Spatio-Temporal Changes and Driving Mechanisms of Vegetation Net Primary Productivity in Xinjiang, China from 2001 to 2022

Abstract

Net primary productivity (NPP), a key indicator of terrestrial ecosystem quality and function, represents the amount of organic matter produced by vegetation per unit area and time. This study utilizes the MOD17A3 NPP dataset (2001–2022) to analyze the spatio-temporal dynamics of NPP in Xinjiang and projects future trends using Theil-Sen trend analysis, the Mann–-Kendall test, and the Hurst Index. By integrating meteorological data, this study employs partial correlation analysis, the Miami model, and residual analysis to explore the driving mechanisms behind NPP changes influenced by climatic factors and human activities. The results indicate that: (1) The average NPP in Xinjiang has increased over the years, displaying a spatial pattern with higher values in the north and west. Regions with increasing NPP outnumber those with declining trends, while 75.18% of the area shows un-certain future trends. (2) Precipitation exhibits a stronger positive correlation with NPP compared to temperature. (3) Climate change accounts for 28.34% of the variation in NPP, while human activities account for 71.66%, making the latter the dominant driving factor. This study aids in monitoring ecological degradation risks in arid regions of China and provides a scientific basis for developing rational coping strategies and ecological restoration initiatives.

1. Introduction

As a crucial component of the global ecosystem, vegetation plays a key role in regulating the global carbon balance and maintaining climate stability [1,2,3], which is a major indicator of regional ecological conditions [4,5]. In the context of global climate change and intensified human activities, vegetation dynamics tend to change significantly, directly or indirectly affecting local climate stability and regional ecological balance [6,7,8,9]. Net primary productivity (NPP) of vegetation refers to the amount of carbon fixed by photosynthesis in green plants per unit of time and area, after accounting for respiratory consumption. NPP visually reflects the natural growth of vegetation [10,11,12]. The dynamics of vegetation are primarily driven by climatic factors and human activities, indicating the extent of vegetation recovery and degradation, which play an important role in global change and carbon balance [13,14].

Many scholars have conducted extensive research on NPP. Since the mid-20th century, the concept of NPP has been defined and studied in various countries worldwide. Multiple models have been developed to measure, assess, and analyze NPP across different regions [15,16]. Subsequently, the Biosphere Program, Global Change and Terrestrial Ecosystems, and the Kyoto Protocol have all identified the study of vegetation NPP as a core component [17]. In recent years, numerous scholars have explored the spatial and temporal patterns of NPP in arid zones. Furthermore, they have examined evolutionary trends and driving factors from various perspectives, resulting in significant research findings [18,19]. Currently, China faces severe environmental issues such as desertification, particularly in arid and semi-arid regions, which are ecologically fragile and exhibit significant ecological degradation, leading to considerable socio-economic losses [20]. Xinjiang, a typical arid region in northwest China, is characterized by widespread deserts and low vegetation cover. To effectively monitor the health of regional ecosystems and enhance sustainable development, NPP has become an extremely essential indicator.

Recent research on NPP in Xinjiang has primarily focused on the following three aspects: (1) Development history of NPP monitoring. Initially, the determination of vegetation NPP was based on field observations on sample plots [21]. In recent decades, large-scale NPP estimation models have been developed both domestically and internationally, leveraging remote sensing technology. These models can be categorized into two main types: One approach involves empirical models based on in situ measurements, such as the Miami model [22], the Thornthwaite model [23], the Chikugo model [24], the Zhu Zhihui model [25], and the Zhou Guangsheng model [26]. Additionally, process models, such as CASA [27,28,29] and the Biome-BGC model [30], are typically calibrated using in situ observations. (2) Spatio-temporal patterns and evolutionary trends. Currently, studies on the NPP distribution patterns of vegetation in Xinjiang encompass scales ranging from the entire country to the Northwest Arid Region and small topographic units. Liu et al. analyzed the spatio-temporal dynamics, stability, and persistence of grassland NPP in China from 2000 to 2015, discovering that areas with significant grassland NPP reduction were primarily located in northern Xinjiang [31]. Li and Pan used linear trend analysis, standard deviation, and the Hurst index to examine the spatio-temporal characteristics of NPP in the Northwest Arid Region [32]. Liu et al. investigated the spatio-temporal differentiation characteristics of vegetation NPP in the Yili River Basin, employing statistical methods and focusing on seasonal changes [33]. (3) Research on NPP driven by climate change and human activities. Fang et al. examined the effects of precipitation, temperature, and CO₂ concentration on NPP changes in Xinjiang by using the AEM model to divide different scenarios and quantify their relative contributions [34]. Yang et al. analyzed the impacts of land use and land cover changes on vegetation NPP in Xinjiang, finding that forest expansion was the main driver [35]. Huang et al. used the Biome-BGC grazing model to conclude that grazing activities had a strong negative effect on vegetation NPP in Xinjiang grassland [36]. However, changes in vegetation NPP are influenced by both climate change and human activities. To further synthesize and analyze the driving mechanisms of vegetation NPP changes in Xinjiang, Zhang et al. used the Miami model to study the dynamics of grasslands in Xinjiang from 2000 to 2014. The results showed that human activities dominated the increase in NPP, while the decrease in NPP was co-driven by climate change and human activities, with their contributions being almost equal [37]. Jiang et al. applied an innovative method based on partial derivatives and residuals to calculate the relative contributions of climate factors and human activities, finding that increased human activities had a more pronounced effect on NPP [38].

In summary, existing studies exploring vegetation NPP in Xinjiang tend to focus on large-scale geographic regions such as China and the Northwest Arid Zone, or on smaller-scale topographic regions such as the Tianshan Mountains and the Yili River Valley. However, there is a notable lack of research on vegetation NPP at the meso-scale in Xinjiang. Additionally, although research on the impact mechanism of a single climate factor on vegetation productivity is relatively mature, there is relatively little research on the comprehensive impact of multiple factors. Anthropogenic influences are often limited to land use, with a relative lack of research on the quantitative ranking of the contributions of different driving factors.

Therefore, this study focuses on Xinjiang as the research area to analyze the spatial and temporal changes, as well as future trends, of vegetation NPP from 2001 to 2022. The analysis employs Remote Sensing (RS) and Geographic Information System (GIS) technologies, utilizing the NPP dataset MOD17A3HGF along with 1 km annual temperature and precipitation datasets of China on a monthly basis. Additionally, the study delves into the driving mechanisms behind vegetation NPP changes in Xinjiang, considering the impacts of climatic factors and human activities. It quantitatively assesses the relative contributions of these factors to NPP changes. The ultimate goal is to promote ecological conservation and rational resource utilization in Xinjiang, providing a scientific foundation for long-term monitoring of ecological degradation risks in arid and semi-arid regions of China.

2. Materials and Methodology

2.1. Study Area

Xinjiang (73°40′–96°18′ E, 34°25′–48°10′ N) is located in northwestern China, covering an area of 1.66 × 10⁶ km² (Figure 1). The region features an elevation range from −192 to 8545 m and is characterized by “three mountains and two basins”: the Altai Mountains in the north, the Kunlun Mountains in the south, and the Tian Shan Mountains in the middle, with the vast Junggar Basin and Tarim Basin situated between these mountain ranges. Xinjiang’s geography results in a typical temperate continental arid climate. It is far from the sea, deeply inland, and surrounded by mountains. The average annual temperature ranges from 9 to 12 °C, with significant temperature differences, ample light and heat resources [39]. Moreover, annual precipitation is scarce and un-evenly distributed, ranging from 150 to 200 mm in northern Xinjiang to less than 100 mm in southern Xinjiang.

The region’s special geographic location, topographic conditions, and arid climate result in vegetation dominated by desert types. This vegetation has a single structure and low coverage, making the ecological environment sensitive and vulnerable. Due to the harsh natural conditions, cities in Xinjiang are mainly concentrated in oasis areas, with a scattered overall distribution. These natural conditions also limit the region’s socio-economic development.

2.2. Data

The vegetation NPP data used in this study were obtained from the global NPP dataset MOD17A3HGF (https://ladsweb.modaps.eosdis.nasa.gov, accessed on 26 October 2023) provided by NASA’s MODIS from 2001 to 2022. This dataset has a spatial resolution of 500 m and is in TIF format. Before data analysis, pre-processing steps such as mask processing, coefficient conversion, and removal of invalid values were conducted.

Temperature and precipitation data were obtained from the National Earth System Science Data Center (http://www.geodata.cn, accessed on 10 November 2023). The study used the 1-km resolution monthly average near-surface temperature and precipitation dataset for the Chinese region from 2001 to 2022 [40]. Annual average temperature and precipitation data were derived from these monthly datasets. To ensure that the correlation analysis between vegetation NPP and meteorological factors was conducted at the same resolution, the vegetation NPP data were resampled to match the spatial resolution of the meteorological data.

2.3. Methodology

To better study the spatial and temporal evolution characteristics of vegetation NPP and its driving mechanisms in Xinjiang, we implemented three key steps. Firstly, this study pre-processed the NPP remote sensing dataset, temperature data, precipitation data, DEM data, and ecosystem data. Secondly, we used Theil-Sen trend analysis, the Mann–Kendall significance test, and the Hurst index to analyze the spatial and temporal characteristics of NPP changes in Xinjiang and predict future trends. Finally, this study employed correlation analysis and residual analysis to investigate the driving mechanisms of vegetation NPP changes in Xinjiang under the influence of climatic factors and human activities (Figure 2).

2.3.1. Trend Analysis and Significance Tests

In this study, the Theil–Sen Median trend analysis method was used to calculate the year-to-year trend change of vegetation NPP in Xinjiang, and the significance test was performed by the Mann–Kendall statistical test [41,42]. The Sen trend was calculated by the formula:

$$K=\mathrm{median}{\displaystyle \frac{{NPP}_j-{NPP}_i}{j-i}}$$

(1)

In the above equation, i and j are time series, K > 0 indicates an upward trend in the time series, and K < 0 indicates a downward trend in the time series.

The Mann–Kendall test can reveal the trend of characteristic quantities within a specific time series and detect sudden changes in the characteristics of the time series. Additionally, it does not require the data to be normally distributed, nor does it require the trend to be linear. The test is calculated as follows:

Where n is the length of the time series and f is the function sign, which is defined as follows:

$$f({NPP}_j-{NPP}_i)=\left\{\begin{array}{c}+1,{\rm N}P{P}_j-NP{P}_i>0\\ \begin{array}{c}\\ \end{array}0,NP{P}_j-NP{P}_i=0\\ -1,NP{P}_j-NP{P}_i<0\end{array}\right.$$

(2)

$$S={\displaystyle \sum _{i=1}^{n-1}{\displaystyle \sum _{j=i+1}^{n}f(NP{P}_j-NP{P}_i)}}$$

(3)

Standardize the statistic S according to the following formula:

$$Z=\left\{\begin{array}{c}{\displaystyle \frac{S}{\sqrt{Var(S)}}}(S>0)\\ \begin{array}{c}\\ \end{array}0(S=0)\\ {\displaystyle \frac{S}{\sqrt{Var(S)}}}(S<0)\end{array}\right.$$

(4)

In the above equation, Var(S) represents the variance of S. For a given level of significance (p < 0.05 or p < 0.01), |Z| > 1.96 and |Z| > 2.58 indicate that the trend passes the significance test at confidence levels of 0.05 and 0.01, respectively.

2.3.2. Analysis of Future Trends

The Hurst index is one of the most effective methods for quantitatively describing long-range dependence within a time series and has been successfully applied to vegetation studies to determine the persistence of vegetation trends [43,44].

For the time series {V(t)} (t = 1, 2, ...), the mean sequence is defined as follows:

$$V_p={\displaystyle \frac{1}{p}}{\displaystyle \sum _{t=1}^p{V}_{(t)}}$$

(5)

Define cumulative deviation:

$$V_{(t,p)}={\displaystyle {\sum }_{t=1}^p(V_{(t)}-V_{(p)})(1\le t\le p)}$$

(6)

Calculate the extreme difference:

$$R_{(p)}=\mathrm{Max}(V_{(t,p)})-(\mathrm{Min})(V_{(t,p)})(1\le t\le p)$$

(7)

Calculate the standard deviation:

$$S_{(p)}=\sqrt{{\displaystyle \frac{1}{p}}{\displaystyle \sum _{t=1}^p{{\displaystyle (V_t-V_p)}}^2}}$$

(8)

The Hurst exponent H can be defined by R, S, and p, as follows:

$$R\left(p\right)/S\left(p\right)=c·p^H$$

(9)

where c is a constant. H value is obtained through the least squares fitting method.

Whether the time-series NPP is completely random or persistent can be determined according to the value of H, as follows:

(1)

0 < H < 0.25, indicating the time-series data has a strong anti-continuous (SAC) characteristic, and the future trend will be opposite to the previous trend. Meanwhile, the closer the value of H is to 0, the stronger the difference will be;

(2)

0.25 < H < 0.5, indicating the time-series data has a weak anti-continuous (WAC) characteristic, and the future trend will be opposite to the previous trend;

(3)

H = 0.5, indicating that the time-series data is a random sequence, and there is no long-term correlation between the future trend and the previous trend;

(4)

0.5 < H ≤ 0.75, indicating the time-series data has a weak continuous (WC) characteristic, and the future trend will be consistent with the previous trend;

(5)

0.75 < H ≤ 1, indicating the time-series data has a strong continuous (SC) characteristic, and the future trend will be consistent with the previous trend. Meanwhile, the closer the value of H is to 1, the stronger the consistency will be.

2.3.3. Partial Correlation Analysis

The second-order partial correlation coefficient was used to analyze the relationship between vegetation NPP, precipitation, and temperature. The significance of the correlation was tested using the t-test. Vegetation NPP was considered significantly correlated with climatic conditions when the significance level was p < 0.05; otherwise, the correlation was considered non-significant.

$${{\displaystyle r}}_{xy,z}={\displaystyle \frac{{{\displaystyle r}}_{xy}-{{\displaystyle r}}_{yz}}{\sqrt{(1-{{\displaystyle {{\displaystyle r}}_{xy}}}^2)(1-{{\displaystyle {{\displaystyle r}}_{yz}}}^2)}}}$$

(10)

In the above equation, $r_{xy,z}$ is the partial coefficient of x to y, z = [z₁, z₂,…, z_n] is the control variable, $r_{xy}$, $r_{xz}$ and $r_{yz}$ are correlation coefficient between two factors.

The partial correlation coefficients between vegetation NPP and precipitation and air temperature were tested for t-significance, respectively, and the t-test was calculated as:

$$T={\displaystyle \frac{{{\displaystyle r}}_{xy,z}}{\sqrt{1-r_{xy,z}}}}\sqrt{n-m-1}$$

(11)

The significant cases can be categorized as “extremely significant negative correlation (ESNC)”, “significant negative correlation (SNC)”, “non-significant negative correlation (NSNC)”, “non-significant positive correlation (NSPC)”, “significant positive correlation (SPC)”, and “extremely significant positive correlation (ESPC)” (

Table 1. Significance Level of Correlation and Corresponding Parameters.

Significance Level	r-Value	T-Value
Extremely Significant Negative Correlation (ESNC)	r < 0	T < 0.01
Significant Negative Correlation (SNC)		0.01 ≤ T ≤ 0.05
Non-Significant Negative Correlation (NSNC)		T ≥ 0.05
Non-Significant Positive Correlation (NSPC)	r > 0	T ≥ 0.05
Significant Positive Correlation (SPC)		0.01 ≤ T ≤ 0.05
Extremely Significant Positive Correlation (ESPC)		T < 0.01

).

2.3.4. Estimation of Potential NPP

The Miami model is based on reliable measured data on net vegetation productivity and corresponding mean annual temperature and mean annual precipitation data from five continents. The net productivity model is constructed using the least squares method for simulation [45]. This study clarifies, from a physiological and ecological perspective, that the main influencing factors of plant productivity and biomass formation are temperature and moisture. By calculating the annual rainfall and average annual temperature in Xinjiang, the climatic production potential of plants is determined, allowing for a quantitative analysis of the impact of climate change on vegetation [23,46,47]. The calculation formula is as follows:

$$V_{NPP-P}=\min (V_{NPP-\theta },V_{NPP-R})$$

(12)

$$V_{NPP-\theta }=3000/(1+{\mathrm{e}}^{1.1315-0.119\theta })$$

(13)

$$V_{NPP-R}=3000(1-{\mathrm{e}}^{-0.000664R})$$

(14)

In the above equation, V_NPP-θ indicates the potential net primary productivity of vegetation (g C·m⁻²∙a⁻¹), and V_NPP-R is the potential vegetation net primary productivity (g C·m⁻²∙a⁻¹). Where, R and θ are the annual precipitation (mm) and the mean annual temperature, respectively. According to Liebig’s Law of Least Factor, the minimum of the two was chosen as the potential vegetation primary productivity V_NPP-P (g C·m⁻²∙a⁻¹).

2.3.5. Residual Analysis and Contribution Margin Calculation

Factor separation and quantification in this study were based on residual analysis of three NPP values [48,49]: (1) actual NPP derived from MOD17A3HGF data (V_NPP-A); (2) potential NPP estimated using the Miami model (V_NPP-P), which represents the vegetation NPP under the influence of climate change alone; and (3) anthropogenic NPP change (V_NPP-H), calculated as the difference between actual NPP and potential NPP (V_NPP-H = V_NPP-A $-$ V_NPP-P). The slope values of the NPP change indicators for the three vegetation types were then calculated separately using Theil–Sen median trend analysis.

The three slope values were then combined to categorize the effects of climate change and human activities on vegetation NPP changes into six scenarios (

Table 2. Different Scenarios for Assessing the Relative Roles of Climate and Human Activities in Vegetation Degradation or Restoration.

Vegetation Situation	KP	KH	Driven by Climate Change	Driven by Human Activity	Clarification
Vegetation restoration KA > 0	>0	>0	${\displaystyle \frac{\left|\Delta V_{NPP-P}\right|}{\left|\Delta V_{NPP-P}\right|+\left|\Delta V_{NPP-H}\right|}}\times 100\%$	${\displaystyle \frac{\left|\Delta V_{NPP-H}\right|}{\left|\Delta V_{NPP-P}\right|+\left|\Delta V_{NPP-H}\right|}}\times 100\%$	Both of the two factors dominated restoration (combined-restoration)
	>0	<0	100	0	Climate-dominated restoration (climate-restoration)
	<0	>0	0	100	Human activities-dominated restoration (human-restoration)
Vegetation degradation KA < 0	<0	<0	${\displaystyle \frac{\left|\Delta V_{NPP-P}\right|}{\left|\Delta V_{NPP-P}\right|+\left|\Delta V_{NPP-H}\right|}}\times 100\%$	${\displaystyle \frac{\left|\Delta V_{NPP-H}\right|}{\left|\Delta V_{NPP-P}\right|+\left|\Delta V_{NPP-H}\right|}}\times 100\%$	Both of the two factors dominated degradation (combined-degradation)
	<0	>0	100	0	Climate-dominated degradation (climate-degradation)
	>0	<0	0	100	Human activities-dominated degradation (human-degradation)

).

3. Results

3.1. Spatio-Temporal Characteristics of Vegetation NPP in Xinjiang

3.1.1. Inter-Annual Variations of NPP

From 2001 to 2022, the mean vegetation NPP in Xinjiang reflected a generally increasing trend with fluctuations (Figure 3). The NPP increased from 157.90 g C·m⁻² in 2001 to 169.24 g C·m⁻² in 2022, an increase of 11.34 g C·m⁻² over the period, corresponding to a rate of change of 0.84 g C·m⁻² per year (p < 0.05). Interannual fluctuations were observed, with the highest mean NPP recorded in 2016 at 203.40 g C·m⁻² and the lowest in 2008 at 151.03 g C·m⁻², resulting in a fluctuation range of 52.37 g C·m⁻².

3.1.2. Spatial Distribution and Change Characteristics of NPP

The distribution of vegetation NPP in Xinjiang exhibited significant spatial heterogeneity, characterized by higher values in the northern and western regions compared to the southern and eastern regions (Figure 4). During the study period, the multi-year average vegetation NPP in Xinjiang was 156.33 g C·m⁻². According to Jiang et al. [19], NPP in Xinjiang can be divided into high-value (>200 g C·m⁻²) and low-value (<200 g C·m⁻²) areas. The high-value areas, which comprised 28.10% of the study region, were primarily located in the Ili Valley, Tianshan Mountains, Altay Mountains, and Tacheng area. In contrast, the low-value areas, covering 71.90% of the region, were pre-dominantly distributed in the Junggar Basin hinterland and the periphery of the Tarim Basin.

Over the past 22 years, the rate of change in vegetation NPP in Xinjiang ranged from −19.1 to 19.6 g C·m⁻²·a⁻¹ (Figure 5a). Spatially, regions where NPP increased (K > 0) comprised 81.62% of the total area, while those where NPP decreased (K < 0) comprised 18.38%. The Mann–Kendall (MK) significance test was employed to determine increasing and decreasing NPP trends (Figure 5b) and summarized in a statistical table (

Table 3. Statistics of Changes in NPP in Xinjiang from 2001 to 2022.

Type of Change	Basis of Judgment	Proportion (%)
Extremely Significant Reduction (ESR)	Sen < 0, |Z| > 2.58	0.30
Significant Reduction (SR)	Sen < 0, |Z| > 1.96	0.47
No Significant Change (NSC)	|Z| ≤ 1.96	75.99
Significant Increase (SI)	Sen > 0, |Z| > 1.96	10.61
Extremely Significant Increase (ESI)	Sen > 0, |Z| > 2.58	12.63

). The regions with “NSC” in NPP constituted the largest proportion (75.99%) of the study area, followed by regions with “ESI” (12.63%) and “SI” (10.61%). These areas were mainly distributed in the Irtysh River, the northern slope oasis zone of the Tianshan Mountains, the Tarim River basin, the Aksu River basin, the Yarkand River oasis zone, and the Kumkul Basin. The percentages of regions with “SR” and “ESR” in NPP were relatively small, at 0.47% and 0.30%, respectively. These areas are located sporadically in urban centers, such as Urumqi, Yining, Kashgar, and Hotan, likely due to urbanization [50].

3.1.3. Future Trends of NPP

The spatial distribution of future vegetation NPP trends in Xinjiang was analyzed using the Hurst index (Figure 6a). The Hurst index for vegetation NPP in Xinjiang ranged from 0.08 to 1, with an average value of 0.42, showing significant spatial heterogeneity. The regions with continuous changes in vegetation NPP (H > 0.5) accounted for 14.15% of the study area, while areas with anti-continuous changes (H < 0.5) accounted for 74.85%. Overall, future changes in Xinjiang’s vegetation NPP are mainly characterized by anti-continuous. Spatially, the high Hurst index values are primarily distributed on the southern slopes of the Tianshan Mountains, the western and southern parts of the Ili Valley, and sporadically in the Altay region. Conversely, the low Hurst index values are mainly located in the western mountains of the Junggar Basin, the central Tianshan Mountain, and the central Ili River Valley in northern Xinjiang. In these areas, future vegetation NPP trends are expected to be opposite to past trends, indicating inverse behavior.

The future change trend in Xinjiang (Figure 6b) was analyzed using a superposition of the Hurst index and the trend of vegetation NPP [51]. Furthermore, the future change trend of vegetation NPP in Xinjiang can be classified into five categories (

Table 4. Statistics of the Future Change Trend of NPP in Xinjiang.

Type of Change	Basis of Judgment	Proportion (%)
Reduced Strong Anti-Continuous (R·SAC)	Sen < 0; |Z| > 1.96; Hurst ≤ 0.35	0.15
Reduced Weak Anti-Continuous (R·WAC)	Sen < 0; |Z| > 1.96; 0.35 < Hurst < 0.5	0.51
Reduced Weak Continuous (R·WC)	Sen < 0; |Z| > 1.96; 0.5 < Hurst ≤0.65	0.21
Reduced Strong Continuous (R·SC)	Sen < 0; |Z| > 1.96; Hurst > 0.65	0.03
Not significant change (Uncertainly)	|Z| ≤ 1.96	75.18
Increase Strong Anti-Continuous (I·SAC)	Sen < 0; |Z| > 1.96; Hurst ≤ 0.35	4.42
Increased Weak Anti-Continuous (I·WAC)	Sen < 0; |Z| > 1.96; 0.35 < Hurst < 0.5	14.31
Increase Weak Continuous (I·WC)	Sen < 0; |Z| > 1.96; 0.5 < Hurst ≤ 0.65	4.81
Increased Strong Continuous (I·SC)	Sen < 0; |Z| > 1.96; Hurst > 0.65	0.37

): (1) “Strongly Positive Direction”: This includes two categories—” I·SAC” (0.37%) and “R·SAC” (0.15%), which are sporadically distributed throughout the region. (2) “Weakly Positive Direction”: This consists of “I·WC” (4.81%) and “R·WAC” (0.51%), distributed in the southwestern Tarim Basin, the southern slopes of the Tianshan Mountains, and Tacheng in the western Junggar Basin. (3) “Strongly Negative Direction”: Vegetation will show a significant degradation trend, including “R·SC” and “I·SAC”, with a total area of 4.45%. These areas are principally distributed in the Irtysh River Basin, the central and northern Tianshan Mountains Range, and the Kunlun Mountain Range. (4) “Weakly Negative Direction”: This includes “R·WC” and “I·WAC”, covering 14.52% of the region and distributed throughout Xinjiang. (5) “Uncertain Direction”: This category covers 75.18% of the area, indicating that the future direction of vegetation NPP is un-certain and highly susceptible to climate factors, human activities, and other influences.

3.2. Analysis of the Driving Mechanism of Vegetation NPP Change in Xinjiang

3.2.1. Correlation Analysis of NPP with Temperature and Precipitation

This study performed a pixel-by-pixel analysis to investigate the correlation between NPP in Xinjiang and the variations in annual average temperature and annual precipitation. From 2001 to 2022, the range of partial correlation coefficients between NPP and annual average temperature in Xinjiang was −0.78 to 0.85 (Figure 7a,b). The regions with a positive correlation covered 62.90% of the total area, with “ESPC” (p < 0.01) and “SPC” (p < 0.05) comprising 4.05%. These regions were mainly located in the Ili Valley, Tianshan Mountains, and Kunlun Mountains. The Ili Valley has relatively abundant rainfall, while the high altitude and low temperature in the Tianshan and Kunlun Mountains make vegetation growth more sensitive to temperature, leading to a higher correlation between NPP and temperature. However, the regions with a negative correlation accounted for 37.10% of the total area, with “ESNC” and “SNC” comprising 2.20%, pre-dominantly around Wusu and Shawan in the mid-southern Tacheng region and around the Tarim River.

The partial correlation coefficients between NPP and annual precipitation in Xinjiang range from −0.76 to 0.92 (Figure 7c), and the positive correlation is dominant, with 86.84% of the area. Among them, the area with “ESPC” accounts for 12.50%, mainly distributed in the Ili River Valley, the northern slope of Tianshan Mountain, the western section of Kunlun Mountain Range and the Tacheng area. The area with “SPC” accounts for 15.78%, concentrating in the Irtysh River Basin, the edge of Junggar Basin and the western area of the Tarim Basin edge. The area with a negative correlation accounted for 13.16%, mainly distributed in the Altay Mountains, the northern section of the Tianshan Mountains and Kumkul Basin, where the precipitation is larger than in other regions, and the increase in precipitation is rather un-favorable to the growth of vegetation in the region [52] (Figure 7d).

Overall, both the annual average temperature and annual average precipitation in Xinjiang showed pre-dominantly positive correlations with NPP. However, the majority of the regions did not exhibit significant correlations, suggesting that changes in vegetation NPP in Xinjiang were not completely driven by climate change. Therefore, it is necessary to analyze the residuals between the actual and predicted values of NPP to further investigate the underlying factors influencing NPP variability.

3.2.2. Driving Analysis of NPP Changes

The Miami model was employed to simulate the predicted values of potential NPP from 2001 to 2022. Based on residual analysis, NPP was influenced only by climate change (V_NPP-P) and human activities (V_NPP-H) were isolated. The overall distribution of potential vegetation NPP (V_NPP-P) in Xinjiang is greater in the north than in the south and greater in the west than in the east, under the influence of temperature and precipitation alone (Figure 8a). Notably, the high values have generally shifted to the southwest and northeast, concentrating in the oases at the southern edge of the Tarim Basin, the eastern Ili region, and the Altay Mountains (Figure 8b).

The Theil–Sen median trend analysis was used to calculate the K_P and K_H of V_NPP-P and V_NPP-H, respectively. The areas with K_P < 0 accounted for 75.44% of the total area, mainly concentrated in Northern Xinjiang, where the decline in V_NPP-P is most pronounced north of Sayram Lake (Figure 8c). The areas with K_P > 0 accounted for 24.56% of the total area, among which the southern slopes of the Tianshan Mountains, the Tarim River Basin, and its oasis zone were most obviously affected by climate change. The area with K_H > 0 accounts for 83.46% of the total area, with the southern slopes of the Tianshan Mountains, the Tarim River Basin, and its oasis zone most obviously affected by climate change (Figure 8d). The areas with K_H < 0 account for 16.54% of the total area, concentrated around urban agglomerations on the northern slopes of the Tianshan Mountains (Urumqi-Changji, Shihezi-Manas, and Usu-Karamayi-Kuitun) and the three southern Xinjiang pre-fectures (Kizi-Kashgar-Hotan).

The driving mechanisms of vegetation change in Xinjiang were classified according to the K_A, K_P, and K_H (Figure 9). In terms of vegetation restoration, the areas driven by both climate change and human activities together (combined restoration) account for 11.99% of the total area, distributed in the Altay Mountains, the southern slopes of the Tianshan Mountains, and the oasis zone at the southwestern edge of the Tarim Basin. The proportion of areas where vegetation is restored by climate change alone (climate restoration) is 9.25%, similar to the distribution of the combined restoration, but geographically more to the north. The proportion of areas where vegetation is restored by human activities alone (human restoration) is as high as 60.23%, which is significantly larger in the northern part of the region compared to the southern part, mainly concentrated in the Altai region, the southern edge of the Junggar Basin, and dispersed in the Yarkant River Basin and the Kumkul Basin. Regarding vegetation is degradation, the areas degraded by both climate change and human activities (combined degradation) account for 3.56% of the total area, scattered in the central part of Northern Xinjiang. The proportion of areas where vegetation is degraded by climate change alone (climate degradation) is 11.64% and by human activities alone (human degradation) is 3.32%. The areas degraded by climate change are relatively concentrated in the Ili Valley and the Tacheng area in northern Xinjiang, while areas degraded by human activities are mainly concentrated in the Tarim River Basin. Overall, the proportion of areas jointly driven by climate change and human activities is 15.56%, while areas driven by climate change alone and human activities alone are 20.91% and 63.55%, respectively. This indicates significant spatial heterogeneity in the effects of climate change, human activities, and their combined impacts.

3.2.3. Relative Contributions of Climate Change and Human Activities to Variations of NPP

To further clarify the relative contributions of climate change and human activities to vegetation changes, this study classified the study area into regions of vegetation restoration and degradation based on NPP trends. In regions of vegetation restoration, the areas where climate change was the dominant factor (with a relative contribution exceeding 50%) accounted for 17.86% of the total. These areas were primarily located in the central and southern parts of the study area, such as the southern slope of the Tianshan Mountains and the three southern prefectures of Xinjiang (Figure 10a). The areas where human activities were the dominant factor in vegetation restoration constituted 82.14% and were widely distributed in the northern region of Xinjiang (Figure 10b). In regions of vegetation degradation, climate change and human activities accounted for 74.43% and 25.57% of the area, respectively. Climate-driven degradation was mainly concentrated in the Ili River Valley and the northwestern part of the Tacheng area (Figure 10c). In contrast, areas dominated by human activities were primarily located in populous centers such as Urumqi, Aksu, and their surrounding regions (Figure 10d). Overall, the relative contribution of climate change to NPP variation was 28.34%, while human activities accounted for 71.66%. Obviously, human activities emerge as the principal driver of vegetation changes in Xinjiang. Although climate, as a fundamental natural condition, affects a broad area, its intensity is substantially weaker than that of human activities.

4. Discussion

This study focuses on Xinjiang, a representative arid region in China, analyzing the spatio-temporal dynamics and driving mechanisms of vegetation NPP from 2001 to 2022. Compared to previous studies [53,54,55], this research captures the most recent impacts of climate change and human activities on Xinjiang’s vegetation NPP by extending the study period to 2022, providing more relevant and timely insights. Importantly, un-like some previous studies that concentrated on single aspects of NPP changes in Xinjiang [34,35,36,56], this research employs the Miami model and residual analysis to quantitatively assess the relative contributions of climate change and human activities to NPP. It explores the dominant factors driving NPP changes in Xinjiang, thereby enriching the comprehensive integration of multiple factors in the study of vegetation NPP from a human-land relationship perspective and offering scientific support for regional ecological conservation and management strategies.

4.1. Validation of Potential NPP Simulated by the Miami Model

To further validate the simulated potential NPP in this paper, the results of other studies were compared with ours (

Table 5. Comparisons of the values simulated in this study with those of other studies.

Study Area	Study Period	Model	NPP-P (g C·m−2)
Xinjiang (this study)	2001–2022	Miami	323.21
Xinjiang	2001–2020	Miami	355.78
Xinjiang	2000–2015	Miami	324.53
Xinjiang	2000–2020	improved CASA	323.11
Xinjiang	2000–2013	improved CASA	300–350

). For instance, Zhang et al. and Zhao et al. used the Miami model, which is consistent with this study, to simulate the potential NPP in Xinjiang for different study periods, and obtained results of 355.78 g C·m⁻² and 324.53 g C·m⁻², respectively, which are slightly higher than our results [57,58]. In addition, another study showed the potential NPP value of Xinjiang was in a range of 300–350 g C·m⁻² for potential NPP from 2000 to 2013 by using the improved CASA model [59], whereas Bi et al. simulated potential NPP in Xinjiang from 2000 to 2020 on this basis as 323.11 g C·m⁻² [60], which is highly consistent with our results. Overall, the potential NPP simulated by the Miami model presented high correspondence with the other previously reported results around the study area. These studies further demonstrate the reliability of our results when taking the differences in model parameters, applied methods, and spatial-temporal variations into consideration.

4.2. Relationships between Climate Change and Vegetation NPP

The spatial and temporal evolution pattern of vegetation NPP in Xinjiang is closely related to hydrothermal changes [61,62,63]. The results of this study indicate that both positive and negative correlations between NPP and air temperature coexist, while the response of NPP to precipitation pre-dominantly reveals a positive correlation. This finding is consistent with the results of most studies [64,65,66].

On a temporal scale, the NPP of vegetation in Xinjiang showed a greening trend during the study period, which was associated with the warming and humidifying climate trend in the region [67]. The increase in precipitation in Northwest China over recent decades has been particularly pronounced [68], and precipitation is a crucial determinant of vegetation growth in arid and semi-arid regions. Sufficient moisture drives plant nutrient uptake and promotes plant growth by influencing soil nutrient transport and increasing soil microbial activity [69,70]. The results of this study also revealed extreme annual variations in NPP, which align with findings from several studies [28,38,71], and are linked to oscillations at high and low levels in the “warm humidification” pattern in Xinjiang [72]. The highest value of vegetation NPP occurred in 2016, due to increased temperatures and the highest precipitation levels since 1981, providing optimal conditions for vegetation growth [73]. Conversely, the lowest NPP was recorded in 2008, a year marked by a severe drought in northern Xinjiang, which led to a sharp decline in NPP [72,74]. Additionally, the low-temperature environment caused by the freezing rain and snow disasters in winter further inhibited vegetation growth [75].

On a spatial scale, the multi-year average NPP of vegetation in Xinjiang exhibited a pattern where values were higher in the northern region compared to the southern region and higher in the western region compared to the eastern region. Xinjiang is located deep inland, with annual precipitation typically below 200 mm, displaying typical characteristics of an arid region. According to Liebig’s Law of the Minimum, precipitation serves as the primary limiting factor for vegetation growth in this region [76]. Drought and water scarcity can induce dehydration of vegetation cell protoplasm, altering cell membrane structure and permeability, disrupting plant metabolism, and thereby impacting photosynthetic efficiency, respiration intensity, and soil microorganism succession. The southern Xinjiang region hosts the Taklamakan Desert, the world’s second largest mobile desert [77]. The Tianshan Mountains to the north and the Kunlun Mountains to the south obstruct water vapor entry, resulting in low precipitation and high evaporation, contributing to reduced NPP in southern Xinjiang [41]. In western Xinjiang, influenced by westerly circulation, warm and humid airflow from the North Atlantic enters the region. As it ascends due to barriers such as the Tian Shan and Altai Mountains, adiabatic cooling occurs, leading to precipitation formation. This process enhances vegetation NPP along the Ili Valley and Tian Shan Mountains in northern Xinjiang [64]. Conversely, the central and eastern parts of northern Xinjiang, centered around the Gurbantunggut Desert and surrounded by mountains on three sides, are distant from oceanic moisture sources, limiting water vapor transport and resulting in lower NPP. Consequently, there is an overall higher NPP in the western part compared to the eastern part of Xinjiang [78].

Interestingly, this study also found that the decline in vegetation V_NPP-A in the Ili Valley and Tacheng area during the study period was mainly caused by climate change. This decline is likely related to rising temperatures and a decreasing trend in precipitation in recent years, which inhibited vegetation growth and development [35,57,79,80]. The northern slopes of the Tianshan Mountains are another key area of climate change-driven degradation, where vegetation growth and recovery are limited by low precipitation, high evapotranspiration, and severe soil salinization throughout the year [81]. In contrast, due to global warming, the glaciers of the three major mountain systems in Xinjiang have melted to varying degrees [82]. This melting has gradually converted bare land previously covered by snow and ice into grassland, favoring vegetation recovery [58].

4.3. Relationships between Human Activities and Vegetation NPP

More importantly, the results of this study show that human activities are the main driver of changes in vegetation NPP in Xinjiang from 2001 to 2022. This finding is consistent with several other studies [38,83,84] and is primarily manifested through the promotion of vegetation restoration. This trend may be related to the ecological restoration projects and conservation measures implemented by China in recent years. The Grain for Green Program (GTGP) was officially launched in 1999 by the Chinese government, and Xinjiang began comprehensive implementation in 2002. By 2020, Xinjiang had converted 85,680 ha of farmland to forests and 55,960 ha to grasslands. Additionally, in 2011, the Grassland Ecological Protection Subsidy and Incentive Mechanism was fully implemented, with an investment of over 150 billion yuan, promoting continuous recovery of grassland ecology and significantly mitigating the un-controlled use and over-exploitation of grasslands [35,85,86]. These engineering measures have played a positive role in the recovery of vegetation productivity in the region. Locally, the Altay region, a key area for systematic management of mountains, water, forests, lakes, grasslands, and sands, has implemented measures such as ecological corridor construction, mine management, and ecological migration. These efforts have restored 1000 ha of vegetation habitats in the north and restored 50,073 ha of new degraded grasslands in the south, leading to the continuous restoration of vegetation productivity. Furthermore, projects such as the Tarim River Basin Governance Project, the Ten Major Forests Projects in Xinjiang, and the Junggar Basin Sand Prevention and Control Project have effectively promoted the recovery of vegetation in oasis areas around the southern edge of the Junggar Basin and the Tarim Basin [38].

Although various ecological projects in Xinjiang have achieved significant ecological benefits, the ecological damage associated with economic development, driven by Xinjiang’s role as a land transport hub and gateway to Central and West Asia, cannot be ignored. This study’s results indicate that significant degradation of vegetation NPP is concentrated in Urumqi, Kashgar, Aksu East, and surrounding areas. In recent years, with the acceleration of industrialization and urbanization, the proportion of land used for construction has been increasing [87,88], encroaching on grassland and forest resources. Land use changes in the region have shifted from vegetation cover to impermeable surfaces. Additionally, the increasingly prominent urban heat island effect and intensifying aridity have led to a decrease in vegetation NPP and an increased risk of degradation [50]. Meanwhile, the vegetation NPP in the Altay Mountains and the southern slopes of the Tianshan Mountains showed a decreasing trend during the study period, likely related to overgrazing [89,90,91]. Overgrazing, a major grassland utilization mode, can lead to grassland degradation and even desertification when overloaded. Although efforts to restore grasslands in Xinjiang continue, restoration is more challenging in these high-altitude areas. The increased grazing pressure in these regions has added stress to the grassland ecosystems, resulting in decreased vegetation NPP.

Therefore, the long-term promotion and implementation of effective ecological restoration projects and protection measures, which synergize the positive coupling between socio-economic development and vegetation change in Xinjiang, are crucial for controlling and reducing the risk of vegetation degradation. These efforts are also essential for actively responding to the adverse effects of climate change on vegetation.

4.4. Uncertainty and Prospects

The mechanisms driving changes in net primary productivity remain un-certain. Zhao et al. suggest that the increasing trend of grassland NPP in Xinjiang during the study period was primarily driven by climate change, with “warming and humidification” contributing to vegetation restoration [57]. Furthermore, existing research has shown that since the beginning of the 21st century, the dominant role of climate change in increasing Xinjiang’s vegetation NPP has gradually diminished, while the positive impact of human activities on vegetation has become increasingly significant, occasionally even surpassing that of climate change [38,92]. Liang et al. found that the restoration of grasslands in Xinjiang is primarily driven by human activities, with the contributions of climate change and human activities to grassland NPP degradation being nearly equal [37]. However, different from theirs, our findings indicate that human activities are the main drivers of vegetation improvement, whereas climate change is the primary factor contributing to vegetation degradation in the region. The study reveals that vegetation responds to both the improvement and degradation caused by climate change, indicating that Xinjiang experiences both “warming-wetness” and “warming-dryness”, which contrasts with the commonly perceived trend of “warming-wetness” in Xinjiang [93]. Since 2000, under the backdrop of “wet-dry transitions” and intensified extreme climate events [94], variations in temperature and precipitation have significantly influenced vegetation NPP. Climate change has become the dominant factor in NPP degradation, highlighting the increasing ecological negative effects in Xinjiang.

In this study, the relative impacts of climate change and human activities on vegetation degradation and restoration were quantitatively analyzed using the Miami model. The results were relatively consistent with those of previous studies [37], providing valuable reference for arid and semi-arid regions. However, the study has some limitations. Vegetation change results from the combined influence of multiple factors, including climate change and human activities, which are not limited to temperature and precipitation. Other factors such as solar radiation, atmospheric CO₂ concentration, drought, and terrestrial water storage also play significant roles [2,95], and the quantification process has some deficiencies.

Moreover, due to the heterogeneity of the study area, the approach to simulating potential NPP needs to be adapted to local conditions. Future studies should select multiple models for comparative analyses to identify the optimal model. Additionally, the study will further increase and refine the indicators of climatic factors and human activities and explore the changing mechanisms of different vegetation types. This will help deconstruct the driving mechanisms of vegetation NPP change in more detail and provide a scientific basis for promoting ecological civilization construction in arid and semi-arid regions and achieving regional sustainable development.

5. Conclusions

Based on the NPP dataset and meteorological data for Xinjiang from 2001 to 2022, this study analyzes the spatial-temporal patterns and evolution of vegetation NPP. Using partial correlation analysis, the Miami model, and residual analysis, the relative contributions of climate change and human activities to vegetation NPP changes are quantitatively assessed, exploring the driving mechanisms behind NPP changes in Xinjiang under the influence of these factors. The results show that:

(1)

On the time scale, the multi-year mean values of vegetation NPP in Xinjiang generally showed a fluctuating and increasing trend. Spatially, the distribution of average values of NPP showed obvious heterogeneity. The distribution pattern of NPP was higher in the north than in the south and higher in the west than in the east, where the areas with increasing NPP were significantly larger than those with decreasing NPP. However, in terms of future development trend, the proportion of regions with a negative development trend of vegetation NPP in Xinjiang is 18.97%, with a high risk of degradation; 75.18% of regions show an un-certain development trend, which needs to be focused on in the future.

(2)

NPP is pre-dominantly positively correlated with the average annual temperature and annual precipitation in Xinjiang. Precipitation exerts a more significant effect, demonstrating extremely significant positive correlations mainly in regions such as the Yili Valley, the northern slope of the Tianshan Mountains, and the Tacheng area. However, approximately 18% of the total area shows non-significant correlations. This indicates that climate change does not entirely dominate the variation in vegetation NPP in Xinjiang, as 72.05% of the area exhibits an insignificant correlation.

(3)

Vegetation NPP changes in Xinjiang are driven by both climate change and human activities. Specifically, climate change accounts for 28.34% of the influence on NPP, while human activities have a more significant impact, accounting for 71.66%. In areas of vegetation NPP restoration, human activities are the dominant factor, whereas climate change plays a lesser role. Conversely, in areas of vegetation NPP degradation, climate change significantly outweighs human activities, highlighting the serious challenge of climate change in Xinjiang.