Introduction

Forest land, as one of the most critical natural resources on Earth1, functions as a vital carbon sink1 and plays an indispensable role in mitigating climate change2, regulating climate3, maintaining ecological balance4,5, preserving biodiversity6, and providing essential ecosystem services7 such as water conservation, soil retention, and wind erosion control8. Such as the carbon sequestration and oxygen release functions of forests are particularly crucial for addressing climate change, with carbon storage per unit area in forest land being two to four times greater than that in open ecosystems such as croplands and grasslands9. Moreover, forest ecosystems exhibit higher aboveground biomass and biodiversity compared to other land cover types10. However, with economic development and social progress, the ecological space of forest land is increasingly encroached upon by construction, agricultural expansion, and other land use demands, resulting in large-scale forest fragmentation, severe soil erosion, and ecological degradation8,11.

Notably, changes in forest land have become a key indicator in land use/land cover change (LUCC) studies, providing critical insights into the impacts of anthropogenic activities on ecosystems12. Studies have shown that, from 1990 to 2020, global forest area has continued to decline, with a loss of approximately 178 million hectares of forest land. In response, many countries have successfully undergone forest transitions, leading to sustained growth in forest areas13. As China is currently in a phase of high demand for land resources, scientifically managing forest land resources is of paramount importance. Since the implementation of the Natural Forest Protection Program and the Grain for Green Project in China in 1999, forest area has shown an overall increasing trend nationwide14. However, forest areas in certain regions, particularly in ecologically fragile zones, continue to face pressures from deforestation and land degradation15. Consequently, understanding the spatio-temporal dynamics, distribution patterns, and driving factors of forest land is essential for effectively managing forest resources and achieving sustainable development.

Changes in forest land over time are characterized by gradual adjustments in forest area, while spatially they manifest as continuous evolution in landscape structures and patterns16. These changes are driven by a complex interaction between natural and socio-economic factors8,17. Natural factors, such as topography, geomorphology, and climate, govern the distribution and dynamics of forest land and form the fundamental basis for forest cover patterns18. These natural factors can be divided into stable factors (e.g., topography, geomorphology, slope, aspect) and inter-annual variability factors (e.g., temperature and precipitation)19. Specifically, topographic gradients—including elevation, slope, and aspect—exert considerable influence on forest land distribution18. The impact of the changes in climatic factors on the forest ecosystem is reflected in plant physiology, tree species migration, seed transmission, fire, and induced diseases and pests20,21. Natural factors exert cumulative effects over time, socio-economic factors, such as population growth, economic development, and policy interventions, tend to have more immediate and dynamic impacts on forest land changes22. These factors not only influence the expansion or contraction of forest resources but also shape the structural and functional integrity of ecosystems.

As one of China’s most important natural forest regions and ecological barrier zones, the Da-Xiao Liangshan Mountains play a pivotal role in the country’s ecological security framework23. The region’s complex terrain and diverse climatic conditions result in distinct topographic gradient differentiation in forest land distribution. Against the backdrop of resource consumption, rapid urbanization, and the simultaneous implementation of ecological conservation projects, forest land changes in localized areas are becoming increasingly complex, necessitating detailed and refined research. This study focuses on the Da-Xiao Liangshan Mountains and utilizes high-resolution land use and land cover data from 1980 to 2020. By applying kernel density analysis, topographic gradient classification, the PLUS model, landscape pattern indices, and standard deviational ellipse theory, this research investigates the spatio-temporal changes, topographic gradient differentiation, fragmentation, and driving factors of forest land in the study area. The results will shed light on the spatio-temporal dynamics of forest land in the region and clarify the relative contributions of natural and socio-economic factors to forest land changes. These findings will improve our understanding of forest transformation processes, provide insights into the drivers of forest development, and offer valuable data and theoretical guidance for addressing environmental challenges, adjusting regional forest management policies, and promoting the sustainable management and ecological restoration of forest land in the Da-Xiao Liangshan Mountains.

Materials and methods

Overview of the study area

The Da-Xiao Liangshan Mountains is located in the southwestern border region of China (Fig. 1), covering 23 counties (cities) in Sichuan Province’s Liangshan Yi Autonomous Prefecture, Leshan City, Yibin City, and Lijiang City in Yunnan Province, with a total area of approximately 79,359.61 km2. The region lies on the eastern edge of the Hengduan Mountains and serves as the transition zone between China’s first and second topographic steps23, with altitudes ranging from 252 to 5,902 m and a general topographic slope from northwest to southeast. The major mountain ranges include Xiaoxiangling Mountains, Daliang Mountains, Luoji Mountains, Maoniu Mountains, Lunan Mountains, and Jinping Mountains, characterized by complex terrain and diverse landforms24. The Da-Xiao Liangshan Mountains has fragile ecosystems, frequent geological disasters, and significant human-environment conflicts. It serves as an important ecological barrier in the upper reaches of the Yangtze River. The area is the largest Yi ethnic group settlement in China and is also one of the country’s deeply impoverished regions25. As of the end of 2020, the region had a permanent population of 6.8051 million (with the Yi ethnic group accounting for approximately 47.79%), a GDP of 147.595 billion RMB, and an urbanization rate of only 21.83%. Since the late 20th century and early 21st century, with the implementation of China’s “Reform and Opening-up” and “Grain for Green” policies, as well as the ongoing process of urbanization, the rapid growth of population and economic scale in the study area has posed significant threats to the sustainable development of the regional ecological environment.

Fig. 1
figure 1

Geographical location map of the Da-Xiao Liangshan Mountains. The base map is produced using the standard map of the Ministry of Natural Resources (http://bzdt.ch.mnr.gov.cn/index.html), with the approval number GS (2020) 4630, and no modifications have been made to the map boundaries. The maps covered in this thesis we have drawn through the ArcGIS 10.7 (https://www.esri.com/en-us/home) software. Same below.

Full size image

Data sources

The data sources used in this study are as follows: (1) The land use data were obtained from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn/), with a resolution of 30 m. This dataset was derived from visual interpretation of Landsat remote sensing imagery and has an overall accuracy exceeding 90%26, which meets the requirements of this study. (2) Digital Elevation Model (DEM) data was sourced from the Geospatial Data Cloud (http://www.gscloud.cn/), with a resolution of 30 m. Slope, aspect, and relief degree of land surface. were calculated based on this data. (3) The data for annual mean temperature, annual precipitation, annual wind speed, annual ground temperature, annual evaporation, annual relative humidity, annual sunshine hours, population density, GDP per city, accumulated temperature above 10℃, and NDVI were all obtained from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn/), with a resolution of 1 km. (4) Data for railways, protected areas, roads, township government locations, and water systems were obtained from the 1:1,000,000 scale fundamental geographic information dataset published by the National Geomatics Center of China (https://www.webmap.cn/).

Research methods

Kernel density analysis

Kernel density analysis treats the distribution pattern of the study object as a probability distribution and uses a continuous density curve to describe the distribution pattern. This method objectively analyzes the spatial distribution characteristics of geographical elements27. The formula is as follows:

$${f_x}=\frac{1}{{nh}}\sum K \left( {\frac{{d{\text{(}}x,{x_i}{\text{)}}}}{h}} \right)$$
(1)

In the formula, \({f_x}\) represents the forest land kernel density value. K is the kernel function. h denotes the search radius. n is the number of forest land points within the search radius; \(d{\text{(}}x,{x_i}{\text{)}}\) represents the distance between the forest land point xand the sample point\({x_i}\).

Standard deviation elliptic theory

The standard deviational ellipse method is a spatial analysis technique that reflects the spatial distribution characteristics and spatiotemporal evolution of the study object by describing fundamental parameters such as centroid, azimuth, major axis, and minor axis28. This method is used in this study to analyze the spatiotemporal distribution and evolution characteristics of forest land. The formulas for calculating the key parameters are as follows29.

$$\overline {X} =\frac{{\sum\nolimits_{{i=1}}^{n} {{w_i}} {x_i}}}{{\sum\nolimits_{{i=1}}^{n} {{w_i}} }},\overline {Y} =\frac{{\sum\nolimits_{{i=1}}^{n} {{w_i}} {y_i}}}{{\sum\nolimits_{{i=1}}^{n} {{w_i}} }}$$
(2)
$${\text{tan}}\alpha =\frac{{\left( {\sum\limits_{{i=1}}^{n} {{w_i}^{2}{{\overline {x} }_i}^{2} - } \sum\limits_{{i=1}}^{n} {{w_i}^{2}{{\overline {y} }_i}^{2} - } } \right)+\sqrt {{{\left( {\sum\limits_{{i=1}}^{n} {{w_i}^{2}{{\overline {x} }_i}^{2} - } \sum\limits_{{i=1}}^{n} {{w_i}^{2}{{\overline {y} }_i}^{2} - } } \right)}^2}+4\sum\limits_{{i=1}}^{n} {{w_i}^{2}{{\overline {x} }_i}^{2}{{\overline {y} }_i}^{2}} } }}{{2\sum\limits_{{i=1}}^{n} {{w_i}^{2}{{\overline {x} }_i}^{2}{{\overline {y} }_i}^{2}} }}$$
(3)
$${\sigma _x}=\sqrt {\frac{{\sum\nolimits_{{i=1}}^{n} {{{\left( {{w_i}{{\overline {x} }_i}\cos \alpha - {w_i}{{\overline {y} }_i}\sin \alpha } \right)}^2}} }}{{\sum\nolimits_{{i=1}}^{n} {{w_i}^{2}} }}} ,{\sigma _y}=\sqrt {\frac{{\sum\nolimits_{{i=1}}^{n} {{{\left( {{w_i}{{\overline {x} }_i}\sin \alpha - {w_i}{{\overline {y} }_i}\cos \alpha } \right)}^2}} }}{{\sum\nolimits_{{i=1}}^{n} {{w_i}^{2}} }}}$$
(4)
$$S=\pi {\sigma _x}{\sigma _y}$$
(5)

In formulas (2) to (5), \(\overline {X}\)​ and Y​ represent the coordinates of the centroid; \({w_i}\)​ is the weight; \(\alpha\) is the azimuth of the standard deviational ellipse; \({\overline {x} _i}\) and \({\overline {y} _i}\)​ are the deviations of the coordinates of each study object from the mean center; \({\sigma _x}\)​ and \({\sigma _y}\) represent the standard deviations along the x-axis and y-axis, respectively; and S denotes the area of the ellipse.

Quantitative indicators of forest land fragmentation

Landscape indices provide a concise representation of landscape pattern characteristics. Through various levels of landscape metrics, they reveal the composition and spatial distribution of the landscape, widely used in land use landscape pattern studies30. Referring to previous research31, seven landscape indices (Table 1) were calculated using Fragstats software to assess forest fragmentation in the Daliang and Xiaoliang Mountains: number of patches (NP), patch density (PD), cohesion index (Cohesion), landscape shape index (LSI), landscape division index (Division), splitting index (Split), and aggregation index (AI).

Table 1 Landscape index and explanation of forest land in the study area.
Full size table

Topographic data processing and distribution index

To analyze the spatio-temporal variation of forest land across topographic gradients in the Da-Xiao Liangshan Mountains, four topographic elements were selected: elevation, slope, aspect, and relief degree of land surface. Elevation and slope were classified using the natural breaks method in ArcGIS (Table 2), while aspect was classified based on previous research findings32. The extraction of relief degree of land surface, which is subject to scale effects, was conducted considering that the Da-Xiao Liangshan Mountains lie in the transitional zone between the Tibetan Plateau and the Yunnan-Guizhou Plateau, where the terrain is highly variable. Referring to the maximum window scale for terrain relief extraction in adjacent areas, and following the moving window method proposed by previous research findings33, the appropriate window size (51 × 51 pixels) was determined using ArcGIS spatial analysis tools. Under this scale, the relief degree of land surface value for the region ranged from 0 to 2,047 m.

To eliminate the influence of differences in the area of various topographic features on forest land distribution, a standardized, dimensionless distribution index was used to describe the distribution of forest land across different topographic gradients, as per the following formula34:

$$P=\left( {{A_{ec}}/{A_c}} \right)/\left( {{A_e}/A} \right)$$
(6)

In the formula, P​ represents the distribution index of forest land across different topographic gradients; A is the total area of the study region; \({A_e}\)​ is the total area of the e topographic gradient in the study region; \({A_c}\) is the total forest land area in the study region; and \({A_{ec}}\) is the forest land area within the e topographic gradient. When P ​> 1, it indicates that the topographic gradient is a dominant area for forest land distribution.

Table 2 Grading classification of terrain factors in the study area and their area proportion /%.
Full size table

Land use expansion driver model

The land expansion analysis strategy in the PLUS model utilizes the Random Forest Classification (RFC) algorithm to explore the expansion of each land use type and its driving factors individually. This allows for the determination of the development probability for each land use type over a specific period and the contribution of driving factors to the expansion of each land use type35. The RFC algorithm is capable of handling high-dimensional data and addressing multicollinearity among variables, determining the probability \(P_{{i,k}}^{d}\) of k land use type appearing in cell i. The calculation formula is as follows36:

$$P_{{i,k}}^{d}\left( x \right)=\frac{{\sum\nolimits_{{n=1}}^{M} {I\left( {{h_n}\left( x \right)=d} \right)} }}{M}$$
(7)

In the formula, d takes a value of 0 or 1, where 1 indicates that another land use type has been converted to type k, and 0 indicates that the land use type has changed to any other type except k; x is a vector composed of multiple driving factors; I represents the indicator function of the ensemble of decision trees; \({h_n}\left( x \right)\) is the predicted land use type for the n decision tree in the vector x; and M is the total number of decision trees.

Information on the software used

We utilized Microsoft Office Excel 2016 (https://www.microsoftstore.com.cn/), Fragstats 4.2 (https://fragstats.org/), ArcGIS 10.7 (https://www.esri.com/en-us/home), Geodetector (http://geodetector.cn/index.html), PLUS v1.3.5 (https://github.com/HPSCIL/Patch-generating_Land_Use_Simulation_Model), GeoDa 1.22 (https://geodacenter.github.io/), Origin 2024 (https://www.originlab.com/), to perform data processing, analysis, visualization, and other tasks in this study. Addi-tionally, Zotero 7 (https://www.zotero.org/ ) was employed for managing and citing references.

Results and analysis

Analysis of forest land dynamics change

Overall, from 1980 to 2020, forest land was one of the dominant land use types in the Da-Xiao Liangshan Mountains, with an area exceeding 45,470.13 km2 (Table 3), accounting for approximately 57.64% of the total area. During the study period, the forest land area showed a trend of initial decline, followed by an increase, and then another decrease. The proportion of forest land in the total area increased from 57.65% (45,752.12 km2) in 1980 to 57.81% (45,872.53 km2) in 2020, with the lowest value observed in 2000 at 57.30% (45,470.13 km2). Over the 40-year period, the forest area cumulatively increased by 0.26% (120.41 km2).

Table 3 Changes in land use area in the Da-Xiao Liangshan mountains from 1980 to 2020 /km2.
Full size table

In terms of the spatial distribution and changes in forest land (Fig. 2), the forested areas in the Da-Xiao Liangshan Mountains were primarily concentrated in the western, southeastern, and northeastern mountain regions, while forest coverage was relatively sparse in the densely populated eastern, central, and southern edge areas. Between 1980 and 2020, forest land expansion mainly occurred in the central, eastern, northeastern, and southeastern regions of the study area, with most of this increase concentrated between 2000 and 2020, particularly from 2000 to 2010, which coincided with the implementation of the “Grain for Green” policy. During the same period, forest land reduction was primarily observed in the river valley regions of the western, southern, central, and northeastern parts of the study area. From 1980 to 2020, the forest land reduction was concentrated in the central, southwestern, and northwestern regions, whereas from 2000 to 2020, the reduction was more prevalent in the northeastern and southeastern regions, which corresponds to the construction of cascade hydropower stations in these areas.

Fig. 2
figure 2

Spatial-temporal changes of forest land in the study area from 1980 to 2020.

Full size image

At the county level within the Da-Xiao Liangshan Mountains, between 1980 and 2020, 58.17% (26,611.11 km2) of the forest land was distributed across the western and southern counties of Muli, Yanyuan, Ninglang, Yongsheng, Huili, and Mianning (Fig. 3a). Among these, Muli, Yanyuan, and Ninglang had the highest forest land proportions, accounting for 19.43% (8,862.04 km2), 10.65% (4,919.26 km2), and 9.70% (4,424.76 km2) of the total forest land, respectively. In contrast, the eastern counties of Ningnan, Puge, Yuexi, Pingshan, Zhaojue, Butuo, and Jinyang only contained 11.45% (5,238.11 km2) of the total forest land.

From 1980 to 2020, 12 counties, including Pingshan, Xichang, Yanyuan, Huidong, Ningnan, Puge, Zhaojue, Xide, Mianning, Yuexi, Ganluo, and Leibo, showed an increasing trend in forest land(Fig. 3b). In contrast, other counties experienced varying degrees of forest land reduction. Among these, Yanyuan, Ganluo, Zhaojue, and Mianning saw significant forest land growth, with increases of 1.65% (79.77 km2), 5.42% (66.40 km2), 8.56% (55.68 km2), and 2.11% (51.00 km2), respectively. On the other hand, counties such as Huili, Muli, Yongsheng, Huaping, and Mabian experienced notable forest land reductions, with decreases of 2.42% (69.59 km2), 0.69% (61.86 km2), 1.41% (44.58 km2), 0.82% (36.64 km2), and 2.27% (35.58 km2), respectively.

Fig. 3
figure 3

County scale distribution and changes of forest land in the study area from 1980 to 2020.

Full size image

Analysis of forest land transfers

Based on the forest land transitions in the Da-Xiao Liangshan Mountains (Table 4), significant changes in forest land occurred over the 40-year period. Between 1980 and 2000, forest land showed an overall decreasing trend, with a total of 455.94 km2 of forest land being converted, primarily to grassland (409.25 km2) and cropland (43.77 km2). The total area of forest land gained during this period was 174.17 km2, mainly from the conversion of grassland (113.61 km2) and cropland (59.74 km2). Compared to the first 20 years, the changes in forest land between 2000 and 2020 were even more pronounced, with 3,274.79 km2 of forest land converted, primarily to grassland (1,767.81 km2), cropland (1,398.71 km2), and water bodies (69.16 km2). During the same period, 3,678.11 km2 of land was converted to forest land, primarily from grassland (2,222.73 km2) and cropland (1,436.62 km2). Notably, most of the forest land increase occurred between 2000 and 2010, with a net gain of 469.31 km2 of forest land. This indicates that, following the implementation of the “Grain for Green” policy in the early 21st century, the overall forest land area in the Da-Xiao Liangshan Mountains has increased. However, from 2010 to 2020, the rapid urbanization process in the region has led to a reduction in forest land.

Table 4 Transfer of forest land in the study area from 1980 to 2020 /km2.
Full size table

In terms of land use transitions from 1980 to 2020 across the Da-Xiao Liangshan Mountains region (Fig. 4), frequent land conversions occurred between different land-use types, particularly in the central, eastern, and western parts of the study area. Regarding forest land gains, the conversion of cropland to forest land was concentrated in the southwestern counties of Huili, Huidong, Ningnan, and Puge, the northeastern counties of Mabian, Ganluo, and Pingshan, and the central counties of Xichang and Xide. The conversion of grassland to forest land was notably concentrated in the northern counties of Mianning, Yuexi, and Ganluo, the central counties of Xichang, Yanyuan, and Dechang, as well as Muli County in the northwest and Huili County in the southwest. This distribution indicates that the increase in forest land between 1980 and 2020 was closely linked to the “Grain for Green” policy, with significant increases observed in the agricultural regions of the northeastern, southwestern, and central parts of the study area.

In terms of forest land loss, the conversion of forest land to cropland was concentrated in the northeastern counties of Mabian and Pingshan, the central counties of Xide and Xichang, and the southwestern counties of Huili and Huidong. The conversion of forest land to grassland was mainly concentrated in the northeastern counties of Ebian, Mabian and Ganluo, the eastern counties of Jinyang, Puge and Butuo, the southwestern counties of Huili, Huidong and Ningnan, the southwestern counties of Yongsheng and Huaping, as well as Muli County in the northwest. This indicates that while these areas were also influenced by the “Grain for Green” policy, intense human activities led to localized reductions in forest land.

Fig. 4
figure 4

Spatial distribution and direction of land use transfer in the study area from 1980 to 2020. CL: Cropland; FL: Forest land; GL: Grassland; WA: Water area; BA: Built-up area; UL: Unused land.

Full size image

Analysis of forest land spatial clustering

To analyze the spatial clustering of forest land in the Da-Xiao Liangshan Mountains, kernel density analysis (with a search radius of 5 km) was conducted to calculate the kernel density values of forest land distribution from 1980 to 2020. The values were classified into five categories using the natural breaks method (No.·km2): Very low [0.01, 0.30], low (0.30, 0.45], medium (0.45, 0.58], high (0.58, 0.71], and very high (0.71, 0.97]. The results indicate significant spatial heterogeneity in forest land distribution across the region (Fig. 5), with a pattern of low density in the central and eastern areas and high density in the southeastern, northeastern, and western areas.

Fig. 5
figure 5

Spatial distribution of forest land kernel density levels in the study area from 1980 to 2020.

Full size image

The areas classified as high and extremely high kernel density accounted for approximately 26.36% and 21.53% of the total area, respectively (Fig. 6), primarily concentrated in the northeastern Daliang Mountain Range, southeastern Luoji–Lunan Mountains, and the western Jinping, Maoniu, and Hengduan Mountain Ranges (Fig. 5), where the terrain is rugged and forest coverage is extensive. Medium and low-density areas made up approximately 21.48% and 19.04% of the total area, mainly distributed in the northeastern, central, and southern transitional zones between mountains and river valleys. Extremely low-density areas covered about 10.83% of the region, predominantly located in the eastern mountainous areas, the central Anning River Valley, and the densely populated Yanyuan Basin. In terms of the changes in forest land kernel density categories from 1980 to 2020 (Fig. 6a), both extremely low and extremely high-density areas experienced a reduction, while low, moderate, and high-density areas expanded. The reduction in extremely low-density areas was concentrated in the eastern counties of Xide, Zhaojue, and Yuexi between 2000 and 2010, corresponding to the implementation of the “Grain for Green” policy in these regions. The area transitioning from extremely low to low density reached 1,174.71 km2 (Fig. 6b). The reduction in extremely high-density areas was mainly observed in the southeastern counties of Huili and Huidong during the same period, correlating with the rapid agricultural development in these areas, with 1,283.50 km2 of extremely high-density forest land transitioning to high density. The expansion of moderate-density areas was primarily concentrated in Xide County in the east and the southeastern counties of Puge and Ningnan between 2000 and 2010, mostly transitioning from low-density areas (1,146.77 km2). Overall, the spatial changes in forest land kernel density from 1980 to 2020 in the Da-Xiao Liangshan Mountains Mountains were strongly influenced by the “Grain for Green” policy.

Fig. 6
figure 6

Changes and transfer of kernel density at different grades of forest land in the study area from 1980 to 2020.

Full size image

Trends in the spatial and temporal evolution of forest land

From the changes in the standard deviational ellipse parameters of forest land from 1980 to 2020 (Table 5), the azimuth fluctuated within the range of 50.749° to 50.781°. Overall, the azimuth showed a decreasing trend, indicating a shift in the direction of forest land expansion from a “northeast-southwest” orientation towards an “east-west” orientation. The minor axis shortened from 110.792 km in 1980 to 110.465 km in 2020, while the major axis decreased from 150.975 km in 1980 to 150.812 km in 2020. These changes in the axes suggest that the distribution range of forest land has contracted, with relatively stable distribution in the north-south direction, but a shrinking trend in the east-west direction. The eccentricity increased slightly from 1.363 in 1980 to 1.365 in 2020, indicating a growing spatial aggregation of forest land in the study area.

Table 5 Changes in standard deviation ellipse and centroid parameters of ESV in the study area.
Full size table

According to the spatial range and trajectory changes of the standard deviational ellipse, the forest land expansion from 1980 to 2020 predominantly followed a northeast-southwest orientation. In terms of the centroid shift of the standard deviational ellipse (Table 5; Fig. 7), the centroid moved northeastward by 0.84 km over the study period, with the exception of a 0.14 km southeastward shift between 1990 and 2000. The most significant centroid movement occurred between 2000 and 2010, during which it shifted 0.72 km to the northeast.

Fig. 7
figure 7

Distribution and variation of standard deviation ellipse and center of gravity of forest land in the study area.

Full size image

Analysis of forest land fragmentation

From 1980 to 2020, various landscape indices of forest land in the Da-Xiao Liangshan Mountains show significant changes (Table 6). The number of patches, patch density, cohesion index, landscape shape index, and aggregation index of forest land generally exhibited a fluctuating downward trend. Specifically, the number of patches decreased from 5,814 to 5,684, patch density (No./km2) dropped from 0.1271 to 0.1239, the cohesion index fell from 99.9744 to 99.9720%, and the aggregation index declined from 97.3230 to 97.2332%. This indicates that while the number of forest land patches and patch density decreased from 1980 to 2020, the degree of aggregation also weakened. Additionally, the landscape shape index, landscape division index, and isolation index of forest land increased from 191.8400, 72.96%, and 3.6985% in 1980 to 198.4960, 77.31%, and 4.4073% in 2020, respectively. This suggests that the forest land shapes in the study area have become more irregular, and the degree of fragmentation and isolation has intensified. Overall, the forest land in the Da-Xiao Liangshan Mountains exhibits significant spatial aggregation, with a relatively high degree of fragmentation and cohesion.

Table 6 Landscape index of forest land in the study area from 1980 to 2020.
Full size table

Topographic gradient distribution and changes of forest land

Despite the overall increase in forest land area due to the influence of urbanization and the “Grain for Green” policy, the degree of fragmentation has worsened. Using 2020 as a representative year, further analysis of the topographic gradient distribution of forest land in the Da-Xiao Liangshan Mountains reveals a clear vertical zonation in its spatial distribution (Fig. 8). Specifically, forest land area initially increases and then decreases with rising altitude, slope, and relief degree of land surface. Altitude class V, slope class IV, and relief degree of land surface class IV represent transition zones, accounting for 18.33% (8,423.62 km2), 18.59% (8,539.64 km2), and 18.66% (8,576.48 km2) of the total forest land area, respectively. In terms of aspect gradient, except for aspect class I, which contains only 0.01% (707.02 km2) of the total forest land area, forest land is relatively evenly distributed across the other classes, with area proportions ranging from 11.13% (5,113.06 km2) to 13.65% (6,268.21 km2), with the highest concentration in aspect class II. In terms of topographic gradient distribution indices (Fig. 8), forest land is predominantly distributed in altitude classes IV–VIII, slope classes IV–IX, terrain relief classes IV–IX, and aspect classes II–III and VIII–IX (distribution index > 1). This indicates that forest land is mainly concentrated in sunny slope regions with elevations ranging from 1,622 m to 3,976 m, slopes of 8.77° to 44.16°, and relief degree of land surface of 199 m to 969 m.

Fig. 8
figure 8

Distribution of forest land in topographic gradients in the study area.

Full size image

From 1980 to 2020, the changes in forest land area across topographic gradients show that (Fig. 9), except for the increase in forest land area between 2000 and 2010, forest land generally declined during other periods. Forest land increased by 470.12 km2 between 2000 and 2010, but decreased by 65.17 km2, 217.42 km2, and 62.71 km2 between 1980 and 1990, 1990–2000, and 2010–2020, respectively. Notably, during the period 2000–2010, forest land decreased only in aspect class I and terrain relief class I, while all other gradients experienced varying degrees of increase. From 2010 to 2020, forest land increased only in altitude classes V–VII and IX, slope classes IV–IX, aspect classes V–VIII, and relief degree of land surface class IX, while all other gradients showed varying degrees of decline. Overall, from 1980 to 2020, the increase in forest land area was concentrated in sunny slopes at altitudes between 2,017 m and 3,478 m, and 3,976 m and 5,902 m, with slopes of 22.08° to 83.13°, and t relief degree of land surface of 460 m to 2,047 m, while forest land area declined across other gradients.

Fig. 9
figure 9

Changes in forest land in topographic gradients in the study area from 1980 to 2020.

Full size image

Analysis of driving factors of forest land change

Considering the natural and economic conditions of the Da-Xiao Liangshan Mountains, 20 driving factors were selected to analyze forest land area changes from 1980 to 2020. These factors include aspect (X1), altitude (X2), annual evaporation (X3), per capita GDP (X4), mean annual ground temperature (X5), accumulated temperature above 10 °C (X6), NDVI (X7), distance to protected areas (X8), distance to railways (X9), distance to rivers (X10), distance to roads (X11), distance to township government offices (X12), population density (X13), annual precipitation (X14), relief degree of land surface (X15), annual relative humidity (X16), slope (X17), annual sunshine hours (X18), mean annual temperature (X19), and mean annual wind speed (X20). The contribution of these driving factors to forest land expansion in the Da-Xiao Liangshan Mountains from 1980 to 2020 shows that natural and economic factors such as altitude, NDVI, distance to railways, population density, and annual precipitation were the primary drivers, with altitude contributing the most at 0.11 (Fig. 10). By overlaying forest land expansion areas with elevation grid data, it is evident that forest land increase was concentrated in the mountainous regions along the Yalong River in the central part of the study area, the Heishui River in the southeast, and the Niri River in the north. Additionally, the mountain areas in the northwest, east, and northeast of the study area also emerged as secondary hotspots for forest land growth.

Fig. 10
figure 10

Contribution of driving factors for land use expansion in the study area from 1980 to 2020.

Full size image

To further analyze the reasons for forest land changes in the Da-Xiao Liangshan Mountains, it is essential to examine the factors contributing to forest land reduction. Table 4 shows that from 1980 to 2020, most of the forest land loss was converted to grassland and cropland, accounting for 57.26% (1,875.15 km2) and 39.52% (1,294.37 km2) of the total forest land loss, respectively. Conversion to water bodies, construction land, and unused land was minimal, making up only 2.11% (69.00 km2), 1.03% (33.69 km2), and 0.08% (2.68 km2) of the total forest land loss. Thus, the increase in grassland and cropland areas was the primary reason for forest land loss in the region. Therefore, analyzing the driving factors for grassland and cropland expansion can indirectly reveal the causes of forest land area changes.

According to the results (Figs. 10 and 11), the main driver of grassland expansion from 1980 to 2020 was the mean annual ground temperature, contributing up to 0.09, followed by per capita GDP, distance to railways, annual relative humidity, and annual sunshine hours. Grassland expansion was mainly concentrated in the mid, southwestern, southeastern, and northeastern mountainous areas along the river valleys, where the mean annual ground temperature is relatively high. Per capita GDP had the most significant influence on cropland expansion in the region (contributing 0.08), followed by terrain relief, altitude, mean annual ground temperature, and annual precipitation. Cropland expansion occurred predominantly in the low-altitude, densely populated central, southeastern, and northeastern parts of the study area. Overall, the changes in forest land area in the Da-Xiao Liangshan Mountains from 1980 to 2020 were heavily influenced by natural factors such as mean annual ground temperature, elevation, NDVI, and annual precipitation, while economic factors like per capita GDP, distance to railways, population density, and distance to protected areas also played significant roles in forest land area changes.

Fig. 11
figure 11

Overlay of land use expansion regions and their highest driving factors in the study area from 1980 to 2020.

Full size image

Discussion and conclusion

Discussion

Over the past 40 years, forest land in the Da-Xiao Liangshan Mountains has experienced dynamic changes, characterized by an initial decrease followed by an increase, resulting in a net gain of 120.41 km2, Consistenting with previous results on forest land change trends in similar mountainous areas23 and in China as a whole26. This pattern reflects both the effects of human activity and natural factors acting across the topographic gradient. Forest loss was concentrated in the lower elevations, where human activities, such as agricultural expansion, infrastructure development, and timber harvesting, are more pronounced due to the easier accessibility17,37. These activities lead to habitat fragmentation and reduced forest connectivity, weakening ecological stability in these areas38. Conversely, forest gains were observed primarily in mid-to-high elevation zones. The increase in mid-elevation forest cover is primarily driven by ecological restoration projects, such as the Grain for Green Program, which have effectively mitigated land degradation and forest loss in this zone39. In contrast, the expansion of forest cover at higher elevations is likely due to climate-induced upward shifts in the forest line. As elevation increases, the influence of human activity diminishes, while the effects of climate change become more pronounced. For example, rising temperatures create favorable conditions for vegetation to expand to higher altitudes40. Additionally, studies have shown that forest ecosystem vulnerability decreases with increasing altitude, as high-elevation forests generally experience less human disturbance and are thus more ecologically stable41.

At the same time, the degree of fragmentation has intensified, underscoring the challenges posed by expanding human infrastructure and land use. Forest fragmentation not only disrupts habitat continuity but also increases edge effects, altering local microclimates and potentially reducing biodiversity42. This fragmentation also poses challenges to species migration and gene flow, particularly for species that rely on continuous forest cover43. Maintaining connectivity through landscape planning and habitat corridors will be essential for sustaining biodiversity and ecosystem function in the region44. The spatial distribution of forest land in the Da-Xiao Liangshan Mountains demonstrates a clear pattern, with dense forest cover in the southeastern, northeastern, and western regions, while central and eastern areas are more sparsely forested. This spatial heterogeneity is largely influenced by topographic variation, including slope and aspect, which play essential roles in determining forest suitability. Slopes ranging from 8.77° to 44.16° and sunny aspects in particular have favored denser forest distribution, due to optimal sunlight exposure and soil stability that support vegetation growth45. The observed shift of the forest centroid by 0.84 km northeastwards further indicates regional adjustments in forest cover, likely reflecting a gradual migration of forest patches in response to environmental factors.

The study’s findings highlight elevation as a primary factor influencing forest distribution, contributing significantly to forest area changes over the study period. Forest land is concentrated between elevations of 1,622 m and 3,976 m, where suitable climatic conditions and reduced human interference at higher altitudes allow forest patches to remain relatively intact46. The terrain relief, ranging between 199 m and 969 m, also supports vegetation growth by reducing soil erosion in steeper, less accessible areas, thereby sustaining soil stability47. Additionally, factors such as NDVI (Normalized Difference Vegetation Index) and annual precipitation positively correlate with forest density, reflecting the importance of adequate moisture and vegetation cover in supporting forest resilience48. Economic factors, including per capita GDP, proximity to railways, and population density, have also influenced forest land distribution. In regions with higher economic activity, forest land is more likely to be converted to cropland or urban areas, driven by the economic demand for land resources49. Proximity to transportation networks, such as railways, facilitates access to forest resources, further contributing to forest conversion in these areas50. These socio-economic pressures underscore the need for balanced land-use policies that account for economic development while conserving ecological resources8. This study provides valuable insights into the spatiotemporal and topographic dynamics of forest land in the Da-Xiao Liangshan Mountains. The results of the study can provide a relatively scientific theoretical basis for local governments and other decision-making authorities to formulate locally adapted forest land protection strategies, which will be beneficial to the sustainable development of the ecosystem in the study area. However, certain limitations should be acknowledged. First, the analysis relied on remotely sensed land cover data, which, while high-precision, may contain classification uncertainties that could affect the accuracy of forest cover estimates. Second, the study period from 1980 to 2020 captures significant historical trends, but finer temporal data could improve understanding of more recent, rapid changes in forest dynamics. Additionally, this study primarily focused on broad categories of natural and socio-economic drivers; future research should investigate more specific ecological factors, such as species composition and forest health indicators, as well as policy impacts at the local scale to refine conservation strategies. At the same time, this paper used PLUS model and standard deviation ellipse to analyze the spatial differentiation characteristics of the forest land in the study area, and we need to consider adopting a more scientific exploratory spatial data analysis methods for future analysis. Overall, by addressing these limitations and expanding on the socio-ecological interactions at play, future studies can provide more comprehensive guidance for sustainable forest management in the region.

Conclusion

From 1980 to 2020, forest land was one of the dominant land use types in the study area, covering more than 57.65% of the total area. The forest land area initially decreased and then increased, with a net gain of 120.41 km2 over the 40-year period. The spatial distribution of forest land exhibited a pattern of sparse forest cover in the central and eastern regions and dense cover in the southeastern, northeastern, and western regions. During the study period, the degree of fragmentation intensified, and the centroid of forest land shifted 0.84 km to the northeast between 1980 and 2020. Forest land distribution demonstrated significant vertical zonation, being highly concentrated on sunny slopes at altitudes between 1,622 m and 3,976 m, with slopes ranging from 8.77° to 44.16° and relief degree of land surface between 199 m and 969 m. Forest land area generally decreased across different topographic gradients from 1980 to 2020, with the opposite trend observed between 2000 and 2020. Changes in forest land area were strongly influenced by both natural factors (such as altitude, NDVI, and annual precipitation) and economic factors (such as per capita GDP, distance to railways, and population density), with elevation being the primary driving factor, contributing 0.11.