Assessing the Response of the Net Primary Productivity to Snow Phenology Changes in the Tibetan Plateau: Trends and Environmental Drivers

Abstract

Understanding the relationship between climate, snow cover, and vegetation Net Primary Productivity (NPP) in the Tibetan Plateau (TP) is crucial. However, the role of snow cover in influencing the NPP remains unclear. This study investigates the connection between the NPP and snow phenology (SP) across the TP from 2011 to 2020. Interannual trends were assessed using the Theil–Sen non-parametric regression approach combined with the Mann–Kendall test. Additionally, the pathways through which snow cover affects the NPP, considering various environmental factors, were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). Approximately 10.72% of the TP showed a significant decrease in the NPP, accompanied by advancing trends in the Snow Onset Date (SOD) and Snow End Date (SED), as well as a gradual decrease in the Snow Cover Duration (SCD). The PLS-SEM results reveal that precipitation and soil temperature significantly influenced the NPP, with total effects of 0.309 and 0.206 in the SCD structural equation. Temperature had a relatively strong indirect effect on the NPP through its influence on the SOD and SCD, contributing 16% and 10% to the total effect, respectively. Neglecting the mediating effect of SP underestimates the environmental impact on the NPP. This study highlights how environmental factors influence the NPP through snow cover changes as the biomass increases, thereby enhancing our understanding of SP’s impact on the TP.

1. Introduction

The Tibetan Plateau (TP), known for its high elevation and geological complexity, is currently experiencing warming at nearly twice the global average rate against the backdrop of climate change [1]. This warming trend is significantly affecting key components of the ecological environment, including the hydrological cycle and vegetation growth.

Increasing temperatures are altering snow phenology (SP), which refers to the seasonal patterns and timing of snow events, impacting key indicators such as the Snow Onset Date (SOD), Snow End Date (SED), and Snow Cover Duration (SCD) [2]. An uneven spatial distribution of SP was observed in the TP from 2002 to 2022 [3]. Specifically, they found that the SCD significantly decreased in 4.62% of the TP, the SOD showed a marked delay in 2.34%, and the SED advanced significantly in 1.52% of the region. SP plays a crucial role in the TP’s terrestrial ecosystem, substantially influencing vegetation growth [4]. Snow cover can modulate the intensity of light received by vegetation and affect the soil surface temperature [5,6], thus influencing physical processes within the soil, the soil’s hydrothermal structure, and various ecological processes involving flora and fauna. Therefore, changes in SP have significant ecological implications, emphasizing the need for continuous monitoring of SP variations [7].

The Net Primary Productivity (NPP) is a crucial measure of plant growth and directly reflects its ability to convert CO₂ into organic matter [8]. Previous studies have investigated the influence of SP on the NPP. In Northern Sweden, a reduction in shrub productivity was linked to decreased winter snow cover [9]. Similarly, snow manipulation experiments on the Old Man Range in New Zealand demonstrated a clear response of plant growth to snow cover conditions [10]. Research in the Sierra Nevada has also shown that snow cover’s effect on forest productivity varies with elevation [11]. However, the specific mechanisms through which SP affects the NPP remain unclear. A study in Alaska found that snow cover influences vegetation productivity by altering the soil’s hydrothermal conditions [12]. A previous study revealed that the start date of spring snowmelt directly affects the vegetation NPP [13], while delayed snowmelt in temperate grasslands leads to an increased NPP [14]. The delayed response of alpine vegetation productivity to snowmelt is influenced by water and thermal conditions [15].

The abovementioned studies primarily identified the effect of snow cover on vegetation NPP through a correlation analysis and showed that this effect is also modulated by various environmental variables. However, these studies have only compared the contribution of environmental factors to vegetation productivity without considering their indirect impact on snow cover [16]. In reality, vegetation is affected not only by SP but also by other meteorological factors, such as temperature, precipitation, soil moisture, radiation, elevation, soil potential evapotranspiration (PET), and soil temperature. These factors can directly affect vegetation growth and indirectly impact vegetation by affecting snow accumulation and ablation. Currently, the indirect effects of meteorological factors on vegetation through SP, as well as the specific pathways by which SP influences vegetation in the TP, remain unclear. Key variables, including temperature, soil moisture, precipitation, radiation, and PET, were incorporated into our experiment due to their direct and indirect influences on plant productivity and ecosystem processes [17]. Among them, factors such as temperature, radiation, and precipitation are also crucial indicators influencing snow accumulation [18]. Altitude impacts both vegetation and snow cover, with potential indirect effects on snow cover through its influence on vegetation [3]. Soil temperature and soil moisture significantly affect microbial activity within the soil and the processes involved in vegetation growth [7]. Previous studies suggested that vegetation growth can be indirectly affected by changes in soil temperature and moisture [19]. However, whether soil temperature and moisture can indirectly influence vegetation by altering snow conditions remains unclear. In conclusion, we selected these environmental factors for investigation.

Therefore, we aimed to investigate how the NPP responds to SP changes in the TP using remote sensing data from 2011 to 2020. First, the spatial distribution of the NPP and SP during this period was analyzed. The interannual variations in the NPP and various SP indicators were assessed using the Theil–Sen non-parametric regression approach, and the significance of these trends was evaluated using the Mann–Kendall test. A partial correlation analysis was then conducted to explore the correlation and significance between the NPP and SP. Finally, environmental factors were integrated into Partial Least Squares Structural Equation Modeling (PLS-SEM) to elucidate their complex direct and indirect effects on SP and the NPP in the study area. This model aims to clarify the response relationship between the NPP and SP.

2. Study Area and Data Sources

2.1. Study Area

Our study area is the TP (Figure 1), which extends from 73°18′52″E to 104°46′59″E longitudinally and from 26°00′12″N to 39°46′50″N latitudinally, with an average elevation exceeding 4000 m. Elevation plays a key role in vegetation distribution, with alpine steppes and meadows covering more than 60% of the TP [20]. Alpine cold steppes, located between 4100 and 5100 m, are characterized by hardy herbs and dwarf shrubs, while alpine meadows, typically at lower elevations, are often waterlogged [21]. Shrublands are mainly concentrated in the southeast, coniferous and broadleaf forests dominate the south and southeast, and the northwest is largely barren [22]. The TP is highly sensitive to climate change, with temperatures increasing by about 0.4 °C per decade over the past 30 years [23]. The southeastern region experiences warm, moist conditions, contrasting with the cold, dry conditions of the northwest. Annual temperatures range from −6 to 20 °C, and annual precipitation varies between 150 and 800 mm [3]. Snow accumulation on the TP begins in September, increases rapidly, peaks from December to February, and typically ends in May [24].

2.2. Datasets

The MODIS Terra NPP product (MOD17A3HGF, v6) with a 500 m spatial resolution and annual temporal interval was sourced from the NASA Earth Data platform. (https://lpdaac.usgs.gov/products/mod17a3hgfv061/, accessed on 9 November 2023).

Snow phenology variables were obtained from the HMRFS-TP daily snow product (https://poles.tpdc.ac.cn/en/data/52c1ea76-e10e-48de-93d4-468ce15db8fc/, accessed on 11 November 2023), which provides gap-free daily snow cover information for the TP from 2002 to 2021, generated using HMRF modeling on MODIS data. In this study, a complete snow year spans from 1 September to 31 August of the subsequent year. Day of the snow year (DOS) 1 corresponds to 1 September. The SOD and SED refer to the first and last five consecutive days in each snow year when a pixel is identified as being snow-covered, while the SCD represents the duration between the SOD and SED [25].

Vegetation classification data were obtained from the Science Data Bank (https://www.scidb.cn/en/doi/10.11922/sciencedb.398, accessed on 21 December 2023). The data were derived from MODIS product composites and validated against ground-truth data, showing high accuracy at a 500 m spatial resolution. This study focused on five representative vegetation types: alpine steppes, alpine meadows, shrublands, coniferous forests, and broadleaf forests.

Surface solar net radiation data were sourced from the ERA5-Climate Data Store’s CLARA-A3 product (https://forum.ecmwf.int/t/informative-page-in-readiness-for-new-cds-beta/2985, accessed on 8 November 2023), with daily data at a 0.5° spatial resolution. Annual mean radiation data were calculated to generate a time series of radiation data for the study period. These data were resampled to 500 m using nearest-neighbor interpolation.

Elevation data were sourced from NASA’s NASADEM_HGT v001 (https://lpdaac.usgs.gov/products/nasadem_hgtv001/, accessed on 13 July 2023), which offers global elevation data at a 30 m resolution. This dataset, derived from the 2000 Shuttle Radar Topography Mission, combines multiple sources to deliver precise terrain measurements and geolocations. In this study, the data were reprojected to a 500 m resolution.

Meteorological data included precipitation, temperature, and PET. Monthly precipitation data at a 1 km resolution for China were sourced from the Environmental Big Data Platform for Three Poles (http://poles.tpdc.ac.cn/en/data/f-aae7605-a0f2-4d18-b28f-5cee413766a2/, accessed on 30 October 2023) [26]. Temperature data were derived from the monthly average temperature dataset at a 1 km resolution provided by the National TP Data Center (https://data.tpdc.ac.cn/zhhans/data/71a-b4677b66c4fd1a004b2a5-41c4d5bf/?q=%E5%BD%AD%E5%AE%88%E7%92%8B, accessed on 15 July 2023) [27], while PET data were sourced from the A Big Earth Data Platform for Three Poles (http://poles.tpdc.ac.cn/en/data/8b11da09-1a40-4014-bd3d-2b86e6dccad4/, accessed on 30 October 2023). The study period was uniformly selected as the 2011–2020 period, with annual temporal resolution, and resampled to a 500 m spatial resolution using nearest-neighbor interpolation.

Soil data included soil moisture, soil temperature, clay percentage, and sand percentage. Soil moisture data spanning from 2000 to 2020 were obtained from the China 1 km daily dataset based on site observations, which are available at the National TP Data Center (https://data.tpdc.ac.cn/zh-hans/data/49b22de9-5d85-44f2-a7d5-a1ccd17086d2, accessed on 13 July 2023) [28]. Soil temperature data were derived from the ERA5-Land monthly averaged dataset at a 0.1° spatial resolution provided by the Copernicus Climate Data Store (https://cds-beta.climate.copernicus.eu/datasets/reanalysis-era5-land-monthly-means?tab=overview, accessed on 12 September 2024). Clay and sand percentage data were sourced from the National Earth System Science Data Center, National Science & Technology Infrastructure of China (http://www.geodata.cn, accessed on 12 July 2023). The study period for these datasets was uniformly selected as the 2011–2020 period, with annual temporal resolution, and resampled to a 500 m spatial resolution using nearest-neighbor interpolation.

3. Methods

Figure 2 illustrates the flowchart used in this research. To investigate the spatiotemporal variation characteristics of NPP and SP in the TP from 2011 to 2020, the Mann–Kendall test was employed to assess the significance of the decade-long trend, while the Theil–Sen non-parametric regression method was used to calculate the trend. Partial correlation analysis was then conducted to examine the relationship between NPP and SP, determining the correlation coefficient and elucidating the influence mechanism of SP on NPP. Finally, PLS-SEM was employed to explore the regulatory effects and interactions of soil, topography, and climate factors on the changes in NPP and SP.

3.1. Trend Analysis

Theil–Sen non-parametric regression was used to compute the decadal trends in the NPP and SP in the study area. The Theil–Sen method is a robust non-parametric technique for trend analysis, particularly suited for long-term series data. Unlike traditional least-squares linear regression, it is more resilient to measurement errors and efficiently handles clustered data. The Sen slope was calculated as follows:

$$\beta =median\left({\displaystyle \frac{x_i-x_j}{i-j}}\right), 1<j<i<n$$

(1)

where β represents the trend slope; n denotes the length of the time series, with x_i and x_j representing the ith and jth data values in the series, respectively.

After employing Theil–Sen non-parametric regression for trend analysis of SP and NPP, the significance of these trends was further evaluated using the Mann–Kendall test:

$$S=\sum _{i=1}^{n-1}\sum _{j=i+1}^{n}sgn\left(x_j-x_i\right), sgn\left(\theta \right)=\left\{\begin{array}{c}1,\theta >0\\ 0,\theta =0\\ -1,\theta <0\end{array}\right.$$

(2)

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

(3)

where S is the statistic; Var(S) is the variance of S; t_i represents the number of data points in the ith group; and Zc is the standardized statistic. The null hypothesis is tested (H0: β = 0), and when |Zc| > Z1 − α/2, the null hypothesis is rejected. Among them, Z1 − α/2 denotes the standard normal variance, with α representing the significance test level.

3.2. Partial Correlation Analysis

To assess the correlation between a specific SP and NPP while controlling for the influence of other SPs, the simple correlation coefficient and first-order partial correlation coefficient were calculated as follows:

$$r={\displaystyle \frac{{\sum }_{i=1}^n\left(x_i-\stackrel{`}{x}\right)\left(y_i-\stackrel{`}{y}\right)}{\sqrt{{\sum }_{i=1}^n(x_i-\stackrel{`}{x}{)}^2{\sum }_{i=1}^n(y_i-\stackrel{`}{y}{)}^2}}}$$

(4)

$$R_{y{x}_1\cdot x_2}={\displaystyle \frac{R_{y{x}_1}-R_{y{x}_2}\cdot R_{x_1{x}_2}}{\sqrt{1-R_{y{x}_2}^2\cdot 1-R_{x_1{x}_2}^2}}}$$

(5)

where $r$ denotes the simple correlation coefficient between a certain SP; $\stackrel{`}{x}$ and $\stackrel{`}{y}$ represent the mean of the variables; and $R_{y{x}_1\cdot x_2}$ signifies the first-order partial correlation coefficient between the NPP and a certain SP.

Then, the second-order partial correlation coefficient between the NPP and a specific SP for each pixel was calculated as follows:

$$R_{y{x}_1\cdot x_2\cdot x_3}={\displaystyle \frac{R_{y{x}_1\cdot x_3}-R_{y{x}_2\cdot x_3}\cdot R_{y{x}_1\cdot x_2}}{\sqrt{1-R_{y{x}_2\cdot x_3}^2\cdot 1-R_{y{x}_1\cdot x_2}^2}}}$$

(6)

where $R_{y{x}_1\cdot x_2\cdot x_3}$ is the second-order partial correlation coefficient between the NPP and a certain SP.

3.3. Partial Least Squares Structural Equation Modeling (PLS-SEM)

PLS-SEM was used to investigate the impact of various environmental factors (such as temperature, soil moisture, soil temperature, PET, elevation, precipitation, and radiation) on SP and NPP in the TP. These environmental factors exert both direct and indirect effects on the NPP through SP, while soil factors (such as clay and sand percentages) have only direct effects on the NPP. By establishing and validating a theoretical model of multivariate relationships, the direct and indirect impacts of environmental factors on SP and their effects on the NPP were assessed.

A Structural Equation Model (SEM) is a multivariate technique that can simultaneously model the interaction between multiple factors and reveal causal relationships between independent and dependent variables. PLS-SEM, a key branch of SEM, is well suited for exploring new structural models as it does not require normally distributed data and has significant applications in remote sensing [8]. Therefore, this study employed PLS-SEM, and the results were analyzed using SmartPLS 4 software.

The standardized path coefficient, which represents the direct effect of one factor on another, was the primary outcome measure. To determine the proportion of indirect influence, the absolute value of indirect influence was divided by the total absolute values of all indirect influences. This proportion, termed the mediating effect, reflects the extent to which environmental factors indirectly affect the NPP through SP.

The significance of this coefficient was assessed using a resampling procedure with a significance threshold of p < 0.05. Model fit was evaluated using the Standardized Root Mean Square Residual (SRMR) and the coefficient of determination (R²). The SRMR measures the difference between observed and predicted data, with lower values indicating a better model fit. An SRMR below 0.08 denotes a well-fitting model, and all models in this study met this criterion. The SRMR is calculated using the following formula:

$$SRMR=\sqrt{{\displaystyle \frac{{\sum }_{i,j}{\left({\widehat{R}}_{ij}-R_{ij}\right)}^2}{n\cdot p}}}$$

(7)

where ${\widehat{R}}_{ij}$ represents the model-implied correlation between variables i and j, $R_{ij}$ is the observed correlation, $n$ is the number of observations, and $p$ is the number of variables. A lower $SRMR$ value indicates a better fit of the model to the data.

R² quantifies the proportion of variance in the dependent variable explained by the independent variables, reflecting the model’s explanatory power. It is calculated as follows:

$$R^2=1-{\displaystyle \frac{S{S}_{res}}{S{S}_{tot}}}$$

(8)

where $S{S}_{res}$ is the sum of squared residuals (the difference between observed and predicted values), and $S{S}_{tot}$ is the total sum of squares (the variance of the observed data). Higher R² values indicate that the model has greater explanatory power.

The indirect effect between factors was calculated by multiplying the direct path coefficients, while the overall effect encompassed both direct and indirect influences.

4. Results

4.1. Spatial and Temporal Variation Analysis of NPP and SP

4.1.1. The Spatial Patterns of the NPP and SP in the TP

Figure 3 shows the distribution of the ten-year mean NPP and SP across the TP. Overall, the NPP decreased from east to west, with the highest values (>240 gC/m²·a) concentrated along the eastern edge of the plateau, and the lowest values (<210 gC/m²·a) observed in the central and southwestern areas (Figure 3a). Early SODs were prevalent in the central and northeastern regions of the plateau, while later SODs were found in the western and low-elevation areas of the eastern plateau (Figure 3b). The distribution of SEDs was closely related to elevation, with snow cover generally ending later at higher elevations, except on the eastern plateau (Figure 3c). Regions where the SED occurred before the 180th day of the year accounted for 66.61% of the total area, indicating that most snow cover on the TP had melted by the end of June. The SCD also showed a strong correlation with elevation, increasing with higher elevations (Figure 3d). Areas with snow cover lasting fewer than 140 days accounted for 91.18% of the total area, while regions with snow cover lasting over 170 days or persisting throughout the year represented only 3.53% of the total area.

4.1.2. Temporal Trends in the NPP and SP in the TP from 2011 to 2020

Throughout the study period, approximately 16.45% of the TP area exhibited significant temporal trends (p < 0.1) in the NPP. Significant changes were particularly evident in the northeastern and central regions of the study area (Figure 4a). The northern region exhibited an overall upward trend in the NPP, while the southern region experienced a downward trend. Approximately 10.72% of the investigated area showed a notable reduction in the NPP. Regarding snow cover dynamics, 44.61% of the TP displayed significant trends in the SOD, with the most pronounced changes occurring in the northwest and arid eastern regions of the plateau (Figure 4b). On average, the SOD advanced by 1.50 days. Regions with significant SOD advancements accounted for 39.62% of the TP, predominantly located in the northwest and arid eastern parts of the TP, whereas the central region primarily exhibited delayed trends in the SOD. For the SED, significant trends were observed in 16.78% of the total area, with the SED advancing at an average rate of 0.14 days per year. Only 5.59% of the study area exhibited significant trends in the SCD (Figure 4d). Among these, 55.47% experienced a notable reduction in the SCD. Notably, in the northeastern, northwestern, southwestern, and central-western margins of the plateau, the SCD decreased at a rate exceeding 4 days per year, whereas in the central part of the plateau, the SCD mainly showed an increasing trend.

4.2. The Correlation between the NPP and SP in the TP from 2011 to 2020

Figure 5 shows the partial correlation between the NPP and SP in the TP from 2011 to 2020. Statistically, regions where the NPP showed a significant negative correlation with the SOD encompassed 35.60% of the study area based on correlations that passed the significance test at p < 0.05. The NPP in the TP generally displayed a negative association with the SED. Among the areas where correlations passed the 0.05 significance test, 55.98% of the total area exhibited a significant negative relationship between the NPP and SED. Similarly, the NPP in the TP was negatively correlated with the SCD. In regions where this correlation was significant at 0.05, 55.86% of the region exhibited a marked negative association between the NPP and SCD.

4.3. Meteorological Factors That Have an Indirect Impact on the NPP through SP

Figure 6 presents the results of the SEM analysis, including standardized regression path coefficients and the proportions of indirect effects. The model demonstrated a good fit, with R² values of 0.774, 0.718, and 0.718 and SRMR values of 0.008, 0.003, and 0.001 for the SCD, SED, and SOD, respectively. All path coefficients were statistically significant at the 0.05 level, indicating that the model adequately fits the data and aligns with the expectations, thereby providing strong explanatory power [3].

First, the meteorological factors influenced SP as follows: precipitation and temperature had a negative impact on the SOD but a positive effect on the SED and SCD. Conversely, radiation and PET positively influenced the SOD but negatively affected the SED and SCD. Soil moisture negatively impacted all SP parameters, while elevation and soil temperature positively affected each SP parameter.

Second, precipitation emerged as the primary factor influencing the NPP, with a total effect of 0.309 in the SCD structural equation. The soil temperature also significantly affected the NPP, with a total effect of 0.206 in the SED and SOD structural equation. Temperature and precipitation significantly impacted the SCD, with total effects of 0.183 and 0.275, respectively. Meteorological conditions further influenced the NPP through their effects on SP parameters. Temperature had a relatively strong indirect effect on the NPP through its influence on the SOD and SCD, contributing 16% and 10% to the total effect, respectively. Other factors also exhibited notable indirect effects on the NPP through snow cover. For example, while radiation had a low total impact on the NPP, its proportion of indirect effect was higher than that of temperature in the SED and SCD structural equation.

5. Discussion

5.1. The Mediating Effect of the SP on the TP’s NPP

Mediation effects are commonly used to analyze the complex processes and mechanisms through which independent variables influence dependent variables [29]. In this context, snow cover is considered a mediating variable for quantifying its mediating effect and exploring how SP mediates the impact of various environmental factors on the NPP. This research identifies SP as a key mediating factor influencing the effects of different environmental factors. Snow cover impacts photoperiod-sensitive plants, leading to significant differences in vegetation growth before and after the SOD [9]. The onset of snow cover creates an insulating layer that alters the soil surface temperature and thereby influences vegetation growth [30]. While the SCD also affects vegetation by influencing the soil surface temperature, its primary role is to protect the buds of overwintering plants. Insufficient snow cover during winter can cause plant damage, adversely affecting vegetation growth in the following growing season [31]. Compared to its insulating effects on temperature and light, the SED primarily influences vegetation growth by affecting runoff and the length of the growing season. Snowmelt increases surface runoff and the duration of time remaining after snowmelt affects the duration of the vegetative period [7]. In the TP, the mediating effects of the SOD and SCD on the influence of temperature on the NPP are relatively pronounced.

Rising temperatures can enhance vegetation growth, especially in spring and summer, by increasing the activity of plant photosynthetic enzymes and slowing chlorophyll degradation, thereby boosting vegetation productivity [32]. However, if the temperature rises too quickly or too high, it may negatively impact vegetation [17,33]. Generally, increasing spring temperatures in the TP is believed to advance the greening period of alpine vegetation [30]. Rapid temperature increases lead to a delayed SOD and reduced SCD, which, in turn, decrease the soil water content in the subsequent growing season and causes soil temperature loss, affecting vegetation growth and productivity [33]. This study shows that the efficiency of the indirect effect of temperature on the NPP through this process is 10% of the direct effect of temperature on the NPP, suggesting that neglecting the mediating effect of snow cover can introduce a degree of error in studies of temperature’s impact on the NPP in the TP.

Simultaneously, radiation has a significant mediating effect on the relationship between radiation and the NPP. This may be due to the abundant radiation in the TP region, which can cause vegetation to reach light saturation easily, thus reducing the direct effects of radiation on the NPP while maintaining a high indirect effect through its influence on SP parameters. The direct and indirect influence coefficients of radiation on the NPP are relatively small, indicating that radiation does not play a primary role in influencing the NPP. This finding aligns with the conclusions of existing studies [31]. Given the weak impact of radiation on the NPP, the significance of the mediating effect of SP in this process should be interpreted with caution.

Extensive research on the TP has examined how vegetation interacts with key climatic factors such as temperature, precipitation, SP, and terrain elevation, with some studies attributing variations in vegetation growth to these topographic and climatic influences [17,30,31,34]. A previous study found that rising surface temperatures lead to early snowmelt, which warms the vegetation growth environment in alpine areas, promoting vegetation growth under favorable conditions [35]. However, the role of snow cover in this process remains unclear. There is a lack of evaluation regarding the mediating effect of SP on the processes through which environmental factors indirectly affect vegetation. This study addresses this gap by focusing on the mediating effects of SP on the NPP.

5.2. The Mediating Effect of Snow Cover on the Temperature–NPP Relationship Variations with Vegetation Types

This study observed that the mediating effect of the SOD on the relationship between temperature and NPP varies with changes in vegetation type (Figure 7). Generally, as the community biomass increases, the mediating effect of temperature on the NPP through snow cover decreases, indicating that the influence of the SOD also varies across different vegetation types. Previous studies have shown that the effects of warming on plant species are non-linear and depend on the biome [36,37], and the current findings align with these studies. The mediating effect of the SOD on the NPP was greater in steppes compared to shrubs or woodland areas. This suggests that global climate change influences vegetation through alterations in SP, which is a particularly pronounced effect in alpine meadows. These findings align with the conclusions drawn by Theurillat [38].

Snow cover can affect soil temperature during winter, and variations in its state may impact nutrient cycling and soil carbon sequestration processes [37]. A previous study found that initial snow cover in alpine regions has a stronger correlation with the growing season of steppes and meadows compared to shrubs and woodlands [39]. The present study observed that in alpine steppes and meadows, snow cover significantly mediates the effect of temperature on the NPP. Winter snow benefits photosynthesis by keeping the temperature beneath the snow above the critical point for photosynthesis, thereby enhancing its insulating effect. This temperature maintenance leads to asynchronous phenology between aboveground and underground components, profoundly affecting vegetation growth [40]. For instance, studies have shown that low-altitude plants in the Arctic, such as moss tundra and shrubs, frequently experience snow cover. In these vegetation types, the insulating properties of snow cover have a greater mediating effect on the temperature–NPP relationship [41]. In contrast, in high plant communities such as coniferous forests and woodlands, the canopy snowmelt process exerts a weaker effect on the soil temperature, and thus, changes in soil temperature have a reduced impact on the NPP in the short term [42]. This observation explains why the mediating effect of snow cover is smaller in these areas compared to low vegetation areas.

This research offers a novel insight into the impact of alpine vegetation on climate within the context of global change. Structural differences between various types of wasteland and grassland in the subalpine and alpine zones likely lead to differing ecosystem responses [17]. The finding that the mediating effect of snow cover varies with the vegetation type highlights the complex response mechanisms in the TP, revealing how vegetation phenology in alpine regions is influenced by changes in snow cover.

The dynamic changes in SP and the NPP are affected by many factors. Besides those examined in this study, other variables also play crucial roles in shaping SP. Light-absorbing aerosols, such as black carbon and dust, can settle on snow surfaces, reducing the snowpack’s albedo, which increases solar radiation absorption and accelerates snowmelt [43]. Although this factor is potentially significant for SP, it was not included in the SEM analysis because of a lack of data on relevant spatiotemporal scales.

5.3. Limitations and Prospects

Additionally, some studies have indicated that snow cover has a lagged effect on the NPP, and neglecting this lag could lead to an underestimation of snow cover influences the NPP [44]. This study did not account for the delayed effects of snow cover on the NPP, which could introduce some inaccuracies. Future research should incorporate these delayed effects to more accurately explore how snow cover mediates the influence of various meteorological factors on the NPP.

Vegetation greenness can alter the redistribution of snow cover by affecting the balance of solar radiation [45]. In the TP, vegetation directly impacts snow cover [3], indicating a bidirectional relationship where vegetation affects SP and vice versa. Furthermore, studies have shown that snow cover can affect plant growth by modifying soil moisture conditions. In the TP, from the humid to arid zone, snow cover increasingly plays a crucial role in regulating vegetation through changes in soil moisture [19]. These observations suggest a complex interplay among meteorological factors, snow cover, and the NPP. Our study focused on how meteorological factors affect SP and how SP subsequently influences vegetation without considering the mutual interactions among these factors. Future research should delve deeper into the intricate reciprocal relationships between meteorological factors, SP, and the NPP in the TP.

6. Conclusions

This study analyzed multi-platform geographical data from the TP spanning from 2011 to 2020. Notably, approximately 16.45% of the regional NPP exhibited significant changes. The NPP was positively correlated with the SOD, but 35.60% of the region displayed a significant inverse relationship between the NPP and SOD (p < 0.05). A significant negative relationship between the NPP and SED (p < 0.05) was found in 55.98% of the study area.

Meteorological factors can affect the NPP both directly and indirectly through their impact on SP. Different vegetation types experience varying impacts from snow cover, resulting in diverse mediating effects of snow cover on the relationship between meteorological factors and the NPP. The mediating effect of SP in the temperature–NPP relationship is significantly influenced by vegetation type, with temperature’s indirect effect on the NPP through the SOD being particularly pronounced. This phenomenon may be attributed to the varying insulating effects of snow cover on ground temperature across different vegetation types.

This study enhances the understanding of the complex interactions among climate factors, snow cover, and the NPP in the TP region. It provides valuable insights for analyzing changes in the NPP in the context of climate change in the TP region.