Exploring Impacts of Land Use and Cover Changes on Ecosystem Services on the Qinghai-Xizang Plateau

Abstract

The Qinghai-Xizang Plateau (QXP), as Asia’s “Water Tower” and a global climate regulation hub, provides essential ecosystem services that sustain global ecological security and the well-being of 2 billion people. However, the fine-scale relationship between land use changes and ecosystem services on the QXP, as well as the underlying driving mechanisms, remain under researched. Therefore, the InVEST model was used in this study to assess the distribution of key factors of ecosystem services—habitat quality (HQ). A four-quadrant diagram method was proposed to quantitatively identify the dynamic relationship between land use changes and habitat quality. The XGBoost-SHAP algorithm was used to determine the main driving factors affecting the conversion between coordinated and conflicting areas. This study finds the following: (1) the proportion of land area undergoing land use change on the QXP is approximately 7.5%, with significant grassland degradation; (2) habitat quality distribution shows a high edge-low northern pattern, and forest land has the greatest impact on habitat quality changes; (3) in the past 20 years, the relationship between land use changes and habitat quality has shown a trend of coordinated development, with a 50:1 ratio of coordination to conflict evolution. The relationship between cities and water source areas has experienced the most dramatic changes; moreover, in the conversion between coordination and conflict regions, natural factors are the main drivers, followed by socio-cultural factors. This study reveals the dynamic relationship between land use changes and habitat quality interactions on the QXP, providing a scientific basis for regional sustainable development and related planning.

1. Introduction

Ecosystems constitute the fundamental pillar of Earth’s life-support systems, delivering direct and indirect provisioning services essential for human livelihoods [1,2]. However, rapid socio-economic development has intensified human–land conflicts, with unsustainable land use practices exacerbating ecosystem degradation [3]. This anthropogenic pressure has precipitated biodiversity depletion and ecosystem unbalance [4,5,6], critically impairing their structural integrity, functional capacity, and service efficiency [7,8,9]. Consequently, ecosystem degradation and the collapse of biodiversity have been identified by the Global Risks Report of the World Economic Forum as the third most critical global economic risk over the next decade. Responding to this urgency, enhanced conservation and ecosystem restoration have been prioritized by the UN’s 2030 Agenda for Sustainable Development as essential Sustainable Development Goals, highlighting the forthcoming decade as a pivotal window for addressing ecological risks and initiating comprehensive recovery [10,11]. Within this framework, it is crucial to evaluate how ecosystem services are affected by land use change, as this understanding is fundamental for refining land use planning and formulating effective ecological conservation strategies, particularly in regions with vulnerable ecological characteristics.

The QXP, a central area within the ecologically sensitive Third Pole region, serves as an ecological security buffer for China, South Asia, Central Asia, and Southeast Asia [12]. This mega-ecosystem demonstrates extraordinary sensitivity to global climate change, serving as both a critical amplifier and early warning indicator of planetary environmental shifts [13]. Given this ecological prominence, the QXP has emerged as a focal area in global change research, owing to its unparalleled geophysical attributes and bio-geographic uniqueness. Its land use/cover change (LUCC) patterns fundamentally govern the structural–functional stability of this ecological barrier, with cascading impacts on regional socio-ecological sustainability and global environmental equilibrium. Over the past two decades, policy initiatives such as the “Western Development Strategy” and the “Poverty Alleviation Campaign” have facilitated human activities in this region, notably urban expansion and infrastructure development, which have contributed to economic growth. However, human activities have triggered considerable transformations in land use patterns and land cover characteristics. Given the inherently fragile ecological state of the region, such transformations have significantly impacted habitat quality (HQ) across the QXP [14,15]. HQ serves as a direct proxy for biodiversity integrity [16]. It is also an important indicator for measuring the health of an ecosystem, reflecting the environment’s ability to support the survival and reproduction of species [17], and has become one of the foremost challenges in global conservation efforts. Empirical studies reveal that urbanization and agricultural intensification account for over 80% of habitat degradation in specific hotspots [18]. Spatially explicit analyses of LUCC dynamics provide critical insights into biodiversity patterns, given the inherently geographical nature of ecological processes [19]. Current research on plateau ecosystem responses to LUCC shifts predominantly focuses on the following: typological studies, from single-species habitats [20] to river basins [21]; spatial scales; provincial analyses [22]; and urban clusters [23], using resolutions including 10 km [24,25], 1 km, and 250 m [26] grids, and so on. These studies seldom integrate systemic interactions between LUCC and HQ at a plateau-wide scale, nor do they leverage explainable machine learning to dissect drivers—a gap that our quadrant-XGBoost framework addresses. Despite these considerations, substantial research gaps remain. Studies on the entire plateau at refined spatial and temporal resolutions are still insufficient, hindering a comprehensive understanding of ecosystem service responses to LUCC. Such research gaps constrain effective exploration of underlying ecosystem service mechanisms. Therefore, clarifying the response relationships between ecosystem services and LUCC in the QXP is critically important for guiding sustainable regional development strategies.

Land use represents the most direct anthropocentric interface with ecosystems, serving as a pivotal driver of ecosystem service alterations. Quantitative assessment of LUCC impacts on ecosystem services constitutes a critical nexus for bridging land management and ecological conservation [27]. Precise quantification becomes imperative to decipher the cascading environmental consequences of such transformations [28,29]. Yang et al. [30] and Gong et al. [31] documented enhanced soil–water conservation and windbreak services through cropland-to-grassland/forest conversion. Grassland-to-barren-land transitions exhibited significant ecosystem service depletion. Cropland restoration to grasslands substantially increased sand fixation capacity [32,33]. The transformation from forested regions into agricultural lands can potentially cause local temperature increases and precipitation decreases as a result of enhanced surface albedo and diminished surface roughness [34]. Simkin et al. [35] anticipated the implications of expanding urban areas on future biodiversity. Nevertheless, significant constraints exist within current research frameworks, which mostly emphasize isolated interactions between single human activities and individual ecological factors rather than conducting macro-level assessments and comprehensive analyses of both human and ecological systems in an integrated manner [36]. Additionally, ecosystem service assessments based on different research scales and objectives are difficult to compare and even more challenging to integrate into national-scale ecosystem conservation and land use decision making [37]. Therefore, performing a fine-resolution quantitative evaluation of how LUCC influences ecosystem services in the QXP can establish a solid scientific basis for ecological management and spatial planning within this region. Such analyses could effectively support coordinated and sustainable integration among economic development, ecological conservation, and human livelihoods.

To address the aforementioned limitations, this study proposes the use of a four-quadrant model integrated with 30 m resolution remote sensing data to quantitatively identify spatiotemporal dynamics and interactions between LUCC and HQ on the QXP from 2000 to 2020. Additionally, the XGBoost-SHAP method is employed to identify the primary factors influencing transitions between coordinated and conflicting relationships. This research specifically enhances previous studies by (1) quantitatively evaluating the spatiotemporal evolution patterns of LUCC across the QXP from 2000 to 2020 based on detailed land use change maps and trajectory analyses; (2) quantitatively assessing the spatiotemporal variations of HQ using the InVEST HQ module during the same period; and (3) utilizing the four-quadrant model to explore dynamic relationships between LUCC and HQ on the plateau, and subsequently analyzing key driving factors underpinning the transitions between coordinated and conflicting regions. These findings are intended to provide critical insights for guiding practical land use management and spatial planning decisions on the QXP.

2. Materials and Methods

2.1. Study Area

The QXP, often termed the “Third Pole of the Earth” and the “Water Tower of Asia,” is the source of nine major Asian rivers and delivers vital ecosystem services and resources to more than 1.5 billion inhabitants [38]. This region encompasses rich natural resources and contains nearly all ecosystem types found throughout China. As an essential ecological safeguard and strategic reserve of natural resources, the plateau significantly contributes to global biodiversity conservation and climate stability, offering an important platform for natural ecosystem research (Figure 1).

Nevertheless, the plateau’s mountainous climate, defined by cold temperatures, low precipitation, and overall aridity, restricts land availability and its sustainable usage. Intensified agricultural practices further heighten ecological pressures. Patterns of land use and land cover are integral to maintaining the structural and functional stability of this ecological security barrier; moreover, changes in these patterns substantially affect environmental quality. Over the past four decades, built-up areas in the QXP have expanded threefold, while road infrastructure has quadrupled, posing considerable threats to local ecological conditions. Therefore, gaining a comprehensive understanding of how land use transformations impact ecological conditions in the QXP is imperative for supporting sustainable development on both regional and global levels.

2.2. Data Sources

This study utilizes the China Land Cover Dataset (CLCD), which was collaboratively created by Yang et al. (https://zenodo.org/record/4417809 (accessed on 15 July 2025)) [39]. With a 30 m spatial resolution, continuous coverage spanning three decades, and an overall accuracy rate of 79.31%, the dataset effectively fulfills the demands of this analysis. Compared with datasets such as MCD12Q1, ESACCI_LC, FROM_GLC, and Globe Land30, the CLCD exhibits superior accuracy. It categorizes land cover into nine distinct classes, specifically the following: farmland, forest, shrubland, grassland, water, ice/snow, barren land, and built-up areas. The dataset used to identify driving factors integrates natural elements—including annual mean precipitation, annual mean temperature, soil characteristics, elevation, slope, and proximity to protected zones—as well as socio-economic indicators like GDP, population density, and nighttime illumination data. Detailed information regarding these data sources is summarized in

Table 1. Data sources of driving factors.

Types	Description	Source
Natural factors	Digital elevation model (DEM)	https://lpdaac.usgs.gov/products/astgtmv003/ (accessed on 15 July 2025)
	Precipitation (PER)	https://www.geodata.cn/main/ (accessed on 15 July 2025)
	Temperature (TEM)	https://www.geodata.cn/main/ (accessed on 15 July 2025)
	SLOPE	Derived from DEM data
	Soil organic carbon (SOC)	https://gaez.fao.org/pages/hwsd (accessed on 15 July 2025)
	The distance to the nature reserve (DIS)
Humanistic and social factors	The proportion of industry GDP (GDP)	https://www.resdc.cn/ (accessed on 15 July 2025)
	Population density (POP)	https://www.worldpop.org/ (accessed on 15 July 2025)
	Nighttime lights (NTL)	http://www.geodata.cn (accessed on 15 July 2025)

. Ultimately, all datasets were uniformly reprojected to the UTM Zone 46 coordinate system.

2.3. Methods

2.3.1. Land Use Change Map

To investigate the spatial variability of land use types over the research period, a mapping-based approach was adopted to measure conversions among land use categories. The calculation formula applied is presented below:

$$\mathrm{C}=10\mathrm{A}+\mathrm{B}$$

(1)

In this expression, C refers to the raster code indicating land use conversion, A corresponds to the initial land use category code, and B signifies the subsequent land use category code.

2.3.2. Change Trajectory Method

The land use change trajectory analysis method, which characterizes temporal variations of land use dynamics, provides a robust technique for capturing continuous and discrete patterns of change. Temporal land use shifts are represented through trajectory codes, where each raster is assigned numeric or alphabetic codes denoting its respective land use category at specified time intervals. The computational formula utilized for this analysis is as follows:

$$T_{ij}={\left(G1\right)}_{ij}\times 9^{n-1}\times +{\left(G2\right)}_{ij}\times 9^{n-2}+\cdots +{\left(Gn\right)}_{ij}\times 9^{n-n}$$

(2)

where Tij represents the trajectory code value of the raster at row i and column j in the trajectory result raster image, indicating the change process of different land use types, with no mathematical significance. n is the sequence number of the research sample. (G1)ij, (G2)ij, …, (Gn)ij are the land use type code values at different time nodes on the raster images.

2.3.3. HQ Assessment

Field surveys of biological samples offer a more precise and quantitative evaluation of HQ and its suitability for species survival. However, the unique environmental conditions of the QXP present significant challenges for conducting field surveys, and the findings are not readily generalizable to other regions. Consequently, this study employs the HQ module of the InVEST model to assess HQ and reflect the region’s biodiversity levels. The principal formula for this assessment is as follows:

$$Q_{xi}=H_i\left[1-\left({\displaystyle \frac{D_{xi}^z}{D_{xi}^z+k^z}}\right)\right]$$

(3)

where Q_xi refers to the HQ value for land use category i at a given raster $x$ location. Hi denotes habitat suitability specific to category i. Dxi indicates the cumulative stress impacting raster $x$ within land use category $i$. $z$ represents a normalization constant, typically assigned a default value of 2.5. $k$ stands for the half-saturation parameter, usually set at 0.5. Taking into account the actual characteristics of the study area and referring to the settings of the numerical values for the sensitivity of HQ and threat source factors in References [40,41], the main parameters required for running this module and the parameter settings for other modules are shown in

Table 2. Main parameters used in the InVEST model (HQ) module.

Function	LUCC	Habitat Suitability of Different Land Types	Sensitivity to Cropland	Sensitivity to Barren Land	Sensitivity to Impervious Surfaces
HQ	Cropland	0.9	0.8	0.8	0.6
	Forest	1	0.9	0.6	0.6
	Shrub	0.8	0.8	0.9	0.8
	Grassland	0.8	0.7	0.9	0.7
	Water	0.9	0.7	0.6	0.7
	Snow/Ice	0.7	0.8	0.9	0.8
	Barren	0	0	0	0
	Impervious	0	0	0	0
	Wetland	0.6	0.6	0.7	0.7

.

2.3.4. Four-Quadrant Method

The four-quadrant method, originally a time management theory, has recently been applied across various disciplines. This method involves analyzing and weighing two distinct attributes, categorizing them into four quadrants, and arranging them according to the sequence of their guiding content. The items are sorted in a counterclockwise direction based on the importance of their research objectives. In this study, a four-quadrant analytical framework was designed based on land use variations and HQ dynamics, classifying their interactive evolutionary processes into four separate categories (Figure 2). The formula applied for this classification is expressed as follows:

$$\left({\mathrm{change} \mathrm{map},\mathrm{HQ}}_{\mathrm{t}1}{-\mathrm{HQ}}_{\mathrm{t}0}\right)=\left\{\begin{array}{l}(\mathrm{change},+),\mathrm{I},\mathrm{Coordination}\\ (\mathrm{no} \mathrm{change},+),\mathrm{II},\mathrm{Good} \mathrm{for} \mathrm{habitat} \mathrm{quality}\\ (\mathrm{no} \mathrm{change},-),\mathrm{III},\mathrm{Degradation}\\ (\mathrm{change},-),\mathrm{IV},\mathrm{Conflict}\end{array}\right.$$

(4)

In the first quadrant, the relationship between land use change and HQ is mutually coordinated, which represents an ideal state. The second quadrant is beneficial to the habitat, where the HQ improves without any changes in land use, further consolidating the ecological environment development in the QXP. The third quadrant represents a situation where HQ declines without changes in land use, indicating a state of degradation. The fourth quadrant shows a decline in HQ due to land use changes, which may be related to economic construction activities, where the relationship is in a conflict state.

2.3.5. XGBoost-SHAP Analysis Based on Raster Data

XGBoost is an efficient gradient-boosting decision tree algorithm. Its core idea is to iteratively train weak classifiers (decision trees) and adjust the sample weights based on the prediction errors from the previous round, gradually optimizing model performance. XGBoost prevents overfitting by introducing regularization terms, supports parallel computation to accelerate training, and approximates the loss function using second-order Taylor expansion to improve accuracy. The algorithm excels in feature selection, handling missing values, and model interpretability, making it widely applicable to classification, regression, and ranking tasks. In this study, the XGBoost model is used to predict the impact of various factors on the interaction between land use change and HQ.

The Shapley Additive Explanations (SHAP) method is an interpretability technique derived from cooperative game theory, aiming to quantify and interpret feature importance within machine learning models. It calculates each feature’s marginal contributions to the overall predictions, ensuring a fair and consistent attribution of feature significance through principles such as symmetry, additivity, dummy property, and efficiency. SHAP provides both global and local interpretations and supports various models. The output is the SHAP value for each feature in every sample, with positive values indicating a contribution that promotes the prediction and negative values indicating a suppression of the prediction. In this study, the Shapley values are calculated to measure the driving effects of various factors on the impact of coordination and conflict regions, as well as their importance ranking. The formula for calculating the Shapley value is as follows:

$${\varnothing }_i={\displaystyle \sum _{S\subseteq N\backslash \left|i\right|}\frac{\left|S\right|!(\left|N\right|-\left|S\right|-1)!}{\left|N\right|!}}(v(S\cup \left\{i\right\})-v(S))$$

(5)

In the formula, $N$ represents the set of all features; $S$ is any feature subset that does not contain feature $i$; $\left|S\right|$ is the number of features in set $S$; $v(s)$ is the contribution of feature set $S$ to the model’s prediction output; and $v(S\cup \left\{i\right\})$ is the contribution of feature set J, which includes feature $S\cup \left\{i\right\}$, to the model’s prediction output.

3. Results

3.1. Spatial Changes in Land Use and Land Cover in the QXP

The QXP region experienced a 7.5% change in land use during the study period, indicating a relatively stable overall transformation. From 2000 to 2020, the region underwent extensive construction activities, resulting in spatial alterations in LUCC, as illustrated in Figure 3. Notable changes were predominantly observed in the central plateau, south of the Nyainqêntanglha Mountains, and around Qinghai Lake. However, the fragile natural environment and characteristic mountainous terrain of the QXP pose challenges to development in these areas. Other regions exhibited minimal spatial changes, lacking significant trends in contiguous, large-scale transformations. From 2000 to 2020, the land use patterns on the QXP were significantly altered due to accelerated transportation and energy infrastructure development, urbanization, overgrazing, industrial–agricultural transformation, and population growth. This phenomenon was particularly pronounced around the Qaidam Basin ((e) 2000–2020 B). Notably, extensive land expansion occurred in the peri-urban areas, such as the southern part of the Western Sichuan Plateau and the northern Hengduan Mountain region. Conversely, peripheral mountainous regions of the plateau exhibited minimal land use transitions, attributable primarily to harsh environmental constraints limiting extensive agriculture, industrialization, and human settlement.

3.2. Temporal Changes in Land Use and Land Cover of the QXP

To clearly explore the land use changes in the QXP, the conversion of the same land use type between adjacent years was removed during the time-series statistical process, resulting in land use transition maps for each year (Figure 4). Across all four periods, the changes remained relatively stable. Based on the scale of land use conversions, interannual variation characteristics, and the impact of the Ecological Conservation Redline (ECR) policy implemented in 2017, the time-series land use change can be divided into two distinct phases. From 2000 to 2015, the transition amounts in the three periods were relatively similar, indicating stable changes during this period. However, from 2015 to 2020, the area of land use change was the largest, suggesting that land use transitions were more frequent during this period. Between 2000 and 2015, there were relatively fewer conversions overall, with the largest area being the mutual conversion between grassland and abandoned land. This result suggests that, under human influence, grassland degradation was severe during this period on the QXP. Between 2015 and 2020, land use transitions were frequent, with the conversion between grassland and abandoned land continuing to be the largest conversion type. There was also a conversion between ice and snow regions and abandoned land. Since 2015, grassland areas have sharply decreased, while water bodies and abandoned land areas have significantly increased. During this period, the Western Development Strategy attracted substantial investments, driving the construction of infrastructure projects such as roads, airports, and energy. In some regions, key environmental governance projects were initiated to strengthen the protection of plateau wetlands and nature reserves, implementing natural resource conservation and ecological restoration measures. Urbanization accelerated in some cities, such as Lhasa and Xining.

The land use transformations from the last two decades were quantified through trajectory analysis (Figure 5). Land use categories labeled A to I correspond, respectively, to cultivated land, forest, shrubland, grassland, water bodies, ice/snow areas, abandoned land, impermeable surfaces, and wetlands. The line thickness represents the magnitude of conversion, while the arrows indicate the direction of land use transformation. The notation ‘A > B’ indicates the conversion from land use type A to type B, where the land use type before the ‘greater-than’ symbol changes to the type after it. This study revealed that, over the past 20 years, the QXP experienced the most substantial changes in grassland, ice and snow, and abandoned land. Notably, the most significant conversions occurred between grassland and abandoned land, followed by conversions between ice and snow areas and abandoned land. Although other land use types also experienced conversions, these were of lesser magnitude. Grassland exhibited the largest outflow, whereas abandoned land showed the largest inflow over the past two decades. Changes in impermeable surfaces and wetlands were minimal. Both impermeable surfaces and wetlands constitute a minor proportion of the land use types on the QXP, with wetland resources being well-preserved under ecological management. The changes in impermeable surfaces are mainly influenced by development policies and the constraints of the QXP’s topography and climate, with minimal changes compared to other land use types.

3.3. HQ Analysis of the QXP from 2000 to 2020

3.3.1. Changes in HQ of the QXP from 2000 to 2020

HQ assessments for the QXP were conducted using the InVEST habitat quality module, and the results are depicted in Figure 6. Higher values represent improved HQ and more effective ecological protection. Due to the inherently sensitive and fragile ecosystem of the plateau, shifts in land use significantly impact habitat stability, as intensification of land use modifications typically exacerbate habitat threats. From 2000 to 2020, the habitat quality on the QXP exhibited significant spatial variations with pronounced regional disparities. The areas of lower values were primarily distributed in the northern regions, moderate values in the central-southern regions, and higher values along the marginal areas of the plateau. This spatial pattern reveals a gradual decline in habitat quality from the southeast to the northwest. Over the two decades studied, the Kunlun, Qilian, and Altun Mountain ranges consistently presented low HQ, whereas relatively high HQ persisted in regions such as the Hengduan Mountains and the southern Xizang Valley. Further analysis of HQ dynamics across four distinct time intervals indicated limited habitat deterioration from 2000 to 2005, predominantly concentrated in northern areas, specifically around the Kunlun, Qilian, and Altun Mountains, with only scattered instances in the southern plateau. The total affected area was roughly 2%, implying relatively stable habitat conditions during this interval. From 2005 to 2010, habitat degradation remained focused primarily near the Qilian Mountains, showing similar spatial distributions to the preceding period, though with greater intensity and a denser occurrence of declines in northern regions compared to those of the south. The 2010–2015 period differed from the preceding stages, as it featured an inverted U-shaped zone of decline along the northern margin of the southern Xizang Valley. In the 2015–2020 period, the concentrated area of continuous decline shifted to the junction of the Karakoram and Qiangtang Plateau, while the remaining areas exhibited a more scattered distribution.

3.3.2. Changes in HQ for Different Land Use Types

Table 3. Average HQ index for each land use type.

Land Use Type	2000	2005	2010	2015	2020
Cropland	0.60	0.62	0.61	0.62	0.62
Forest	0.91	0.90	0.92	0.91	0.91
Shrub	0.90	0.88	0.88	0.89	0.86
Grassland	0.80	0.82	0.83	0.83	0.82
Water	0.85	0.91	0.86	0.89	0.88
Snow/Ice	0.88	0.86	0.89	0.88	0.88
Barren	0	0	0	0	0
Impervious	0	0	0	0	0
Wetland	0.80	0.79	0.81	0.82	0.81

summarizes HQ indices corresponding to different land use categories from 2000 to 2020. Among these, the highest mean HQ was observed in forests and shrublands, followed by water bodies and grasslands, whereas impermeable surfaces exhibited the lowest mean HQ index. Throughout the entire research period, notable fluctuations in HQ predominantly occurred in shrublands and water bodies, while croplands, grasslands, impermeable surfaces, and fallow lands exhibited relatively stable patterns. Although HQ changes were minor between 2000 and 2005, they intensified considerably after 2010, with water bodies showing the most prominent alterations. From 2010 onwards, intensified human activities on the QXP, together with increased grassland degradation, desertification, and biodiversity losses, collectively contributed to reduced HQ across multiple land use types. However, subsequent ecological conservation measures implemented later effectively alleviated further deterioration in HQ.

3.3.3. Interactive Effects of Land Use Change and HQ on the QXP from 2000 to 2020

To precisely delineate interactions—either conflicting or coordinated—between land use changes and HQ dynamics, the four-quadrant model was applied to evaluate their evolving relationship over four intervals: 2000–2005, 2005–2010, 2010–2015, and 2015–2020 (Figure 7). Throughout this 20-year period, substantial shifts occurred in the relationship between land use modifications and HQ, exhibiting a dispersed distribution without a clear concentrated pattern from 2000 to 2015. However, in the subsequent period from 2015 to 2020, a pronounced pattern of concentrated and contiguous distribution became evident. Specifically, between 2000 and 2005, regions of conflict between land use development and HQ were predominantly situated at the peripheries of lakes, along roadways, and in the vicinity of the Qilian Mountains. Areas that were beneficial to HQ were interspersed with conflict zones, particularly concentrated at the intersection of the Karakoram Mountains and the Qiangtang Plateau ((a) 2000–2005A). In contrast, degraded areas constituted a smaller proportion and were primarily located around Xining ((a) 2000–2005 C). Between 2005 and 2010, the dynamics of the relationship between land use change and HQ exhibited patterns analogous to those observed in the preceding phase. Areas of conflict were predominantly situated at the western extremity of the Qiangtang Nature Reserve, whereas regions conducive to HQ were identified along the peripheries of lakes and within forested areas.

From an analysis of the proportion of the four types at each stage, it is evident that the coordination area consistently represented the largest share, whereas the degraded area constituted the smallest. The area of coordination significantly exceeded that of conflict, at a ratio approaching 50:1 (Figure 8). Notably, coordinated regions showed limited overall change, experiencing only slight declines during the 2015–2020 interval. Generally, regions supportive of HQ gradually increased, while conflicting zones expanded continuously over the study duration. Although the proportion of degraded areas remained small during the first three intervals, it reached its highest level in the fourth interval, reflecting a pronounced impact of land use changes on HQ at that time. Specifically, the periods from 2005 to 2010 and from 2015 to 2020 demonstrated more significant impacts of land use alterations on HQ, compared to the less pronounced changes during 2010–2015.

3.3.4. Major Driving Factors of Coordination and Conflict Areas

The factors influencing coordinated and conflicting interactions between land use changes and ecosystem services on the QXP from 2000 to 2020 were analyzed and ranked utilizing the XGBoost and SHAP methods. The impacts of each factor on model outcomes are displayed in Figure 9A,B. Figure 9A illustrates the distribution of each feature’s influence, demonstrating how variations in feature values affected model predictions, with larger absolute SHAP values indicating stronger feature impact. According to the analysis, natural environmental variables played a more decisive role in shaping the dynamics of coordinated and conflicting areas, whereas socio-economic variables had comparatively limited predictive influence. Except for temperature and GDP, the increase in the values of other factors generally had a positive impact. Figure 9B displays the ranking of the importance of each feature for driving factors of coordination and conflict areas, with the results showing that natural factors dominated the classification changes in coordination and conflict areas in the QXP over the past 20 years. According to the importance ranking, temperature, terrain, soil, and slope are key variables, while population, GDP, and nighttime lights are the least important, with relatively small values.

The natural environment in the QXP is harsh, with factors such as altitude and temperature significantly constraining the conversion of coordination and conflict classification areas in the region. In high-altitude areas, HQ is relatively poor, and low temperatures hinder large-scale industrial and agricultural activities. The interaction between land use change and HQ is sensitive and prone to significant changes due to external factors. Soil organic carbon (SOC) is vital for sustaining biodiversity, promoting material cycling, and enhancing soil fertility and structure. The QXP’s unique geography fosters SOC accumulation, where healthy SOC levels deliver critical ecosystem services and improve disturbance resistance. Slope gradients serve as a fundamental natural constraint on the plateau’s ecosystem patterns and land use evolution, governing material transport (e.g., soil erosion, snowmelt runoff) and energy distribution (temperature, sunlight), thereby shaping vertical vegetation zonation and soil freeze–thaw dynamics. Simultaneously, slopes act as a threshold for land use suitability, dictating the spatial distribution of agricultural/pastoral activities and the feasibility of infrastructure projects. The plateau exhibits extreme slope heterogeneity, with steep and gentle terrains coexisting; in addition, steep slopes dominate certain regions, directly influencing transitions between coordinated and conflicting land use zones. The QXP has a typical plateau mountain climate, with scarce rainfall throughout the year, making its importance lower than that of other natural factors. The overall development level of the QXP is slightly below the national average, with sparse population density and no obvious aggregation effect. Economic development (nighttime light index) is concentrated in plateau cities such as Lhasa and Xining. Human and social factors mainly drive the transformation of coordination and conflict types in urban and surrounding areas, while the effect in other regions is not significant. Therefore, the overall driving force for the transformation of coordination and conflict areas in the QXP is not significant, with its importance being much lower than that of natural factors [42].

4. Discussion

4.1. The Impact of Land Use Change on HQ

The QXP, as a unique integrated regional system, has land use changes affecting the regional economy, society, and ecosystem. These changes not only alter the HQ and socio-economic characteristics of the land system within the region, but also constrain the orderly development of the regional spatial environment. This study shows that land use change has had a profound impact on HQ, with the greatest influence coming from forest and shrubland. While Wang et al. [43] found forests most impactful in arid Altay, our study further reveals that shrublands in the QXP show higher sensitivity to HQ (Table 3), likely due to their marginal distribution in alpine zones. Among the various types of land use conversions, the exchange between grasslands and abandoned land is the most frequent. The severe degradation of grasslands has a multi-faceted impact on HQ, including a decrease in vegetation cover, deterioration of soil quality, weakened water regulation capacity, reduced biodiversity, exacerbation of climate change effects, restricted land use, socio-economic losses, and an increased risk of natural disasters. These ecological transformations have significantly influenced not only the stability and health of ecosystems, but also human welfare and socio-economic sustainability [44]. The grassland-to-abandoned-land conversions predominantly occur in the piedmont zones of the Kunlun and Himalayan Mountains. Areas characterized by degradation and conflict in the four-quadrant interaction analysis between land use change and habitat quality are likewise concentrated within these identical regions. It is widely accepted that socio-economic components substantially affect ecosystem services [45]. Beyond the immediate effects of land use alterations, HQ is also influenced by policy measures, demographic changes, and cultural contexts [46]. Since the implementation of the Western Development policy in 2000, rapid infrastructure expansion, urban growth, industrial realignment, and enhancements to living standards have taken place across the QXP. These developments, especially infrastructure projects, have had notable repercussions on HQ [47]. Moreover, demographic expansion and economic development impose substantial constraints on HQ, placing increased stress on ecosystems and diminishing their capacity to deliver ecosystem services. The cultural perspectives and environmental consciousness among ethnic populations residing on the QXP also significantly shape HQ. Cultural values and perceptions directly determine resource use behaviors, subsequently affecting ecosystem conditions. The ecological conservation policy framework has significantly strengthened the ecosystem barrier functions of the region. The Grain-to-Green Program facilitated large-scale conversion of cropland to grassland and forest, effectively enhancing soil retention and water conservation capacities, while the Ecological Conservation Redline policy, exemplified by initiatives like the Three-River-Source National Park, systemically safeguarded biodiversity and ensured the provision of core ecosystem services. Concurrent energy and industrial transition policies promoted clean energy development through solar and wind power projects, substantially reducing environmental pressures in vulnerable glacier and permafrost zones. Infrastructure development adopted an ecology-first approach, prioritizing ecological corridor construction and enforcing mandatory environmental impact assessments to minimize risks of habitat fragmentation and soil erosion. Together, these policy interventions have not only amplified ecosystem service functions, but also effectively mitigated the escalating conflicts between land use changes and HQ across the QXP, counteracting the worsening degradation trends in the region.

Land use change maps partially reflect variations in the intensity of human activities. These activities have transformed the current status of land use, subsequently influencing HQ. Xining and Lhasa, two principal cities on the QXP, are home to over half of the plateau’s population. The inhabitants of the surrounding areas maintain strong connections to these cities, with the plateau’s modernization being heavily reliant on their developmental progress. Studies indicate that land use changes within Xining and Lhasa, and their consequent effects on HQ, often exhibit conflicting patterns, with Xining demonstrating a more pronounced impact. Conversely, land use changes in the peripheral areas of these cities generally enhance HQ. Although human activity intensity is elevated in urban regions, the scarcity of land resources results in inevitable conflicts between urban development and HQ. The interactive effects of land use change and HQ show a large proportion of coordination and are good for habitat quality areas. The coordination and good for habitat quality areas represent favorable states of the interaction between land use change and HQ, and are predominantly distributed in foothill regions, the central and peripheral areas of QXP, with overlap with the other three areas. Nature reserves focused on ecological governance and environmental protection, as well as certain towns, also exhibit significant trends of coordination. On the other hand, the distribution of degradation and conflict areas is mainly concentrated in the Kunlun Mountains, the Qilian Mountains, and the northern plateau. In recent years, activities such as regional development, population aggregation, and economic growth have caused the areas of conflict and degradation around the city of Lhasa in the southern plateau to expand rapidly. Additionally, research highlights the evolving relationship between land use changes around lakes and rivers and their effects on HQ in the QXP. Over the course of the four study periods, the interplay between land use changes surrounding water sources and HQ transitioned from initially being advantageous to exhibiting a pattern of conflict and degradation. This trend suggests that, over time, anthropogenic activities have increasingly impacted the regions adjacent to lakes and rivers, thereby intensifying the tension between human activities and land dynamics in these areas. Consequently, these factors contribute to the ongoing evolution of relationships between land use modifications and HQ. The relationship between geographic environment and the impact of land use change on HQ exhibits spatial heterogeneity. In the arid Qaidam Basin and surrounding regions, areas experiencing conflict and degradation between land use change and HQ expanded continuously during the study period. In contrast, the humid southeastern QXP saw a gradual reduction in such conflict and degradation zones over the past two decades. Industrial development dynamics in plateau regions drive changes in the land-use–HQ relationship. For example, in the Golmud Salt Lake area, conflict and degradation zones shift spatially with development activities. Meanwhile, in Nyingchi City—a tourist hub and ecologically fragile area—the foothills surrounding the urban center show a marked increase in conflict and degradation zones linked to land use change and HQ.

4.2. Land Use Control Policies

Land use changes are influenced by various conflicts between socio-economic factors and natural resources. These changes can have diverse effects on HQ. A pressing concern for the QXP is the development of land use control policies that are tailored to the region’s unique conditions and aligned with the principles of sustainable development. This study identified critical land use categories affecting HQ and pinpointed regions where HQ was markedly impacted by land use changes. By systematically investigating the interplay between HQ dynamics and land use variations across the QXP, this study provides targeted recommendations for localized land use management.

The QXP, known as the Roof of the World and Water Tower of Asia, plays a crucial role in global ecological security and climate stability due to the health of its ecosystems. Forests and shrublands are key components of the region’s ecosystems, performing irreplaceable functions in maintaining biodiversity, preventing soil erosion, and regulating the climate. When developing land use control policies for the plateau, it is essential to prioritize ecological protection and restoration, scientifically demarcate protected areas, and implement stringent management measures. In particular, within the Qiangtang Nature Reserve and the Sanjiangyuan Nature Reserve, greater attention should be paid to the HQ of forests and grasslands, constantly improving the living environment within these reserves. Optimizing the use of natural resources, scientifically determining the carrying capacity of grasslands, and promoting environmentally friendly industries are critical. While ensuring economic development, the environmental carrying capacity should be comprehensively considered, with adjustments made to the industrial structure as needed to reduce the impact of pastoral activities on HQ, thus creating a new socio-ecological development model. In cities like Xining and Lhasa within the QXP, it is necessary to reasonably address urban land use demands. Urban planning should closely integrate the land division content outlined in national spatial planning, ensuring that enough ecological land is reserved and enhancing the self-regulating capacity of urban ecosystems. Land use changes in the surrounding areas of these cities can positively affect HQ. These areas can accommodate some land functions from within the cities, alleviating land use conflicts inside urban areas and promoting more harmonious development.

4.3. Limitations and Future Outlook

This study explores the impact of land use changes on HQ over the past 20 years on the QXP. Spatially, it classifies the relationships into four categories: coordination, beneficial to habitat, conflict, and degradation. It identifies the areas of coordination and conflict between land use change and HQ during each time period and reveals the driving factors behind the transformation between coordination and conflict regions. This study will be beneficial for relevant departments to conduct more precise land resource management and environmental governance. However, this research also has limitations. The sensitivity of HQ data and the threat source indicators come from other studies. The sensitivity and threat parameters are based on similar studies, which may introduce some subjectivity in certain regions. The QXP is particularly sensitive to climate, altitude, and other factors, making its ecological environment extremely fragile. The discrepancies in HQ evaluation indicators may lead to inconsistencies in the HQ results in specific regions. Therefore, future research should take into account local realities when setting indicators. Additionally, we overlooked random errors in time selection. One methodological constraint of this study is the use of five-year intervals for the period from 2000 to 2005, potentially limiting the robustness and generalizability of the findings. Additionally, although policy influences are recognized as significantly affecting HQ on the QXP, the current analysis did not quantitatively incorporate policy impacts or explicitly account for human–environment interaction dynamics [48]. Future research should address these limitations by integrating policy considerations into HQ analyses and by comprehensively quantifying human–environment interaction mechanisms. To quantify the impact of policy factors on the dynamic relationship between land use and HQ in the Tibetan Plateau, it is essential to develop an integrated “Policy-Driven-Land Response-Ecological Feedback” analytical framework. This approach combines policy text mining with spatial positioning techniques to quantitatively assess the spatiotemporal effectiveness of conservation measures, including ecological conservation redlines and national park systems. The methodology integrates land use change modeling (CLUE-S/FLUS) and habitat assessment tools (InVEST) to simulate different policy scenarios—such as comparing ecological-protection-focused approaches with development-oriented strategies—thereby evaluating how varying levels of policy enforcement influence land conversion patterns and corresponding habitat degradation outcomes across the region’s fragile ecosystems. This will provide more precise guidance for ecosystem management measures, achieving the harmonious development goal of regional development and ecological protection.

5. Conclusions

Between 2000 and 2020, the land use change area in the QXP was 7.5%, with an overall trend of stability. The most significant land use conversion occurred between grasslands and abandoned lands, with a severe degradation of grasslands. Habitat quality in the QXP varies from north to south, with low-value, medium-value, and high-value zones. Among the different land use types, forest land has the most significant impact on habitat quality. From 2000 to 2020, land use changes and habitat quality in the QXP have become more coordinated, though there is significant spatial heterogeneity across different time periods. The ratio of coordination to conflict evolution is 50:1. Conflict areas continuously change over time, mainly distributed along the edges of lake water sources, roadsides, the foothills of the Kunlun and Qilian Mountain ranges, and within plateau cities. The relationship between land use changes and habitat quality around plateau cities and water sources has undergone continuous evolution and requires focused attention. Natural factors predominantly drive the conversion of coordinated and conflicting areas, especially temperature and terrain, which play a strong role in driving the transformation between conflict and coordination. In future national land spatial planning practices, it is essential to consider natural factors in the implementation process to ensure the harmonious coexistence of humans and nature and achieve leapfrog development in the QXP.