Introduction

The ecological vulnerability of the Himalayan region has become a focal point of global concern due to the compounded effects of climate change, human activities (such as land-use changes, urbanization, and deforestation), and frequent natural disasters1,2. This region, characterized by its unique geology and ecology, is particularly susceptible to the impacts of climate change3. Over the past decades, extreme weather events like heavy rainfall, droughts, and floods, along with geological hazards such as earthquakes and landslides, have severely affected local ecosystems4,5. Additionally, issues like glacial melt and river flow changes have further intensified ecological vulnerability. The accelerating climate change, marked by rising temperatures, altered precipitation patterns, and increased extreme weather events, is exacerbating the region’s ecological fragility, threatening not only ecosystem stability but also water resources, agricultural productivity, and human livelihoods. Ecological vulnerability is closely linked to climatic factors but also influenced by geological, ecological, and socio-economic conditions6,7. Particularly in the Himalayas, increasing population pressure and the growing conflict between humans and natural resources are worsening the impact of human activities on the environment. Understanding the region’s ecological vulnerability and its driving forces is crucial for developing effective strategies for ecological conservation and adaptation in the face of ongoing global climate change8.

Land-use change is one of the most direct manifestations of human influence on the environment, impacting ecosystem types, spatial distribution, and driving changes in ecosystem services9. Landscape patterns, reflecting the spatial distribution of natural ecosystems, are shaped by both natural factors and human activities10,11. Changes in these patterns inevitably affect ecosystem quality and evolution12. With the increasing human impact, fragmentation of protected area landscapes is becoming a significant issue, leading to the degradation of ecosystem structure and function13. Analyzing the spatiotemporal dynamics of landscape patterns is vital for ensuring the sustainable development of protected habitats14. Despite the essential role of protected areas in preserving species and maintaining ecological processes, the combined effects of historical changes and overexploitation have resulted in ecological degradation in many areas. Recent studies on protected area landscapes focus on the dynamics of landscape pattern changes, their driving factors, optimization, and the simulation of landscape succession15. Methods like principal component analysis are useful for analyzing the impact of factors such as climate, hydrology, economy, and policy on landscape patterns, especially in large, complex protected areas. This research helps reveal the combined effects of natural and human factors, predict landscape trends, and provide guidance for the sustainable use of landscapes and biodiversity conservation16.

Compared to other mountainous regions, the ecological degradation in the Himalayas is more pronounced and continues to worsen17. Recent changes in the environment, driven by both climate change and human activity, have caused significant alterations in cross-border ecosystems, with problems such as glacial retreat, expansion of glacial lakes, and forest degradation becoming more severe18. In particular, the southern slopes of the Himalayas face increasing population pressure, land conflicts, and drastic changes in regional ecosystems, all of which threaten biodiversity and drastically reduce ecosystem service functions19,20. The growing risk of flood disasters, influenced by climate change and urbanization, has further intensified the region’s ecological degradation. Furthermore, climate change and hydrological disasters have had a marked impact on local monsoon crop patterns, directly exacerbating food security risks. While numerous studies on ecosystem vulnerability have been conducted, most focus on external disturbances, using models that assess exposure, sensitivity, and adaptive capacity21. However, many studies overlook the spatial heterogeneity of ecosystems and the interactions between various factors. Although global regression models and spatial analysis techniques have been applied in vulnerability studies, there is still a lack of comprehensive analysis regarding the spatial non-stationarity of climate and human activity factors22,23.

This study examines changes in landscape composition within the Himalayan cross-border protected areas from 2000 to 2020, comparing land-use changes and landscape patterns across these years and discussing the key driving factors behind these changes. Using Mount Everest as a case study, we developed an “exposure-sensitivity-resilience” ecological vulnerability assessment model and applied spatial principal component analysis to evaluate the region’s ecological vulnerability. The results highlight significant spatial-temporal differences between the northern and southern slopes of Mount Everest. By assessing ecological vulnerability, this study identifies critical areas of fragility, offering scientific support for targeted conservation and development strategies and providing references for the sustainable use of protected area resources.

Research area overview and data sources

Research area overview

The Himalayan Transboundary Protected Area (27°25′10″– 29°19′12″N, 84°26′51″– 88°22′11″E) is located on the border between the People’s Republic of China and Nepal, in the middle of the Himalayas on the southern margin of the Qinghai-Tibet Plateau (Fig. 1). The region has been affected by strong tectonic movements that have created many world-class peaks, including six peaks above 8000 m, namely the world’s first peak Mount Qomolangma (8848.68 m), Mount Khangchendzonga (8586 m), Lhotse (8516 m), Makalu (8463 m), Cho Oyo (8201 m) and Shisha Pangma (8027 m)24. Due to the high altitude, steep terrain, strong ecosystem sensitivity, and weak resilience of the region, China and Nepal have successively established a series of nature reserves and national parks since the 1970s to protect fragile mountain ecosystems25. Qomolangma National Nature Preserve on the north slope of Mount Everest was established in 1988 and was classified as a national protected area in 1994, with a total area of 33,819 km2. On the southern slopes of Mount Everest, in the Himalayan region of northern Nepal, six nature reserves and national parks have been established from west to east, namely Manaslu Conservation Area, Langtang National Park, Gaurishankar Conservation Area, Sagarmatha National Park, Makalu Barun National Park and Kanchenjunga Conservation Area. The total area is 11,760 km2.

Fig. 1
figure 1

Location of study area. This original map was created by the author using ArcGIS v10.8 software (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview). Terrain data was processed through raster mosaicking, projection, mask extraction, and hill-shading to create the base map (https://cmr.earthdata.nasa.gov/search/concepts/C1546314043-LPDAAC_ECS.html). Boundary information for China’s Qomolangma National Nature Preserve and Nepal’s Manaslu Conservation Area, Langtang National Park, Gaurishankar Conservation Area, Sagarmatha National Park, Makalu Barun National Park, and Kanchenjunga Conservation Area was sourced from Protected Planet (https://www.protectedplanet.net/en). Peak location data was obtained from OpenStreetMap (https://www.openstreetmap.org/#map=5/38.01/-95.84). These elements were then exported and refined in Adobe Illustrator (version 28.0) to overlay and label geographic names, producing the final map.

Full size image

Data sources and processing

The data sources utilized in this research are detailed in Table 1. The remote sensing image data of 2000, 2010 and 2020 are derived from the MODIS image data provided by the Google Earth Engine (GEE) website, with a resolution of 500 meters26. The GEE platform is currently the most advanced geographic big data analysis and visualization platform, which makes remote sensing monitoring of surface vegetation and water bodies break through the limitations of difficult data acquisition, large local storage, and low processing efficiency27. By running the input code, the MODIS image data can be loaded directly and effectively, the remote sensing data preprocessing and band calculation can be carried out, and the annual maximum normalized difference vegetation index (NDVI) and the annual average modified normalized difference water index (MNDWI) in a large area can be obtained efficiently28. The meteorological data (temperature and precipitation) are derived from the CRU TS climate data set provided by the National Center for Atmospheric Sciences (NCAS) in the United Kingdom and are processed into raster data using the inverse distance weighting method. The population density data is derived from the WorldPop global population data set, the DEM data is derived from the US Geological Survey, the total primary productivity data of vegetation is derived from the National Aeronautics and Space Administration (NASA), and nighttime lighting data Obtain ‘NPP-like-VIIRS’ nighttime lighting data from the Harvard Dataverse platform. Due to the different sources and resolutions of the data, all the data are converted into raster data for projection conversion and resampling, and the spatial resolution is 0.5 km × 0.5 km29,30.

Table 1 Remote sensing data sources.
Full size table

Research methods

Analysis of landscape pattern indices

The study area has rich patch types, a large number of patches, and high biodiversity, but it has the characteristics of strong ecological vulnerability. Referring to the FRAGSTATS guidelines and a large number of landscape pattern index studies (Table 2), this paper selects five indicators: patch density (PD), largest patch index (LPI), landscape shape index (LSI), contagion index (CONTA) and Shannon’s evenness index (SHEI) to analyze the current situation of landscape pattern in the study area, focusing on reflecting the current situation of landscape pattern in the ecologically fragile areas of the Himalayan cross-border region31,32.

Table 2 Landscape pattern index formula and its ecological significance.
Full size table

Ecological vulnerability assessment model

This study assesses the ecological vulnerability of the northern and southern slopes of Mount Everest, considering both natural and anthropogenic disturbance factors33. Ecological vulnerability is evaluated through three main aspects: exposure, sensitivity, and adaptive capacity. Exposure is measured by population density and a comprehensive mountain hazard risk index, reflecting the external disturbances affecting the ecosystem. Sensitivity is captured by indicators related to climate, water, topography, and vegetation, such as temperature, precipitation, slope, water index (MNDWI), and vegetation index (NDVI). For adaptive capacity, net primary productivity (NPP) indicates ecosystem resilience, while the nighttime light index represents socio-economic development, indirectly reflecting local governments’ commitment to ecological protection (Fig. 2).

Fig. 2
figure 2

Framework of the ecological vulnerability evaluation indicator system for the North and South Slopes of Mount Everest. This figure, created by the author using Microsoft PowerPoint 2019 (Microsoft Corporation, Redmond, WA, USA), illustrates the framework for assessing ecological vulnerability in the Himalayas through the VSD model.

Full size image

To ensure comparability across indicators, all data are standardized using the range standardization method. Indicators are classified as positive or negative based on their relationship with ecological vulnerability34. Positive indicators, such as population density, hazard risk, temperature, precipitation, and slope, correlate with higher vulnerability, while negative indicators, including water index, vegetation index, NPP, and nighttime light, reflect lower vulnerability35. This approach facilitates a comprehensive and consistent assessment of ecological vulnerability across the region36.

Positive Evaluation Indicators: \(\:{Y}_{ij}=\frac{{X}_{ij}-{X}_{min}}{{X}_{max}-{X}_{min}}\)

Negative Evaluation Indicators: \(\:{Y}_{ij}=\frac{{X}_{max}-{X}_{ij}}{{X}_{max}-{X}_{min}}\)

In this equation, \(\:{Y}_{ij}\) represents the normalized value of the \(\:i\) raster data for indicator \(\:j\) with a range of 0 to 1, while \(\:{X}_{ij}\) denotes the original value of the ith raster data for indicator \(\:j\). \(\:{X}_{max}\) and \(\:{X}_{min}\) are defined as the maximum and minimum values of the original data for indicator \(\:j\:\)at the \(\:i\) raster.

Ecological vulnerability results from the interaction of various factors. To identify the most significant influences on regional vulnerability, Principal Component Analysis (PCA) is applied to nine indicators in the ecological vulnerability assessment system37,38. PCA is a multivariate statistical technique that reduces complex data while retaining essential information39. Using ArcGIS, spatial PCA is performed, and based on a cumulative contribution rate exceeding 90%, the first four principal components are selected to represent the original nine variables. This data reduction simplifies the analysis while preserving key ecological insights (Table 3).

Table 3 Eigenvalues, contribution rates, and cumulative contribution rates of each principal component.
Full size table

This study assesses the ecological vulnerability of the northern and southern slopes of Mount Everest, focusing on the impacts of natural and human-induced disturbances. The evaluation relies on three core dimensions: exposure, sensitivity, and adaptive capacity40. Exposure measures external pressures on the ecosystem, combining indicators like population density and a risk index for mountain hazards41. These factors help gauge the extent of human activity and natural threats affecting the region. Sensitivity is assessed through four main environmental factors-climate, water, terrain, and vegetation42. Key indicators include average annual temperature, annual precipitation, slope, the Modified Normalized Difference Water Index (MNDWI), and the Normalized Difference Vegetation Index (NDVI). These metrics capture the ecological response to changes in climate, landscape, and water availability43. Adaptive capacity considers the ecosystem’s resilience and the region’s socio-economic development. Net Primary Productivity (NPP) is used to measure the ecosystem’s self-regulation ability, while the nighttime light index reflects the area’s economic capacity to support ecological conservation. To ensure comparability across varied metrics, all data were standardized, distinguishing between factors that increase (e.g., population density) and decrease (e.g., NPP) vulnerability. A spatial principal component analysis was applied to synthesize these factors into a comprehensive ecological vulnerability index. This index reflects the overall ecological health and resilience of the Himalayan region, guiding targeted conservation efforts44,45.

The ecological vulnerability index (EVI) is defined as the weighted sum of m principal components, and the weight is expressed by the corresponding contribution rate of each principal component46. The specific equation is as follows:

$$\:EVI={\sigma\:}_{1}{X}_{1}+{\sigma\:}_{2}{X}_{2}+{\sigma\:}_{3}{X}_{3}+\dots\:+{\sigma\:}_{n}{X}_{n}$$

In the formula, \(\:{X}_{i}\) is the \(\:i\) th principal component; \(\:{\sigma\:}_{i}\) is the corresponding contribution rate of the \(\:i\) th principal component; \(\:n\:\)is the number of selected principal components. The larger the value of EVI is, the higher the degree of ecological vulnerability is, and the ecosystem is vulnerable to damage. In contrast, a lower EVI value indicates a more stable ecosystem47. The calculation formula of the contribution rate is as follows:

$$\:{\sigma\:}_{i}=\frac{{\lambda\:}_{i}}{\sum\:_{i=1}^{n}{\lambda\:}_{i}}$$

In the formula, \(\:{\sigma\:}_{i}\) is the corresponding contribution rate of the \(\:i\) th principal component; \(\:{\lambda\:}_{i}\) is the eigenvalue of the \(\:i\) th principal component.

The standardization of the obtained EVI index can better solve the comparability of ecological vulnerability assessment results between different years, and can also analyze the spatial and temporal evolution of ecological vulnerability more clearly and intuitively. In this study, combined with the current situation of the ecological environment and the degree of ecological vulnerability of the Himalayan transboundary, the ecological vulnerability was divided into five levels by using the Jenks natural discontinuity point classification method, followed by slight (< 0.35), mild (0.35 ~ 0.55), moderate (0.55 ~ 0.72), severe (0.72 ~ 0.84) and extreme (> 0.84) vulnerability areas48. The comprehensive evaluation index of ecological vulnerability is constructed to express the comprehensive characteristics of ecological vulnerability, to express the overall ecological vulnerability state of the Himalaya transboundary by means of intuitive quantification. The equation is specified as follows:

$$CEVI = \mathop \sum \limits_{{i = 1}}^{n} T_{i} \times \frac{{A_{i} }}{S}$$

In the formula, \(\:CEVI\) is the comprehensive index of ecological vulnerability, \(\:{T}_{i}\) is the value of the \(\:i\) vulnerability level, \(\:{A}_{i}\) is the area of the \(\:i\) vulnerability level area, \(\:S\) is the total area of the study area, and \(\:n\) is the number of vulnerability levels. By calculating the comprehensive index of ecological vulnerability, we can intuitively show the overall change trend of transboundary ecological vulnerability in the Himalayas, and can also compare the ecological vulnerability of each nature reserve49.

CATREG model

GATREG is a non-parametric multivariate regression analysis model that allows the analysis of categorical and numerical variables50. The classification regression (CATREG) selected in this study was analyzed, the discrete variables were selected and the pre-variables were selected into the regression. Category quantization refers to quantifying the category variables by a specific nonlinear transformation, and then iterating directly to find the optimal equation. The CATREG model is a classical linear regression model that applies and converts variables51. In the simple linear regression model, a response variable is predicted from the m predictors in x, trying to find a linear combination Xb that has the greatest correlation with it. The optimal scaling involved makes the feasible nonlinear function analysis between Z and \(\:\sum\:_{j=1}^{m}\left({b}_{j}{\phi\:}_{j}\left({X}_{i}\right)\right)\). Its expression is as follows:

$$\left\| {X^{*} b - Z^{*} } \right\|^{2} ~\,where~\,\left\| {X^{*} b - Z^{*} } \right\| = \sqrt {\left( {X^{*} b - Z^{*} } \right)^{T} \left( {X^{*} b - Z^{*} } \right)}$$

The categorical variable \(\:{h}_{j}\) defines the binary indicator matrix \(\:{G}_{j}\) of n rows and \(\:{l}_{j}\) columns, where \(\:{h}_{ij}\) defines \(\:{g}_{ir\left(j\right)}\) as follows:

$$\left\{ {\begin{array}{*{20}l} {g_{{ir}} \left( j \right) = 1,h_{{ij}} = r} \\ {g_{{ir}} \left( j \right) = 0,h_{{ij}} \ne r} \\ \end{array} } \right.$$

where \(\:r\)=1, 2… \(\:{l}_{j}\), the run index, indicates the category number in \(\:\text{j}\). If the category quantification can be expressed in terms of \(\:{y}_{j}\), the transformed variable can be written as \(\:{G}_{j}{y}_{j}\), for example, the weighted sum of the predictor variables is expressed as:

$$\mathop \sum \limits_{{j = 1}}^{m} b_{j} G_{j} y_{j} = X^{{\text{*}}} b$$

This is the same as the standard linear model and GATREG is equivalent to a linear regression model which can be expressed as follows:

$$\:{Z}^{\text{*}}={X}^{\text{*}}b+\epsilon\:$$

In the formula, \(\:{X}^{\text{*}}\) represents the coefficient matrix, \(\:{Z}^{\text{*}}\) is the observation vector, \(\:b\) is the normalized coefficient vector, and \(\:\epsilon\:\) is the error vector. In the optimal scaling process, the qualitative variables are transformed into quantitative variables. Categorical variables are quantified to reflect the characteristics of the original category, and nonlinear transformations allow the analysis of variables at various levels to find the best fitting model.

Results and analyses

Features of spatiotemporal changes in land use and landscape pattern

Change in land use from 2000 to 2020

By calculating the area and area proportion of different land use types in the Himalayan cross-border region from 2000 to 2020, this paper analyzes the interannual variation characteristics of land use in the study area during the past 20 years. The land use types include 8 types: Cropland, Forest, Shrub, Grassland, Water, Settlement, Naked land, and Glacier. At the same time, taking 10 years as a cycle, the study period is divided into two cycles, namely 2000–2010 and 2010–2020, to further analyze the characteristics of inter-annual changes in land use in the study area (Fig. 3).

Fig. 3
figure 3

The distribution and change map of land use types in the study area from 2000 to 2020. This figure is an original creation by the author, developed using ArcGIS v 10.8 (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview) to analyze land use data from Global Land Analysis & Discovery (https://glad.umd.edu/dataset/GLCLUC2020). The process involved raster mosaicking, mask extraction, raster-to-vector conversion, and area calculations for various land types. Subsequently, we used Origin 2024 (OriginLab Corporation, Northampton, MA, USA) to create a Sankey diagram illustrating land use trends from 2000 to 2020, and Adobe Illustrator Artwork 28.0 (Adobe Inc., San Jose, CA, USA) to finalize the composition.

Full size image

The land use of the Himalayan transboundary area is mainly Naked land and Shurb, with about 68% of Naked Land in the northern and eastern alpine plateau regions, and about 10% of Shurb in the southern region. From the figure, it can be seen that about 1349.25 km2 (2.94% of the total area) of land use types in the Himalayan transboundary area have changed. Overall, the bare land decreased by 266.59 km2, the shrub decreased by 99.33 km2, the construction land increased by 205.95 km2, and the water area increased by 92.56 km2. According to the specific analysis, the glacier transfer area is the largest, with a total of 566.34 km2 transferred, mainly into bare land 512.00 km2 (1.24%) and water area 10.23 km2 (0.08%). The loss of glaciers is mainly distributed in the Himalayan transboundary area of Shisha Pangma, Khangchendzonga, and so on. This is followed by bare land, which turns mainly to water 114.33 km2 (0.24%), mainly concentrated near glaciers. Shrubs are mainly converted to construction land 72.91 km2 (0.16%), and forest is mainly converted to construction land 50.00 km2 (0.11%), etc. In general, the land use types in the Himalayan transboundary area have changed, and the changes in bare land and construction land are more obvious. Glacier-Naked land, Naked land-Glacier or Water, Shrub-Settlement and Water- Naked land are the main types of land transformation in Himalayan transboundary areas.

Changes in landscape pattern from 2000 to 2020

Based on the theory and method of landscape pattern, using the landscape pattern analysis software FRAGSTAT 4.2, patch density (PD), largest patch index (LPI), landscape shape index (LSI), contagion index (CONTA) and Shannon’s evenness index (SHEI) were selected to analyze the landscape pattern changes in the study area based on the land use distribution map of the Himalayan transboundary area in 2000, 2010 and 2020 (Fig. 4).

Fig. 4
figure 4

Landscape pattern index of the study area from 2000 to 2020. The landscape pattern indices (PD, LPI, LSI, CONTAG, and SHEI) for the years 2000, 2010, and 2020 were calculated using Fragstats 4.2–64 (http://www.umass.edu/landeco/research/fragstats/fragstats.html), the landscape pattern index formula is shown in Table 2. The results were then mapped in ArcGIS v10.8 (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview) and refined using Adobe Illustrator Artwork 28.0 for final composition. It is mainly calculated from the data presented in the Table 1.

Full size image

From 2000 to 2010, the SHEI of land use landscape pattern in the study area decreased from 0.5027 to 0.4983, and the SHEI increased to 0.507 from 2010 to 2020, indicating that the fragmentation degree of land use landscape types in the Himalayan transboundary area increased first and then decreased with the complexity of landscape types. The increase in the PD for Settlement and Water in the eight land classes indicates an increase in fragmentation, while no significant change in the PD for the other land classes indicates little change in the degree of fragmentation. The CONTAG was 63.02 in 2000, increased to 63.28 in 2010, and decreased to 62.06 in 2020, when the maximum degree of aggregation of plaques increased first and then decreased. The LPI of the study area is low, where the index of bare land is much greater than the other land types, indicating the highest dominance of Naked land in the area. The LSI value of the shrub in the study area is the largest, indicating that the longer and irregular the patch boundary is the higher the degree of landscape fragmentation.

From a spatial and temporal point of view, the PD value increases from north to south, and the degree of landscape fragmentation on the southern slope of Everest is much higher than that on the northern slope. The LPI value is also gradually decreasing from north to south, that is, the dominant species in the landscape of the northern slope of Everest is bare land. There is a trend of increasing the LSI values from north to south, with the southern slopes showing a higher degree of landscape fragmentation than the northern slopes. The CONTAG value of the northern slope is higher than that of the southern slope, and the aggregation degree of the northern slope is higher. The SHEI values were higher on south slopes than on north slopes, and the patch types increase, or the proportion of each type of patch tends to be similar.

Features of spatiotemporal dynamics of ecological vulnerability

Spatiotemporal distribution characteristic of ecological vulnerability

The ecological vulnerability index of the study area was calculated based on the spatial principal component analysis model. After standardizing the index, the index is graded according to the established Himalayan transboundary ecological vulnerability grading standard, and the spatial and temporal distribution map of Himalayan transboundary ecological vulnerability is obtained (Fig. 5). From the figure, it can be seen that the ecological vulnerability in the Himalayan transboundary study area is obviously different. The extremely fragile area of ecological vulnerability accounts for a high proportion and a large area is distributed on the northern slope. Especially in 2000, the extremely fragile area accounted for 55.13% of the total area of the study area. It can be seen that the ecological environment of the Himalayan transboundary protection area is easy to be destroyed, extremely sensitive to external interference, and difficult to recover itself. It is urgent to adjust and protect; the ecosystem of the southern slope is more stable than that of the northern slope. The slightly vulnerable areas and mildly vulnerable areas are widely distributed, and only a few areas are severely vulnerable areas and extremely vulnerable areas. Through the comparison of different years, it can be found that the extremely vulnerable areas are decreasing, especially in the eastern part of the northern slope of the study area, the extremely vulnerable areas are transformed into severely vulnerable areas, and gradually tend to integrate, indicating that the ecological environment of the Himalayan transboundary protected areas is getting better.

Dynamic change characteristics of ecological vulnerability

The dynamic changes of transboundary ecological vulnerability in the Himalayas are specifically analyzed as follows (Fig. 5). From 2000 to 2010, the inflow area of the slight-vulnerable area and the mild-vulnerable area was greater than the outflow area, indicating that the trend of the high ecological vulnerability area flowing into the low ecological vulnerability area. It is found that the outflow area of moderate-vulnerable area, severe-vulnerable area and extreme-vulnerable area is large, especially the outflow area of extremely vulnerable area is 3281.25 km2, accounting for 23% of the area of extreme-vulnerable area in 2000. At the same time, the area flowing into the severe-vulnerable area is the largest, which is 3939.75 km2, and its main source is 662.25 km2 flowing out of the moderate-vulnerable area and 3277.5 km2 flowing out of the extreme-vulnerable area. From 2010 to 2020, the inflow areas of slight-vulnerable area, mild-vulnerable area and moderate-vulnerable area increased significantly compared with the previous period, flowing into 391.25 km2, 1296 km2 and 1762.25 km2 respectively, indicating that the trend of high ecological vulnerability areas flowing into low ecological vulnerability areas remained unchanged. The severe-vulnerable area and the extreme-vulnerable area showed a higher outflow area, which were 3157.5 km2 and 3421 km2 respectively, accounting for 33% and 15% of the severe vulnerable area and the extremely vulnerable area in 2010. Among them, the severe-vulnerable area mainly flows to the moderate-vulnerable area, accounting for 53% of the outflow area, and the extreme-vulnerable area mainly flows to the severe-vulnerable area, accounting for 99% of the outflow area. Meanwhile, the area transferred to the severe-vulnerable area is the largest, which is 3524.25 km2. Its main source is 123.25 km2 transferred out of the moderate-vulnerable area and 3401 km2 transferred out of the extreme-vulnerable area.

Fig. 5
figure 5

Changes in ecological vulnerability levels in the Himalayan transboundary area from 2000 to 2020. This figure was created by the author using ArcGIS v10.8 (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview). software to visualize indicators such as population density, mountain disaster risk index, climate, water environment, topography, vegetation, annual average temperature, annual precipitation, slope, modified normalized water index, and normalized vegetation index. Principal component analysis was conducted using the raster calculator, and the results were overlaid to produce ecological vulnerability maps for the Himalayan region for the years 2000, 2010, and 2020. These maps were then compared to assess the index changes across the different years. Finally, the image was composed using Adobe Illustrator 28.0 (Adobe Inc., San Jose, CA, USA). It is mainly calculated from the data presented in the Table 1.

Full size image

Evaluation of comprehensive index of ecological vulnerability

The comprehensive index of ecological vulnerability in 2000, 2010, and 2020 was 4.05, 3.99, and 3.86, respectively (Fig. 6). From 2000 to 2020, the overall ecological vulnerability of the Himalayan transboundary region showed a downward trend, and the ecosystem became more stable and healthier.

In 2000, 2010, and 2020, the mean value of the comprehensive index of ecological vulnerability on the northern slope was 4.51, while the mean value on the southern slope was 2.34. The comprehensive index of ecological vulnerability on the northern slope of Mount Qomolangma was significantly higher than that on the southern slope, so the overall ecological environment quality on the northern slope was significantly lower than that on the southern slope (Fig. 7). Through the trend line, it can be clearly seen that the ecological vulnerability comprehensive index of the northern and southern slopes shows a downward trend, and the overall ecological environment situation tends to be stable and improved. By comparing the comprehensive index of ecological vulnerability of nature reserves and national parks across the Himalayas, the ecological vulnerability of nature reserves and national parks varies greatly. The comprehensive index of ecological vulnerability of nature reserves and national parks in 2020 is significantly lower than that in the previous two years. The ecological environment of nature reserves and national parks has been improved between 2000 and 2020. The average value of the comprehensive index of ecological vulnerability in the three years was calculated and sorted. The order from high to low was Qomolangma National Nature Preserve (4.51) > Sagarmatha National Park (3.12) > Manaslu Conservation Area (3.02) > Kanchenjunga Conservation Area (2.58) > Langtang National Park (2.28) > Makalu Barun National Park (1.78) > Gaurishankar Conservation Area (1.69). Among them, the average value of the comprehensive index of ecological vulnerability of Qomolangma National Nature Preserve is the largest, which is 4.51. The average value of the comprehensive index of ecological vulnerability in Gaurishankar Conservation Area is the smallest, which is 1.69. the largest average value of the southern slope of the study area appears in the Sagarmatha National Park, which is 3.12. As an important way to climb Everest from the southern slope, Sagarmatha National Park has always been a tourist attraction for climbers and tourists. It has been disturbed more frequently and its ecological vulnerability is relatively high. Among them, the average value of the comprehensive index of ecological vulnerability of Mount Qomolangma Nature Reserve is the largest, which is 4.51. The average value of the comprehensive index of ecological vulnerability in Gaolisanka Nature Reserve is the smallest, which is 1.69. The largest average value of the southern slope of the study area appears in the Sagarmatha National Park, which is 3.12. As an important way to climb Everest from the southern slope, Sagarmatha National Park has always been a tourist attraction for climbers and tourists. It has been disturbed more frequently and its ecological vulnerability is relatively high.

Fig. 6
figure 6

Interannual variation of Himalayan transboundary ecological vulnerability index. This figure is an original work by the author, created using Origin 2024 (OriginLab Corporation, Northampton, MA, USA). It presents the results of a linear analysis of ecological vulnerability indices, which were evaluated to assess the relationship between key environmental factors.

Full size image
Fig. 7
figure 7

Radar chart of the comprehensive index of the ecological vulnerability of Himalayan transboundary nature reserve.

Full size image

The main factors affecting the distribution of landscape pattern

On the basis of the significance test, the measurement of its relative importance is helpful to explain the contribution value of CATREG model to regression. The R2 value of the fitting model was between 0.002 and 0.557, and all the fitting models passed the F test (P < 0.05), which was statistically significant. The tolerance of all variables is high enough to ensure that multicollinearity problems are excluded. For the same landscape index, the regression coefficients and significance of different influencing factors are different. On the basis of significance test, the relative importance measure is helpful to explain the contribution value of CATREG model to regression (Table 4). Compared with other important values, Tmp (0.38), Worldpop (0.28), Pre (0.24), and NDVI (0.18) are the main factors affecting PD. The importance of Tmp shows a downward trend, while the importance of pre-shows an upward trend. Pre (0.43), Tmp (0.26), and NDVI (0.26) were the main factors affecting LPI. The importance of pre increased first and then decreased, the importance of NDVI decreased, and the importance of MNDWI decreased first and then increased. Pre (0.44), NDVI (0.30), and Slope (0.28) were the main factors affecting LSI. The importance of pre increased first and then decreased, the importance of NDVI decreased, and the importance of Slope decreased first and then increased. Pre (0.61) and Slope (0.26) were the main factors affecting CONTAG, and the importance of Pre showed a downward trend. Pre (0.65), NDVI (0.26), and MNDWI (0.22) are the main factors affecting SHEI. The importance of pre-shows a downward trend, the importance of NDVI increases first and then decreases, and the importance of MNDWI decreases first and then increases.

Table 4 The importance of different environmental factors on the distribution of landscape pattern. Note: All fitted models exhibit R² values between 0.002 and 0.557 and pass the F-test (P < 0.05), indicating statistical significance. The tolerance levels of all variables are sufficiently high to avoid multicollinearity. Based on significance testing, the relative importance of variables helps explain the contribution of the CATREG model to the regression, with  higher importance indicating greater influence.
Full size table

Conclusions and discussions

Conclusions

Under the pressures of global climate change and socio-economic challenges, the transboundary regions of the Himalayas are experiencing increasing ecological vulnerability, necessitating a comprehensive analysis of driving factors for effective conservation strategies. This study examines ecological landscape changes from 2000 to 2020 in these regions, focusing on spatial and temporal patterns of vulnerability. The findings show that land use has shifted due to the reduction of glacial cover and the expansion of barren land, exacerbated by global warming and increased moisture. Landscape fragmentation increased initially but decreased over time, with barren land dominating the landscape, particularly on the northern slope of Mount Everest, which exhibited lower fragmentation than the southern slope. Ecological vulnerability has generally decreased over the past two decades, with the proportion of highly vulnerable areas decreasing by 9.83%, while the share of less vulnerable areas increased. The composite ecological vulnerability index dropped from 4.05 in 2000 to 3.86 in 2020, indicating overall ecological improvement. Spatially, the southern slope has lower ecological vulnerability than the northern slope, with 59% of the southern slope categorized as mildly or moderately vulnerable in 2020, compared to 60% of the northern slope being extremely vulnerable. Vulnerability hotspots have shifted, with increasing vulnerability observed at the boundaries of protected areas, especially between Sagarmatha National Park and the Everest Natural Reserve. The average ecological vulnerability index for 2000–2020 showed significant improvement across all protected areas, with the highest index in the Everest Natural Reserve (4.51) and the lowest in the Gauri Sankar Conservation Area (1.69). Analysis using the GATREG model identified key drivers of landscape metrics, including annual temperature, population density, and precipitation, which influenced patch density (PD), largest patch index (LPI), landscape shape index (LSI), contagion (CONTAG), and the Shannon entropy index (SHEI).

Discussions

This study analyzes landscape changes and ecological vulnerability in the transboundary region of the Himalayas from 2000 to 2020. Over this period, climate change and human activities have caused significant land-use shifts, including reduced glacier cover, increased bare land, and tree line shifts, resulting in the migration of alpine vegetation and biodiversity, particularly above 2000 m. The northern slope of Mount Everest is more ecologically vulnerable than the southern slope, mainly due to drier conditions and insufficient precipitation hindering ecological recovery. Despite these challenges, the overall ecological environment of the Everest region shows signs of stability. This study is the first to use the VSD model and landscape pattern driving force analysis to assess ecological vulnerability in this region and identify key driving factors, with the goal of promoting sustainable development in cross-border areas.

  1. (1)

    Ecological vulnerability in the Himalayas

The Everest region and the broader Himalayan area face significant ecological vulnerability due to the dual impacts of global climate change and human activities. These factors exacerbate environmental issues, testing the resilience and capacity of the ecosystem. Understanding how climate change and human interventions affect ecological vulnerability is complex, as it involves not only the adaptability of ecosystems but also challenges for mountain communities, which manifest differently across various temporal and spatial scales. Specifically, ecosystems above 2000 m exhibit heightened vulnerability, with land use, topography, meteorological conditions, and soil moisture as key contributing factors52. Climate change has significantly impacted flagship species’ habitats, leading to habitat compression and altered geographic distributions53. Notably, land use changes have profound implications for species, with projections suggesting that approximately 30% of snow leopard habitats may disappear due to rising tree lines and shrinking high-altitude areas54. In Langtang National Park, climate change has markedly affected the cryosphere, impacting both local ecosystems and community livelihoods55. Despite these challenges, findings from this study indicate a trend toward decreased ecological vulnerability in the Everest region at the landscape level, suggesting that the ecological environment is stabilizing and improving.

  1. (2)

    Landscape pattern evolution and sustainable development

This study highlights the similarities and differences in landscape fragmentation and ecosystem vulnerability within the Himalayan region. Research demonstrates that landscape fragmentation exhibits complex dynamics across various temporal and spatial scales, aligning with our finding that fragmentation initially increases before decreasing56,57. Notably, fragmentation of built-up areas and water bodies has risen, while bare land remains the dominant land cover, complementing previous work on land use patterns and ecosystem dynamics58. Additionally, we observed that shrubland is the most fragmented, with irregular and significantly fragmented patch boundaries, echoing Sharma on the impact of forest fragmentation on biodiversity59. Our analysis reveals that the southern slope of Everest experiences greater landscape fragmentation than the northern slope, with increased complexity in the south, which contrasts with the dominance of bare land in the north. This regional disparity offers new insights, highlighting the varied impacts of human activities and climatic conditions on landscape structure60. The findings indicate that changes in landscape structure are significantly influenced by human activity and climate change, with different landscape types exhibiting varying levels of vulnerability and diversity, providing a scientific basis for future landscape management policies. Landscape fragmentation and land use changes pose threats to ecosystems and sustainable development, particularly due to the expansion of built-up areas and increased bare land, which negatively affect biodiversity and ecological functions. The differences in landscape complexity between the southern and northern slopes emphasize the need for regionally tailored management that accounts for ecological heterogeneity. Fragmentation leads to habitat isolation, compromising ecological connectivity and ecosystem services. Therefore, enhancing ecological corridors, vegetation restoration, and sustainable agriculture is essential to mitigate fragmentation impacts on vulnerable ecosystems. International cooperation is crucial for addressing climate change and promoting sustainable development in this region.

  1. (3)

    The future of protected areas in the Himalayas

The Himalayan region is home to 55 national parks and nature reserves, with 49 located within 100 km of international borders61. These transboundary protected areas are crucial for conserving cross-border species, promoting sustainable development, and enhancing peaceful cooperation62. Focusing on the example of Nepal and China, both nations prioritize ecological protection and sustainable development in the Everest region, establishing various parks and implementing environmental policies to foster innovative conservation models. In 1973, Nepal enacted the National Parks and Wildlife Conservation Act, establishing the Department of National Parks and Wildlife Conservation to regulate activities such as overgrazing, deforestation, and wildlife poaching within protected areas. In 1988, China established the Mount Everest Nature Reserve, and in 2018, it implemented management regulations to enhance biodiversity protection and resource management, strengthening the capabilities of conservation authorities. Additionally, in 2020, Nepal introduced a ban on plastic products under 30 microns thick in specific areas to improve the ecological environment. Looking ahead, the two countries plan to merge Sagarmatha National Park and Mount Everest National Park, creating the world’s highest international peace park, aligning with the United Nations Sustainable Development Goal 17, which emphasizes the importance of international cooperation for sustainable development63,64.