Investigation of the Effects of Climate Change and Human Activities on the Spatio-Temporal Trends of Vegetation in the Source Region of the Yellow River in China

Abstract

The dynamic changes in vegetation significantly impact the sustainability, safety, and stability of ecosystems in the source region of the Yellow River. However, the spatiotemporal patterns and driving factors of these changes remain unclear. The MODIS NDVI dataset (1998–2018), together with climatic records from meteorological stations and socio-economic statistics, was collected to investigate the spatiotemporal characteristics of vegetation coverage in the study area. For the analysis, we employed linear trend analysis to assess long-term changes, Pearson correlation analysis to examine the relationships between vegetation dynamics and climatic as well as anthropogenic factors, and t-tests to evaluate the statistical significance of the results. The results indicated the following: (1) From 1998 to 2018, vegetation in the source region of the Yellow River generally exhibited an increasing trend, with 92.7% of the area showed improvement, while only 7.3% experienced degradation. The greatest vegetation increase occurred in areas with elevations of 3250–3750 m, whereas vegetation decline was mainly concentrated in regions with elevations of 5250–6250 m. (2) Seasonal differences in vegetation trends were observed, with significant increases in spring, summer, and winter, and a non-significant decrease in autumn. Vegetation degradation in summer and autumn remains a concern, primarily in southeastern and lower-elevation areas, affecting 25% and 27% of the total area, respectively. The maximum annual average NDVI was 0.70, occurring in 2018, while the minimum value was 0.59, observed in 2003. (3) Strong correlations were observed between vegetation dynamics and climatic variables, with temperature and precipitation showing significant positive correlations with vegetation (r = 0.66 and 0.60, respectively; p < 0.01, t-test), suggesting that increases in temperature and precipitation serve as primary drivers for vegetation improvement. (4) Anthropogenic factors, particularly overgrazing and rapid population growth (both human and livestock), were identified as major contributors to the degradation of low-altitude alpine grasslands during summer and autumn periods, with notable impacts observed in counties with higher livestock density and population growth, indicating that for each unit increase in population trend, the NDVI trend decreases by an average of 0.0001. The findings of this research are expected to inform the design and implementation of targeted ecological conservation and restoration strategies in the source region of the Yellow River, such as optimizing land-use planning, guiding reforestation and grassland management efforts, and establishing region-specific policies to mitigate the impacts of climate change and human activities on vegetation ecosystems.

1. Introduction

Against the backdrop of accelerating climate change and expanding anthropogenic pressures, terrestrial ecosystems are experiencing unprecedented challenges [1]. Vegetation, as a fundamental component of global terrestrial ecosystems, serves critical ecological functions including climate regulation, hydrological cycle modulation, and soil conservation [2,3]. Vegetation growth patterns have been widely recognized as sensitive indicators of ecosystem health, stability, and sustainability, showing pronounced responsiveness to climatic variations, natural disturbances, and human activities [4]. Despite its ecological significance, the complex spatiotemporal patterns of vegetation dynamics and their driving mechanisms remain insufficiently characterized. Therefore, comprehensive investigation of vegetation dynamics across multiple temporal and spatial scales, particularly their responses to climatic and anthropogenic drivers, is essential for developing evidence-based strategies for sustainable ecosystem management.

Spatiotemporal analysis of regional vegetation coverage provides valuable insights into ecosystem dynamics. The Normalized Difference Vegetation Index (NDVI) has emerged as a robust indicator for monitoring vegetation cover changes, as it quantitatively reflects key ecological parameters including net primary productivity, canopy density, and biomass accumulation [5]. Extensive research utilizing NDVI datasets has been conducted across diverse geographical regions to examine vegetation dynamics and their driving mechanisms [6,7,8,9]. Methodological approaches such as correlation analysis [10], residual analysis [11], and predictive modeling [12,13] have been effectively employed to investigate NDVI variations and their underlying determinants. These studies consistently demonstrate that vegetation dynamics are governed by the complex interplay between climatic factors and anthropogenic influences. The climatic impact on NDVI exhibits pronounced spatial heterogeneity, characterized by non-linear relationships between environmental variables and vegetation growth. While increased precipitation generally enhances soil moisture availability, thereby facilitating vegetation growth, the effects of temperature fluctuations show seasonal and regional variability. For instance, elevated summer temperatures may intensify evapotranspiration rates, potentially inhibiting vegetation productivity [14]. Concurrently, expanding anthropogenic pressures have become increasingly significant in shaping vegetation patterns, particularly through rapid urbanization processes that lead to land-use conversion and intensive grazing practices. Notably, vegetation cover changes are most pronounced in economically developed regions with high population density, while remaining relatively stable in areas with minimal human disturbance [15].

The source region of the Yellow River, serving as a crucial water conservation zone within the basin, plays a pivotal role in maintaining both regional ecological security and basin-wide water resource sustainability [16]. The vegetation dynamics in this ecologically sensitive area have been undergoing significant transformations under the combined pressures of climate change and anthropogenic activities. In recent years, driven by climate warming, the interannual and seasonal variability of precipitation in the source region of the Yellow River has become increasingly pronounced. Long-term observational records reveal that precipitation in this area follows an irregular fluctuation pattern, thereby intensifying the spatial heterogeneity of water distribution [17]. In addition to climatic influences, human activities have increasingly shaped vegetation patterns in the region. Rapid population growth has led to land expansion and intensified resource use, while changes in livestock management, such as increased herd sizes and grazing intensity, have accelerated alpine meadow degradation. Industrial development and urbanization have further contributed to land-use conversion, habitat fragmentation, and localized ecological degradation. These socio-economic pressures, interacting with climatic variability, create complex spatiotemporal heterogeneity in vegetation dynamics [18]. Moreover, unsustainable grazing practices and extensive livestock trampling have markedly accelerated the fragmentation of alpine meadows in the region. Extensive research has been conducted to understand the spatiotemporal patterns and driving mechanisms of vegetation changes in this region. Initial investigations focused on altitudinal gradients and land-use patterns [17], while subsequent studies employed advanced analytical approaches, including correlation analysis to quantify vegetation responses to environmental changes [19,20] and residual analysis which compares observed and model-predicted values to identify unexplained variations and thereby separate climatic and anthropogenic contributions [18].

More recent advancements have incorporated eco-hydrological modeling to simulate hydrological processes and analyze the complex interactions among multiple environmental factors [21]. Despite these advancements, the current research landscape exhibits several limitations. While numerous studies have extensively examined the climate-vegetation relationship in the Yellow River’s source region [22,23,24], the anthropogenic dimension remains relatively underexplored. Although previous studies have investigated the impacts of human activities on vegetation dynamics in the Yellow River Basin [25], they have primarily focused on ecological restoration projects. Residual analysis is a statistical method that separates the contributions of climate and human activities to vegetation changes by analyzing the deviations between observed and climate-predicted vegetation indices. However, the relationships between key socioeconomic factors—such as regional economic development, population density, and livestock density—and vegetation change have not yet been systematically examined. By explicitly addressing these socio-economic drivers in addition to climatic factors, this study provides new insights into the mechanisms shaping vegetation dynamics in the source region of the Yellow River. This highlights a methodological and theoretical gap in fully integrating socio-economic factors into vegetation change analyses, which our study addresses. Furthermore, by providing a framework for combining climatic and socio-economic drivers in vegetation dynamics research, this study offers guidance that may be applicable to other alpine or ecologically sensitive regions globally, enhancing the understanding of vegetation response mechanisms under combined environmental and human pressures.

Building upon existing research, this research aims to: (a) investigate the temporal and spatial trends of NDVI at both annual and seasonal scales from 1998 to 2018, (b) analyze regional differences in vegetation responses across different sub-areas of the Yellow River source region, and (c) quantify and compare the relative impacts of climatic factors (temperature and precipitation) and anthropogenic factors (population growth and livestock density) on vegetation dynamics. The findings are expected to provide a scientific basis and reference for sustainable ecological conservation and restoration strategies in the source region of the Yellow River.

2. Materials and Methods

2.1. Study Area

The source region of the Yellow River is located in the northeastern part of the Qinghai–Tibet Plateau, with geographical coordinates ranging from 95°50′ to 102°30′ E and 33°12′ to 36°48′ N. The headwaters of the Yellow River originate at the administrative boundary between Maduo County and Qumalai County in Qinghai Province, with its hydrological boundary extending to the Tangnaihai Hydrological Station in Xinghai County, which serves as the regional outlet [26]. The basin spans approximately 120,000 km² across six prefectures in three provincial-level administrative regions (Qinghai, Gansu, and Sichuan). The region features complex and diverse topography, with elevations ranging from 2605 to 6200 m, and the terrain generally decreases in elevation [1].

The climate is characterized as a semi-humid to semi-arid sub-frigid plateau climate, with relatively low precipitation and temperatures, influenced by both the plateau climate and geographical location. The natural ecosystems within the Yellow River’s source region are particularly vulnerable due to their high ecological sensitivity and fragility, establishing this area as a critical conservation zone within the broader Yellow River Basin ecosystem.

2.2. Data Sources

This study utilizes NDVI remote sensing satellite data from 1998 to 2018, obtained from the National Tibetan Plateau Data Center (https://data.tpdc.ac.cn, accessed on 19 October 2025). The data are derived from the MODIS vegetation index product MOD13Q1, with a temporal resolution of 16 days and a spatial resolution of 250 m. After format conversion and projection transformation, the maximum value composite (MVC) method was applied to generate seasonal and annual NDVI datasets for the study area. Meteorological data were collected from seven weather stations within the source region of the Yellow River (Figure 1) Digital Elevation Model (DEM) data were sourced from the GDEMV3 dataset (https://www.gscloud.cn/, accessed on 19 October 2025), with a spatial resolution of 30 m. These data were mosaicked, clipped, and processed to extract and classify elevation information for the study region. The 2020 land use data were derived from Landsat remote sensing imagery and obtained from the Resource and Environment Science and Data Center (https://www.resdc.cn/, accessed on 19 October 2025), provided as an annual dataset with a spatial resolution of 1 km, representing detailed land cover and use categories across the study region (Figure 2). An eco-climatic zoning map of the study area was also referenced to facilitate zonal analysis. The DEM data were reclassified into six elevation zones (<3250 m, 3250–3750 m, 3750–4250 m, 4250–4750 m, 4750–5250 m, and 5250–6250 m) for subsequent analysis. Population and economic data were extracted from provincial statistical yearbooks.

The period 1998–2018 was selected for this study because it provides a sufficiently long and continuous dataset for analyzing long-term vegetation trends and their responses to climatic and socio-economic drivers. MODIS NDVI and corresponding climatic data are complete and consistent over this period, ensuring robust trend analyses. More recent data (2019–2025) were not fully available or validated at the time of analysis, but future studies could extend the investigation to include these years to capture ongoing vegetation dynamics.

The climatic analysis focused on temperature and precipitation. This selection was guided by their established ecological significance in controlling alpine vegetation growth and, pragmatically, by the availability of continuous, long-term records from meteorological stations. Although variables like solar radiation, soil moisture, and evapotranspiration are recognized as influential, comprehensive spatial datasets for these parameters were not available for the entire study period and region. Therefore, they were excluded to maintain a consistent and reproducible analytical scope.

2.3. Methods

2.3.1. Maximum Value Composite (MVC) Method

Maximum Value Composite (MVC) method effectively mitigates the influence of factors such as cloud cover, atmospheric water vapor, and solar angle [27].

$$NDV{I}_m=Max(NDV{I}_i)$$

(1)

where $NDV{I}_i$ represents the 16-day or monthly NDVI, and $NDV{I}_m$ represents the seasonal or annual NDVI composites.

In this study, the MVC method was employed to process the original 16-day NDVI data, generating seasonal and annual NDVI datasets. The seasons are categorized as follows: spring (February to May), summer (June to August), autumn (September to November), and winter (December to February) [25]. In addition to the F-test, the non-parametric Mann–Kendall test was also applied to the seasonal NDVI time series to robustly verify the significance of the observed trends, as it is less sensitive to data distribution and outliers.

2.3.2. Linear Trend Analysis

In vegetation studies, univariate linear regression is commonly employed for trend analysis [28]. The calculation formula is as follows:

$${\theta }_{\mathrm{slope}}={\displaystyle \frac{n\times {\displaystyle \sum _{i=1}^{n}i}\times NDV{I}_i-{\displaystyle \sum _{i=1}^{n}i}{\displaystyle \sum _{i=1}^{n}N}DV{I}_i}{n\times {\displaystyle \sum _{i=1}^n{i}^2}-{\left({\displaystyle \sum _{i=1}^{n}i}\right)}^2}}$$

(2)

In the equation, ${\theta }_{\mathrm{slope}}$ represents the slope of the trend line, indicating the annual rate of change in NDVI; $NDV{I}_i$ denotes the NDVI value for the $i$ year; $i$ is the year variable, $i=1,2,3,\dots ,21$; $n=21$, represents the total number of monitoring years. When ${\theta }_{\mathrm{slope}}>0$, it indicates an increasing trend in vegetation NDVI; conversely, when ${\theta }_{\mathrm{slope}}<0$, it signifies a decreasing trend.

In this study, the univariate linear regression method was used to calculate the overall trends of vegetation, temperature, and precipitation data over the entire study period. The trend maps reflect the changes in NDVI within the study area over the multi-year time series [18,29]. The NDVI trend represents the annual change rate of NDVI. The results of the linear trend analysis were subjected to significance testing using the F-test method (p = 0.01). The formula for the F-test is as follows:

$$F={\displaystyle \frac{S_R}{S_E/(n-2)}}$$

(3)

$$S_R={\displaystyle \sum _{i=1}^n{\left({\widehat{y}}_i-\overline{y}\right)}^2}$$

(4)

$$S_E={\displaystyle \sum _{i=1}^n{\left(y_i-{\widehat{y}}_i\right)}^2}$$

(5)

In the equation, $S_R$ represents the regression sum of squares, with the regression degrees of freedom equal to 1; $S_E$ denotes the residual sum of squares; $n-2$ represents the residual degrees of freedom. $i=1,2,3,\dots ,21$. $y_i$ represents the value of the variable y for the $i$ year; ${\widehat{y}}_i$ denotes the linear regression value of $i$ with respect to the variable y; $\overline{y}$ is the mean value of the variable y.

2.3.3. Correlation Analysis

To elucidate the influence of climatic factors on vegetation growth, a quantitative analysis of the correlation between growing season NDVI and temperature as well as precipitation was conducted. Based on pixel-level spatial analysis, the correlation coefficients between NDVI and temperature/precipitation were calculated for each grid cell. Among correlation analysis methods, the Pearson correlation coefficient is the most widely applied [30]. Furthermore, to quantify the direct association between vegetation trends and climatic trends at the station scale, a simple linear regression analysis was conducted between the NDVI trends and the temperature and precipitation trends across the seven meteorological stations. The calculation formula is as follows [31]:

$$R_{x,y}={\displaystyle \frac{{\displaystyle \sum _{i=1}^n(x_i-\overline{x})}(y_i-\overline{y})}{\sqrt{{\displaystyle \sum _{i=1}^n{(x_i-\overline{x})}^2}}\sqrt{{\displaystyle \sum _{i=1}^n{(y_i-\overline{y})}^2}}}}$$

(6)

In the equation, $R_{x,y}$ is a statistical metric used to quantify the degree of correlation between two variables, with a value range of [−1, 1]. When $R_{x,y}>0$, it indicates a positive correlation between the two variables; the closer the value is to 1, the stronger the correlation, and the closer it is to 0, the weaker the correlation. Conversely, when $R_{x,y}<0$, it signifies a negative correlation between the two variables; the closer the value is to −1, the stronger the correlation, and the closer it is to 0, the weaker the correlation.

To explicitly account for spatial heterogeneity, temperature and precipitation data were used at the individual meteorological station level rather than interpolated into continuous surfaces. Although interpolation methods have been suggested [3,7], they are less appropriate for the Yellow River source region due to its vast extent and strong heterogeneity among climatic zones, vertical belts, and ecological regions. Instead, NDVI values were extracted to station locations using the ‘Extract Values to Points’ tool in ArcGIS 10.4 with bilinear interpolation from adjacent raster cells, and correlation coefficients between NDVI and climatic factors were tested using the t-test. For anthropogenic drivers, only county-level population and livestock density data were available; therefore, regression analyses were conducted using the average NDVI trend of each county as the dependent variable [32]. Finally, to analyze vegetation trends across different underlying surfaces, zonal statistics were computed. The annual NDVI trend raster was statistically summarized within each zone of the eco-climatic map, land-use map, and reclassified elevation map using the ‘Zonal Statistics as Table’ tool in ArcGIS 10.4, calculating the mean trend and the proportion of pixels with significant changes for each category.

3. Results

3.1. Spatiotemporal Trends of Vegetation

3.1.1. Temporal Trends of Vegetation

Time series analysis of NDVI, a dimensionless value ranging from −1 to 1 with higher values indicating greater vegetation biomass and values around 0 representing bare soil, quantifies the dynamic trends of vegetation responses to climate change across the entire source region of the Yellow River. From 1998 to 2018, the annual average NDVI in the study area showed a significant increasing trend (p < 0.01), with a growth rate of 0.004 per year (Figure 3, Equation (1)). The maximum annual average NDVI was 0.70, occurring in 2018, while the minimum value was 0.59, observed in 2003. The multi-year average NDVI for the source region of the Yellow River was 0.63. Significant seasonal differences in NDVI were observed within the study area (Figure 3c).

The seasonal average NDVI exhibited an inverted “U” shape throughout the year, a pattern that repeated annually. Spring and winter NDVI values were relatively low, with averages of 0.31 and 0.20, respectively, while summer and autumn values were higher, with averages of 0.61 and 0.42 (Figure 3a,c, Equation (1)). Based on the Mann–Kendall significance test, the NDVI in spring, summer, and winter showed a significant increasing trend (p < 0.01), with growth rates of 0.0029, 0.0016, and 0.0017, respectively. In contrast, the NDVI in autumn exhibited a non-significant decreasing trend.

3.1.2. Spatial Trends of Vegetation

A pixel-by-pixel analysis was conducted to examine the vegetation change trends in the source region of the Yellow River and to explore the responses of vegetation to climate change. The results indicate that from 1998 to 2018, vegetation growth improved in 92.7% of the study area, while localized regions experienced vegetation degradation (Figure 4, Equation (2)), accounting for 7.27% of the total area. The degraded areas were primarily distributed in regions with higher population densities. The spatial distribution of annual NDVI exhibited significant regional differences, generally decreasing from the southeast to the northwest, which aligns with the spatial patterns of climate and elevation. Notably, the greatest vegetation increase occurred in areas with elevations of 3250–3750 m, whereas vegetation decline was mainly concentrated in regions with elevations of 5250–6250 m.

The trends in seasonal NDVI changes in the source region of the Yellow River differ between seasons (Figure 5, Equation (2)). The areas with increased vegetation coverage in spring, summer, autumn, and winter accounted for 90%, 75%, 77%, and 82% of the total area, respectively. Among these, the proportions of areas with statistically significant (p < 0.01) increases were 34%, 20%, 15%, and 21%, respectively. Although an overall increasing trend was observed across all seasons, vegetation degradation in summer and autumn remains a notable concern. The areas experiencing vegetation degradation in summer and autumn accounted for 25% and 27% of the total area, respectively.

3.2. Vegetation Changes Under the Influence of Climate Change

It is widely recognized that vegetation growth is directly influenced by climatic variability. In the source region of the Yellow River, both temperature and precipitation exhibited overall increasing trends during 1998–2018, but with pronounced spatial heterogeneity. Temperature rose significantly across most areas (92.8%), with the most pronounced warming observed in the central and eastern parts, where the rate of increase exceeded 0.35 °C per decade (Figure 6A, Equation (2)). In contrast, localized cooling trends were detected in some northern and western areas. Precipitation also showed an overall increasing trend (97.1% of the region), with the eastern and southern parts experiencing the largest increases, up to 3 mm/year (Figure 6B, Equation (2)). However, localized decreases were observed in the western and, particularly, the northwestern parts of the study area.

To further quantify the influence of climatic factors on vegetation dynamics, we performed linear regression analyses between the NDVI trends and the trends of temperature and precipitation across the seven meteorological stations (Figure 7, Equations (2) and (3)). Among the stations, the most pronounced simultaneous increases in NDVI and temperature were observed in the central and eastern parts of the study region (e.g., stations Maqin and Xinghai). The results demonstrate a statistically significant and ecologically relevant positive relationship between temperature trends and NDVI trends (R² = 0.4387, p < 0.01). Conversely, the relationship between precipitation trends and NDVI trends, while also statistically significant (p < 0.01), was notably weaker (R² = 0.1391). One station in the northwest (e.g., station Qumalai) even exhibited a diverging trend, with increasing precipitation alongside a decreasing NDVI trend.

To further elucidate the seasonal dynamics of climate-vegetation interactions, we conducted a correlation analysis between the regional average NDVI and the key climatic variables (temperature and precipitation) for each season. The statistical results, summarized in

Table 1. Correlation analysis between regional average seasonal NDVI and climatic factors during 1998–2018.

Season	Temperature		Precipitation
	r	p-Value	r	p-Value
Spring	0.66	<0.01	0.60	<0.01
Summer	0.69	<0.01	0.66	<0.01
Autumn	0.31	0.16	0.21	0.37
Winter	0.52	<0.01	0.02	0.94

, reveal distinct seasonal patterns. Notably, both spring and summer exhibited strong, statistically significant positive correlations between NDVI and the two climatic factors (temperature: r = 0.66/0.69, p < 0.01; precipitation: r = 0.60/0.66, p < 0.01). In contrast, the relationships weakened substantially during autumn, with non-significant correlations for both temperature (r = 0.31, p = 0.16) and precipitation (r = 0.21, p = 0.37). A divergent pattern emerged in winter, where NDVI maintained a significant moderate correlation with temperature (r = 0.52, p < 0.01) but showed no association with precipitation (r = 0.02, p = 0.94).

3.3. Vegetation Changes Under the Influence of Human Activities

Human activities, especially population growth and livestock expansion, have exerted considerable pressure on alpine grasslands in the source region of the Yellow River. Statistical analysis shows that from 1998 to 2018, the population increased from 250,000 to 390,000, a rise of nearly 60%. Livestock numbers, expressed in sheep units, increased from 5.36 million to 6.10 million, an increase of about 14%. Population growth was concentrated in counties such as Maduo and Dari, while livestock expansion was most notable in Henan, Banma, and Dari counties. Spatially, areas with higher population and livestock density were mainly distributed in the northeastern low-altitude regions, whereas the western high-altitude areas generally exhibited lower densities. These spatial patterns were associated with the spatial distribution of vegetation change trends (Figure 8, Equation (2)).

Regression analysis at the county level was conducted to examine the relationships between human activity on vegetation dynamics (Figure 9). NDVI trends declined significantly with increasing livestock numbers and higher population densities. The regression analysis indicates a weak but statistically significant negative relationship between population trend and NDVI trend (R² = 0.1991, p < 0.01). Specifically, an increase in one unit in the population trend is associated with an average decrease of 0.0001 in the NDVI trend. Statistical tests showed that both factors passed the significance threshold.

Beyond demographic and grazing pressures, broader socio-economic changes also shaped vegetation dynamics. Specifically, the relationship between industrial/urban expansion and vegetation trends, as revealed by Figure 10, requires attention. Analysis of industrial structure shows that the primary industry, dominated by traditional animal husbandry, maintained a large proportion of the economy and continued to grow during 1998–2018 (Figure 10). Meanwhile, the secondary industry, including mining and energy development, expanded rapidly in some counties, while the tertiary industry, especially tourism and services, experienced marked growth in the later years of the study period. At the same time, the urbanization rate rose sharply from 3.27% in 1998 to 16.85% in 2018. This process of economic diversification and urban expansion not only altered land-use patterns but also increased ecological pressure, as more land was converted to settlements, infrastructure, and non-pastoral economic activities.

Overall, the evidence demonstrates spatial and statistical associations between human activities and vegetation dynamics in the source region of the Yellow River during the study period.

3.4. Vegetation Changes Across Different Underlying Surfaces

Vegetation dynamics in the source region of the Yellow River also exhibited significant differences across ecological climate zones, land-use types, and elevation bands, reflecting the combined influence of natural conditions and human activities.

At the scale of ecological climate zones, vegetation improvement was generally dominant, but the extent varied considerably among subregions (Figure 11A(1),A(2)). The Qilian–Qinghai Lake area and the eastern Qingnan Plateau showed the most pronounced increases in NDVI, with more than 70% of the area experiencing improvement. By contrast, the western parts of the plateau, such as the Maduo Basin, exhibited a higher proportion of degradation, with more than 30% of the grassland area showing NDVI decline.

Land-use type also played a critical role in shaping vegetation dynamics (Figure 11B(1),B(2)). Grasslands, which account for the largest proportion of land cover in the study area, showed the most notable changes. Approximately 64% of grassland areas exhibited improved NDVI trends, particularly in meadow pastures with relatively abundant soil moisture. However, degraded patches were widely distributed in steppe grasslands and degraded meadow zones. In contrast, cropland areas showed relatively stable NDVI. Woodland and wetland areas exhibited overall improvement.

Elevation exerted an additional layer of influence on vegetation dynamics (Figure 11C(1),C(2)). At lower elevations (<3500 m), NDVI showed weaker improvement. Mid-elevation bands (3500–4200 m) exhibited the most significant improvement, accounting for more than 70% of the vegetation increases. At higher elevations (>4200 m), vegetation dynamics were more complex, with extensive permafrost degradation and harsher hydrothermal constraints.

4. Discussion

4.1. Climatic Changes and Vegetation Dynamics

The spatiotemporal trends in NDVI from 1998 to 2018 provide strong evidence that climate is a predominant driver of vegetation dynamics in the Yellow River Source Region. In summary, our spatial and seasonal analyses together reveal that climate exerts a dual influence on vegetation dynamics. Spatially, this is manifested as a clear dichotomy: on the one hand, warming combined with increased precipitation in the eastern and central parts of the region has significantly promoted vegetation growth. On the other hand, warm–dry conditions in the northwestern part have constrained productivity and driven vegetation degradation. This pattern, where vegetation degradation was concentrated in the northwestern warm–dry zones, underscores that the benefits of climate change are not uniform and reveals the regulatory role of regional climatic background on vegetation response [17]. These findings emphasize the critical importance of considering both spatial heterogeneity and seasonal variability when evaluating the impact of climate change on alpine ecosystems.

The overall greening trend observed in our study aligns with extensive research on the Tibetan Plateau, where climate warming and precipitation increases have been widely reported as key drivers of enhanced vegetation productivity [23,24,33]. However, our results add critical nuance to this general understanding. The observed vegetation degradation during summer and autumn suggests that warming alone cannot guarantee vegetation improvement when water availability becomes a limiting factor. This finding challenges studies that emphasize universally positive effects of temperature increases in alpine regions [31], and instead strongly supports the growing consensus that the ecological consequences of warming are highly contingent on local moisture conditions [14]. The coupling between climate and vegetation is most pronounced during the primary growing seasons (spring and summer), becomes decoupled in autumn, and is primarily driven by temperature rather than precipitation during the winter period, as confirmed by our seasonal correlation analysis. This seasonal dynamic can be attributed to the fact that spring precipitation is crucial for triggering growth initiation, and adequate summer moisture supports peak biomass accumulation. In contrast, autumn precipitation may be insufficient to offset elevated evapotranspiration demands under warmer temperatures, leading to water stress and premature senescence. The winter correlation with temperature likely reflects the role of warmer conditions in mitigating cold stress and influencing snow cover dynamics.

Furthermore, the divergent responses across different eco-climatic zones and elevation gradients reinforce the centrality of water–energy balance, echoing earlier findings focused on altitudinal gradients and land-use patterns in this region [22,25]. The most significant vegetation improvement in mid-elevation bands (3500–4200 m) reflects favorable climatic conditions and moderate human disturbance, whereas the degradation in high-altitude areas, despite warming trends, is linked to harsher hydrothermal constraints and extensive permafrost degradation. Ultimately, the spatial and seasonal patterns illustrate that the matching of hydrothermal conditions plays a decisive role in vegetation dynamics. This principle of water–temperature co-limitation is fundamental to projecting the future trajectory of alpine ecosystems under ongoing climate change.

4.2. Human Activities and Vegetation Dynamics

Our analysis confirms that human activities constitute a critical driver of vegetation dynamics in the Yellow River Source Region, operating alongside and interacting with climatic factors. These spatial patterns correspond closely with vegetation change: regions with high human and livestock pressure often showed vegetation degradation, while sparsely populated higher-altitude areas tended to experience vegetation recovery. This pattern of degradation near human settlements and recovery in remote areas is consistent with findings from other fragile ecosystems on the Plateau [18]. Regression analysis at the county level confirmed the negative impact of human activity on vegetation dynamics, demonstrating statistically significant negative relationships between NDVI trends and both livestock numbers and population densities. The effect was particularly strong in the northeastern parts of the region, where grassland ecosystems are more accessible and thus more intensively utilized.

The negative correlations observed between NDVI trends and both population and livestock density align with previous research on the Tibetan Plateau that identified grazing intensity as a major driver of vegetation decline [34,35]. Our results regarding livestock expansion are further supported by [36], which also documented grassland degradation due to increasing herd sizes. Our findings reinforce this understanding, particularly through the spatial demonstration of degradation in accessible low-altitude zones. Rapid population growth and livestock expansion directly intensified grazing pressure on low-altitude grasslands, leading to the degradation observed in these areas. In contrast, counties with lower human and livestock density generally exhibited positive NDVI trends, reflecting less anthropogenic disturbance.

Beyond pastoral activities, socio-economic restructuring and urbanization also influenced vegetation change. The rapid growth of secondary industries such as mining and energy development, together with the notable increase in urbanization rate from 3.27% to 16.85% between 1998 and 2018, placed additional ecological pressure on the fragile alpine environment. This trend of industrial expansion impacting fragile environments is not unique to our study area and has been reported in other regions of western China [37]. Furthermore, the expansion of secondary/tertiary industries and urbanization has predominantly exerted negative impacts on vegetation cover by fragmenting and converting natural landscapes, despite the concurrent overall greening trend observed in the region. This process of economic diversification and urban expansion not only altered land-use patterns but also increased ecological pressure, as more land was converted to settlements, infrastructure, and non-pastoral economic activities. Previous research [23] emphasized that industrial expansion and urban development exacerbate ecological risks in alpine grasslands. However, unlike studies that focus primarily on grazing [33], our integration of demographic, livestock, and economic indicators provides more comprehensive evidence that multiple human drivers—rather than grazing alone—jointly determine vegetation dynamics in the Yellow River Source Region.

The combined evidence demonstrates that human activities are an essential driver of vegetation dynamics in the source region of the Yellow River. Together, these socio-economic factors have played a critical role in shaping spatial patterns of vegetation degradation and recovery during the study period. This comprehensive understanding of multiple anthropogenic pressures highlights the need for integrated management approaches that address not only grazing pressure but also the impacts of economic development and urbanization in this ecologically sensitive region.

4.3. The Interplay of Climatic and Anthropogenic Drivers

Building upon the separate analyses of climatic and anthropogenic factors, a synthesis of our findings provides a more holistic understanding of the vegetation dynamics in the Yellow River Source Region. In summary, vegetation changes in the source region were shaped by a combination of ecological heterogeneity and anthropogenic disturbance. This interplay is vividly demonstrated by the divergent responses across different underlying surfaces, as detailed in Section 3.4.

Vegetation improvement was most pronounced in favorable climate zones, meadow pastures, and mid-elevation bands, where the beneficial effects of warming and increased precipitation coincide with relatively favorable hydrothermal conditions and moderate human disturbance. Conversely, degradation was concentrated in areas where climatic limitations overlapped with high-intensity land use. For instance, the degradation in the western Maduo Basin indicates that regional climatic constraints, combined with intensive grazing pressure, contributed to localized vegetation deterioration. Similarly, at lower elevations, degradation was concentrated in areas of high population and livestock density, whereas the improvement in mid-elevation bands reflects favorable climatic conditions and moderate human disturbance. The complexity at higher elevations, where warming trends facilitated vegetation growth in some areas but was counteracted by harsher hydrothermal constraints and extensive permafrost degradation, further underscores the primacy of the local water-energy balance.

This integrated perspective confirms that neither climate nor human activities act in isolation. The overall greening trend is largely a climate-driven process, yet its spatial pattern is significantly modified by anthropogenic pressures. The most severe degradation occurs where climatic stressors (e.g., water limitation in the northwest, harsh conditions at high altitudes) and human pressures (e.g., overgrazing in the northeast, land conversion in low-altitudes) converge. These results highlight the importance of considering underlying surface conditions when evaluating vegetation dynamics and their driving mechanisms, as these conditions dictate the local sensitivity and response to broader climatic and anthropogenic forcings.

4.4. Limitations

While this study provides insights into the vegetation dynamics of the Yellow River Source Region, several limitations should be considered. First, the analysis of climatic drivers was constrained by the sparse distribution of meteorological stations, which may not fully capture the heterogeneity of the complex terrain and could introduce uncertainty in the spatial representativeness of the station-based regression results. Second, the use of county-level socioeconomic data represents a coarse spatial scale that may not perfectly match the pixel-level NDVI trends, potentially obscuring fine-scale human-environment interactions. Finally, while temperature and precipitation are primary drivers, the exclusion of other factors like solar radiation, soil moisture, and actual evapotranspiration due to data constraints limits a more comprehensive mechanistic understanding. Future research should therefore prioritize the integration of spatially interpolated climate data, higher-resolution gridded population and livestock datasets, and satellite-derived soil moisture and solar radiation products to better disentangle the complex interplay of factors governing vegetation change in this sensitive alpine ecosystem.

5. Conclusions

This study systematically investigated the spatiotemporal trends of vegetation and quantified the impacts of climatic and anthropogenic factors in the Yellow River Source Region from 1998 to 2018. The principal conclusions are as follows:

(1)

The region experienced a significant overall greening trend, yet this recovery is spatially heterogeneous and seasonally vulnerable. While vegetation improved across 92.7% of the area, the persistent and extensive degradation observed during the summer and autumn seasons, affecting approximately one-quarter of the region, underscores that the ecosystem’s recovery is fragile and susceptible to seasonal climatic stressors, particularly in southeastern low-altitude areas.

(2)

Climate change is the dominant driver of the large-scale greening pattern, but its effectiveness is contingent on hydrothermal coordination. Warming temperatures and increased precipitation jointly promoted vegetation growth, explaining the overarching positive trend. However, the observed degradation in the northwest and during summer-autumn reveals that warming alone is insufficient and can even be detrimental where water availability is limited, demonstrating that the matching of hydrothermal conditions is ultimately decisive for vegetation dynamics.

(3)

Human activities are critical drivers of localized degradation, acting as key modifiers of the spatially heterogeneous vegetation pattern. Rapid population growth, livestock expansion, and associated socio-economic development have significantly exacerbated grassland degradation, particularly in accessible low-altitude regions. These anthropogenic pressures interact with climatic limitations, shaping the specific spatial pattern of vegetation degradation and recovery.

In light of our findings that vegetation degradation is most pronounced in low-altitude grasslands with high livestock density during summer and autumn, we recommend implementing a rotational grazing system combined with seasonal grazing bans in these vulnerable areas. Specifically, local governments could establish a grazing calendar that restricts livestock access to alpine meadows during peak degradation periods (summer and autumn) and designate alternative forage supply areas or provide subsidies for supplementary feed. Such measures would directly alleviate grazing pressure, promote natural vegetation recovery, and enhance the ecological resilience of grasslands in the source region of the Yellow River.