The study area covered the entire region of the Yinshan Mountains. This important mountain range is located in northern China, from east to west, includes, Wolf Mountain, Wula Mountain, Serteng Mountain, Daqing Mountain. The mountain range is 1200 km long, 50-100 km wide from north to south, with an average altitude of 1800 m. The northern and southern slopes of the mountains are asymmetrical, with the northern slope dipping gently into the Inner Mongolia Plateau and the southern slope dropping to the Yellow River Hetao Plain with a drop of more than 1000 m. The Yinshan Mountains represent the boundary between the warm temperate and mid-temperate climate zones in Inner Mongolia, and the difference in the climates in the north and south separates the grassland from the desert steppe. The unique topography and climate change in the Yinshan Mountains have greatly enriched the plant species range and ensured the diversity of the region’s medicinal plant resources, forming an obvious uneven distribution pattern of the region’s medicinal plant resources.

The Yinshan Mountains involved 40 counties and the area was large, and the field survey could not fully cover the entire scope of the 40 counties. Based on the grid technology, this study divided the Yinshan Mountains into 10 km×10 km, 20 km×20 km, 30 km×30 km, 40 km×40 km, 50 km×50 km, 60 km×60 km and different scales of a DEM data space grid system. The types of traditional Chinese medicine resources in each grid were summarized, counted, and used as a statistical basis for richness.

Global Moran’s I analysis is a valuable tool for detecting spatial correlations based on observations and their respective spatial locations. This tool allows researchers to assess whether a given index exhibits spatial autocorrelation within the study area (Ebdon, 1985). Among them, Moran’s I is the most commonly used to measure spatial autocorrelation and is an important research metric for measuring the potential interdependence between observations of variables in the same region. The value range of Moran’s I is (-1, 1) (Mitchell, 2005). Z-values and P-values were used to test the applicability and significance of the index. In the process of spatial calculation, the spatial weight is normalized to prevent the Moran’s I index from exceeding [-1, 1] (Bivand and Wong, 2018). Moran’s I is calculated as follows:

where n is the total number of geographical units distributed in the study area, x is the attribute value of the observation variable, \overline{x}is the average value of the attribute value of the observation variable, and w_ij is the spatial weight.

Local Moran’s I serves as a breakdown of the global spatial autocorrelation index, enabling a finer assessment of dissimilarities between specific spatial regions and their neighboring areas. The size of the Local Moran’s Index provides a more intuitive measure for comparing differences in MPDs across districts (Bivand and Wong, 2018; Anselin, 1995). However, this method of analysis does not represent a specific geographic area at a significance level.

We analyzed the local spatial autocorrelation index using clustering and outlier analysis in spatial analysis methods and obtained five distinct outcomes: High-High Cluster (H-H), High-Low Outlier (H-L), Low-High Outlier (L-H), Low-Low Cluster (L-L), and Not Significant (Liu et al., 2018). These outcomes offer valuable insights into the spatial distribution patterns and significance levels of MPD throughout the study area, enhancing the understanding of its geographic variability.

In spatial statistics, Hotspot analysis, also known as Getis-Ord Gi* analysis, primarily serves to detect spatial clusters of attribute elements with either high or low values. It helps identify whether each regional unit conforms to the spatial distribution pattern of high- or low-value clusters (Zhang et al., 2020). This analytical method calculates Gi* value statistics and corresponding P-values for each element within the dataset, enabling the identification of spatial clustering locations for high-value or low-value MPD. The principle of this analytical method holds that if an analytical element can become a hot area of statistical significance, it should exhibit a high value while being spatially surrounded by other elements with high values (Getis and Ord, 1992). The results are expressed at different levels of significance, namely 99% confidence, 95% confidence, 90% confidence, and no significance. If the value of Gi* is positive and significant, the region is rich in diversity of medicinal plant resources and is a high-value spatial aggregation area; if the value of Gi* is negative and significant, the region is scarce in diversity of medicinal plant resources and is a low-value spatial aggregation area.

Our aim is to not only predict the distribution of MPD under the current climate conditions but also to gain insights into how various socioeconomic development pathways might influence the future spatial distribution of MPD. To achieve our objectives, we incorporated a comprehensive set of data, including 19 bioclimatic variables and elevation data (Supplementary Table 1). This multifaceted approach allowed us to explore the potential shifts in MPD patterns and their correlations with different socioeconomic development scenarios in the future. Current and future climate data were downloaded from the Worldclim database at a resolution of 2.5 arcmins (Zhang and Wang, 2023). Future climate data were selected from four climate scenarios (SSP126, SSP245, SSP370, and SSP585) in the BCC-CSM2-MR model for the 2030s (2021-2040) and the 2070s (2061-2080), which represent future carbon emissions from low to high, respectively (Huang et al., 2022a; Zhang et al., 2022a). Elevation data were from SRTM elevation data. Notably, because there was no information on future projected elevations, we assumed that there was no change in current or future elevations. Because of the large number of species in this study, we used the most widely accepted data, namely, all of the aforementioned data were used for projections of MPD distributions under current and future climatic conditions, respectively.

The MaxEnt model, a widely recognized tool for assessing species distributions, has been used in existing research. However, because of the extensive dataset gathered for this study and the potential risk of results being influenced by complex data deviations, a rigorous screening process was undertaken. This meticulous approach aimed to identify the most representative medicinal plant species distributions, focusing on three distinct categories (Supplementary Table 2). First, we considered the “Widely Distributed medicinal plants in the Yinshan Mountains,” which included species found at many sampling points, specifically those exceeding 60 individual species of medicinal plants. Second, the “Rare medicinal plants in the Yinshan Mountains” category encompassed species with a more limited presence, typically falling within the range of 10 to 30 sampling sites for a single species. Last, we examined the “Endangered medicinal plants in Yinshan Mountains,” comprising plants listed in the Endangered Plant Protection List due to their critical status.

The main objective of this study was to reveal the spatial distribution characteristics of MPD. Consequently, we did not to employ the “Jackknife” method, a commonly used in general MaxEnt model studies. This method is typically used to assess the impact of various ecological factors on the distribution of individual species or to create response curves for quantifying the relationship between ecological factors and the probability of individual species distribution. Instead, we directed our efforts toward the overarching goal of understanding MPD’s spatial distribution patterns. MaxEnt is set as follows: the maximum number of iterations of the model is 10⁵, the convergence threshold is 0.0005, and “Crossvalidate” is used as the model repeated run type, running a total of 10 times (Kapuka et al., 2022). To ensure the precision of our model predictions, we employed the area under the curve (AUC) as an evaluation metric. In general, an AUC value exceeding 0.75 signifies robust model predictions, indicating a high degree of reliability in the results (Parichehreh et al., 2020). Consequently, we scrutinized the predicted outcomes for each species to confirm that the AUC values surpassed the 0.75 threshold. Logistic regression yielded outputs ranging from 0 to 1, aligning closely with the probability of species distribution, which aligns with our research objectives. Thus, we opted for Logistic regression as the output method while retaining default settings for the remaining parameters. Subsequently, we conducted calculations for each species in the ASC format using ArcGIS 10.8, followed by the conversion of results to the tiff format. These individual species maps were then superimposed (Zhang et al., 2022a). Last, to ensure consistency, the superimposed layers underwent a normalization process.

NPP is the total amount of organic matter accumulated per unit of time and per unit of area by a plant during the primary production stage (Wang et al., 2019; Li et al., 2023). The CASA model is one of the most commonly used models for NPP estimation, which calculates NPP mainly from the absorbed light use efficiency (ε) and absorbed photosynthetically active radiation (APAR) by vegetation (Huang et al., 2022b). The calculation expression is as follows:

where APAR(x,t) is the effective photosynthetic radiation that pixel x absorbed during month t (MJ·m^-2 month^-1), and ε(x,t) is the pixel x’s actual light utilization efficiency during month t (gC·MJ⁻¹) (Qiu et al., 2023). The expression of APAR and ε are as follows:

The NPP can be validated using the measured-value validation method and the relative validation method (Wang et al., 2023). However, due to the lack of measurement data, an indirect validation method was used. The indirect validation involved a comparison between the NPP inversion results generated by the CASA model and the NPP data from MOD17A3HGF. Specifically, we randomly selected 100 points within the study area and extracted the NPP values for 2017, 2019, and 2021 from both the MOD17A3HGF dataset and the NPP estimates provided by the CASA model at these selected points. The accuracy of the CASA model was measured using the coefficient of determination (R²), as shown in Eq. (5) (Fang et al., 2021).

where n represents the number of variables; y_i is the NPP value estimated by the CASA model; y_di is the MOD17A3 NPP value; y_a is the average of the MOD17A3 NPP values.

NDVI is commonly used to detect vegetation growth and vegetation cover (Zhang et al., 2022b; Shi et al., 2024; Zhou et al., 2025). Because the growth period for most wild medicinal plants is between June and September, we aimed to deepen the understanding of vegetation coverage during this critical phase by extracting NDVI data for the Yinshan Mountains from June to September in 2017, 2019, and 2021. This extraction was performed using the BandMath tool available in ENVI 5.3. In our analysis, NDVI values below 0.1 signified barren areas, typically covered by rocks, sand, or snow, with minimal surface vegetation. To reduce the influence of NDVI variations on the formulation of regression equations for exposed surfaces, we uniformly categorized NDVI ranges from [-1.0, 0.1] as areas devoid of vegetation coverage, (0.1, 0.5] as areas with low vegetation coverage, and (0.5, 1] as areas characterized by high vegetation coverage (Wei et al., 2023).

We calculated the global autocorrelation (Moran’s I) and its standardized statistic Z for the study area across different grid sizes, providing insights into the richness of these valuable resources in the region. At different grid scales, the species richness of traditional Chinese medicine resources in the Yinshan Mountains was spatially positively correlated and significantly different (Supplementary Table 3). When compared, it was found that the species richness of traditional Chinese medicine resources showed a high spatial correlation at the grid scale of 60 km × 60 km.

There were differences in the spatial distribution of MPDs at different spatial scales (Figure 2). At 60 km×60 km, 50 km×50 km, and 40 km×40 km on a large spatial scale, the H-H aggregation areas were concentrated in the eastern region of Yinshan Mountains. At 30 km×30 km, 20 km×20 km, and 10 km×10 km on a spatial scale, the H-H aggregation area also appeared in the central region of Yinshan Mountains. With the decrease of spatial scale, the H-L and L-H outliers had scattered distribution on spatial scale and showed an increasing trend, which means that different spatial scales can indicate different aggregation degree of MPD in the same region. This finding has important guiding significance for attaining a refined understanding of the degree of aggregation of MPD resources.

Hotspot analysis can display spatial MPD distributions in different regions at the level of significance. Our hotspot analysis results for the MPD at different spatial scales were similar to those obtained using local Moran’s I analysis (Figure 3). At different spatial scales, no cold spot distribution was found in the Yinshan Mountains, indicating that the area has a rich diversity in medicinal plants. However, it is important to highlight that while cold spot aggregation areas are absent across these different spatial scales, there are instances where regions transition from being hotspot aggregation areas to yielding insignificant results.

The results of MPD analysis were categorized into four distinct groups by using ArcGIS 10.8: unsuitable (0-0.25), low suitability (0.25-0.5), moderate suitability (0.5-0.75), and high suitability (0.75-1). The results are shown in Figure 4. In the case of widely distributed medicinal plants, high and moderately suitable areas were predominantly situated in the eastern region, and low suitability zones were primarily concentrated in specific areas within the central and eastern regions. Notably, there were unsuitable areas between the central and eastern regions. For rare medicinal plants, high and moderately suitable regions also tended to cluster in the eastern part of the area, with low suitability areas dispersed in select locations across the central and eastern regions. Endangered medicinal plants exhibited a pattern: high suitability areas were primarily located in the eastern region, and moderate and low suitability regions were scattered across specific zones in the central and eastern areas. In summary, highly and moderately suitable regions for all three categories were predominantly in the eastern section of the Yinshan Mountains, and low suitability areas were concentrated in the central and eastern regions. The Yinshan Mountains form a boundary between monsoon and non-monsoon regions, and some regions in the western and central Yinshan Mountains are non-monsoon regions. Despite the presence of the Yellow River in the region, it still faces challenges due to insufficient precipitation and other factors conducive to plant growth. Consequently, the overall suitability for local medicinal plant distribution is reduced, leading to a lack of diversity in medicinal plant species within the area.

Our MPD analysis results under future climatic conditions are shown in Figure 5 (Supplementary Table 4). The purple, blue, and orange boxes represent the diversity of widely distributed, rare, and endangered medicinal plants, respectively.

The first is the widely distributed medicinal plants. Compared with the current habitat, in the 2030s, the area of suitable habitat under the SSP126, SSP245, SSP370, and SSP585 development pathways increased in the order of 754.89 km², 866.72 km², 698.97 km², 698.97 km², and decreased in the order of 1062.43 km², 587.13 km², 838.76 km², and 1873.23 km², respectively. These data suggest that the SSP245 development pathway is the development situation with the highest suitable area for MPD. By the 2070s, the area of suitable habitat under the SSP126, SSP245, SSP370, and SSP585 development pathways increased in the order of 615.09 km², 1090.39 km², 1845.27 km², 1677.52 km², and decreased in the order of 950.60 km², 1230.18 km², 1230.18 km², and 866.72 km², respectively. These data suggest that the SSP585 development pathway is the development situation with the highest suitable area for MPD.

The second is rare medicinal plants. Compared with the current habitat, in the 2030s, the area of suitable habitat under the SSP126, SSP245, SSP370, and SSP585 development pathways increased in the order of 335.50 km², 279.59 km², 671.01 km², 1369.98 km², and decreased in the order of 810.80 km², 978.55 km², 531.22 km², and 335.50 km², respectively. These data suggest that the SSP585 development pathway is the development situation with the highest suitable area for MPD. By the 2070s, the area of suitable habitat under the SSP126, SSP245, SSP370, and SSP585 development pathways increased in the order of 195.71 km², 1090.39 km², 111.83 km², 335.50 km², and decreased in the order of 754.88 km², 922.64 km², 2656.08 km², and 1342.02 km², respectively. These data suggest that the SSP245 development pathway is the development situation with the highest suitable area for MPD.

Finally, the distribution trend of endangered medicinal plant species. Compared with the current habitat, in the 2030s, the area of suitable habitat under the SSP126, SSP245, SSP370, and SSP585 development pathways increased in the order of 698.97 km², 726.93 km², 810.80 km², 643.05 km², and decreased in the order of 1118.35 km², 1090.39 km², 894.68 km², and 950.60 km², respectively. These data suggest that the SSP370 development pathway is the development situation with the highest suitable area for MPD. By the 2070s, the area of suitable habitat under the SSP126, SSP245, SSP370, and SSP585 development pathways increased in the order of 726.93 km², 950.60 km², 1034.47 km², 726.93 km², and decreased in the order of 447.34 km², 1565.69 km², 866.72 km², and 1342.02 km², respectively. These data suggest that the SSP126 development pathway is the development situation with the highest suitable area for MPD.

The average vegetation NPP in Yinshan Mountains in 2017, 2019 and 2021 estimated by the CASA model showed a fluctuating trend. Specifically, in 2017, the average vegetation NPP was 362.47 g CM^-2·a^-1, while in 2019, it increased to 428.99 g CM^-2·a^-1, and in 2021, it decreased slightly to 384.72 g CM^-2·a^-1. To further analyze these trends, we divided the vegetation NPP into three categories: low value areas (0-200 g CM^-2·a^-1), median value areas (200-600 g CM^-2·a^-1), and high value areas (600 g CM^-2·a^-1 or above), as depicted in Figure 6. The low-value areas in Yinshan Mountains were mainly distributed in the western region; the median-value areas, mainly in the central and eastern regions; and the high-value areas, mainly in some areas of the central and eastern regions. These findings are similar to the MPD spatial distribution characteristics. This suggested that regions with high carbon sequestration potential were more conducive to the growth of medicinal plants, potentially contributing to the rich diversity of medicinal plant species in these areas. Despite the absence of measured NPP data, this study conducted a comparison between NPP values obtained from MODIS NPP products and those derived from the CASA model. The results revealed a correlation coefficient (R²) of 0.8935 (Figure 7). These findings indicated a strong correlation between the NPP estimations produced by the CASA model and the results from the MOD17A3HGF dataset. Consequently, the NPP estimations derived from the CASA model in this study exhibit high accuracy and are suitable for NPP-related research within the Yinshan Mountains.

Using ArcGIS, we reclassified the extracted NDVI into three distinct categories: non-vegetation coverage areas, low vegetation coverage areas, and high vegetation coverage areas (Figure 8). As shown in Figure 8, high vegetation coverage areas are predominantly concentrated in the central and eastern regions of Yinshan Mountains, while low vegetation coverage areas and non-vegetation coverage areas are primarily situated in the western region of Yinshan Mountains. This distribution pattern closely resembles the hotspot areas and NPP high-value area of MPD in the Yinshan Mountains. It suggests that areas rich in vegetation have a heightened capacity for ecosystem carbon sequestration, creating favorable conditions for the survival of most medicinal plants. This, in turn, likely plays a pivotal role in shaping the diversity of medicinal plant species. By comparing vegetation coverage in 2017, 2019, and 2021, notable trends emerge. Vegetation coverage in the western region of the Yinshan Mountains exhibited a consistent decline from 2017 to 2021, and the central and eastern regions have generally maintained relative stability. However, some areas transitioned from high vegetation coverage to low vegetation coverage. As a result, it becomes imperative to bolster conservation efforts in the central and eastern regions while simultaneously focusing on initiatives to improve the ecological health of the western region in the future.

Medicinal plants play a crucial role in treating various diseases worldwide, particularly in developing countries where they are integral to traditional medicine and healthcare (Romeiras et al., 2023). However, the increasing global population growth and the continual expansion of residential areas have led to increased human interference in natural ecosystems, resulting in severe ecological damage. Human activities have emerged as a primary driver of accelerated biodiversity loss on a global scale. Despite the abundant diversity of medicinal plants in China, the expansion of the human population and ecological degradation pose serious threats to these valuable wild resources. The establishment of nature reserves represents a pragmatic and effective approach to safeguarding medicinal plant resources, especially those classified as endangered species (Zhang and Gong, 2016). In this study, we employed spatial analysis methods to investigate species richness and MPD in the Yinshan Mountains across varying spatial scales. The findings underscore the importance of multi-scale analysis in providing comprehensive assessments of regional diversity and offering visual guidance for diversity reserve planning.

In September 2015, the United Nations Summit on Sustainable Development formally endorsed the “2030 Agenda for Sustainable Development,” outlining global sustainable development goals (Zhang et al., 2022c). These goals encompass three vital dimensions, namely, society, economy, and environment, and include specific biodiversity-related targets. Additionally, with the rapid global expansion of the Chinese medicine market, the demand for Chinese medicine products is expected to surge in the near future (Wang et al., 2022). Therefore, implementing measures that ensure the continued production of medicinal plants to maintain their availability is necessary (Fajinmi et al., 2023). This study not only examined the spatial distribution patterns of medicinal plants in the Yinshan Mountains and their distribution changes under climate change scenarios but also provides a foundational research framework for advancing biodiversity goals and the sustainable development of MPD in line with future development objectives.

Species distribution models are primarily constructed based on occurrence records and environmental variables, providing probabilistic estimations of suitable habitats for species (Wen et al., 2021; Guo et al., 2024). Species distribution modeling relies on various statistical techniques, of which MaxEnt is currently the most widely used model (Kapuka et al., 2022; Dong et al., 2023). Research has demonstrated that the MaxEnt model offers stability, speed, and accuracy superior to that of other models and can yield better predictive results than other models when species distribution data is scarce (Shen et al., 2023). While species distribution models have proven invaluable in predicting future extinction risks and informing policy decisions, the choice of Global Circulation Models (GCMs) and different scenarios for gas emissions can introduce uncertainties in future distribution predictions (Song et al., 2023). Consequently, integrated models that consider these factors can enhance the accuracy of species distribution predictions, representing a promising direction for diversity model research.

As ecological concerns grow increasingly prominent, there is a heightened focus on the carbon sequestration capacity of vegetation. China has established relevant policies with ambitious goals, aiming to achieve a carbon peak by 2030 and carbon neutrality by 2060. NPP assumes a central role in the terrestrial carbon cycle and serves as a sensitive indicator of ecosystem performance, locally and globally (Bao et al., 2016). NPP quantifies the accumulation of aboveground and belowground organic matter within vegetation over a specific period, typically one year, offering insights into ecosystem productivity capacity (Liu et al., 2017). Consequently, accurate NPP estimation becomes pivotal. In this study, we employed MOD13A2 imagery (1 km) to predict NPP values for the Yinshan Mountains over the past three years (2017, 2019, 2021) using the CASA model. A comparison of the CASA estimates with MOD17A3HGF data at randomly selected sample points revealed a high degree of correlation. This suggests the reliability of the experiment’s results, which can serve as a foundational framework for informing policy decisions and promoting the sustainable utilization of natural resources in the Yinshan Mountains. Although the absence of measured NPP data precluded direct model comparisons in this study, future data collection efforts will be essential for refining and optimizing the CASA model. Additionally, as satellite sensor technology continues to advance, such as with MODIS, Landsat, RapidEye, and Sentinel-2, providing higher spatial resolutions (Fang et al., 2021), future research will explore prediction results and accuracy across various spatial resolutions, further enhancing and refining the CASA model.

This study delved into the diversity of medicinal plants within the Yinshan Mountains, revealing a noteworthy correlation among NPP, NDVI, and MPD. NPP is a direct source of food and energy for organisms (Yan et al., 2021). Notably, Zhang et al. (2023) used MODIS data and demonstrated the similarity in spatial distribution patterns and strong correlations between NPP and NDVI in the Dabie Mountains (Zhang et al., 2023). This suggests that NDVI can effectively depict the distribution and variation characteristics of NPP within a region. Correlation studies further underscored the highly significant positive relationship between NDVI and annual precipitation across all vegetation types (Sun et al., 2020). The Yinshan Mountains, at the boundary between monsoon and non-monsoon regions, include areas in the western and central regions classified as non-monsoon regions. The overall scarcity of precipitation and unfavorable growth conditions in these areas result in a low NDVI. Additionally, some studies have proposed a significant positive correlation between NDVI and plant diversity (He et al., 2021). In summary, the insufficient precipitation and unsuitable growth conditions in portions of the western and central Yinshan Mountains contribute to lower NPP and MPD levels in these areas.

There is stratified heterogeneity in the spatial distribution of MPD in the Yinshan Mountains. Regarding vegetation type and vegetation cover, the Yinshan Mountains have a strong vertical distribution pattern, and the eastern part is located in a typical grassland area with high vegetation cover and medicinal plants such as Polygala tenuifolia and Echinops davuricus (Chinese Academy of Sciences Inner Mongolia and Ningxia Comprehensive Expedition Team, 1985). The baseband mainly consists of Stipa Bungeana grassland and Stipa grandis grassland, with local areas having Bothriochloa ischaemum grassland distribution. As the elevation rises, scrub gradually appears, and forests appear on the shady slopes above 1400 m above sea level, and grasslands dominate on the sunny slopes. Betula platyphylla forests occur at altitudes of 1600-2000 m and with increasing altitude are gradually replaced by Picea crassifolia forests, Picea wilsonii forests, and Picea meyeri forests. The central part of the Yinshan Mountains is located in the desert steppe area, and the vegetation cover is slightly lower than that of the east, with medicinal plants such as Artemisia frigida and Iris tenuifolia. The baseband mainly consists of Stipa klemenzii grassland and Stipa breviflora grassland, and with increasing altitude, sunny slope shrubs dominate. Platycladus orientalis forests dominate between 1200-1500 m above sea level, and as the elevation continues to rise, Pinus tabuliformis forests, Juniperus rigida forests, and Betula platyphylla forests appear. The western part of the Yinshan Mountains has a low vegetation cover with medicinal plants such as Convolvulus tragacanthoides and Prunus mongolica. The baseband mainly includes, for example, Convolvulus tragacanthoides desert and Reaumuria songarica desert. The medicinal plants include, for example, Convolvulus tragacanthoides, Zygophyllum xanthoxylum, and Reaumuria songarica. As the altitude continues to rise, Stipa klemenzii grassland and Lonicera microphylla shrub appear. This difference may limit the development of MPD in the western Yinshan Mountains. Regarding climatic conditions, the Yinshan Mountains are the boundary between monsoon and non-monsoon regions, and parts of the western and central regions are non-monsoon regions. Although part of the Yellow River flows through this area, there remains a lack of precipitation and other conditions suitable for plant growth. From the perspective of human factors, the ecological damage in the western region is serious, which may lead to a decrease in MPD.

In summary, the MPD of the Yinshan Mountains is subject to a combination of ecological factors such as vegetation types, climatic conditions, and human activities. Over time, changes in climatic conditions and human activities, among other factors, can lead to varying degrees of MPD excursions. Therefore, relevant departments should establish nature reserves and dynamic monitoring stations to protect MPD. Meanwhile, the policy of returning farmland to forest and grassland should be implemented to improve the ecological environment and reduce the influence of human factors.