Spatio-Temporal Dynamics of Wetland Ecosystem and Its Driving Factors in the Qinghai–Tibet Plateau

Abstract

Globally, wetlands have suffered severe degradation due to natural environmental changes and human activities. The wetlands on the Qinghai–Tibet Plateau (QTP) play a unique and critical ecological role, making it essential to understand their spatiotemporal dynamics and driving forces for effective conservation. Based on multi-source remote sensing data and Partial Least Squares Structural Equation Modeling (PLS-SEM), this study comprehensively quantified the spatiotemporal changes in wetlands and their key driving factors on the QTP from 1990 to 2020. The results show a net increase in total wetland area (including both natural and artificial wetlands) of approximately 538.72 km² per year over the 30-year period. Spatially, wetland expansion was most pronounced in the central–western and northern parts of the plateau, primarily driven by the conversion of grasslands, barren lands, and snow/ice cover, while localized degradation persisted in eastern regions. The PLS-SEM demonstrated an excellent fit (R² = 0.962) and identified human activities—such as ecological restoration policies and infrastructure development—as the dominant direct driver of wetland expansion (path coefficient = 0.918). Climate change, improved vegetation cover, and cryospheric loss also contributed positively to wetland gains (path coefficients = 0.056, 0.044, and 0.138, respectively). This study provides a transferable framework for understanding complex wetland dynamics and their drivers in alpine regions under global environmental change, which is crucial for designing more effective wetland conservation strategies.

1. Introduction

Wetlands play a vital role in hydrological regulation, biodiversity conservation, and the maintenance of regional ecological balance [1,2]. However, against the background of global climate change and rapid urbanization, these critical ecosystems are experiencing significant degradation. It is estimated that over the past 150 years, more than 50% of the world’s wetlands have been altered, degraded, or lost altogether [3]. Climate change has profoundly impacted wetland ecosystems by influencing their spatiotemporal distribution, structure, and function through complex interactions [4,5,6]. Concurrently, wetlands are increasingly threatened by human activities stemming from long-term groundwater exploitation, population growth, aquaculture, and other processes associated with urbanization [7,8,9].

Global wetlands have experienced severe disturbance [10]. Current global coastal wetlands have experienced significant shrinkage [11,12]. Model projections by Spencer et al. [13], based on an improved DIVA wetland change model, indicated that by 2100, global coastal wetlands may suffer a loss of 37% to 78% due to sea-level rise and increasing anthropogenic pressures. However, the degradation of inland wetlands is much more severe than that of coastal wetlands [8]. Wetlands in the Sanjiang Plain of Northeast China are facing continual reduction and degradation [14,15]. Based on satellite monitoring from 1984 to 2023, wetland ecosystems across the American West are undergoing a functional shift characterized by a significant decline in semi-permanent wetlands and a concurrent increase in seasonal and temporary systems, reflecting a broader trend of landscape drying and altered hydroperiods [16]. Tobore and Bamidele [17] have quantified a wetland loss rate of 3.53% in the Ogun River Basin from 1999 to 2030, through the integration of the CA-Markov model and remote sensing technology. Wetlands also constitute a core functional system in global alpine regions. However, alpine wetland ecosystems have likewise suffered severe shrinkage, drying, degradation, and landscape fragmentation [18].

The Qinghai–Tibet Plateau (QTP), acclaimed as the “Asian Water Tower” and the “Third Pole of the Earth,” harbors extensive and diverse wetland ecosystems. It contains unique alpine wetlands that account for 20% of China’s total wetland area, with its lakes alone making up half of the country’s total lake surface area [19]. The QTP is widely recognized as a climate-sensitive region, exhibiting pronounced thermal responses and exerting considerable influence on downstream hydrology [20]. Its ecosystems are characterized by structural simplicity and high fragility, rendering them especially vulnerable to both climate change and human disturbance [9], including activities such as grazing, infrastructure development, land use changes, and pollution [21]. Despite these challenges, these wetlands are indispensable for regional ecological security, serving critical functions in water conservation, biodiversity maintenance, climate regulation, and carbon sequestration. In recent decades, accelerated climate warming and increasing anthropogenic pressures have triggered significant alterations in the extent, distribution, and structure of these fragile ecosystems [9]. Consequently, quantifying the spatio-temporal dynamics of wetlands and elucidating their driving mechanisms are essential for comprehending their dynamic changes and underlying processes.

Numerous studies have explored the drivers of wetland evolution across varying spatial scales. For instance, Tian et al. [22] employed grid-based geographically weighted regression (GWR) to attribute wetland loss across China primarily to GDP growth, population density, and elevation. Zhang et al. [23] utilized the Google Earth Engine (GEE) platform and long-term Landsat imagery to demonstrate that temperature and precipitation are dominant factors influencing changes in the Maqu alpine wetlands. Similarly, Bai et al. [24] highlighted the significant role of human activities in driving landscape pattern evolution in the Zoige Plateau wetlands. Based on global wetland datasets, Assefa and Eneyew [25] map the spatial distribution of diverse wetland types across Ethiopia, quantify a 9% decline in wetland coverage alongside a 20% increase in open water between 2000 and 2020, and identify agricultural expansion, urbanization, water abstraction, and policy gaps as key drivers of these changes. Phillips [26] based on a structured landscape response attribution (LRA) protocol integrating historical analysis, plausibility assessment, and elimination of alternative drivers, quantitatively attributes estuarine marsh loss to climate-induced sea-level rise and event-based storm surges, while explicitly accounting for human modifications such as ditching. These multi-scale investigations provide a foundational framework and methodological reference for wetland classification, dynamic monitoring, and causal analysis.

However, most studies have relied on statistical analyses and machine learning techniques [27,28,29]. In contrast, structural equation modeling (SEM) involves the integrated and refined application of path, factor, variance, and regression analyses, enabling the quantitative identification of both direct and indirect influences of driving factors. Since wetland degradation can be attributed to the interplay between human activities and the natural environment, SEM proves suitable for analyzing the complex interactions among factors affecting wetland change. For example, the study by Wang et al. [30], based on partial least squares structural equation modeling (PLS-SEM), systematically revealed the spatiotemporal dynamics and driving mechanisms of wetlands in Wuhan. Furthermore, Shi et al. [31] integrated PLS-SEM with the PLUS model to elucidate the spatiotemporal dynamics of wetland changes in the Sanjiang Plain—where total wetland area increased but natural wetlands declined—and identified the distinct dominant roles and direct/indirect pathways of influence exerted by topography, urbanization, and climatic factors across different periods.

Therefore, this study investigates the spatiotemporal dynamics and driving mechanisms of wetland area changes across the QTP based on multi-source geographic data. The specific objectives are to: (1) analyze the spatiotemporal patterns of wetland change on the QTP during 1900–2020; (2) develop an integrated analytical framework to identify the principal drivers of wetland changes; (3) quantify the direct, indirect, and interactive effects of natural and anthropogenic factors on wetland dynamics.

2. Materials and Methods

2.1. Study Area

The QTP, defined by longitudes 73° to 104.5° E and latitudes 25° to 40° N (Figure 1), is the highest and most extensive plateau on Earth, often referred to as the “Roof of the World” and the “Third Pole”. With an average elevation exceeding 4500 m, it spans approximately 250 × 10⁴ km² across western China and extends into parts of India, Nepal, Bhutan, and Pakistan [32]. The plateau serves as the headwaters for many of Asia’s major river systems, including the Yangtze, Yellow, Indus, Brahmaputra, and Mekong, supplying freshwater to hundreds of millions of people downstream. The QTP is recognized as a significant wetland distribution area in China, comprising approximately 20% of the nation’s total wetland area [2]. Climate is characterized by its complexity, primarily influenced by the Indian summer monsoon, the East Asian monsoon, and the mid-latitude westerlies, which create distinct wet and dry seasons [33,34,35]. The mean annual temperature ranges between −5 and 15.5 °C [36], and annual precipitation on the QTP is unevenly distributed, decreasing from over 2000 mm in the southeast to less than 50 mm in the northwest.

2.2. Database

This study integrates land use/land cover (LULC), meteorological, Normalized Difference Vegetation Index (NDVI), and socio-economic datasets spanning the period from 1990 to 2020. We utilized the 30 m annual land cover datasets developed by Yang and Huang [37] from China, which achieves a baseline overall accuracy of 79.31% and outperforms other similar datasets [38]. This dataset has been updated to 2024 and is widely used in land use and related studies [39,40,41]. Wetland and ice/snow cover classes were extracted from these raster datasets for further analysis. Meteorological data, including annual average temperature [42], precipitation [43], and potential evapotranspiration [44], were derived from monthly raster datasets with a resolution of 1 km obtained from the National Tibetan Plateau/Third Pole Environment Data Center (TPDC) (https://data.tpdc.ac.cn/, accessed on 25 November 2024), which is located in China. The NDVI raster dataset was derived from the maximum NDVI values between 1990 and 2020, calculated using corresponding bands of Landsat 5/7/8/9 satellites. The Landsat satellite series are developed and operated by the National Aeronautics and Space Administration (NASA) of the United States (US). All calculations were performed via the Google Earth Engine (GEE) cloud-based remote sensing platform, with processes including cloud masking and removal of anomalous pixels. The original NDVI data had a resolution of 30 m, which was resampled to 1 km resolution for computational convenience. Furthermore, socio-economic data included Nighttime Light (NT; 1 km resolution) [45] and population density (1 km resolution) [46], were sourced from China. All raw datasets were reprojected to an Albers Equal Area Conic projection and resampled to a standardized 1 km spatial resolution to ensure coordinated analysis.

2.3. Methodology

2.3.1. Mann–Kendall Test and Sen’s Slope Estimator Test

The Mann–Kendall (MK) test [47,48], a non-parametric statistical method, is widely used for detecting monotonic trends in time series data [49]. Theil-Sen Median trend analysis, also known as Sen’s slope estimation, is an effective and robust non-parametric statistical method well-suited for analyzing trends in long-term time series data. This method is highly resilient to measurement errors and outliers, ensuring accurate estimates of trends [50]. To quantify the magnitude of a detected trend, the Sen’s slope estimator test is commonly applied in conjunction with the MK test.

This study mainly uses these two methods to calculate the temporal variation trends of wetland area. The Mann–Kendall test statistic $S$ is calculated as

$$S=\sum _{i=1}^{n-1}\sum _{j=i+1}^{n}sgn\left(x_j-x_i\right)$$

(1)

where $n$ is the number of data points, $x_i$ and $x_j$ represent the data values in the time series at positions $i$ and $j$ ($j$ > $i$), respectively, and $sgn\left(x_j-x_i\right)$ denotes the sign function, which is defined as follows:

$$sgn\left(x_j-x_i\right)=f\left(x\right)=f\left(x\right)=f\left(x\right)=\left\{\begin{array}{c}1, if x_j>x_i\\ 0, if x_j=x_i\\ -1, if x_j<x_i\end{array}\right..$$

(2)

The variance is calculated as follows:

$$Var\left(S\right)={\displaystyle \frac{n\left(n-1\right)\left(2n+5\right)-\sum _{i=1}^m{t}_i(t_i-1)(2t_i+5)}{18}}$$

(3)

where $n$ is the number of data points, $m$ is the number of tied groups, and $t_i$ denotes the size of the tie at range $i$. A tied group refers to a set of sample data sharing identical values. For sample sizes where $n$ > 10, the standard normal test statistic $Z_s$ is calculated as follows:

$$Z_s=\left\{\begin{array}{c}{\displaystyle \frac{S-1}{\sqrt{Var\left(S\right)}}}, if S>0\\ 0, if S=0\\ {\displaystyle \frac{S+1}{\sqrt{Var\left(S\right)}}},if S<0\end{array}\right.$$

(4)

A positive value of $Z_s$ indicates an increasing trend, while a negative value suggests a decreasing trend. The trend test is performed at a specific significance level α. When $\mid Z_s\mid$ > $Z_{1-a/2}$, the null hypothesis is rejected, indicating a significant trend in the time series. The value of $Z_{1-a/2}$ is derived from the standard normal distribution table. In this study, significance levels of α = 0.01 and α = 0.05 were used. At the 5% significance level, the null hypothesis of no trend is rejected if $\mid Z_s\mid$ > 1.96; at the 1% significance level, it is rejected if $\mid Z_s\mid$ > 2.576.

Sen’s slope estimator test is a nonparametric procedure used to estimate the slope of a trend in N pairs of data samples:

$$Q_i={\displaystyle \frac{x_i-x_j}{i-j}}for i= 1, \dots ,N$$

(5)

where $x_i$ and $x_j$ represent the data values at time $i$ and $j$ ($i$ > $j$), respectively.

2.3.2. Land Use Transition Matrix

The land use transition matrix is a two-dimensional matrix derived from the dynamic transition relationships of land use conditions at different time points within the same region. It graphically represents the dynamic changes in land use types over time, thereby elucidating the specific processes underlying land use change [51]:

$$B_{ij}=\left[\begin{array}{ccc}{B}_{11}& \cdots & B_{1n}\\ ⋮& \ddots & ⋮\\ B_{n1}& \cdots & B_{nn}\end{array}\right]$$

(6)

where $B_{ij}$ is the area transformed from land use type $i$ to land use type $j$ over a given period, and $n$ is the number of different land use types.

This study employed land use/cover (LUC) data from 1990, 2000, 2010, and 2020 to construct land use transition matrices to characterize land use dynamic changes across distinct periods and analyze contemporary spatial distribution patterns.

2.3.3. The PLS-SEM Model

Partial Least Squares Structural Equation Modeling (PLS-SEM) is utilized to estimate causal relationships among latent constructs, which cannot be directly observed but are inferred from one or more measurable indicator variables. A key advantage of PLS-SEM is its strong performance with small sample sizes and its minimal dependence on stringent distributional assumptions, particularly concerning multivariate normality. Typically, PLS-SEM consists of two main components: the measurement model, which specifies the relationships between latent constructs and their observed indicator variables, and the structural model, which defines the causal pathways between exogenous and endogenous latent constructs [52,53].

Wetlands form in a variety of landscapes, and their overall physical structure and hydrology are determined by climatic and geomorphological factors [54]. Simultaneously, human activities (such as policies, urbanization, industrial development, and agricultural production) are significant factors influencing wetland development [55,56]. The role of vegetation is crucial for wetland restoration [57]. And for high-altitude wetlands, the impact of glacial melt cannot be overlooked [58]. Based on existing literature and the ecological context of the QTP, we developed a PLS-SEM framework to analyze the influence of latent variables—including climate change, human activities, NDVI, and snow/ice cover—on wetland changes. The observed variables are temperature, precipitation, evapotranspiration, population density, and nighttime light index. For specific details regarding the calculation process, please refer to Hair et al. [59].

3. Results

3.1. The Spatiotemporal Variation of the Wetlands

From 1990 to 2020, the wetland area of the QTP underwent significant changes (Figure 2). Overall, wetlands exhibited an expanding trend, with the total area increasing at 538.72 km²/a (p < 0.001) (

Table 1. Trends in wetland area, temperature, precipitation, evapotranspiration, population density, nighttime light, NDVI, and snow/ice cover on the QTP from 1990 to 2020.

Factor	Factor
Wetland (km2/a)	538.7188 ***
TEM (°C/a)	0.0248 ***
PRE (mm/a)	0.0580
ET (mm/a)	0.0602 ***
POP (pop/km2/a)	0.0553 ***
NL (/a)	0.0006 ***
NDVI (/a)	0.3835 *
Snow/ice (km2/a)	−13.0228

Note: *** represents p < 0.001, * represents p < 0.05.

). The total wetland area increased from 4.30 × 10⁴ km² in 1990 to a peak of 5.68 × 10⁴ km² in 2020, representing a net gain of 1.47 × 10⁴ km².

Figure 3 illustrates the spatiotemporal dynamics of wetland area on the QTP over the 30-year period, derived by subtracting the 1990 wetland layer from that of 2020. Spatially, wetland expansion was most pronounced in the central–western and northern parts of the plateau, particularly in the Qiangtang Plateau, Qaidam Basin, and Hoh Xil regions, while localized degradation persisted in eastern regions. Changes at the level of individual wetlands, however, varied considerably and call for case-specific analysis. Notable expansion occurred in Ayakkum Lake, Aqqikkol Lake, and Siling Co, while Zonag Lake underwent substantial areal reduction. In contrast, Taijinar Lake and Qarhan Salt Lake exhibited more complex dynamics, with both gains and losses occurring in different subareas.

Figure 4 further reveals distinct decadal patterns in wetland transitions across the QTP. The initial decade (1990–2000) witnessed modest net gains of 1594 km², succeeded by a pronounced expansion phase adding 5259 km² during 2000–2010. Subsequent growth stabilized at 2929 km² in the 2010–2020 period. Crucially, land cover conversions—primarily from grassland, barren terrain, and snow/ice zones—accounted for 9891 km² of the aggregate increase, representing 66.9% of the total areal change.

3.2. Changes in Driving Factors

Figure 5 illustrates the temporal dynamics of the driving factors. As shown in Table 1, temperature exhibited a significant warming trend of 0.0248° C/a with high statistical confidence (p < 0.001). This temperature rise stimulated regional evapotranspiration, which increased steadily by 0.0602 mm/a (p < 0.01). The resulting extension of ice-free seasons fostered favorable conditions for wetland proliferation. Although precipitation showed a non-significant upward trend of 0.0580 mm/a, it nonetheless contributed meaningfully to wetland replenishment.

Simultaneously, human activities intensified markedly throughout the study period. Population density increased by 0.0553 pop/km²/a (p < 0.001), alongside a rise in nighttime illumination, which grew by 0.0006 per year (p < 0.001). Notably, snow and ice cover on the QTP diminished substantially at a rate of 13.02 km²/a, consistent with regional warming patterns. This retreat of frozen landscapes may accelerate wetland expansion through enhanced meltwater contributions. Vegetation changes accompanied these dynamics, exhibiting a moderate greening trend. NDVI increased significantly by 0.3835 thousandths per year (p < 0.05), indicating enhanced plant productivity within wetland corridors. This enhancement likely represents a positive feedback response to increased hydrological enrichment across the plateau.

3.3. Driving Factors of Wetland Area Dynamics

3.3.1. Correlation Diagnostics

Before conducting PLS-SEM, Spearman correlation analysis was utilized to examine the relationships between wetland area and seven potential drivers, following the confirmation of non-normality in the dataset. Figure 6 illustrates the correlation coefficients for QTP from 1990 to 2020. All drivers demonstrated positive correlations with wetland expansion. Among the meteorological drivers, temperature exhibited the strongest correlation (ρ = 0.61, p < 0.001), followed by evapotranspiration (ρ = 0.47, p = 0.01) and precipitation (ρ = 0.29). Regarding land surface factors, NDVI showed a moderate correlation (ρ = 0.73, p = 0.001), whereas snow and ice cover displayed a negligible correlation (ρ = 0.03). Despite the expectation that melting ice and snow in the QTP would impact changes in wetland area, the weak correlation suggests it may not have a direct effect. Socioeconomic drivers exhibited near-perfect positive correlations, with nighttime lights (ρ = 0.95) and population density (ρ = 0.96), both significant at p < 0.001.

3.3.2. Analysis of Driving Factors Based on PLS-SEM

The constructed model, which examines the factors influencing wetland ecosystems, demonstrates an excellent overall fit and robust internal quality. Key goodness-of-fit indices indicate that the standardized root mean square residual (SRMR = 0.080) is slightly above the critical threshold of 0.08, while the normed fit index (NFI = 0.867) approaches the ideal benchmark of 0.9. These results suggest that the model effectively captures the covariance structure of the observed data.

Table 2. Assessment of the validity and reliability of PLS-SEM.

Factor	Outer Loadings	Cronbach’s Alpha	AVE	CR (rho_c)	HTMT
Climate		0.669	0.649	0.832	0.699
Temperature	0.981
Precipitation	0.397
Evapotranspiration	0.908
Human Activity		0.977	0.977	0.988	0.979
Nighttime Lights	0.989
Population Density	0.988
NDVI					0.712
Snow/ice					0.061

presents the performance evaluation of the PLS-SEM through validity and reliability tests. In relation to the measurement model, all latent variables meet acceptable reliability and validity criteria. For the human activity construct, the composite reliability (CR = 0.988) and average variance extracted (AVE = 0.977) substantially exceed the established thresholds (>0.7 and >0.5, respectively). The outer loadings for the indicators—nighttime lights (NL = 0.989) and population (POP = 0.988)—further confirm that these indicator variables adequately represent the latent construct. The climate construct also satisfies the necessary criteria, with a composite reliability of 0.832 and an average variance extracted of 0.649; notably, the temperature indicator (TEM) shows a strong outer loading of 0.981. The HTMT ratios for all constructs, with the exception of Human Activity (0.979), were below the established threshold of 0.85, demonstrating adequate discriminant validity among the latent variables.

Path analysis depicted in Figure 7 reveals interactions among driving factors and wetland area. The path coefficients from factors to wetland area and their total effects are provided in

Table 3. The path coefficients and total effects of driving factors on wetland area.

Factor	Path Coefficients (α)	Total Effects (β)
Climate	0.056	0.095
Human Activity	0.918	0.917
NDVI	0.044
Snow/ice	0.138

, while the effect coefficients among factors are summarized in

Table 4. Effects between factors.

Path	Effects (γ)
Climate → Ice/snow	0.271
Climate → NDVI	0.030
Human Activity → Ice/snow	−0.242
Human Activity → NDVI	0.555

, and indirect effects on wetland area are presented in

Table 5. Specific indirect effects.

Path	Specific Indirect Effects (δ)
Climate → Ice/snow → Wetland	0.037
Climate → NDVI → Wetland	0.001
Human Activity → Ice/snow → Wetland	−0.032
Human Activity → NDVI → Wetland	0.031

. Climate factors exert a weak direct positive effect on wetland area (α = 0.055), and also generate an indirect positive effect by accelerating snow/ice melt (γ = 0.271), resulting in a specific indirect effect of 0.037. In contrast, the specific indirect effect mediated by NDVI remains limited (δ = 0.001). It is worth noting that climate warming directly promotes wetland expansion and contributes to it through reduced snow/ice cover, while also promoting vegetation growth (γ = 0.030), thereby further enhancing its overall positive impact. The total effect (0.095), though amplified compared to the direct effect, remains limited, indicating a modest overall contribution of climate to wetland dynamics. In contrast, human activities emerge as the primary driver, exerting a strong direct effect on wetland area (α = 0.918), which far exceeds the influence of other variables, as indicated by a substantial effect size (f² = 8.740). Although human activities directly stimulate wetland expansion and exert a promoting effect on vegetation (γ = 0.707; δ = 0.031), their negative impact on snow/ice cover (γ = −0.232; δ = −0.032) indirectly counteracts this positive influence. Nevertheless, the overwhelming total effect of human activities (β = 0.917) still highlights their dominant role in shaping wetland dynamics.

The reduction in snow/ice cover has a positive effect on the increase in wetland area in the QTP, with a path coefficient of 0.138. Similarly, NDVI acts as a supportive mediator, demonstrating a positive direct effect on wetland area (α = 0.044). This suggests that wetland expansion is associated with vegetation growth, potentially reflecting synergistic eco-hydrological processes.

4. Discussion

4.1. The Uniqueness of Wetland Changes in QTP

From 1990 to 2020, the wetland area on the QTP showed a significant increasing trend, with a notable acceleration during the 2010s. This observation is consistent with the findings reported by Liu and Zhao [60]. The total wetland area on the QTP increased from 4.30 × 10⁴ km² in 1990 to 5.68 × 10⁴ km² in 2020, resulting in a net gain of 1.47 × 10⁴ km²—an increase of approximately 34%. This pattern sharply contrasts with the widespread global trend of wetland loss, but aligns with changes observed in alpine wetlands of regions such as Central Asia and South America [58,61], highlighting the ecological uniqueness of high-altitude wetlands.

It is widely recognized that climate change has profound impacts on global ecosystems. Over the past five decades, the warming rate on the QTP has been approximately twice the global average [9]. Against this backdrop, wetlands on the plateau have undergone positive changes, and their overall vulnerability has decreased [62]. In some cases, ecological functions may even continue to improve under rapid climate change [63].

However, this positive trend should not lead to complacency. As shown in Figure 3, wetland shrinkage has occurred in the eastern part of the QTP, and the wetland vulnerability index of QTP increases from west to east [62]. This reminds us that while the overall condition of wetlands on the plateau is improving, localized ecological degradation must not be overlooked. Targeted measures should be taken based on specific regional conditions.

4.2. The Influence of Human Activities on the Wetland Area

The expansion of wetlands in the QTP is primarily attributed to human activities—a phenomenon closely linked to China’s national ecological restoration policies. Key initiatives include the “Returning Grazing Land to Grassland” project implemented between 2004 and 2012, and the compensation policy introduced in 2009 to incentivize grassland conservation through fencing degraded areas and promoting balanced grazing practices.

As a result, human activity intensity in high-altitude grassland regions decreased by 16.1% from 2000 to 2017 [64], even as the overall population density of the plateau increased. This seemingly contradictory trend can be explained by the urban concentration of populations, which reduced anthropogenic pressure on ecologically sensitive areas. It also suggests that ecological resettlement policies have contributed to alleviating grazing pressure and supporting the natural recovery of degraded wetlands. The most pronounced wetland expansion occurred between 2000 and 2010, a period that aligns with the pilot establishment of the Three-River-Source National Park. This synchronicity indicates that ecological projects and protection policies introduced since the early 2000s have been effective in reversing environmental degradation [56]. Specific measures—such as scientific grazing management, wetland reconstruction, and desertification control—have directly improved the efficiency of wetland restoration [65].

Although urban expansion is often considered a major driver of wetland loss worldwide [62], in the unique context of the QTP, improved regional accessibility and tourism development have facilitated the construction of artificial water bodies and irrigation systems, further contributing to wetland expansion under regulated conditions [66]. Simultaneously, industrial development, through the construction of hydraulic engineering projects, can lead to the expansion of wetland areas in certain regions. A representative example is the Qaidam Basin on the Qinghai–Tibet Plateau, which has developed a circular economy industrial system emphasizing the comprehensive utilization of salt lake resources. In the Qarhan Salt Lake area—the largest saline region in the Qaidam Basin—wetland areas, including both artificial wetlands such as salt pans and natural wetlands such as lakes, have exhibited an increasing trend [67]. This expansion is driven by salt lake resource exploitation, industrial production demands, and tourism development.

However, it is crucial to critically examine the ecological sustainability of this wetland expansion. While the increase in surface water area may create novel habitats for certain species [68], these anthropogenic wetlands are often tied to intensive resource extraction (e.g., groundwater mining for salt lake industries) and may lead to unintended consequences such as water table depletion, soil salinization, and habitat fragmentation for native species [69]. Therefore, while contributing to the net gain in wetland area, such human-induced expansion may not represent a sustainable ecological gain in the long term. Long-term monitoring and assessment are essential to evaluate the trade-offs between economic benefits and potential ecological degradation associated with these activities. Thus, the case of the Qaidam Basin highlights the unique and complex role of human activities in shaping wetland dynamics on the QTP. Sustainable development practices and careful water resource management are essential to mitigate the negative impacts of both climate change and human activities on the fragile wetland ecosystems of the QTP.

4.3. Systematic Effects of Climatic Conditions

Climate warming has initiated a notable warm-humidification trend on the QTP since the 1950s. This ongoing warming has intensified regional hydrological cycling, particularly by amplifying surface runoff generation [70]. As a result, total river discharge across the plateau has demonstrated a significant upward trend [71], which substantially contributes to wetland expansion due to the increased water surplus in low-lying basins. Meanwhile, increased precipitation is identified as the primary factor driving rapid lake expansion [72,73].

The role of enhanced evapotranspiration is more complex. Traditional hydrological theory suggests that rising temperatures enhance evapotranspiration, leading to soil moisture deficits and net water loss [74,75]. Previous studies have confirmed that increased evapotranspiration acts as the primary driver of wetland degradation in the Three-River Source Region [76], indicating that higher evapotranspiration may cause lake shrinkage in certain local areas of the plateau. However, at the scale of the entire QTP, climate change has fundamentally altered the relationships governing water supply in cryospheric zones. Climate warming promotes lake expansion, which in turn increases lake evaporation and atmospheric moisture supply. This elevated moisture availability—resulting from increased evapotranspiration—may enhance regional precipitation, leading to further lake growth and establishing a positive feedback loop [77].

Thus, climate change—through the combined effects of temperature, precipitation, and evapotranspiration—exerts a positive feedback effect on overall wetland dynamics across the QTP. Understanding how wetlands on the QTP—a region of remarkable complexity—respond to climate change necessitates a comprehensive analysis of the interconnected hydro-climatic-ecological system. Overall, the synergistic effects of warming-enhanced runoff, increased precipitation, and evapotranspiration-mediated moisture recycling are transforming the QTP into a progressively wetter system, despite localized moisture deficits. This illustrates a paradigmatic shift in high-altitude hydrology: climate change is not merely perturbing isolated elements but reconfiguring entire water cycles through positive feedback loops. Such systemic alterations underline the vulnerability and adaptability of alpine wetland ecosystems under global change. Moving forward, safeguarding these fragile yet functionally critical ecosystems will require integrated models and policies that account for cross-scale interactions between climate, cryosphere, hydrology, and ecology—an essential step toward predicting and mitigating future changes in the Third Pole’s life-supporting water towers.

4.4. The Mechanism of Underlying Vegetation and Cryosphere Interactions

The improved vegetation coverage on the QTP has played a positive role in the expansion of the total wetland area in the region. The study demonstrates that vegetation restoration played a facilitative role in wetland recovery in the Source Region of the Yangtze and Yellow Rivers. During the period of 2015–2020, the wetland regulation capacity recovered rapidly from 29.78 mm to 83.78 mm—an increase of 181.3%—which aligned strongly with a concurrent 23% increase in vegetation coverage following restoration efforts [78]. This suggests that enhanced vegetation cover, likely driven by both natural recovery and ecological engineering interventions, contributed to improved hydrological conditions that supported wetland replenishment and functional recovery, highlighting the synergistic interaction between vegetation dynamics and wetland hydrological processes in alpine ecosystems. Existing research underscores that vegetative restoration serves as a critical driver for successful wetland recovery, acting as an essential initial step that catalyzes subsequent ecological and functional development [79]. Consequently, restoring vegetation is not only ecologically beneficial but also vital for maximizing long-term economic returns and resilience benefits from wetland restoration efforts.

Although no strong direct correlation was observed between cryospheric cover dynamics and wetland area changes across the QTP, the reduction in snow/ice cover still exerts a positive influence on wetland expansion. Zhou et al. [73] demonstrated that snow and glacier melt significantly contribute to lake expansion on the QTP. This contribution depends on factors such as watershed ice cover, runoff pathways, and local topography. Furthermore, soil freeze–thaw cycles play a crucial role in these hydrological processes. Since the 1980s, the active layer of permafrost on the QTP has gradually thinned [80,81], indicating substantial meltwater release from ground ice into local hydrological systems. Li et al. [65] highlighted that groundwater has become the dominant source of runoff in the major drainage basins of the QTP. Thus, cryospheric melt has increased runoff, thereby enhancing water availability for wetlands. These findings collectively underscore the critical yet complex role of cryospheric changes in shaping wetland dynamics on the Qinghai–Tibet Plateau. While a direct large-scale correlation remains elusive, the mechanistic linkages—such as meltwater supplementation, groundwater recharge, and permafrost thaw—reveal a coherent hydrological narrative: cryospheric retreat is an important facilitator of wetland expansion under a warming climate. This highlights the necessity of incorporating cryosphere-hydrology interactions into future climate adaptation strategies, as the sustainability of wetland ecosystems—and the biodiversity and pastoral livelihoods they support—increasingly depends on the delicate balance of ice, water, and thaw in this fragile yet vital region.

4.5. Prospects for Protecting Wetlands on the Qinghai–Tibet Plateau

Based on the findings of this study, we propose several targeted measures for the future conservation and sustainable management of wetland ecosystems on the QTP. First, the success of ecological restoration policies (e.g., “Returning Grazing Land to Grassland”) should be consolidated and optimized through more precise zoning and community-based management, particularly in the vulnerable eastern regions where degradation persists. Second, a differentiated protection strategy must be implemented. This involves establishing strict protection zones in the east for rehabilitation, promoting sustainable grazing and ecotourism in the central region, and focusing on water resource conservation and maintaining cryospheric stability in the expanding western wetlands. Simultaneously, we note that the expansion of wetland area on the Qinghai–Tibet Plateau is primarily derived from the conversion of grasslands, which may have implications for grassland biodiversity. Future research should incorporate detailed ecological and biodiversity indicators to further elucidate the functional consequences of these land cover changes. Third, the ecological sustainability of anthropogenically expanded wetlands, such as those linked to salt lake industries, requires stringent oversight. Long-term monitoring of groundwater extraction and its impacts is essential to prevent water depletion and soil salinization. Finally, building a climate-resilient adaptive management framework is paramount. This includes developing early warning systems based on hydro-climatic models and preparing for potential future shifts in the water balance. Implementing these measures, supported by robust interdisciplinary monitoring, will be crucial for safeguarding the ecological functions of the “Asian Water Tower” under ongoing climatic and anthropogenic pressures.

5. Conclusions

This study systematically quantified the spatiotemporal dynamics and driving mechanisms of wetland changes on the QTP from 1990 to 2020. The key findings are as follows: (1) Comprehensively characterized the spatiotemporal patterns of wetland area changes. The wetland area on the QTP experienced a significant net expansion at a rate of 538.72 km²/a. Spatially, the most pronounced gains occurred in the central–western and northern regions, primarily driven by the conversion of grassland, barren land, and snow/ice cover. (2) Quantified the main driving factors of wetland changes. The integrated PLS-SEM framework identified human activities as the dominant direct driver of wetland expansion (path coefficient = 0.918), with climate change, improved vegetation coverage and cryospheric loss also contributing positively. (3) Analyzed the interactions among factors. Climate change directly promotes wetland expansion and also contributes indirectly by accelerating snow/ice melt and vegetation changes (with specific indirect effects of 0.037 and 0.001, respectively). While exerting an overwhelming direct positive impact, human activities also indirectly influence wetlands through their negative effect on cryospheric cover and positive effect on vegetation (with specific indirect effects of −0.032 and 0.031, respectively). These findings establish a transferable analytical framework for deciphering alpine wetland dynamics and provide a scientific basis for formulating targeted conservation strategies under escalating climatic and anthropogenic pressures.