Impact of landscape pattern on habitat quality in typical karst areas and analysis of its driving factors

YangPingping1,2,3Email18085905817@163.comEmailpingping_yang0320@163.com

BanZhongnian1,2

ZhouZhongfa1,2,3✉Email2439724086@qq.com

ZhangHaoru1,2Email78501306@qq.com

1School of Karst ScienceGuizhou Normal University550001GuiyangGuizhouChina

2State EngineeringTechnology Institute for Karst Desertification Control550001GuiyangGuizhouChina

Guizhou Karst Mountain Land Ecology and Land Use Observation and Research Station, MNR561301GuanlingChina

Yang Pingping^1,2,3, Ban Zhongnian ^1,2, Zhou Zhongfa^1,2,3*, Zhang Haoru^1,2

(1.School of Karst Science,Guizhou Normal University,Guiyang,Guizhou 550001,China;2.State Engineering Technology Institute for Karst Desertification Control,Guiyang,Guizhou 550001,China;3.Guizhou Karst Mountain Land Ecology and Land Use Observation and Research Station,MNR,Guanling 561301,China)

*Corresponding author(s). E-mail(s): 18085905817@163.com;Contributing authors: pingping_yang0320@163.com; 2439724086@qq.com; 78501306@qq.com.†These authors contributed equally to this work.

Abstract

The heterogeneity of karst landforms is strong. The superimposition of natural fragmentation and human activities further divides the habitats. The mechanism by which the fragmentation of the landscape pattern affects species migration, gene flow, and ecosystem functions remains unclear. This study is based on land use data and natural geographic data (DEM, karst landforms, soil types, slope, soil layer thickness, comprehensive vegetation coverage, bare rock exposure rate, and rocky desertification degree) of Guizhou Province from 2000 to 2020, this study analyzed various landscape pattern indices and habitat quality during the 20-year period, by exploring the effects of different landscape pattern indices on habitat quality at global and local scales. The results showed that the level of land use in Guizhou Province increased and the structure stabilized during the 20-year period (LPI: 58.3044→66.9573), and landscape fragmentation decreased (AI: 88.3015→88.7391) and connectivity weakened (CONTAG: 56.5393→55.9076). Habitat quality had large spatial distribution differences, with an overall annual average decline rate of 5.59% from 2000 to 2020. There was a strong local spatial autocorrelation between landscape pattern index and habitat quality. Globally, AI, COHESION, CONTAG and LPI were positively correlated with HQ (AI: r = 0.45; COHESION: r = 0.34; CONTAG: r = 0.39; LPI: r = 0.42), while LSI, PD, SHDI and SPLIT were negatively correlated with HQ (LSI: r=-0.45; PD: r=-0.44; SHDI: r=-0.52; SPLIT: r=-0.28). From a local perspective, “H-H” and “H-L” types are interspersed in non-karst areas and karst trough valleys, “L-H” and “L-L” types are interspersed in karst trough valleys and karst plateaus, and “H-L” and “L-H” types are interspersed in karst canyons, peak-tufted depressions, and faulted basin areas. Habitat quality degradation is the result of the nonlinear superposition of natural vulnerability (bedrock exposure rate, soil thickness) and human activities (land expansion for construction, agricultural intensification). After 2010, the driving mechanism in Guizhou Province shifted from “nature-led” to “human-land interaction”. The results of this study can help to further understand the ecological environmental problems in Guizhou Province, and provide theoretical references for the formulation of ecological environmental protection policies and ecological functional zoning in Guizhou Province.

Keywords:

Habitat quality

Landscape pattern

Different karst development levels

1 Introduction

Ecosystem services refer to the necessary resources and environments that ecosystems provide for human survival, focusing on the response of biomes, biospheres and ecosystems to natural and anthropogenic-induced global changes (Xiang et al., 2019; Zhu et al., 2022). Among them, human-activity-induced land use/land cover (LULC) evolution is one of the important causes of ecosystem service changes (Seto et al., 2012; Song et al., 2018; Kong et al., 2022). Meanwhile, degradation of regional ecosystems brought about by land use changes, and reduction or even loss of habitat quality (HQ) are still continuing to occur (Mitchell et al., 2015; Zheng et al., 2023). Guizhou is an important ecological security barrier in the upper reaches of the Yangtze and Pearl Rivers, and was included in the first batch of national ecological civilization pilot zones in China in 2016. 2022, it was given the position of a pioneer zone of ecological civilization construction (Guizhou has achieved remarkable results in implementing the big ecological strategy, 2024; Guizhou staged a “Green Attack” -- A survey report from the National Ecological Civilization Pilot Zone (Guizhou), 2021). Guizhou Province, as a typical karst region, has a fragile ecological environment (Zhang et al.,2022). There is an obvious correlation between HQ and LULC changes. The landscape pattern index (LSPI) can clearly describe the characteristics of LULC changes and provide targeted strategies for managing the ecological environment in this region (Tang et al.,2022). However, areas with different degrees of karst d·evelopment exist in Guizhou Province, and there is a lack of partitioning in existing studies to explore the driving mechanisms of LSPI changes and HQ changes. Therefore, exploring the effects of LSPI changes on HQ and analyzing the underlying mechanisms in areas with different degrees of karst development are important for ecological conservation and related policy formulation in Guizhou Province.

Habitat quality (HQ) is an important indicator of the level of biodiversity in a region, which reflects the ability of ecosystems to provide suitable conditions for the continued survival of individuals and populations (Xiang et al.,2023; Tobisch et al., 2023). Currently, HQ is mainly assessed by calculating indices such as habitat quality index, vegetation cover index, water network density index, land stress index, and pollution loading index (Wang et al., 2022; Rehitha et al., 2022; Zhang et al., 2022), of which through the InVEST model, due to the fact that it can directly provide the quantitative HQ results, it is now widely used (Xie et al., 2022; Priadka et al.,2022; Fearon et al., 2023; Smith et al., 2023). In addition, previous studies have analyzed the evolution characteristics of HQ at regional, urban (cluster), watershed and economic belt scales, explored the direct or indirect impacts of changes in HQ on ecosystem services, and proposed some rational landscape plans for the conservation and enhancement of biodiversity (Yohannes et al., 2021; He et al., 2023; Wang et al., 2023). However, there is a lack of analysis of the current status of HQ in regions with different degrees of karst development, which limits the proposal of regional biodiversity conservation and ecosystem protection optimization policies. Therefore, it is crucial to assess the changes in spatial and temporal patterns of HQ in areas with different degrees of karst development.

Changes in landscape patterns have direct and indirect effects on ecosystem services, and the response of different ecosystem services to landscape patterns varies (Wang et al., 2017; Liu et al., 2022; Wen et al., 2022). The landscape pattern index (LSPI) is an important tool used in landscape ecology to quantitatively characterize landscape structure and function, which can reflect aspects of landscape composition, structure, diversity, spatial configuration and shape complexity (Antrop 2004). On the one hand, LSPI affects regional HQ by influencing regional climatic factors, hydrological conditions, and climatic conditions; on the other hand, landscape pattern affects regional HQ by controlling regional material exchange processes and the flow of ecological elements (Li et al., 2023). Previous studies have mostly investigated the influence of LSPI on HQ at regional, urban (cluster), watershed, and economic zone scales based on correlation analysis and geoprobe models (Liu et al., 2020; Liang et al., 2022;Lin et al., 2022; Xu et al., 2023) to analyze the relationship between a single type of LSPI and HQ in time and space. The limitation lies in the lack of synthesizing multiple models to study the effects of multiple types of LSPI on HQ, and the elaboration of the relationship between the two is relatively simple and cannot clearly indicate the relationship between the two. In addition, there are areas with different degrees of karst development in Guizhou Province, and it is not clear whether the relationship between LSPI and HQ is related to the degree of karst development.

Therefore, this study focuses on the scientific issue of the influence mechanism and driving factors of landscape patterns on habitat quality in karst areas. By comprehensively applying multi-source remote sensing data, landscape ecology models and spatial statistical methods, it systematically analyzes the spatio-temporal evolution laws and intrinsic correlations of habitat quality (HQ) and landscape pattern index (LSPI) in Guizhou Province from 2000 to 2020. The aim is to explore: 1. How do the characteristics such as fragmentation and connectivity of the landscape pattern in karst areas affect the quality of the habitat? 2. How do natural geographical factors (such as DEM, karst landforms, soil types, etc.) and human activities collaboratively drive the spatiotemporal heterogeneity of HQ?

2 Materials and methods

2.1 Study area

Guizhou Province is located in the karst region of Southwest China, 73.8% of the whole territory is karst landscape, with outstanding ecological vulnerability. in 2020, the area of rocky desertification and soil erosion accounted for 8.81% and 26.68% of the national land area respectively (Land Spatial Planning of Guizhou Province (2021–2035)). It belongs to the subtropical temperate and humid monsoon climate zone, with an average annual precipitation of 682-1,134 mm, and an average annual temperature of 14–16 ℃. The soil zones in Guizhou Province mainly belong to the red soil-yellow soil zone, characterized by the widespread spread of carbonate rocks. The vegetation types in the region are rich and diverse, with obvious subtropical nature, and the vegetation coverage is higher in non-karsted areas than in karsted areas, with abundant forest resources (Zhang et al., 2022; Luo et al., 2024).

The total area, proportion and prevention and control tasks of rocky desertification in Guizhou Province still rank first in the country, and the total area of biodiversity conservation hotspots reaches 37,700 square kilometers, accounting for 21.37% of the total area of the study area. Areas with extremely important ecological functions such as water conservation, soil and water conservation, and biodiversity, as well as key development areas in central Guizhou and northwest Guizhou provinces, are highly overlapping with sensitive and vulnerable areas of ecological environment, and the task of coordinating ecological environment protection and development is very difficult. Evaluating the relationship between LSPI and HQ, and providing a systematic understanding and targeted analysis of the ecological status of areas with different degrees of karst development in Guizhou Province can provide scientific guidance for the formulation of ecosystem protection policies in Guizhou Province.

This study is based on the development degree of karst landforms and the classification of landform types in Guizhou Province, and combines the grades of rocky desertification to form a comprehensive zoning system, as shown in Fig. 1.The geographical distribution and geomorphic features of the six major regions are as follows: 1. Non-karst area: Mainly distributed in Qiandongnan and Qiandongnan regions, etc. This area is dominated by clastic rock mountains and hills, with very few carbonate rocks. The soil layer is relatively thick (> 50cm), and the vegetation coverage is high (forest coverage rate > 60%).2. Karst trough valley area: Mainly distributed in northern and central Guizhou, this area features long and strip-shaped valleys with peak clusters/peak forests on both sides. The valley bottom is flat, the soil layer is thin (20-50cm), and the exposure rate of bedrock is medium (10–30%), making it prone to mild rocky desertification.3. Karst plateau area: Mainly distributed in the central and western parts of Guizhou Province and the southwestern part of Guizhou Province, this area is a karst plateau surface at an altitude of 1,000 to 1,400 meters, with widespread karst mounders and deprestices. The surface is short of water, the soil layer is extremely thin (< 20cm), and the exposure rate of bedrock is over 40%.4. Peak Cluster and depression areas: Mainly distributed in southern and southwestern Guizhou, this area features a combination of dense conical peak clusters and closed depressures. Underground rivers are well-developed, and the soil layer at the bottom of the depressures is slightly thicker (30-60cm). The bedrock on the peak cluster slopes is exposed by more than 50%.5. Fault depression basin area: Mainly distributed in western and southwestern Guizhou and other places, the topographic feature of this area is a basin controlled by a fault zone, with steep edges and flat bottoms. The soil layer inside the basin is relatively thick (40-80cm), and the potential risk of rocky desertification is high.6. Karst canyon area: Mainly distributed in the middle and lower reaches of the Beipan River and the Wujiang River, etc. The topographic features of this area are deep canyons, steep bank slopes (> 45°), vertical height difference > 500m, extremely thin soil layer (< 10cm), and exposed bedrock rate > 60%.

Fig. 1

Maps of study area and distribution of different karst zones

2.2 Data sources

The land use/cover (LULC) data used in the study were from Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (https://www.resdc.cn/Introduction.aspx), with a resolution of 30m. By figure of heaven and earth to provide the data extraction in guizhou province administrative divisions (https://cloudcenter.tianditu.gov.cn/administrativeDivision/).

Table 1
presents the data used in this study and their sources. Among them, the natural geographic data include DEM, karst landform, soil type, slope, soil layer thickness, comprehensive vegetation coverage, bedrock exposure rate and degree of rocky desertification. These data come from the Geospatial Data Cloud, the Rocky Desertification Prevention and Control Center of Guizhou Province, the Center for Resources and Environmental Sciences and Data of the Chinese Academy of Sciences, and the Georemote Sensing Ecological Network. Then, these data are imported into the corresponding administrative regions of each county using ArcGIS 10.8 and rasterized. Table.1 Data sources
Type	Name	Units	Data Source
Basic Data	Guizhou Province Land Use Data	/	Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences
	Guizhou Province Land Use Data	/	(https://www.resdc.cn/Introduction.aspx)
	Administrative Boundaries of Guizhou Province	/	Administrative Division Data of China _ Approval Number :GS(2024)0650
	Administrative Boundaries of Guizhou Province	/	(https://cloudcenter.tianditu.gov.cn/administrativeDivision/)
Natural Drivers	Digital Elevation Model (DEM)	m	Geospatial data cloud
	Digital Elevation Model (DEM)	m	(http://www.gscloud.cn)
	Slope	°	Extraction based on DEM
	Soil Type	/	Resources and Environmental Science and Data Center, Chinese Academy of Sciences
	Soil Type	/	(https://www.resdc.cn/Default.aspx)
	Karst landform	/	Guizhou rocky desertification control center
	Soil layer thickness	/	Guizhou rocky desertification control center
	Comprehensive coverage of vegetation	/	Guizhou rocky desertification control center
	The exposure rate of bedrock	/	Guizhou rocky desertification control center
	Rocky desertification type	/	Guizhou rocky desertification control center

2.3 Research methodology

2.3.1. Rate of change in land use dynamics

It is used to quantitatively describe the extent and rate of area change of a land use/land cover (LULC) type over a specific time period. The formula is as follows:

$\:LC=\frac{{U}_{b}-{U}_{a}}{{U}_{a}}\times\:\frac{1}{T}\times\:100\%$	(1)
$\:LCG=\left(\frac{{\sum\:}_{i=1}^{n}\varDelta\:L{U}_{i-j}}{2{\sum\:}_{i=1}^{n}\varDelta\:L{U}_{i}}\right)\times\:\frac{1}{T}\times\:100\%$	(2)

where LC represents the dynamic rate of change of a specific LULC type, Ua and Ub represent the area of the type in the initial and final stages respectively, LCG represents the dynamic rate of change of the integrated LULC, LUi represents the area of the LULC in the initial stage, ΔLUi-j represents the absolute area of the LULC that was converted to the other types during the study period, and T represents the length of the study period (in years) ( Liu et al., 2018).

2.3.2. Land use transfer matrix

It is used to show the dynamic transformation process between LULC types in the study area, including transformation direction and transformation structure. The formula is as follows:

$\:{G}_{\text{xy}}=\left[\begin{array}{cccc}{G}_{11}&\:{G}_{12}&\:\cdots\:&\:{G}_{1n}\\\:{G}_{21}&\:{G}_{22}&\:\cdots\:&\:{G}_{2n}\\\:\cdots\:&\:\cdots\:&\:\cdots\:&\:\cdots\:\\\:{G}_{n1}&\:{G}_{n2}&\:\cdots\:&\:{G}_{nn}\end{array}\right]$	(3)
where n represents the number of LULC types, x and y (x,y = 1, 2, ..., n) represent the LULC types in the initial and final stages, respectively, and G_xy represents the transformation area from LULC type x to LULC type y (Yang et al., 2018)Landscape pattern index

In this study, a 1km*1km cell grid was used as the unit of analysis, combining with previous studies (Jia et al.,2019; Xue et al.,2023), different types of landscape indexes were selected for analysis (Table 2), and all the LSPI calculations were realized in Fragstats 4.0.

Table.2 Chosen landscape pattern index

Category	Index	Description	Formula	Annotation
Landscape area index	Maximum Patch Index (LPI)	The higher the dominant degree of the largest patch in the landscape, the better the habitat quality	$\:\text{LPI}=\frac{\text{MAX}\left({a}_{ij}\right)}{{A}_{}}\times\:100\%$	a_ij = area of patch_ij (m²); A = Total landscape area (m²)
Landscape quantity index	Patch density (PD)	Number of patches per 100 hectares of landscape	$\:\text{PD}=NP/A$	NP = the total number of patches in the landscape; A = Total landscape area (m²)
Landscape shape index	Landscape Shape Index (LSI)	It mainly reflects the complexity of patch shape in landscape.	$\:\text{LSI}=\frac{\text{0.25}L}{\sqrt{{A}_{}}}$	A = Total landscape area (m²)
Landscape aggregation index	Aggregation index(AI)	The higher the degree of landscape fragmentation, the better the habitat quality	$\:AI=\frac{1}{2}\left[1-\frac{{\sum\:}_{i=1}^{m}{\sum\:}_{j=1}^{m}\frac{{g}_{ij}}{{max}\left({g}_{ij}\right)}{\left(1-{p}_{ii}\right)}^{2}}{{\left(1-\frac{1}{m}\right)}^{2}}\right]$	m = the total number of patch types in the landscape; g_ij =the total boundary length between patch type i and patch type j; max(g_ij) = the maximum possible boundary length between all pairs of patch types; p_ii =the proportion of the internal boundary of patch type i.
Landscape connectivity index	Cohesion index (COHESION)	The stronger the connectivity within the landscape, the worse the habitat quality.		a_ij = the area of patch j in Class i landscape; P_ij is the perimeter (m) of patch j in a Class i landscape.
Landscape spread index	Spread index (CONTAG)	It can describe the agglomeration degree or extension trend of patch types in the landscape, including spatial information.	$\:CONTAG={\sum\:}_{i=1}^{m}{\sum\:}_{j=1}^{m}\frac{{g}_{ij}\times\:\left({\pi\:}_{ij}+{\pi\:}_{ji}\right)}{2\times\:{m}^{2}}$	m = the total number of patch types in the landscape; g_ij = the adjacency probability between plaque type i and plaque type j; π_ij is the frequency adjacent to plaque type i and plaque type j; π_ji is the frequency adjacent to patch type j and patch type i.
Landscape diversity index	Patch Richness Density (PRD)	It refers to the richness of patch types (or land cover types) per unit area, which can reflect the diversity and heterogeneity of the landscape	$\:\text{PRD}=R/A$	R = the total number of plaque types in the study area; A = Total landscape area (m²)
Landscape fragmentation index	Fission index (SPLIT)	This metric evaluates both the spatial distribution pattern of the specified patch type and its isolation from other patch types in the landscape.	$\:SPLIT=\frac{1}{2}{\sum\:}_{i=1}^{m}\frac{{g}_{ij}}{{A}_{i}}$	m = the total number of patch types in the landscape; gij = the adjacency probability between plaque type i and plaque type j;

2.3.3. InVEST model habitat quality assessment

The InVEST Habitat Quality module, by combining landscape type sensitivity and external threat intensity in a comprehensive calculation, considers HQ as a continuous variable under the premise of considering factors such as the distance of influence of the coercive factor, spatial weight and the degree of legal protection of the land, etc., and takes into full consideration the impacts of changes in the land cover pattern and land cover mode on the quality of the habitat in conducting the assessment. The calculation formula is as follows:

$\:{D}_{xj}={\sum\:}_{r=1}^{R}{\sum\:}_{y=1}^{{Y}_{r}}({w}_{r}/{\sum\:}_{r=1}^{R}{w}_{r}){r}_{y}{i}_{rxy}{\beta\:}_{x}{S}_{jr}$

(4)

Where: D_xj is the degree of habitat degradation of raster x in habitat type j; R is the number of threat sources; W_r is the weight of threat source r; Y_r is the number of rasters of threat elements; r_y is the coercive value of raster y; ß_x is the accessibility of threat elements to raster x (the value of ß_x is determined to be between 0 and 1 according to the degree of its legal protection); S_jr is the sensitivity of habitat type j to threat source r; i_rxy is the Stress level of raster y to raster x. There are two kinds of linear decay and exponential decay:

linear decay: $\:{i}_{rxy}=1-(\frac{{d}_{xy}}{{d}_{rmax}}$	(5)
exponential decay: $\:{i}_{rxy}={exp}(\frac{-2.99{d}_{xy}}{{d}_{rmax}}$	(6)

Where: d_xy is the straight line distance between grid x and grid y; d_rmax is the maximum coercive distance of the threat sourcer.

HQ is calculated by the formula:

$\:{Q}_{xj}={H}_{j}\left[1-\left(\frac{{D}_{xj}^{z}}{{D}_{xj}^{z}+{k}^{z}}\right)\right]$

(7)

Where, Qxj is the HQ index of x grid in habitat type j; Hj is the habitat suitability of habitat type j (0 ≤ Hj ≤ 1); k is the half-saturation constant, and half of the maximum habitat degradation degree is taken, which is generally set as 0.5; z is a normalized constant, usually 2.5. In this paper, InVEST 3.8.0 version is applied to carry out HQ model calculation, referring to relevant literature at home and abroad(Liu et al.,2021 ; Yuan et al.,2023),combined with the actual situation of Guizhou Province, a series of processing, such as vectorization, reclassification, raster calculation and data summary, were carried out for the LULC data of 3 periods in ArcMap10.8 platform.The selected land class was assigned a value of 1, and the remaining land class was assigned a value of 0. Five land classes, namely cultivated land, urban land, rural residential area, other construction land and unused land, were extracted as threat sources, and the related maximum stress distance, weight and attenuation type were determined (see Table 3),the suitability of other different types of habitat and sensitivity to stress factors are shown in Table 4.

Table 3
Threat factor weighting table
Threat factor	Maximum influence distance/km	weight	Decaying linear dependence
Plowland	1	0.7	linearity
Urban construction land	8	1	exponent
Rural residential land	6	0.8	exponent
Other construction land	4	0.4	linearity
Unutilized	4	0.4	linearity

Table 4
Threat factor sensitivity table
Land class types	Habitat suitability	plowland	Urban construction land	Rural residential land	Other construction land	unutilized
Paddy field	0.4	0	0.7	0.6	0.6	0.5
Dry land	0.3	0	0.7	0.6	0.6	0.5
Forest land	1	0.7	0.8	0.3	0.3	0.6
shrubbery	0.8	0.8	0.6	0.4	0.4	0.6
Open forest land	0.7	0.7	0.7	0.5	0.5	0.6
Basal forest land	0.6	0.6	0.7	0.5	0.5	0.6
High cover grassland	0.8	0.8	0.5	0.7	0.7	0.5
Medium coverage grassland	0.7	0.7	0.7	0.5	0.5	0.3
Low cover grassland	0.6	0.5	0.7	0.5	0.5	0.4
River and canal	0.8	0.8	0.6	0.6	0.8	0.3
lakes	0.9	0.7	0.7	0.7	0.4	0.5
Reservoir pit	0.6	0.7	0.6	0.7	0.5	0.5
Bottom land	0.6	0.6	0.7	0.3	0.5	0.5
Urban land	0	0	0	0	0	0
Rural residential area	0	0	0	0	0	0
Other construction land	0	0	0	0	0	0
Special land	0	0	0	0	0	0
Marshland	0.1	0.1	0.1	0.1	0.1	0.1
Bare land	0.1	0.1	0.1	0.1	0.1	0.1
Bare rock stony land	0.1	0.1	0.1	0.1	0.1	0.1

2.3.4. Spatial autocorrelation analysis

With the technical support of GIS10.8 software, this paper uses a 1km*1km cell grid to resample the study area in a gridded manner, and applies a bivariate spatial autocorrelation analysis model to analyze the spatial correlation characteristics of LSPI and HQ.

GeoDa software can perform exploratory spatial data analysis on raster data, which is commonly used for autocorrelation statistics and outlier indication analysis (Yang et al.,2024; Chasco 2025). Local spatial autocorrelation analysis was performed using local indicators of spatial association (LISA) to express local clustering and discrete effects (Zhu et al., 2020), and the calculation formula was as follows:

$\:{I}_{sr}=\frac{n{\sum\:}_{i=1}^{n}{\sum\:}_{j=1}^{n}{w}_{ij}\left(\frac{{y}_{i.s}-\overline{{y}_{s}}}{{\sigma\:}_{s}}\right)\left(\frac{{y}_{i.r}-\overline{{y}_{r}}}{{\sigma\:}_{r}}\right)}{\left(n-1\right){\sum\:}_{i=1}^{n}{\sum\:}_{j=1}^{n}{w}_{ij}}$

(8)

Where:I_sr is the bivariate global autocorrelation coefficient of unit area HQ value s and landscape pattern index r; y_i.s and y_i.r are the unit area HQ value and landscape pattern index of the ith evaluation unit, Ϭ_s and Ϭ_r are the variances, w_ij is the weight matrix established based on spatial adjacency, and n is the number of evaluation units.

A Lisa clustering diagram was generated using the obtained data to analyze the spatial correlation between LSPI and HQ, where high-high clustering and low-low clustering denote positive correlation and high-low clustering and low-high clustering denote negative correlation.

2.3.5. Geographically weighted regression model (GWR)

Geographically weighted regression (GWR) is a spatial analysis technique that explores the spatial and temporal variation of objects and their drivers at a given scale by creating local regression equations at points on each spatial scale ( Tian et al.,2023; Liao et al.,2023). The model has higher accuracy and is not limited by the global perspective. The geographically weighted regression model takes the following form:

$\:{y}_{i}={\beta\:}_{0}\left({u}_{i},{v}_{i}\right)+{\sum\:}_{k=1}^{p}{\beta\:}_{k}\left({u}_{i},{v}_{i}\right){x}_{ik}+{\epsilon\:}_{i}$

(9)

where (u_i,v_i) are the coordinates of sampling point i and ß_k(u_i,v_i) is the kth parameter of the regression equation on sampling point i, which is a function of geographic location and is obtained by utilizing the weighted function approach in the estimation process, i = 1,2,...,n.

The local regression coefficients obtained from the geographically weighted regressions indicate the extent to which the explanatory variables affect the dependent variable (Gao et al.,2023). The multiple natural geographic factors obtained in the linear regression model were used as explanatory variables, and HQ was used as the dependent variable, and imported into the geographically weighted regression model for geographically weighted regression analysis to obtain the spatial variability maps of the regression coefficients of the natural geographic factors corresponding to the HQ, so as to explore the effects of the multiple natural geographic factors on the HQ.

3 Results

3.1 Spatial and temporal changes in land use

3.1.1 Temporal change of land use

According to Table 5, the overall landscape type of land use in the study area did not change much between 2000 and 2020, cropland and forested land generally showed an increasing trend, with a dynamic rate of 1.72% and 0.44%, respectively, and remained the main land use type in the whole study area. Meanwhile, the area of constructed land also shows an increasing trend and grows at a dynamic rate of 0.39% during the period of 2010–2020. In terms of ecological land use, the total area decreases, with the areas of shrubland and grassland rapidly decreasing at a dynamic rate of -1.90% and − 0.96%, respectively. In contrast, the water area shows a general increasing dynamic trend with a relatively stable rate of change of 0.15%.

The combined land use dynamics for the periods 2000–2010 and 2010–2020 were 6.78% and 7.21%, respectively, indicating an increase in the intensity of human activities. All land use types changed during the study period, and land use changes were more rapid between 2010 and 2020.

Fig. 2

Distribution of land use types in Guizhou Province from 2000 to 2020

Table 5
The dynamic rate of change of different LULC classes in the study area from 2000 to 2020.
LULC type	2000–2010	2010–2020	2000–2020
Cropland	1.87%	-1.43%	0.44%
Forest	-0.96%	2.68%	1.72%
Shrub	-0.54%	-1.36%	-1.90%
Grassland	-0.61%	-0.35%	-0.96%
Water	0.08%	0.06%	0.15%
Barren	0.00%	0.00%	0.00%
Impervious	0.16%	0.39%	0.55%

3.1.2 Spatial changes in land use

Figure 3 shows the trend of spatial change.Between 2000 and 2010, most land use changes occurred in areas such as karst canyons, peak-tufted depressions and fault-trapped basins, and were mainly conversions between farmland, forests and shrublands.The hotspot areas for conversions shifted to karst plateaus and troughs and valleys in the period of 2010–2020, and the types of land use changes increased, with arable land converted out of the area accounting for 43.01%.

During the period of 2000–2010, the conversion area of forest land occupies the largest proportion of the converted area, which is 40.03%. During the period 2010–2020, cropland became the main type of conversion, with a conversion share of 43.01% (Figs. 3 and 4). Specifically, during the period 2000–2020, 308.97 km² of cropland was converted to construction land, accounting for 5.47% of the total area transferred out of cropland. During this period, the transfer of ecological land (including grasslands, forests and waters), whose converted areas were 25.56 km², 21.49 km² and 4.90 km² respectively, also contributed to the sustained increase in the area of construction land. However, only 2.13 km² of non-ecological land was converted into ecological land, and it was relatively dispersed, with weak ecological restoration effects; 308.97 km² of arable land was converted into construction land, which directly exacerbated habitat fragmentation.

Fig. 3

The land-use transfer matrix of different years:(a)2000–2010;(b)2010–2020

Fig. 4

The diagram of interannual variation trend of land-use

(Note: This figure exclusively examines the types of land use that result in conversion.)

3.2 Landscape pattern change characteristics

The spatio-temporal features of LSPI were selected for analysis, including AI, COHESION, LPI, CONTAG, PRD, PD, LSI and SPLIT, and ArcGIS was used to generate the spatio-temporal distribution map of the above LSPI, as shown in Fig. 5.

The AI (Aggregation Index) is used to measure the degree of aggregation of patches in the landscape or the continuity of the landscape, and a low value of AI indicates that the patches in the landscape are small and dispersed, and there is a lack of large, continuous patches and poor connectivity between patches.The mean value of AI increased from 88.3015 in 2000 to 88.7391 in 2020, indicating that at the landscape level, land use activities are characterized by continuous aggregation of landscapes.Higher AI values are mainly concentrated in non-karst areas, while lower AIs are mainly concentrated in karst canyons and faulted basin areas, which have fertile soils, are suitable for agriculture, and have lush vegetation growth, and therefore human production activities are more frequent, for example, activities such as agricultural reclamation, urbanization, and road construction, which can lead to more natural landscape Broken.

COHESION reflects the aggregated and dispersed state of patches in the landscape and quantifies the perception of physical connectivity of habitats by organisms as they disperse in a binary landscape. During the study period, the mean values of COHESION were 96.6891 in 2000, 96.8361 in 2010, and 96.7856 in 2020, showing a slightly increasing trend but little fluctuation. The distribution characteristics are the same as those of AI, indicating the influence of human activities on the landscape pattern. Non-karst areas have better vegetation cover and higher biodiversity, so their landscape connectivity is better.

LPI is used as an indicator of the degree of dominance of a particular landscape type in the landscape as a whole.Higher values of LPI indicate the existence of one or several very dominant patches in the landscape, while lower values of LPI indicate that there is no particularly dominant patch in the landscape and the landscape is more homogeneous.The LPI increased from 58.3044 in 2000 to 66.9573 in 2020, indicating that the degree of landscape dominance has continued to increase.The spatial distribution pattern of LPI is similar to that of AI and COHESION, while the low value area has decreased significantly and different landscape types tend to be evenly distributed.

CONTAG is the degree of aggregation or expansion of different patch types within a landscape, and a higher CONTAG indicates that some dominant patches in the landscape have good connectivity. CONTAG decreased from 56.5393 in 2000 to 55.9076 in 2020, indicating that the connectivity of the landscape continues to weaken. The low-value zones of CONTAG have obviously decreased, but there are still localized concentrations in mountainous areas such as the non-karstic areas and karst trough areas, and the higher value zones are mostly distributed in the main urban cores of karst plateau zones, where the aggregation or expansion of different landscapes is significantly stronger.

SHDI is usually used to describe the spatial distribution and degree of diversity of different types of patches in the landscape, the SHDI value equal to 0 indicates that there is no diversity of patch types in the landscape, the larger the SHDI value indicates that the higher the diversity of patch types in the landscape, the more uniform the distribution of each patch type in the landscape.Between 2000 and 2020, the SHDI value decreased slightly from 3.5272 in 2000 to 3.5125 in 2020, which indicates a decrease in land use diversification.During the study period, land use became more intensive and land use structure became more stable; at the spatial scale, high values of SHDI were found in areas with frequent human activities, such as karst plateaus, karst canyons and faulted basins, where there are more developed urban agglomerations or contiguous centralized croplands with significantly higher anthropogenic impacts and higher landscape diversity, whereas the areas with low values of SHDI were mainly concentrated in the mountainous areas or the basins.

PD is an important indicator used in landscape ecology to describe the degree of landscape fragmentation, a higher PD value indicates a higher degree of landscape fragmentation, and a lower PD value indicates a more continuous landscape with a lower degree of fragmentation.Between 2000 and 2020, the PD was reduced from 22.8563 in 2000 to 20.2019 in 2020, which indicates that land-use fragmentation has been reduced. During the study period, land use became more intensive and the land use structure became more stable; at the spatial scale, the high values of PD were in areas with frequent human activities, such as karst plateaus, karst canyons and faulted basins, etc., which have more developed urban agglomerations or continuous concentrated croplands, where the impact of anthropogenic activities is more pronounced, and the landscapes tend to be fragmented, whereas the areas with high PDs were mainly concentrated in the non-karstic areas.

LSI can reflect the degree of zigzagging of the plaque boundary and the regularity of the plaque shape, the larger the LSI value indicates that the plaque shape is more complex, with a zigzagging boundary and irregular shape.Smaller LSI values indicate simpler patch boundaries and regular shapes.LSI decreases from 3.0606 in 2000 to 2.9968 in 2020, which suggests that landscape patch shapes tend to be more regular. Urban and agricultural expansion caused by human activities, on the other hand, can lead to an increase in LSI values and the fragmentation of ecological processes, such as the reduction of biodiversity, the decline of ecological connectivity and the degradation of ecological functions, for example, in karst plateaus, karst valleys and fault basins where higher LSI values are concentrated.

In general, natural landscapes less disturbed by human activities have lower SPLIT, while natural landscapes more affected by human activities have higher SPLIT. the SPLIT decreased from 2.3329 in 2000 to 2.2362 in 2010, and increased to 2.3024 in 2020, indicating that the condition of the natural landscapes has stabilized to some extent. the SPLIT's spatial distribution pattern is similar to that of the LSI, with lower core areas also mainly concentrated in non-karst areas with better ecological environment, and higher values mainly concentrated in areas such as karst plateaus, karst canyons and fracture basins, attributed to the high level of human activities in the area, which cause significant disturbance to the natural landscape.

Fig. 5

Spatial and temporal distribution of LSPI in Guizhou Province from 2000 to 2020

The characteristics of the overall landscape pattern change are as follows: landscape connectivity increased, patch aggregation increased (AI: 88.30→88.74; COHESION: 96.69→96.79); dominant patches centralized (LPI: 58.30→66.96); landscape diversity decreased, and land use tended to intensify (SHDI: 3.53→3.51; PD: 22.86→20.20); regularization and fragmentation of patch shapes eased (LSI: 3.06→3.00; SPLIT: 2.33→2.30). The overall spatial differentiation shows that the high AI and LPI values are concentrated in the non-karst areas, and the low values are distributed in the karst canyon area and the fracture basin area, where human activities are more intensive.

3.3 Characterization of changes in habitat quality

3.3.1 Spatial and temporal changes in habitat quality classes

Between 2000 and 2020, the average habitat quality in the study area declined. It decreased from 0.634 (2000) to 0.606 (2020), a decrease of 4.4%. Using ArcGIS 10.8 software, the habitat quality values obtained were categorized into five classes based on previous studies (Liu et al., 2022): I (0-0.2), II (0.2–0.4), III (0.4–0.6), IV (0.6–0.8), and V (0.8-1). The acreage of the different classes of habitat quality and their percentage of overall habitat acreage was calculated for the three time points from 2000 to 2020 (see Table 6).

Between 2000 and 2020, habitat quality in the study area was dominated by Class IV, although its area decreased continuously throughout the study period from 96911.99 km² to 92667.65 km² and its percentage decreased from 55.03% to 52.62%, and the second percentage was Class II, which showed a small decrease in area from 49624.28 km² to 48268.09 km². All other classes increased in size between 2000 and 2020, with the most significant increases in Class V and Class I, whose proportions increased from 0.36% and 13.44% to 1.39% and 15.15%, respectively.

Table 6
Changes in Area and Percentage of Habitat Quality Classes in the study area from 2000 to 2020.
Habitat quality level	2000		2010		2020
	Area	Percentage	Area	Percentage	Area	Percentage
	km²	%	km²	%	km²	%
Ⅰ	639.65	0.36%	904.07	0.51%	2445.09	1.39%
Ⅱ	49624.28	28.18%	49401.70	28.05%	48268.09	27.41%
Ⅲ	5251.61	2.98%	5285.86	3.00%	6030.39	3.42%
Ⅳ	96911.99	55.03%	95863.48	54.44%	92667.65	52.62%
Ⅴ	23674.00	13.44%	24648.07	14.00%	26680.11	15.15%

Higher HQ areas may benefit from regional geographic factors, such as abundant rivers, forests and wetland resources, or mountainous hills that are less disturbed by human activities. Level V in the study area is mainly distributed in the non-karstic area, which has less ground undulation, a variety of landform types, including hills, low mountains, and mid-mountains, etc., with a more developed surface water system, a variety of soil types, a better vegetation cover, and high biodiversity, and thus a better quality of habitats. Low HQ areas are concentrated in the karst plateau area, which is characterized by rapid urbanization and shows a large continuous distribution with low vegetation cover and a single ecosystem composition, as urban construction areas are often distributed as dense residential areas. Lower HQ areas are concentrated in karst trough valleys, peaked depressions and other areas with thin soil layers and even lower vegetation cover, and land use types are mainly cropland and construction land in karst canyons and faulted basins areas, all with low biodiversity. Therefore, it can be seen that the spatial distribution of HQ is closely related to natural geographic features, land use types and human activities.

Fig. 6

3.3.2 Degree of habitat degradation

Degree of habitat degradation is the extent to which a habitat is disturbed by threatening factors. In this study, the degree of habitat degradation was calculated using the InVEST model for three time points: 2000, 2010, and 2020, and its spatial distribution was mapped (Fig. 7).

In the study area, the degree of habitat degradation was generally low, but increased over time, with the mean value increasing from 0.1002 in 2000 to 0.1081 in 2020. In 2000, the areas with higher degradation were mainly concentrated in and around urban and construction sites in the karst plateau area, which has flatter topography with less undulation compared to the complex topography of canyons and peak forests and has richer groundwater resources. Although the soil in the karst plateau area is relatively thin, it is fertile in some areas, which is suitable for agricultural development. Meanwhile, the ecological environment is relatively good, with high air quality, which is suitable for living. By 2020, however, the areas with a higher degree of degradation have been extended to the karst trough areas, probably due to the increase in demand for natural resources caused by urbanization development and population growth, which exacerbates the environmental pressure on the karst trough areas, and the over-cultivation of agricultural production, the expansion of urbanization, and the irrational use of land may lead to soil erosion, the destruction of vegetation cover, the destruction of wildlife habitat, and the decline of biodiversity. This indicates that the urbanization process has had a serious negative impact on habitats.

Fig. 7

3.4 Analysis of landscape pattern effects on habitat quality

3.4.1 Spatial autocorrelation analysis

The LSPI and HQ data in 2000, 2010 and 2020 were imported into GeoDa software, and correlation analysis was conducted between habitat and LSPI. AI, COHESION, CONTAG and LPI in the landscape pattern index were positively correlated with habitat quality, while the other indexes were negatively correlated, which indicated that the habitat quality is facilitated by the greater aggregation and continuity of patches in the landscape and by the role of dominant patches in the ecological processes of the landscape. All correlation analyses were statistically significant (p < .01). The absolute values of the correlation coefficients for each of the indices, except for CONTAG and SHDI, were increasing, suggesting that although there is some correlation between habitat quality and landscape pattern, the correlation between the two is increasing.

Table 7
Global Moran's I values for landscape pattern index and habitat quality in the study area from 2000 to 2020
Year	Value type	AI&HQ	COHESION&HQ	CONTAG&HQ	LPI&HQ	LSI&HQ	PD&HQ	SHDI&HQ	SPLIT&HQ
2000	Moran′s I	0.451	0.336	0.394	0.415	-0.451	-0.435	-0.515	-0.280
2000	Z value	31.994	34.075	32.050	30.245	-31.930	-31.650	-37.258	-21.724
2010	Moran′s I	0.471	0.371	0.375	0.465	-0.476	-0.473	-0.476	-0.276
2010	Z value	32.019	32.098	32.935	31.492	-32.361	-32.500	-32.418	-22.361
2020	Moran′s I	0.485	0.385	0.385	0.476	-0.490	-0.488	-0.490	-0.290
2020	Z value	32.426	32.533	32.348	31.609	-32.881	-32.880	-32.915	-22.901

The results of the local spatial autocorrelation analysis are shown in Table 8, the cells with insignificant per indicator accounted for more than half of the total number of cells in the study area without spatial autocorrelation.CONTAG had a more balanced distribution of clusters, with the number of cells with positive spatial correlation closest to the number of cells with negative spatial correlation; the rest generally had more high-high and low-low clustering, and fewer high-low and low-high clustering, the number of The number of cells with positive spatial correlation was greater than the number of cells with negative correlation; AI, COHESION, and LPI had significantly more high-high clusters than low-low, high-low, and low-high clusters, and LSI, PD, SHDI, and SPLIT had significantly more low-low clusters than high-high clusters.

Table 8
Statistics of the number and percentage of agglomeration-type patches in 2020
		Not significant	High-High	Low-Low	Low-High	High-Low
AI	Quantity block	86629	33517	17435	10637	9916
AI	Percentage	54.78%	21.20%	11.03%	6.73%	6.27%
COHESION	Quantity block	86627	34318	16243	10236	10710
COHESION	Percentage	54.78%	21.70%	10.27%	6.47%	6.77%
CONTAG	Quantity block	86629	20747	27068	12407	11283
CONTAG	Percentage	54.78%	13.12%	17.12%	7.85%	7.14%
LPI	Quantity block	88629	37041	15925	8513	8026
LPI	Percentage	56.05%	23.42%	10.07%	5.38%	5.08%
LSI	Quantity block	86622	8698	35783	13457	13574
LSI	Percentage	54.78%	5.50%	22.63%	8.51%	8.58%
PD	Quantity block	86624	7480	37432	14674	11924
PD	Percentage	54.78%	4.73%	23.67%	9.28%	7.54%
SHDI	Quantity block	86625	8274	36164	13880	13191
SHDI	Percentage	54.78%	5.23%	22.87%	8.78%	8.34%
SPLIT	Quantity block	86624	7264	38111	14891	11244
SPLIT	Percentage	54.78%	4.59%	24.10%	9.42%	7.11%

The Lisa clustering plot shows that there are five types of localized spatial autocorrelation: “high - high”, “low - low”, “high - low”, “low - high ” and “not significant” (Fig. 8). The “H-H” type indicates a clustering effect of high values of landscape pattern indices and habitat quality. The “H-L” pattern indicates that high values of landscape pattern index and low values of habitat quality have a clustering effect. Differences in natural geographic features and hydrothermal conditions in different units of Guizhou Province led to different responses of different LSPIs to HQ. The “high - high” clustering and “low - low” clustering distribution characteristics of AI, COHESION, CONTAG and LPI on HQ were more obvious in the 1km*1km unit scale of the study area, which indicated positive correlation between LSPI and HQ, while the rest of the index characteristics indicated negative correlation with HQ. The spatial clustering characteristics of the eight types of LSPI on HQ are different, which is also a reflection of spatial heterogeneity.

Due to the spatial correlation, the relationship between LSPI and HQ in the same region follows a certain pattern. The relationship between LSPI and HQ in the same region follows a certain pattern due to the spatial correlation, and is analyzed according to the different karst development areas in Guizhou Province, with “H-H” clustering dominating in non-karst areas and karst troughs and valleys, and “L-L” clustering dominating in karst canyons and faulted basins. In the karst plateau area, the distribution of “L-H” and “H-L” clusters is intertwined, reflecting the complex coupling of anthropogenic disturbances and natural geographic conditions.

Fig. 8

High-low clustering of landscape pattern index and habitat quality

3.4.2 Multiple linear regression analysis

Multiple linear stepwise regression analysis of the relationship between multiple natural geographic factors and HQ was performed using SPSS software. This process included selecting the natural geographic factors and performing data standardization prior to data processing to make the model more accurate. The selection criteria were: regression model R2 ≥ 0.3 and VIF variance inflation factor test < 10 (Kroll et al.,2013; Imdadullah et al.,2016).

According to the model results (Table 9), HQ was linearly related to the physical geography factor:

Q_xi=-0.001*DEM + 0.142*Karst landform+-2.227*Soil type + 1.167*Slope + 2.221*Soil layer thickness + 0.077*Comprehensive coverage of vegetation+-0.276*The exposure rate of bedrock + 7.581*Rocky desertification type

After standardization:

Q_xi=-0.002*DEM + 0.001*Karst landform+-0.011*Soil type + 0.004*Slope + 0.007*Soil layer thickness + 0.005*Comprehensive coverage of vegetation+-0.012*The exposure rate of bedrock + 0.018*Rocky desertification type

From the table, it can be seen that karst landform, slope, soil thickness, comprehensive cover of vegetation, rocky desertification degree and HQ are positively correlated, and the degree of influence is in the following order: karst landform > slope > thickness of soil > comprehensive cover of vegetation > rocky desertification type, whereas DEM, soil type, and exposed rate of bedrock are negatively correlated with HQ, with higher significance for DEM and soil type. Conditions such as low temperature and humidity, low oxygen content and infertile soil are not conducive to plant growth, resulting in lower biodiversity and poorer habitat quality; conditions such as high vegetation cover, fertile soil, soil and water conservation and light are good, which are conducive to plant growth and provide habitat, food and shelter, and the habitat quality of these types of areas is relatively high.

Table 9
Final variable properties
Model	Unstandardized coefficients	Standard errors	Standardized coefficients	Significance	VIF
	-84.429	39.843		0.034
DEM	-0.001	0.002	-0.002	0.975	1.025
Karst landform	0.142	0.818	0.001	0.953	1.05
Soil type	-2.227	1.021	-0.011	0.881	1.135
Slope	1.167	1.335	0.004	0.86	1.162
Soil layer thickness	2.221	1.831	0.007	0.701	1.427
Comprehensive coverage of vegetation	0.077	0.125	0.005	0.362	2.761
The exposure rate of bedrock	-0.276	0.191	-0.012	0.315	3.172
Rocky desertification type	7.581	4.008	0.018	0.237	4.212

3.4.3 Spatial non-equilibrium analysis

The GWR model can reveal the non-stationary spatial characteristics of local coefficient variation, and in this study, the geographically weighted regression analysis of natural geographic factors and HQ in 2020 was conducted with the township as a unit, and the selection criteria of explanatory variables used were the same as those used in the multiple linear stepwise regression analysis, including the DEM, karst geomorphology, soil type, slope, soil thickness, integrated vegetation cover, bedrock exposure rate and rocky desertification degree, and its model fit goodness-of-fit was 0.812, indicating that GWR had a better fit. A positive CV indicates that a certain natural geographic factor in the region has a positive promoting effect on HQ, and a negative CV indicates a disturbing inhibiting effect. The spatial heterogeneity is determined by the local coefficient variation (CV) of the corresponding indicator, and the larger the absolute value of CV, the stronger the facilitating or inhibiting effect, and the smaller the opposite.

Figure 9 reveals the complex spatial heterogeneity relationship between habitat quality (HQ) and natural geographic factors in Guizhou Province. It was found that the effects of different natural factors on HQ showed significant regional differentiation: in non-karst areas, thicker soil layers and higher vegetation cover had a significant positive effect on HQ, while karst plateau areas showed a significant inhibitory effect due to high bedrock exposure rates and steep-slope topography. However, the same factor may have very different effects in different regions, e.g., slope promotes HQ enhancement in the southern gently sloping area, while exacerbates ecological degradation in the northern steeply sloping area. The interaction between the intensity of human activities and natural conditions is more prominent. In the ecologically fragile karst trough valleys and valleys, human disturbances amplify the negative effects of DEM, while in the non-karst areas with better ecological background, the positive effects of vegetation can partially offset some of the urban topographic pressures. The study also identified three types of typical response zones: karst plateau urban agglomerations with “high-low” characteristics, which are the most severely degraded areas; crested depressions with relatively stable “low-high” status by virtue of topographic closure; and non-karst areas with “H-H” transition characteristics. These findings indicate that the changes in habitat quality in karst areas are the result of the joint action of natural background conditions and human activities, and this action has significant spatial heterogeneity, which provides an important basis for the development of differentiated regional ecological protection strategies.

Fig. 9

Spatial distribution of the regression coefficients in GWR

4 Discussion

4.1 Implications of landscape pattern governance for improving habitat quality

Spatial changes in habitat quality are influenced to some extent by landscape pattern (Zhu et al., 2020; Guo et al., 2023). In this study, we found a significant negative correlation (r=-0.49, p < 0.01) between landscape fragmentation (PD: 22.86→20.20) and the decline of habitat quality (average annual decline of 0.14%) in Guizhou Province during the period of 2000–2020 through quantitative analysis, indicating that landscape fragmentation is an important factor contributing to the decline of habitat quality (Evans et al., 2017) .

The study area in which this research was conducted is located in the Karst region of Southwest China, where the ecological environment is fragile and vulnerable to anthropogenic disturbances and natural disasters (Xie et al., 2023). Human activities exacerbate habitat degradation mainly through two pathways: direct coercion and indirect weakening of ecological functions. Land development due to accelerated urbanization and the destruction of forests and grasslands by agricultural development in Guizhou Province have directly led to ecological land fragmentation (Li et al., 2024; Xiang et al., 2025), a process that has resulted in the alteration of original natural landscapes, habitat fragmentation, reduction of habitat connectivity, and reduction of biodiversity, lowering the quality of the habitats, e.g., in 2010–2020, 308.97 km² of arable land in Guizhou Province was converted to construction land, which directly led to a 5.47% decrease in HQ, demonstrating that urbanization indirectly reduces ecological connectivity and negatively affects habitat quality through landscape fragmentation (PD decrease of 22.86 → 20.20); the higher degree of landscape fragmentation and habitat degradation in urban agglomerations in karst plateau zones also suggests that over-exploitation of land often seriously affects the spatial and temporal changes of LSPI and HQ (Zhang et al., 2023; Hu et al., 2023; Ma et al., 2025). In addition, uncontrolled development of agricultural land destroys the ecology of vegetation and leads to extensive vegetation degradation, which may lead to fragmentation and habitat loss and further reduce habitat quality (Dai et al., 2019). For example, basins in the faulted basin area are usually important for agriculture and habitation, but their ecosystems may be strongly affected by human activities, making them ecologically fragile. Vegetation cover is also a key factor influencing habitat quality, and its changes are influenced by a combination of natural and anthropogenic factors and show different degrees of influence and characteristics in different regions ( Ma et al., 2020; Shi et al., 2023; Ma et al., 2025). karst trough valleys and karst plateaus have thin soil layers, low vegetation cover, fragile ecosystems, and weak resistance to disturbance, whereas non-karst areas have thicker soil layers, which are favorable for vegetation growth, diverse vegetation types, and high biodiversity.

Reduced landscape connectivity (AI: 88.30 → 88.74) can impede biological migration, as ecological isolation from canyon closure has been shown to affect habitat quality in multiple ways (Fernandez-Arcaya et al., 2017; Chen et al., 2025), including vegetation changes, increased sensitivity of ecosystems to climate change, and intensified human activities, for example, karst canyon areas and peak-tufted depressions have poor habitat quality due to the reduction of species habitats caused by poor ecological corridor connectivity, which is in line with past studies that have demonstrated the importance of depressions as ecological corridors for maintaining ecological balance, connecting different ecosystems, improving landscape connectivity, and conserving biodiversity (Ding et al., 2023; Jin et al., 2024).

At the same time, the implementation of environmental protection measures will help mitigate landscape fragmentation and promote ecosystem services and functions, thus improving habitat quality (Zou et al., 2022). For example, the promotion of “returning farmland to forests” is conducive to the restoration of ecological land, reducing the extent of landscape fragmentation and improving habitat quality. However, the effect of the policy is constrained by the regional ecological background, and the karst plateau area in this region needs to be combined with soil restoration and development control because of the seriousness of rocky desertification. The in-depth implementation of large-scale land greening, wetland and river and lake protection and restoration, soil and water conservation, biodiversity conservation, comprehensive land improvement and other major ecological protection and restoration projects in Guizhou Province in recent years has also been of great significance to ecological environmental protection, enhancing ecosystem service functions and ensuring ecosystem health and sustainability.

4.2 Drivers of spatial and temporal variation in habitat quality in Guizhou Province

The spatial heterogeneity of regional natural resources and environments create initial spatial patterns of habitat quality. In addition, spatial variations in habitat quality have become increasingly significant under the combined effects of natural and anthropogenic factors (Ma et al., 2025; Chen et al., 2023;Kou et al., 2025). In this study, we focused on the effects of DEM, karst geomorphology, soil type, slope, soil thickness, integrated vegetation cover, bedrock exposure rate, and rocky desertification degree on the habitat quality of areas in Guizhou Province, and the analysis of the results indicated that the degradation of the habitat quality in Guizhou Province is the result of the nonlinear coupling of natural vulnerability and human coercion.

Slope (q = 0.32), vegetation cover (q = 0.28) and soil thickness (q = 0.21) were the three core natural drivers of HQ spatial differentiation. Among them, karst plateau areas with slopes > 25° had lower HQ values than gently sloping non-karst areas due to increased soil erosion; while vegetation cover increased HQ accordingly.From 2010 to 2020, the expansion of construction land directly led to a decrease in HQ by 5.47% through encroachment of ecological land and exacerbation of landscape fragmentation (PD: 22.86→20.20). Meanwhile, in karst trough valleys and valleys with > 40% exposed bedrock, the expansion of construction land increased the rate of HQ degradation, highlighting the superimposed effect of “natural constraints + anthropogenic interference”. The main reason for the decline of HQ in non-karst areas is agricultural intensification, while rocky desertification and development of steep slopes dominate in karst plateau areas, and topography dominates the change of HQ in peaked depressions. 2000–2010, natural factors explained 72% of the change of HQ, and the contribution rate of human activities increased to 65% in 2010–2020, reflecting the change of driving mechanism from “natural-dominated” to “human-induced”.

4.3 Suggestions for future ecological improvement in Guizhou Province

Guizhou has a fragile ecosystem base, serious rocky desertification, and insufficient environmental protection infrastructure, so it is especially important to harmonize the relationship between conservation and development, and adhere to ecological priority and green development. Based on the results of this study, the following measures and recommendations are proposed to protect the habitat quality in similar areas.

Guizhou Province should continue to adhere to the principle of ecological priority as the core orientation of regional development. This means that when formulating policies, plans and development projects, ecological impacts should be considered first to ensure that the ecological environment is not jeopardized. For example, the increase in non-ecological land (arable land and construction land) needs to be controlled in order to minimize the impact of human activities on the ecological environment and promote a virtuous cycle of the ecosystem. At the same time, differentiated zoning management should be carried out to address the characteristics of ecological fragility in Guizhou Province, and appropriate strategies should be adopted for different areas to improve the stability and resilience of the ecosystem. For large areas of ecological land, corresponding graded protection zones and buffer zones should be delineated to protect the integrity of ecological land and reduce its fragmentation. For non-ecological land, especially high-density construction land, ecological planning should be emphasized to reduce the agglomeration of construction land and to combine natural recovery with artificial restoration. For example, these ecological land patches can be embedded through the establishment of urban parks and green corridors to increase connectivity between patches with high habitat quality, which will improve habitat quality. In addition, ecological land protection around construction land in relevant areas should be considered to control the blind expansion of construction land, reduce its damage to ecological land, and mitigate its negative impact on habitat quality (Xu et al., 2018).

4.4 Limitations and outlook

In this study, we quantified the spatial and temporal changes and regional differences of HQ and LSPI in Guizhou Province, and analyzed the correlations between HQ and LSPI. The influence of LSPI on HQ was explored at 1km*1km scale through spatial autocorrelation analysis. However, there are some limitations, firstly, the indirect effect of human activities on HQ may be underestimated due to the data limitation of not integrating multi-source data such as climate, socio-economics and policies. Second, this study did not construct a coupled nature-economy-policy model, which made it difficult to analyze the synergistic effects of multiple factors. Finally, although a zoning governance framework was proposed, the spatial threshold of the ecological protection red line and the restoration priority were not clearly defined. Future research should deepen the study of the practical significance of ecological protection in terms of multidisciplinary data integration and dynamic scenario simulation, provide support for policy transformation based on the results of the study, and propose specific areas where the ecological protection red line or nature reserve should be delineated, so as to provide literature and data support for the formulation of ecological environmental protection and ecological functional zoning policies.

5 Conclusion

Based on the land use data of Guizhou Province, this study uses HQ and LSPI to quantify the regional differences of different natural geographic environments in Guizhou Province, calculated by the InVEST model. The effect of LSPI on HQ was explored at a scale of 1km*1km through multiple linear regression, spatial autocorrelation analysis and geographically weighted regression methods.

It was found that between 2000 and 2020, the landscape types in Guizhou Province were dominated by woodland and cropland, with woodland accounting for more than 48% and cropland accounting for more than 46% of the study area. The changes in the area of each landscape type in Guizhou Province on the time scale were mainly the transformation of cropland and woodland, while the transformation of land use types to construction land was more obvious in urban agglomerations, especially in the urban agglomerations of the karst plateau.The LSPI was affected by human activities and land use types, and in the past two decades, the AI increased from 88.3015 to 88.7391, and the COHESION increased from 96.6891 to 96.7856, LPI increased from 58.3044 to 66.9573, CONTAG decreased from 56.5393 to 55.9076, SHDI value decreased from 3.5272 to 3.5125, PD decreased from 22.8563 to 20.2019, LSI decreased from 3.0606 to 2.9968. SPLIT decreased from 2.3329 to 2.3024.Overall, it shows that the land use intensity in Guizhou province has increased in the past two decades, the land use structure tends to stabilize, and the urban and agricultural expansion brought by human activities will increase the degree of landscape fragmentation and weaken the connectivity of the landscape.The spatial distribution of HQ varies significantly, which is greatly influenced by the type of land use, and the distribution of higher HQ values is the most extensive, accounting for as much as 52.62% of the total. up to 52.62%, and low HQ value distribution accounted for about 1.39%, which commonly appeared in urban agglomerations in the karst plateau. On the time scale, with the rapid urbanization development in some areas of Guizhou Province, the core area of habitat degradation expanded from the urban agglomeration of the karst plateau to the karst trough valleys and valleys, and there was a certain risk of habitat degradation.

There are different correlations between LSPI and HQ, and they show strong local spatial autocorrelation. From a global perspective, AI, COHESION, CONTAG and LPI were positively correlated with habitat quality, while LSI, PD, SHDI and SPLIT were negatively correlated with HQ. From a local perspective, considering the influence of natural geographic features on the correlation between LSPI and HQ, Guizhou Province was divided into six regions: non-karst areas, karst trough valleys, karst plateaus, karst canyons, peak-tufted depressions, and faulted basins. The dominant cluster types corresponding to the same LSPI are consistent in the same regions. “H-H” and “H-L” types are interspersed in non-karst areas and karst trough valleys, “L-H” and “L-L” types are interspersed in karst trough valleys and karst plateaus, and “H-L” and “L-H” types are interspersed in karst canyons, peak-tufted depressions and faulted basins. Habitat quality degradation is the result of a non-linear combination of natural vulnerability and human activities. Slope, vegetation cover and soil thickness are the core natural drivers, while the expansion of built-up land directly contributes to the decline of HQ by increasing landscape fragmentation. 2010 onwards, the contribution of human activities increased to 65%, and the driving mechanism shifted from “nature-led” to “human-land interaction”. The results of this study are of great significance to the realization of ecological environment management and regulation in Guizhou Province, and can provide literature and data support for the formulation of ecological environment protection and ecological function zoning policies.

Additional Files

Additional file 1

References

Antrop M (2004) Landscape change and the urbanization process in Europe. Landsc urban Plann 67(1–4):9–26. https://doi.org/10.1016/s0169-2046(03)00026-4

Chasco C (2025) Basic Spatial Exploratory Spatial Data Analysis in GeoDa Software. https://doi.org/10.20944/preprints202503.2044.v2

Chen X, Xiong J, Guan Y, Zhou J (2025) Ecological Network Construction and Optimization in the Southwest Alpine Canyon Area of China Based on Habitat Quality Assessment. Remote Sensing, 17(11), 1913. https://doi.org/10.3390/rs17111913

Chen Y, He Z, Yue T, Mu W, Qin F (2023) Spatiotemporal evolution and multi-scenario prediction of habitat quality in the Yellow River Basin. Front Ecol Evol 11:1226676. https://doi.org/10.3389/fevo.2023.1226676

Ding G, Yi D, Yi J, Guo J, Ou M, Ou W, Pueppke SG (2023) Protecting and constructing ecological corridors for biodiversity conservation: A framework that integrates landscape similarity assessment. Appl Geogr 160:103098. https://doi.org/10.1016/j.apgeog.2023.103098

Evans MJ, Banks SC, Driscoll DA, Hicks AJ, Melbourne BA, Davies KF (2017) Short-and long‐term effects of habitat fragmentation differ but are predicted by response to the matrix. Ecology 98(3):807–819. https://doi.org/10.1002/ecy.1704

Fearon ML, Wood CL, Tibbetts EA (2023) Habitat quality influences pollinator pathogen prevalence through both habitat–disease and biodiversity–disease pathways. Ecology 104(2):e3933. .https://doi.org/10.1002/ecy.3933

Fernandez-Arcaya U, Ramirez-Llodra E, Aguzzi J, Allcock AL, Davies JS, Dissanayake A, Van den Beld IM (2017) Ecological role of submarine canyons and need for canyon conservation: a review. Front Mar Sci 4:5. https://doi.org/10.3389/fmars.2017.00005

Gao SJ, Mei CL, Xu QX, Zhang Z (2023) Non-Iterative Multiscale Estimation for Spatial Autoregressive Geographically Weighted Regression Models. Entropy 25(2):320. https://doi.org/10.3390/e25020320

Guizhou Daily (2024), November 21 Significant achievements in the implementation of the Grand Ecological Strategy in Guizhou. Guizhou Daily. https://gzszx.gov.cn/gzzxb/web/doc/detail/d_1647243742412832

Guo R, Lyu S, Feng J (2023) The Effects of Urban Landscape Pattern Evolution on Habitat Quality Based on InVEST Model A Case of Harbin City. Int Rev Spat Plann Sustainable Dev 11(2):5–25. https://doi.org/10.14246/irspsd.11.2_5

He Y, Mo Y, Ma J (2022) Spatio-temporal evolution and influence mechanism of habitat quality in guilin city, China. Int J Environ Res Public Health 20(1):748. https://doi.org/10.3390/ijerph20010748

Hu C, Wu W, Zhou X, Wang Z (2023) Spatiotemporal changes in landscape patterns in karst mountainous regions based on the optimal landscape scale: A case study of Guiyang City in Guizhou Province, China. Ecol Ind 150:110211. https://doi.org/10.1016/j.ecolind.2023.110211

Imdadullah M, Aslam M, Altaf S (2016) mctest: An R package for detection of collinearity among regressors. https://digitalcommons.unl.edu/r-journal

Investigation Report on the (2021), July 15 Green Counterattack in Guizhou Province—From the National Ecological Civilization Pilot Zone (Guizhou). Chinese Government Website. http://xw.zgzx.com.cn/2021-07/15/content_9844403.htm

Jia Y, Tang L, Xu M, Yang X (2019) Landscape pattern indices for evaluating urban spatial morphology–A case study of Chinese cities. Ecol Ind 99:27–37. https://doi.org/10.1016/j.ecolind.2018.12.007

Jin B, Geng J, Ding Z, Guo L, Rui Q, Wu J, Ding G (2024) Construction and optimization of ecological corridors in coastal cities based on the perspective of structure-function. Sci Rep 14(1):27945. https://doi.org/10.1038/s41598-024-79433-9

Kong X, Fu M, Zhao X, Wang J, Jiang P (2022) Ecological effects of land-use change on two sides of the Hu Huanyong Line in China. Land Use Policy 113:105895. https://doi.org/10.1016/j.landusepol.2021.105895

Kou J, Liu J, Wang H, Huang S, Zhou Z, Yang C, Huang X (2025) Spatio-temporal evolution in habitat quality and its driving factors in Han River Basin, China. Scientific Reports, 15(1), 26139. | https://doi.org/10.1038/s41598-025-10514-z

Kroll CN, Song P (2013) Impact of multicollinearity on small sample hydrologic regression models. Water Resour Res 49(6):3756–3769. https://doi.org/10.1002/wrcr.20315

Li L, Li X, Niu B, Zhang Z (2023) A study on the dynamics of landscape patterns in the Yellow River Delta Region. Water 15(4):819. https://doi.org/10.3390/w15040819

Li Y, Feng X, Wu LH, Luo GJ, Luo HF (2024) Spatial-temporal evolution and quantitative attribution of habitat quality in typical Karst Counties of Guizhou Plateau. Huan Jing ke Xue = Huanjing Kexue 45(5):2793–2805. https://doi.org/10.13227/j.hjkx.202306238

Liang J, Pan S, Chen W, Li J, Zhou T (2022) Cultivated land fragmentation and its influencing factors detection: A case study in Huaihe River Basin, China. Int J Environ Res Public Health 19(1):138. https://doi.org/10.3390/ijerph19010138

Liao Y, Wu G, Zhang Z (2023) Multi-scale remote sensing assessment of ecological environment quality and its driving factors in watersheds: a case study of Huashan Creek watershed in China. Remote Sens 15(24):5633. https://doi.org/10.3390/rs15245

Lin J, Zhu C, Deng A, Zhang Y, Yuan H, Liu Y, Chen W (2022) Causes of changing woodland landscape patterns in southern China. Forests 13(12):2183. https://doi.org/10.3390/f13122183

Liu C, Yang M, Hou Y, Xue X (2021) Ecosystem service multifunctionality assessment and coupling coordination analysis with land use and land cover change in China's coastal zones. Sci Total Environ 797:149033. https://doi.org/10.1016/j.scitotenv.2021.149033

Liu F, Qin T, Girma A, Wang H, Weng B, Yu Z, Wang Z (2018) Dynamics of land-use and vegetation change using NDVI and transfer matrix: A case study of the Huaihe River Basin. Pol J Environ Stud 28(1):213–223. https://doi.org/10.15244/pjoes/82900

Liu L, Chen X, Chen W, Ye X (2020) Identifying the impact of landscape pattern on ecosystem services in the Middle Reaches of the Yangtze River Urban Agglomerations, China. Int J Environ Res Public Health 17(14):5063. https://doi.org/10.3390/ijerph17145063

Liu Y, Wang S, Chen Z, Tu S (2022) Research on the response of ecosystem service function to landscape pattern changes caused by land use transition: A case study of the Guangxi Zhuang Autonomous Region, China. Land 11(5):752. https://doi.org/10.3390/land11050752

Luo D, Zhou Z, Zhang L, Chen Q, Huang D, Feng Q, Wu L (2024) Evolution and driver analysis of forest carbon stocks in karst mountainous areas of southwest China in the context of rocky desertification management. CATENA 246:108335. https://doi.org/10.1016/j.catena.2024.108335

Ma C, Yang H, Yan Z (2025) Spatio-temporal evolution of habitat quality and its influencing factors in karst areas based on the InVEST model. PLoS ONE 20(3):e0314161. https://doi.org/10.1371/journal.pone.0314161

Ma ZC, Yu HB, Cao CM, Zhang QF, Hou LL, Liu YX (2020) Spatiotemporal characteristics of fractional vegetation coverage and its influencing factors in China. Resour Environ Yangtze Basin 29(6):1310–1321. https://doi.org/10.11870/cjlyzyyhj202006006

Mitchell MG, Suarez-Castro AF, Martinez-Harms M, Maron M, McAlpine C, Gaston KJ, Rhodes JR (2015) Reframing landscape fragmentation's effects on ecosystem services. Trends Ecol Evol 30(4):190–198. https://doi.org/10.1016/j.tree.2015.01.011

Priadka P, Brown GS, DeWitt PD, Mallory FF (2022) Habitat quality mediates demographic response to climate in a declining large herbivore. Basic Appl Ecol 58:50–63. https://doi.org/10.1016/j.baae.2021.11.005

Rehitha TV, Vineetha G, Madhu NV (2022) Ecological habitat quality assessment of a tropical estuary using macrobenthic functional characteristics and biotic indices. Environ Sci Pollut Res 29(31):47629–47646. https://doi.org/10.1007/s11356-022-19295-8

Seto KC, Güneralp B, Hutyra LR (2012) Global forecasts of urban expansion to 2030 and direct impacts on biodiversity and carbon pools. Proceedings of the National Academy of Sciences, 109(40), 16083–16088.https://doi.org/10.1073/pnas.1211658109

Shi N, Yang H, Wang Q, Han Y, Yan H, Xiao N (2023) Spatiotemporal variation of habitat quality and its response to fractional vegetation cover change and human disturbance in the Loess Plateau. Front Ecol Evol 11:1302224. https://doi.org/10.3389/fevo.2023.1302224

Smith JB, Keiter DA, Sweeney SJ, Miller RS, Schlichting PE, Beasley JC (2023) Habitat quality influences trade-offs in animal movement along the exploration–exploitation continuum. Sci Rep 13(1):4814. https://doi.org/10.1038/s41598-023-31457-3

Song XP, Hansen MC, Stehman SV, Potapov PV, Tyukavina A, Vermote EF, Townshend JR (2018) Global land change from 1982 to 2016. Nature 560(7720):639–643. https://doi.org/10.1038/s41586-018-0411-9

Tang S, Song L, Wan S, Wang Y, Jiang Y, Liao J (2022) Long-time-series evolution and ecological effects of coastline length in coastal zone: A case study of the circum-bohai coastal zone, china. Land 11(8):1291. https://doi.org/10.3390/land11081291

Tian A, Xu T, Gao J, Liu C, Han L (2023) Multi-scale spatiotemporal wetland loss and its critical influencing factors in China determined using innovative grid-based GWR. Ecol Ind 149:110144. https://doi.org/10.1016/j.ecolind.2023.110144

Tobisch C, Rojas-Botero S, Uhler J, Kollmann J, Müller J, Moning C, Ewald J (2023) Conservation-relevant plant species indicate arthropod richness across trophic levels: Habitat quality is more important than habitat amount. Ecol Ind 148:110039. https://doi.org/10.1016/j.ecolind.2023.110039

Wang H, Qin F, Zhu J, Zhang C (2017) The effects of land use structure and landscape pattern change on ecosystem service values. Acta Ecol Sin 37:1286–1296. https://doi.org/10.5846/stxb201606291295

Wang H, Yuan W, Ma Y, Bai X, Huang L, Cheng S, Guo W (2023) Spatiotemporal dislocation of ecosystem supply and demand services from habitat quality under different development scenarios. Ecol Ind 157:111230. https://doi.org/10.1016/j.ecolind.2023.111230

Wang S, Lu F, Wei G (2022) Direct and spillover effects of urban land expansion on habitat quality in chengdu-chongqing urban agglomeration. Sustainability 14(22):14931. https://doi.org/10.3390/su142214931

Wen JH, Li R (2022) Temporal and spatial variation of ecosystem service value and its response to landscape pattern change in Guizhou Province, China. Ying Yong Sheng tai xue bao = The. J Appl Ecol 33(11):3075–3086. https://doi.org/10.13287/j.1001-9332.202211.020

Xiang J, Zhang W, Song X, Li J (2019) Impacts of precipitation and temperature on changes in the terrestrial ecosystem pattern in the Yangtze River Economic Belt, China. Int J Environ Res Public Health 16(23):4872. https://doi.org/10.3390/ijerph16234872

Xiang Q, Kan A, Yu X, Liu F, Huang H, Li W, Gao R (2023) Assessment of topographic effect on habitat quality in mountainous area using InVEST model. Land 12(1):186. .https://doi.org/10.3390/land12010186

Xiang S, Xiong K, Zhang B, Li Y, Zhang W, Li R (2025) Spatiotemporal dynamics and driving mechanisms of grassland landscape fragmentation in China’s karst regions. Ecol Ind 179:114185. https://doi.org/10.1016/j.ecolind.2025.114185

Xie B, Zhang M (2023) Spatio-temporal evolution and driving forces of habitat quality in Guizhou Province. Sci Rep 13(1):6908. https://doi.org/10.1038/s41598-023-33903-8

Xie B, Meng S, Zhang M (2022) Evolution of habitat quality and its response to topographic gradient effect in a Karst Plateau: a case study of the Key Biodiversity Conservation Project Area of Wuling Mountains. Int J Environ Res Public Health 20(1):331. https://doi.org/10.3390/ijerph20010331

Xu C, Li C (2023) Correlation between Landscape Pattern and Net Primary Productivity of Vegetation: A Case Study in the Arid and Semi-Arid Regions of Northwest China. Land, 12(11), 2004. https://doi.org/10.3390/land12112004

Xu X, Tan Y, Yang G, Barnett J (2018) China’s ambitious ecological red lines. Land Use Policy 79:447–451. https://doi.org/10.1016/j.landusepol.2018.08.037

Xue S, Ma B, Wang C, Li Z (2023) Identifying key landscape pattern indices influencing the NPP: A case study of the upper and middle reaches of the Yellow River. Ecol Model 484:110457. https://doi.org/10.1016/j.ecolmodel.2023.110457

Yang F, Yang L, Fang Q, Yao X (2024) Impact of landscape pattern on habitat quality in the Yangtze River Economic Belt from 2000 to 2030. Ecol Ind 166:112480. https://doi.org/10.1016/j.ecolind.2024.112480

Yang X, Zheng XQ, Chen R (2014) A land use change model: Integrating landscape pattern indexes and Markov-CA. Ecol Model 283:1–7. https://doi.org/10.1016/j.ecolmodel.2014.03.011

Yohannes H, Soromessa T, Argaw M, Dewan A (2021) Spatio-temporal changes in habitat quality and linkage with landscape characteristics in the Beressa watershed, Blue Nile basin of Ethiopian highlands. J Environ Manage 281:111885. https://doi.org/10.1016/j.jenvman.2020.111885

Yuan W, Chen L, Chen H, Deng S, Ji H, Liang F (2023) Assessing habitat quality at Poyang Lake based on InVEST and Geodetector modeling. Ecol Evol 13(12):e10759. https://doi.org/10.1002/ece3.10759

Zhang S, Wang Y, Li C, Wu Y, Yin Y, Zhang C (2023) The response of rocky desertification to the development of road networks in karst ecologically fragile areas. Int J Environ Res Public Health 20(4). 3130.https://doi.org/10.3390/ijerph20043130

Zhang T, Chen Y (2022) The effects of landscape change on habitat quality in arid desert areas based on future scenarios: Tarim River Basin as a case study. Front Plant Sci 13:1031859. https://doi.org/10.3389/fpls.2022.1031859

Zhang Y, Li R, Jing J (2022) Soil erosion gradient and quantitative attribution in southwest China based on karst development degree. Ecol Ind 144:109496. https://doi.org/10.1016/j.ecolind.2022.109496

Zheng L, Wang Y, Li J (2023) Quantifying the spatial impact of landscape fragmentation on habitat quality: A multi-temporal dimensional comparison between the Yangtze River Economic Belt and Yellow River Basin of China. Land Use Policy 125:106463. https://doi.org/10.1016/j.landusepol.2022.106463

Zhu C, Zhang X, Zhou M, He S, Gan M, Yang L, Wang K (2020) Impacts of urbanization and landscape pattern on habitat quality using OLS and GWR models in Hangzhou, China. Ecol Ind 117:106654. https://doi.org/10.1016/j.ecolind.2020.106654

Zhu S, Li L, Wu G, Liu J, Slate TJ, Guo H, Li D (2022) Assessing the Impact of Village Development on the Habitat Quality of Yunnan Snub-Nosed Monkeys Using the INVEST Model. Biology, 11 (10): 1487. https://doi.org/10.3390/biology11101487

Zou L, Wang J, Bai M (2022) Assessing spatial–temporal heterogeneity of China’s landscape fragmentation in 1980–2020. Ecol Ind 136:108654. https://doi.org/10.1016/j.ecolind.2022.108654

Author StatementFunding: This study is financially supported by the National Natural Science Foundation of China(32560394) Guizhou Provincial Science and Technology Program (Qian Ke He Ping Tai YWZ [2025]001); Guizhou Provincial 2025 Central Government Guided Local Science and Technology Development Fund Project(Qian Ke He zhong Yin Di [2025]031); Guizhou Provincial Key Laboratory Construction Project, (Qian Ke He Ping Tai [2025]014)

Competing Interests:

The authors declare that there are no relevant financial or non-financial competing interests to disclose.

Author Contributions:

Conceptualization, P.Y., Z.B., Z.Z., and H.Z.; Data curation, P.Y. and Z.B.; Formal analysis, P.Y., Z.B., Z.Z., and H.Z.; Funding acquisition, P.Y.; Investigation, P.Y. and Z.B.; Methodology, P.Y. and Z.B.; Project administration, P.Y. and Z.Z.; Resources, P.Y.; Software, P.Y., Z.B., and H.Z.; Supervision, Z.Z.; Validation, P.Y., Z.B., and H.Z.; Visualization, P.Y. and Z.B.; Writing – original draft, P.Y. and Z.B.; Writing – review & editing, P.Y. and Z.B. All authors have read and agreed to the published version of the manuscript.

Table.1 Data sources

Type	Name	Units	Data Source
Basic Data	Guizhou Province Land Use Data	/	Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences
	Guizhou Province Land Use Data	/	(https://www.resdc.cn/Introduction.aspx)
	Administrative Boundaries of Guizhou Province	/	Administrative Division Data of China _ Approval Number :GS(2024)0650
	Administrative Boundaries of Guizhou Province	/	(https://cloudcenter.tianditu.gov.cn/administrativeDivision/)
Natural Drivers	Digital Elevation Model (DEM)	m	Geospatial data cloud
	Digital Elevation Model (DEM)	m	(http://www.gscloud.cn)
	Slope	°	Extraction based on DEM
	Soil Type	/	Resources and Environmental Science and Data Center, Chinese Academy of Sciences
	Soil Type	/	(https://www.resdc.cn/Default.aspx)
	Karst landform	/	Guizhou rocky desertification control center
	Soil layer thickness	/	Guizhou rocky desertification control center
	Comprehensive coverage of vegetation	/	Guizhou rocky desertification control center
	The exposure rate of bedrock	/	Guizhou rocky desertification control center
	Rocky desertification type	/	Guizhou rocky desertification control center

Table.2 Chosen landscape pattern index

Table 3 Threat factor weighting table

Category	Index	Description	Formula	Annotation
Landscape area index	Maximum Patch Index (LPI)	The higher the dominant degree of the largest patch in the landscape, the better the habitat quality	$\:\text{LPI}=\frac{\text{MAX}\left({a}_{ij}\right)}{{A}_{}}\times\:100\%$	a_ij = area of patch_ij (m²); A = Total landscape area (m²)
Landscape quantity index	Patch density (PD)	Number of patches per 100 hectares of landscape	$\:\text{PD}=NP/A$	NP = the total number of patches in the landscape; A = Total landscape area (m²)
Landscape shape index	Landscape Shape Index (LSI)	It mainly reflects the complexity of patch shape in landscape.	$\:\text{LSI}=\frac{\text{0.25}L}{\sqrt{{A}_{}}}$	A = Total landscape area (m²)
Landscape aggregation index	Aggregation index(AI)	The higher the degree of landscape fragmentation, the better the habitat quality	$\:AI=\frac{1}{2}\left[1-\frac{{\sum\:}_{i=1}^{m}{\sum\:}_{j=1}^{m}\frac{{g}_{ij}}{{max}\left({g}_{ij}\right)}{\left(1-{p}_{ii}\right)}^{2}}{{\left(1-\frac{1}{m}\right)}^{2}}\right]$	m = the total number of patch types in the landscape; g_ij =the total boundary length between patch type i and patch type j; max(g_ij) = the maximum possible boundary length between all pairs of patch types; p_ii =the proportion of the internal boundary of patch type i.
Landscape connectivity index	Cohesion index (COHESION)	The stronger the connectivity within the landscape, the worse the habitat quality.		a_ij = the area of patch j in Class i landscape; P_ij is the perimeter (m) of patch j in a Class i landscape.
Landscape spread index	Spread index (CONTAG)	It can describe the agglomeration degree or extension trend of patch types in the landscape, including spatial information.	$\:CONTAG={\sum\:}_{i=1}^{m}{\sum\:}_{j=1}^{m}\frac{{g}_{ij}\times\:\left({\pi\:}_{ij}+{\pi\:}_{ji}\right)}{2\times\:{m}^{2}}$	m = the total number of patch types in the landscape; g_ij = the adjacency probability between plaque type i and plaque type j; π_ij is the frequency adjacent to plaque type i and plaque type j; π_ji is the frequency adjacent to patch type j and patch type i.
Landscape diversity index	Patch Richness Density (PRD)	It refers to the richness of patch types (or land cover types) per unit area, which can reflect the diversity and heterogeneity of the landscape	$\:\text{PRD}=R/A$	R = the total number of plaque types in the study area; A = Total landscape area (m²)
Landscape fragmentation index	Fission index (SPLIT)	This metric evaluates both the spatial distribution pattern of the specified patch type and its isolation from other patch types in the landscape.	$\:SPLIT=\frac{1}{2}{\sum\:}_{i=1}^{m}\frac{{g}_{ij}}{{A}_{i}}$	m = the total number of patch types in the landscape; gij = the adjacency probability between plaque type i and plaque type j;

Table 4 Threat factor sensitivity table

Threat factor	Maximum influence distance/km	weight	Decaying linear dependence
Plowland	1	0.7	linearity
Urban construction land	8	1	exponent
Rural residential land	6	0.8	exponent
Other construction land	4	0.4	linearity
Unutilized	4	0.4	linearity

Table 5 The dynamic rate of change of different LULC classes in the study area from 2000 to 2020.

Land class types	Habitat suitability	plowland	Urban construction land	Rural residential land	Other construction land	unutilized
Paddy field	0.4	0	0.7	0.6	0.6	0.5
Dry land	0.3	0	0.7	0.6	0.6	0.5
Forest land	1	0.7	0.8	0.3	0.3	0.6
shrubbery	0.8	0.8	0.6	0.4	0.4	0.6
Open forest land	0.7	0.7	0.7	0.5	0.5	0.6
Basal forest land	0.6	0.6	0.7	0.5	0.5	0.6
High cover grassland	0.8	0.8	0.5	0.7	0.7	0.5
Medium coverage grassland	0.7	0.7	0.7	0.5	0.5	0.3
Low cover grassland	0.6	0.5	0.7	0.5	0.5	0.4
River and canal	0.8	0.8	0.6	0.6	0.8	0.3
lakes	0.9	0.7	0.7	0.7	0.4	0.5
Reservoir pit	0.6	0.7	0.6	0.7	0.5	0.5
Bottom land	0.6	0.6	0.7	0.3	0.5	0.5
Urban land	0	0	0	0	0	0
Rural residential area	0	0	0	0	0	0
Other construction land	0	0	0	0	0	0
Special land	0	0	0	0	0	0
Marshland	0.1	0.1	0.1	0.1	0.1	0.1
Bare land	0.1	0.1	0.1	0.1	0.1	0.1
Bare rock stony land	0.1	0.1	0.1	0.1	0.1	0.1

Table 6 Changes in Area and Percentage of Habitat Quality Classes in the study area from 2000 to 2020.

LULC type	2000–2010	2010–2020	2000–2020
Cropland	1.87%	-1.43%	0.44%
Forest	-0.96%	2.68%	1.72%
Shrub	-0.54%	-1.36%	-1.90%
Grassland	-0.61%	-0.35%	-0.96%
Water	0.08%	0.06%	0.15%
Barren	0.00%	0.00%	0.00%
Impervious	0.16%	0.39%	0.55%

Table 7 Global Moran's I values for landscape pattern index and habitat quality in the study area from 2000 to 2020

Habitat quality level	2000		2010		2020
	Area	Percentage	Area	Percentage	Area	Percentage
	km²	%	km²	%	km²	%
Ⅰ	639.65	0.36%	904.07	0.51%	2445.09	1.39%
Ⅱ	49624.28	28.18%	49401.70	28.05%	48268.09	27.41%
Ⅲ	5251.61	2.98%	5285.86	3.00%	6030.39	3.42%
Ⅳ	96911.99	55.03%	95863.48	54.44%	92667.65	52.62%
Ⅴ	23674.00	13.44%	24648.07	14.00%	26680.11	15.15%

Table 8 Statistics of the number and percentage of agglomeration-type patches in 2020

Year	Value type	AI&HQ	COHESION&HQ	CONTAG&HQ	LPI&HQ	LSI&HQ	PD&HQ	SHDI&HQ	SPLIT&HQ
2000	Moran′s I	0.451	0.336	0.394	0.415	-0.451	-0.435	-0.515	-0.280
2000	Z value	31.994	34.075	32.050	30.245	-31.930	-31.650	-37.258	-21.724
2010	Moran′s I	0.471	0.371	0.375	0.465	-0.476	-0.473	-0.476	-0.276
2010	Z value	32.019	32.098	32.935	31.492	-32.361	-32.500	-32.418	-22.361
2020	Moran′s I	0.485	0.385	0.385	0.476	-0.490	-0.488	-0.490	-0.290
2020	Z value	32.426	32.533	32.348	31.609	-32.881	-32.880	-32.915	-22.901

Table 9 Final variable properties

		Not significant	High-High	Low-Low	Low-High	High-Low
AI	Quantity block	86629	33517	17435	10637	9916
AI	Percentage	54.78%	21.20%	11.03%	6.73%	6.27%
COHESION	Quantity block	86627	34318	16243	10236	10710
COHESION	Percentage	54.78%	21.70%	10.27%	6.47%	6.77%
CONTAG	Quantity block	86629	20747	27068	12407	11283
CONTAG	Percentage	54.78%	13.12%	17.12%	7.85%	7.14%
LPI	Quantity block	88629	37041	15925	8513	8026
LPI	Percentage	56.05%	23.42%	10.07%	5.38%	5.08%
LSI	Quantity block	86622	8698	35783	13457	13574
LSI	Percentage	54.78%	5.50%	22.63%	8.51%	8.58%
PD	Quantity block	86624	7480	37432	14674	11924
PD	Percentage	54.78%	4.73%	23.67%	9.28%	7.54%
SHDI	Quantity block	86625	8274	36164	13880	13191
SHDI	Percentage	54.78%	5.23%	22.87%	8.78%	8.34%
SPLIT	Quantity block	86624	7264	38111	14891	11244
SPLIT	Percentage	54.78%	4.59%	24.10%	9.42%	7.11%

Model	Unstandardized coefficients	Standard errors	Standardized coefficients	Significance	VIF
	-84.429	39.843		0.034
DEM	-0.001	0.002	-0.002	0.975	1.025
Karst landform	0.142	0.818	0.001	0.953	1.05
Soil type	-2.227	1.021	-0.011	0.881	1.135
Slope	1.167	1.335	0.004	0.86	1.162
Soil layer thickness	2.221	1.831	0.007	0.701	1.427
Comprehensive coverage of vegetation	0.077	0.125	0.005	0.362	2.761
The exposure rate of bedrock	-0.276	0.191	-0.012	0.315	3.172
Rocky desertification type	7.581	4.008	0.018	0.237	4.212