With the objective of investigating the spatial distribution and temporal variations of vegetative cover in the BNNR from 1986 to 2020. Tendency and stability characteristics of vegetation cover have been investigated utilizing the Mann-Kendall test (MK) and the Theil-Sen slope estimator (TS).

TS is considered suitable for investigating the slope tendency in time series data (Sun et al., 2021). This method has advantages including insensitivity to measurement error and group data as well as high computational efficiency (Peng et al., 2015). MK is one of the commonly used tools for nonparametric trend testing, which is not necessary that the data samples follow a specific pattern, as well as can be used to test the trend of data changes under long time series (Yue et al., 2002; Luo et al., 2020). Combining TS and MK will provide a strong anti-interference ability for data noise, and the specific distribution of analysis data is not a prerequisite (Zhang et al., 2022b). Thus, TS-MK was applied to investigate the trend of the dynamic shifts in FVC of the BNNR. The formula of TS equation is as follows (see Equation 6):

Where 1< a< b< n, a and b represent the amounts of years in the time series. FVC_b and FVC_a are the FVC values of time series a and b, respectively. When the slope S_FVC is greater than 0, it indicates a growth tendency; when it is less than 0, it indicates a downward tendency.

The calculation formula of Manna-Kendall test is shown in Equations 7–10:

Where S is the correlation coefficient of the Mann-Kendall test; Z is the significance index, and its value range is (-∞, +∞), which follows the standard normal distribution. Z-value greater than 0 represents a significant increasing tendency, while a Z-value less than 0 represents a significant decreasing trend.

The Hurst exponent was employed in this research to forecast the FVC trend in the BNNR going forward. Hurst originally introduced the Hurst exponent, which is a way to determine if continuous time series data exhibit long-term correlation (Bashir et al., 2020). This exponent was established using R/S analysis on the basis of long-term hydrological observations (Zhang and Jin, 2021). The Hurst index is generally in the range from 0 to 1. When 0< H< 0.5, the future tendency of FVC will be reversed; when H=0.5, the development trend of FVC is unpredictable; when H is greater than 0.5 and less than 1, the future time series of FVC remains in agreement with what has happened previously (Tong et al., 2018; Liu et al., 2022). The selection of the time window for the Hurst exponent requires consideration of the study object and the cloudiness of the study area (Zhu et al., 2021; Fu et al., 2022b). Since the dominant tree species in the study area are all evergreen and have high cloudiness in the growing season. Therefore, to achieve year-by-year time series FVC estimation in the study area, FVC with a time window of the non-growing season was used for Hurst exponent calculation. The Hurst exponent is calculated as shown in Equations 11–15.

Define the FVC time series as FVC(τ), The mean of this time series is:

The cumulative deviation is:

The range R(τ) is:

The standard deviation sequence is:

Calculate the Hurst exponent:

where H represents the Hurst exponent in the range greater than 0 and less than 1.

Afterwards, the Hurst exponent and TS results were overlaid to get coupled data on the changes’ tendency and consistency, and the results were classified as: (1) decrease to increase; (2) consistent decrease; (3) increase to decrease; (4) consistent increase (see Table 1 ).

According to the FVC classification standards and pertinent studies (Liu et al., 2022; He et al., 2023), FVC values were categorized into five classes and corresponded to different landscapes (see Table 2 ). Figure 4A illustrates the spatial distribution of FVC in the BNNR between 1986 and 2020. Areas of higher FVC are situated primarily at the BNNR’s eastern and southern regions. The northern and west-central snow-covered areas of the study area have lower FVC. Compared to 1986, the Class V FVC coverage area in 2000 and 2020 showed significant growth, primarily located in the BNNR’s central and southern regions.

The mean FVC value in the research region has exhibited a tendency to increase in a wavering manner over these thirty-five years (see Figure 5A ). The average FVC increases from 59.40% in 1986 to 68.67% in 2020, an increase of 0.26% per decade. Throughout this 35-year period, the average annual FVC experienced two significant declines, with the turnaround occurring in 1994 and 2001. The decline in FVC may be attributed to climate-related disasters, and the implementation of ecological policies, such as the return of grazing land to forests, has to some extent contributed to the recovery of vegetation. The highest mean FVC of 72.53% occurred in 2018. The lowest mean FVC was 59.41% in 1986.

Further statistics were made on the changing trend of FVC of different classes in BNNR between 1986 and 2020. Among these classes, FVC for class I, II, III, and IV tended to decline, decreasing by 6.19%, 3.64%, 3.93%, and 5.00%, respectively. The corresponding class V showed an obvious upward trend, rising from 39.28% in 1986 to 58.03% (see Figures 4B , 5B ). The improving trend in fast FVC in the BNNR is dominated by the continued growth of high-class FVC.

A total of seven vegetation types (alpine vegetation, coniferous forest, broadleaf forest, meadow, shrub, tussock, and cultivated vegetation) were distributed in the BNNR. The more widely distributed vegetation types are coniferous forest and shrub, accounting for 32.51% and 31.79% of the BNNR, respectively. The FVC of all vegetation types exhibited an upward trend to varying degrees during these 35 years, with the more significant increases mainly in tussock (15.86% increase), cultivated vegetation (14.97% increase), shrub (11.32% increase), and coniferous forest (10.94% increase) (see Figure 5C ).

The results of TS analysis can be classified into three categories based on the slope of FVC changes: S_FVC < -0.0005, -0.0005< S_FVC < 0.0005, and S_FVC > 0.0005, indicating decreasing, no change, and increasing trends in FVC, respectively (see Table 3 ). As illustrated in Figure 6A , the area of the BNNR with an upward trend is 1651.57km² and the area with a downward trend is 619.13km², which account for 59.03% and 22.13% of the area, individually.

Among the area proportion of each trend, 61.48km² and 118.23km² of FVC were significantly reduced and very significantly decreased, accounting for 2.20% and 4.23% of the BNNR (see Figure 6B ). The area where the decrease of FVC occurred is mainly on both sides of the Zhubaluo River and in the southeastern part near the boundary of the reserve. All of these regions have relatively high levels of human activity. However, there are more areas showing significant vegetation recovery than significant vegetation decline. The areas of significant and extremely significant increase of FVC were 286.24km² and 498.44km², accounting for 10.23% and 17.81% of the BNNR (see Figure 6B ). The regions of vegetation restoration are predominantly situated around the snowy mountains in the north and in the southwestern part of the research region. In addition, in the eastern part of the BNNR, regions of vegetation recovery and deterioration are patchy.

The overall trend of vegetation change in the BNNR appears to be persistent, as indicated by the Hurst index research. Specifically, the area with a Hurst index greater than 0.5 accounted for 76.7% of the total area of the study area, while the percentage of the area less than 0.5 was 4.0% (see Figure 7A ). The results of the TS analysis were integrated with the Hurst index to investigate the degree of sustainability of the FVC. The results of the research are depicted in Figure 7B ), where the area transitioning from FVC reduction to growth and sustained growth are 24.17km² and 1551.06km², respectively. Conversely, the area transitioning from FVC improvement to degradation and sustained degradation are 82.71km² and 487.13km², respectively. Since the vegetation in most areas will continue the past trend, the distribution of FVC growth and decline in the future BNNR will basically coincide with the past trend.

Based on the actual situation of the BNNR and related literature, its altitude, slope, and aspect were reclassified to explore the influence of topographic effect on FVC. The area proportion of FVC classes varies significantly with the elevation (see Figure 8A ). The proportion of class I fluctuated greatly at different elevations. In the elevation range from 2500m to 3500m, the proportion of class V increases with the elevation, while the proportion of class I and class II decreases in this elevation range. However, in the area with elevation greater than 3500 m, the proportion of class I area increases with the elevation, especially in the area above 4500m, reaching up to 82%. But the proportion of class IV and class V in this area is even less than 1%. The variation trend of FVC also has an obvious terrain response at different elevations (see Figure 8D ). Both vegetation deterioration (P< 0.05) and recovery (P< 0.05) are primarily located at altitudes intervals below 4500 m, and reach the maximum value between 2500 m and 3000 m. The proportion of regions with vegetation deterioration (P< 0.05) and recovery (P< 0.05) in FVC was 13% and 39%, respectively, at altitudes from 2500m to 3000m. It can be seen that the elevation zones with higher vegetation cover are also the places with the most drastic changes in vegetation cover. In addition, at elevations greater than 3000 m, the percentage of area with significant changes in vegetation decreases with elevation. Ultimately, at elevations greater than 4500m, vegetation is largely unchanged.

The proportion of area in class V increases with slope gradient from 46% for flat slopes to 66% for steep slopes when the slope is less than 35° (see Figure 8B ). Concurrently, class IV, class III, and class II decreased by 6%, 7% and 5% respectively from flat slope to steep slope. The FVC significantly reduced proportion increases with the slope increasing, from 4% for flat slopes (<5°) to 10% for dangerous slopes (>46°) (see Figure 8E ). The proportion of significant increase reaches its peak value in the slope range of 26°~45°. However, the difference in FVC between different slopes is relatively small compared to the difference in FVC between different altitudes and aspects.

The percentage of class V FVC in the BNNR is relatively higher on the shady slopes and semi-sunny slopes compared to other aspects (see Figure 8C ). Additionally, the percentage of significant increases is higher for shady and semi-sunny slopes than for other slope orientations, while the percentage of significant decreases is relatively high for sunny and semi-sunny slopes (see Figure 8F ). Specifically, shady slopes accounted for less than 2% of the area of significant FVC reductions, while sunny slopes accounted for 12% of the area of significant reductions. Conversely, the ratio of each grade of FVC does not differ much on the flats, and fewer areas with significant vegetative changes.

In summary, the distribution and changes in vegetative cover in the BNNR are likely more influenced by elevation compared to slope and aspect. Therefore, the vegetation in the BNNR with an altitude of 2500m to 3500m and located on the sunny slopes needs more attention.

The annual mean temperature of the BNNR stands at 4.2°C, displaying a fluctuating upward trend over the 35-year period. There’s a notable difference of 1.6°C between the highest (5.0°C in 2009) and lowest (3.4°C in 1992) annual mean temperatures (see Figure 9 ). Moreover, the average annual precipitation is recorded at 733.5 mm, with a slowly declining trend but significant inter-annual fluctuations. The disparity between the highest precipitation year (939.4 mm in 1998) and the lowest year (538.9 mm in 2014) is nearly 400 mm. Overall, temperature decreases from east to northwest in the BNNR, primarily influenced by topography. The lowest mean annual temperatures occur in the snow-covered mountainous areas at high elevations, while the highest mean annual temperatures are observed in the valleys at lower elevations. The spatial distribution of annual precipitation varies widely, with areas of higher precipitation distributed only in the southern part of the BNNR and along the banks of the Zhubaluo River.

For the majority of the BNNR (62.65%), there was a positive correlation between FVC and temperature; of this, 27.90% of the region demonstrated a significant positive association (p<0.05) (see Figures 10A, C ). The percentage of area where FVC is negatively correlated with annual mean temperature is 27.36%, with 7.63% of the negatively linked area is explained by a significant negative correlation. The southwestern and west-central sections of the BNNR are primarily home to the places where FVC and temperature have a major positive tie, while the negative correlation regions are primarily situated in the valley area of the BNNR’s northern part, on both sides of the Zhubaluo River in the central part, and in the southeastern part. Combined with the topography, it can be found that the negative correlation area is mostly distributed in the area of low elevation. The correlation between FVC and precipitation is indicated in Figures 10B, D ), the area with positive correlation is 52.64% of the BNNR, while the negative correlation is 37.36%. Among them, significant positive correlation and significant negative correlation made up 7.79% and 3.11% of positive and negative correlations, respectively. The southern part of the BNNR is home to the majority of the regions where FVC and precipitation have a substantial positive association. The regions of negative correlation are located primarily in the northeastern portion. The findings revealed that the correlation between FVC and temperature is stronger in the BNNR compared to annual precipitation. The findings of the significance test confirm this view as well.

The vertical distribution of vegetative cover in the BNNR shows that the region below 3300m is cool coniferous forest, warm coniferous forest and river valley scrub; from 3300m to 4000m is cool coniferous forest, and above 4000m is scrub meadow (Fan and Bai, 2021; Shi et al., 2021a). Different categories of vegetation are strongly affected by the scale of the terrain, and therefore trends also vary (Li et al., 2016). In the altitude range from 2500m to 4000m, the distribution of class IV and class V is predominant, and the sum of the two is more than 70%. And the percentage of significant increase in FVC was also relatively high in this elevation interval. The possible reason for this is that the soil here is dominated by mountain brown loam with high soil fertility. Therefore, this elevation range has a relatively high vegetation cover ability. Conversely, the relatively low-altitude areas below 2500m and above 4500m in the reserve have low vegetation cover, with grassland and scrub dominating the vegetation. The lower areas, below 2500 m, are characterized by dry, hot river valleys with minimal precipitation and high evaporation, making them unsuitable for forests but conducive to shrubs and grasses (Chai et al., 2002; Zhang et al., 2019). Areas above 4500 m are characterized by year-round snow and ice cover, slow accumulation of organic matter, and a shortened growing season for vegetation as the altitude rises (Li et al., 2016). Vegetation degradation is primarily observed at altitudes of 2500 m to 3500 m. According to the relevant data, the areas suitable for agriculture and animal husbandry in the study region are mainly located at low altitude (Wang et al., 2023). This suggests that the incidence of vegetation degradation in some low elevation regions may be related to anthropogenic factors.

The research results revealed that the FVC of the BNNR peaked at slopes ranging from 26° to 45°, with minimal differences in FVC among different slope categories. This may be related to the fact that gently sloping areas are subject to more frequent anthropogenic disturbances, but such disturbances are decreasing as the slope increases. Moreover, the percentage of area significantly reduced of FVC increases with slope, but the change is small. Slope has a relatively limited effect on the distribution and tendency of FVC, and the difference in FVC between different slopes in the BNNR is small. This phenomenon may be due to the fact that with the increase gradually of slope, soil water and nutrients will be more easily lost, so that the percentage of vegetation decline in the higher slope area will be slightly increased (Zhang et al., 2013). Relevant studies show that vegetative coverage in the TPRR is highest at a slope of 35° (Wang et al., 2021b), and have concluded that vegetation in this region is less affected by slope (Huang et al., 2023), which is in agreement with the findings of this research.

In the BNNR, vegetative cover is lower on sunny and semi-shady slopes compared to other slopes. Vegetation degradation predominantly takes place on sunny and semi-shady slopes. Revegetation occurs mainly on shady and semi- sunny slopes. This may be due to differences in radiation, temperature and water evapotranspiration on different slopes (Zhang et al., 2022a), leading to differences in vegetation distribution and growth. Combined with the fact that the main tree species in the BNNR are trees of the Abies and Picea (Shi et al., 2021a), which are better adapted to a cooler and colder climate, the shady slopes are more suitable for growth.

In summary, the variances and shifts in vegetative cover distribution within the BNNR are likely more influenced by elevation than by other factors.

The QTP has long been acknowledged as a region experiencing more pronounced impacts from climate warming (Rangwala and Miller, 2012). Precipitation and temperature are the two primary climate elements that impact the growth of vegetative cover. In the BNNR, which is situated in the eastern section of the QTP, we found association between FVC and temperature was stronger than that of precipitation. This conclusion was further verified in the results of the significance test. Furthermore, this is in agreement with previous findings in the region that temperature is the leading element controlling or influencing vegetation cover increase in the region (Wang et al., 2021a; Guo et al., 2023). Owing to the relatively complicated terrain of the area, there are more pronounced spatial variations in the distribution of precipitation and temperature (Zhang et al., 2018; Guo et al., 2023), resulting in different correlations between FVC and climate elements in different areas. The relationship between temperature and elevation elements tends to have a more pronounced linear negative correlation, leading to possible differences in the response of vegetation dynamics to temperature. Wang et al. on the Tibetan Plateau showed that the correlation between vegetation and temperature varied at different altitudes, especially at altitudes above 2400 m, where temperature, instead of precipitation, played a major role in regulating the vegetation ecosystem (Wang et al., 2020).

Additionally, to some degree, the rise in temperature might encourage the development of plants (Wei et al., 2022). Several studies have shown that in the summer, temperature and tree radial growth are favorably connected (Kang et al., 2021). The growth of dominant tree species such as Abies george and Picea asperata in the BNNR was positively correlated with temperature (Shi et al., 2021a). This may also be one of the reasons for the relatively strong correlation between temperature and vegetation cover. However, it has also been suggested that raised temperature would lead to an increase in plant respiration rate (Hua et al., 2019). Rising temperatures will accelerate snowmelt, potentially yielding diverse effects on vegetation. On the one hand, early snowmelt and longer growing seasons will increase vegetation productivity, but on the other hand, early snowmelt may also lead to low temperatures in the early spring and reduced snowmelt in the summer leading to droughts and a decrease in vegetation productivity, which is mainly related to seasonal and geographic variations (Wang et al., 2013). Vegetation phenology is closely related to climate. It has been suggested that climate warming leads to a longer growing season for vegetation, which promotes plant growth and biomass accumulation (Wheeler et al., 2017). As there is a strong link between vegetation cover and biomass (Qin et al., 2022). Therefore, changes in vegetation phenology as a result of climate warming have had a relatively positive effect on vegetation cover.

Regarding the regional distribution of positive correlations with precipitation, the southern part of the BNNR exhibits a vegetation cover more susceptible to precipitation, possibly linked to variations in the distribution of vegetation categories. In the past, temperature has been more influential on vegetation than rainfall. However, as temperature continues to rise and annual precipitation tends to reduce, the association of vegetative growth with precipitation may become stronger in the future. In the context of global warming, the frequency and magnitude of climate extremes have increased. Hazards such as droughts, temperature extremes, and floods can cause varying degrees of vegetation degradation (Manoranjan et al., 2024). Therefore, the mechanism of vegetation response to global warming in high-altitude mountainous areas needs to be further investigated.