribution in other areas. This suggests that the landscape was relatively stable and the ecological environment improved with strengthened ecological protection, resulting in a low LER.

Table 4 The LER value’s global Moran index in JMMB from 1985–2022.
Full size table
Fig. 9
figure 9

LISA maps of the LER from 1985–2022.

Full size image

Driving force analysis of LER

The contribution of driving factors

With reference to other research results18,19,40,41have been analyzed the influence of natural and human factors on the evolution of LER, and also considered the rapid economic development and environmental changes in JMMB area in recent years. This study according to the actual situation of JMMB, the indexes were selected from the two aspects of natural factors including landform, vegetation cover and social and economic factors of human disturbance for quantitative analysis. Therefore, 8 representative gauges of natural environment and human influence were selected respectively to explore the internal relationship between different factors and LER in the study area. X1, X2, X3, X4, X5, X6, X7 and X8 represented night light, degree of human interference, population density, secondary industry value, slope, normalized vegetation index (NDVI), average annual temperature and annual precipitation, respectively.

According to Table 5, Among the social and economic factors, population density contributes the most during 1990–2022, which were 20.400%, 18.495%, 18.375%, 20.956%, 18.923%, 19.625%, 19.135% and 16.051%, respectively. The second contribution rate was slightly different, in 1990, it was the secondary industry value with a contribution rate of 12.615%; in 1995, it was the night light with a contribution rate of 14.105%; in 2000, it was the night light with a contribution rate of 13.814%; in 2005, it was the level of human interference with a contribution rate of 12.1795%; in 2010, it was the level of human interference with a contribution rate is 13.361%; in 2015, it was the level of human interference with a contribution rate is 13.123%; in 2020, it was the level of human interference with a contribution rate is 12.191%; in 2022, it was the night light with a contribution rate was 13.875%. Night light and degree of human interference also had obvious effects on LER in JMMB, which indicated that the rapid development of JMMB economy changed the regional landscape structure, thereby affected JMMB’s LER. Among the natural factors, slope contributes the most to the LER evolution of JMMB, which were 19.767%, 20.256%, 20.407%, 18.962%, 20.393%, 18.775%, 20.943% and 18.088% respectively. It indicating that slope affected the landform of JMMB and thus caused the change of LER. Secondly, the average annual temperature also had a certain impact on LER, which was more than 15.000%, explaining that the average annual temperature affected the landscape structure of LULC and thus influenced the LER of JMMB. NDVI was the driving factor that had the least influence on LER in JMMB, and its contribution rate to LER was less than 5.000%. Overall, the contribution degree of social and economic factor to JMMB during the study period was slightly stronger than that of natural factors, demonstrating that during the study period, anthropogenic causes significantly influenced the regional evolution of JMMB’s LER, although natural forces also had some influence. However, with the implementation of the mine restoration policy and the continuous advancement of ecological civilization construction, the total contribution rate of human activities decreased with the change of time, while the total contribution rate of natural factors increased. This suggests that as the strategic deployment of ecological protection and restoration deepens, to some extent, the effects of human activity on LER were being lessened. Meanwhile, the change of natural landscape has an increasing influence on LER.

Table 5 The drivers factor of the LER contribution from 1985 to 2022 in JMMB. %.
Full size table

The interactive detection of different driving factors

JMMB’s LER spatial differentiation driver interaction test results (Fig. 10) presented that nonlinear enhancement and double-factor enhancement were shown in the interaction detection results among the driving factors, and there was no interdependence or weakening during 1985–2022, indicating that the impact degree of all interaction factors on the spatial heterogeneity of LER was significantly enhanced compared with that of a single impact factor.

In 1990, X5∩X7, X3∩X5, X2∩X5, X1∩X5 had the large explanatory power and their q values were above 0.8, while the q value of X1∩X3, X2∩X3 and X3∩X4 also were above 0.7. In 1995, X5, X1∩X5, X2∩X5, X3∩X5 and X4∩X5 had a great interaction effect, which were 0.788511, 0.746680, 0.768169, 0.783219 and 0.768661 respectively, followed by X2∩X3, X2∩X7, X3∩X6, which q value were above 0.7. In 2000, the interaction factors with high contribution rate were X5∩X7, X1∩X5, X2∩X5, X3∩ X4∩X5, which reached above 0.7 respectively, and the highest contribution rate was 0.795870. In 2005, X3∩X5 had the highest contribution rate of interaction factor and the highest q value was 0.740959, followed by X2∩X5 and X1∩X3 with q values of 0.721426 and 0.637684 respectively. In 2010, X5∩X7 had the largest explanatory power and the highest value of q was 0.712080, while the interaction of X1∩X5, X2∩X5, X3∩X5 and X4∩X5 also reached above 0.64. In 2015, X5∩X7 had the largest explanatory power with a q value of 0.718553, and the interaction between X2∩X5, X3∩X5 and X4∩X5, X1∩X5 also reached above 0.6. In 2020, X3∩X8 had a large explanatory power with q values of 0.697263, the second strong interaction were X2∩X5 and X4∩X5, which q values reached 0.651676 and 0.625369 respectively. In 2022, X5∩X8 had the strongest interaction with a q value of 0.688573, while other high contribution rates were X1∩X5, X2∩X5, X4∩X5, X3∩X8 and X5∩X7. The explanatory power of X6 single factor was very low, but the interaction of other factors with X6 were above 0.5, suggesting that spatial heterogeneity was more affected by the interaction. Generally speaking, the interaction of several factors rather than a single element drove the spatial differentiation of JMMB’s LER. According to the results, most factors had the strongest explanatory power when interacting with X5, which proved that X5 was the dominant factor affecting JMMB’s LER, and the combined action of social factors and natural factors had the greatest impact on LER, which resulted in the enhancement of explanatory power. It illustrating that multiple factors jointly determined the LER spatial differentiation of JMMB.

Fig. 10
figure 10

The factors interaction detection results of the LER in JMMB.

Full size image

Discussion

The spatiotemporal changes of LER

The LER of JMMB exhibited obvious spatial heterogeneity: high LER in the north and west, and low LER in the south and east, this result was basically consistent with the results obtained by Guo Lijia42 and Wang Feicui43 on the LER in the Beijing-Tianjin-Hebei region. The high-risk areas were characterized by the presence of mineral resources, as well as intensive human mining and reprocessing activities, which resulted in a sequence of adverse ecological and circumstances consequences for the regional ecosystem, as indicated by a higher LER44. The western high-risk areas were situated to the west of Xingtai, Handan and Shijiazhuang, where coal45, coke and limestone ore resources46were prevalent, mining intensity was considerable, and the mining and reprocessing industry was well-developed. This resulted in significant damage to the integrity and stability of the ecosystem, resulting in a high LER. The northern high-risk areas were primarily situated in Zhangjiakou and Chengde, where the mineral resources, including metals, iron, and coal, were abundant and the mining intensity was relatively high. Furthermore, the region was subjected to overgrazing and reclamation, which contributed to the accelerated degradation and desertification of the grassland. Moreover, the region is characterized by low temperatures and low precipitation throughout the year, slow vegetation recovery, and a prolonged restoration period for LULC, which collectively contribute to a sustained high level of LER. In addition, the results of the land transfer matrix indicated that the types of LULC in high-risk areas were complex and the LULC transformation was frequent in the high-risk areas. Among them, the low-risk areas were seated in the east of Xingtai, Handan and Shijiazhuang, where have flat terrain, mainly cultivated and agricultural activities47, high vegetation coverage, and obviously lower LER intensity in cultivated than in mining concentrated areas48, and strong anti-interference ability. At the same time, the urbanization level in this region was relatively high, and the construction land was distributed in continuous concentrated plots, which was not susceptible to human interference and had strong stability and anti-interference49, making the LER in this region at a continuously low level.

In regard to the temporal dimension, LER in the study area exhibited pronounced fluctuations over time and a clear turning point. Before 2015, the mineral resources of JMMB have been subjected to intense exploitation, resulting in the destruction of the original LULC and a subsequent imbalance in the water and soil, which has contributed to the overall weakening of the ecosystem’s stability. Furthermore, the advancement of industrialization promoted the urbanization process of JMMB, which the population in a notable increased and the rapid expansion of the urban land area in JMMB50, Meanwhile, these changes to the regional climate, including a gradual rise in temperature and a decline in rainfall, led to an increase in LER. The results of the geographical detector in the present study were largely in accordance with the previously outlined interpretation. Since 1992, a series of global conferences on sustainable development have been conducted, gradually proposing more targeted approaches and concrete steps for the implementation of sustainable development. Additionally, these conferences have contributed to the formation of a global political commitment to sustainable development51. China, particularly the Beijing-Tianjin-Hebei region, has demonstrated a proactive response to the international call for the achievement of sustainable development and boosted the ecological progress. This has involved the implementation of ecological environmental protection projects, including mine rehabilitation and natural forest protection, in addition to the implementation of ecological restoration measures, including mine pit filling. Moreover, the reclamation and regreening of mine lands have enabled a transformation in the ecological function of the JMMB, evolving from a single form in its early stages to a more complex and stable one, thereby enhancing the ecological quality continuously over time52, LER diminished and environmental quality improved after 2015. In addition, the geodetection results of this study indicate that the contribution of socio-economic factors (anthropogenic interference, population density and secondary industry to LER was gradually decreasing, whereas the influence of natural factors on LER was gradually increasing. This suggests that ecological restoration measures and ecological environmental protection projects have reduced the regional LER to some extent, this was consistent with the research results of Wang Shu36 et al. Furthermore, urban development was now more orderly as a result of support from new urbanization and the national land spatial planning system, which have also led to an increase in awareness of the ecological environment and a greater focus on its protection. As a result, the LER was gradually decreasing to a certain extent53.

Spatial correlation analysis of the LER

During the study period, the LER of JMMB exhibited a distinct aggregation pattern. The area showed a high-high LER accumulation risk area, characterized by abundant mineral resources and frequent human mineral mining activities, which collectively contributed to the elevated LER. Additionally, the mining activities contributed to the economic development in JMMB, influenced the growth of related industries in surrounding areas, and prompted urbanization and industrial development in the surrounding regions. This resulted in a dynamic interrelationship between natural resources and the advancement of social and economic activities in JMMB. Consequently, a high-high LER aggregation pattern emerged within the region. Moreover, the impact of external environmental factors led to a significant alteration of the original LULC type in JMMB, resulting in a high frequency of conversion among cultivated, wooded, and grassland types. This contributed to a notable fragility of the local ecological environment15, which was a key factor in the high-high aggregation of LER in JMMB. Whereas, the area identified as low-low accumulation risk was primarily situated where there was a transition from other LULC types to construction land. Given that this area’s construction land was primarily divided into plots, it demonstrated strong stability and was relatively immune to external disturbances22. Additionally, the mineral resources in this area were limited, and the regional loss degree was low after interference by human activities, resulting in the formation of a low-low accumulation pattern.

Proposals to improve ecological conditions in JMMB

In light of the aforementioned LER analysis of JMMB, this study gives relevant recommendations for ecological governance in an effort to achieve peaceful cohabitation and superior JMMB development between humans and natural, and to promoting high-quality development and high-level protection. For different levels of LER in JMMB, it is necessary to carry out segmentation refinement and differentiation management. First of all, for the high-risk areas of JMMB, the LER changes of JMMB could be assessed regularly to guide the evolution of high-risk areas to lower risk areas, and promote the virtuous cycle of the ecosystem54. For this area, recovery could be carried out while mining, and the high-risk areas can be continuously managed to ensure the ecological quality of the areas. Secondly, for the medium-risk areas of JMMB, ecological degradation caused by environmental change should be mitigated through the establishment of the JMMB regional coordination mechanism and the rational urban development plan that optimizes the size structure of the city’s population55,56, In addition, proactive planning is necessary for the restoration of regional ecological security in order to increase the resilience of regional ecosystems15. Finally, for low-risk areas of JMMB, Low-risk area conservation should be JMMB’s first priority. It should also clearly define red lines for ecological protection, forbid LULC growth, and maintain a healthy ecological environment in low-risk regions52, and summarize and promote the development model of low-risk areas.

In the future development of JMMB, people should pay more attention to the affect of driving factors on LER, and reduce the negative effect of human interference on the ecological environment. Actively carry out related ecological environment protection work in mineral belt, strengthen ecological environment construction, and achieve coordinated development of economic development and ecological environment. The region’s overall ecological function must be taken into consideration in order to meet the needs of long-term development. Similar regions must be managed differently, ecological protection measures must be developed in accordance with local needs, and environmental protection awareness must be raised among the general public. Understanding JMMB’s evolution is crucial for the advancement of ecological security worldwide.

The limitations and the future perspectives

This study analyzed the spatial and temporal evolution pattern of LER and its driving factors, and provided important implications for the national and global ecological environment protection and management of mineral belt. The assessment’s findings can offer a scientific foundation for human beings on which to formulate differentiated LER management strategies and ecological preservation laws based on local conditions51. While the present study has practical significance, there are limitations that must be addressed in future research. First of all, LER is an integral part of the overall ecosystem, landscape index was selected to build the LERA model in this study, LER means can be improved and optimized to increase the ecological content of LER by building a multi-dimensional index model in the future57. Secondly, there is a strong correlation between the changes in LULC structure and the temporal and spatial distribution and dynamics of LER. These relationships can be used to infer the structural traits and geographical differentiation of regional ecosystems28. we can try to select high-precision LULC for LERA, and put forward more constructive suggestions for ecological restoration in the future. Mankind needs to establish differentiated ecological and environmental management systems based on local conditions, strengthen ecosystem security monitoring, promote ecological and environmental protection services, and scientifically promote the world’s ecological protection and restoration.

Conclusion

This study took JMMB as the research object, constructed an LREA model to analyze the LER of JMMB, used the spatial autocorrelation method to assess the LER’s spatial autocorrelation, and used geographic detectors to investigate the driving variables affecting JMMB’s LER. The main conclusions are as follows:

  1. (1)

    From 1985 to 2022, the proportion of cropland area in the study area changed the most, decreased by 8.35%, followed by the proportion of woodland area, increased by 6.19%. The main type of LULC transfer is the conversion of cropland into construction land, and its conversion area is 10136.12km2, indicating that regional social and economic development and human disturbance have a certain impact on the change of LULC structure.

  2. (2)

    From 1985 to 2022, the JMMB’S LER has obvious spatial differentiation, and the high-risk areas are mainly distributed in the north and central-western regions. The low-risk areas are mainly distributed in the eastern and southeastern regions. From 1985 to 2015, JMMB’s LER mainly showed a shift to medium and high risk level, while from 2015 to 2022, the high risk level shifted to other types. The results showed that JMMB’s ecological risk showed a decreasing trend, and the implementation of ecological restoration strategy improved the landscape ecological risk in the study area.

  3. (3)

    From 1985 to 2022, the gradual decline of the Moran’s I indicates that the dependence and convergence of LER spatial distribution are gradually weakened, which is caused by the maladjustment of LER distribution law caused by LULC changes, resulting in the transition from aggregation to uniform distribution. In the past 40 years, JMMB’s LER has obvious spatial aggregation, mainly ‘high-high’ aggregation and ‘low-low’ aggregation, which is due to the concentration and stability of mineral resources production and development activities.

  4. (4)

    Population density and slope are the main factors affecting the spatial distribution of LER. In the interaction factor detection, the positive interaction between slope and average annual temperature, night light and slope, population density and annual precipitation during the study period has a greater impact on regional LER spatial differences, indicating that bivariate enhancement after factor interaction can affect JMMB’s LER spatial distribution to a greater extent. To some extent, this reveals the joint effect of human activities and natural factors on the LER of JMMB, and provides a reference for the coordinated development of economy and ecology in similar regions around the world.