An Analysis of Spatial Variation in Human Impact on Forest Ecological Functions

Abstract

As the cornerstone of terrestrial ecosystems, forests have faced mounting challenges due to escalating human activities, jeopardizing their vital ecological functions and even their existence. It has become an important issue to explore how to promote harmonious coexistence of man and nature, or even to improve the forest ecological function (FEF) through human activities. Thus, in this study, we select the Yellow River Basin (YRB) in China as a typical region. Firstly, we assess the FEF at the county level and reveal their spatial distribution and agglomeration characteristics on the basis of the data from the Ninth National Forest Inventory of China. Then, using multiple linear regression (MLR) and geographically weighted regression (GWR) modeling, we further explore the overall impacts of different human activities on FEF and their spatial differences, respectively. Our findings underscored a moderate deficiency in the county-level FEF in the YRB, with pronounced positive spatial agglomerations. The high–high areas are primarily clustered in the southern and central mountainous areas, whereas low–low areas are distributed in the upstream warm temperate steppe and desert-grassland regions. Human activities exert substantial impacts on FEF, with distinct spatial heterogeneity in the coefficient and significance levels. The trend analysis indicates that FEF is more sensitive to the increase in living land, population density and forest protection in the east–west direction. And in the north–south direction, FEF is more easily affected by agricultural development, population growth and urbanization. This study verifies that natural factors dominate FEF in those regions where human activities are quite scarce, and also reveals that due to the inter-constraint or counteract effects among different human activities, FEF may still ultimately depend on the natural endowments in some populated regions. We point out the core human activity factors affecting FEF after excluding the interference from natural conditions. And we recommend that policymakers prioritize sustainable development strategies that mitigate the adverse impacts of human activities on forest ecosystems while promoting conservation efforts tailored to the unique characteristics of each region.

1. Introduction

Forests are the most important terrestrial ecosystem and are irreplaceable in preserving the human living environment and improving the climate conditions [1]. Forest ecological function (FEF) is a comprehensive concept, which refers to the ecological environment conditions and utility formed by the forest ecosystem and its ecological processes on which human beings rely for survival [2]. Specifically, the ecological service value in carbon sequestration [3], water retention [4], soil erosion control [5], climate regulation, environmental purification, and biodiversity protection [6] have made forests attract a great deal of attention. In recent years, forests face escalating pressure of ecosystem degradation due to intensified human activities associated with urbanization and industrialization, as well as climate change [7,8,9,10]. In response, countries such as China have initiated ecological projects to create artificial forests, targeting both forest coverage expansion and sustainable resource provisioning [11]. According to the Forestry and Grassland Ecology Monitoring and Assessment Report in 2021, China contributed 5.51% to global forest coverage (fifth worldwide) and held a growing stock of 19.493 billion m³, ranking sixth in the world [12]. However, despite its significant total volume, challenges persist in ecological efficiency. The forest ecological function index is 0.57, indicating an overall moderate level that still lags behind the global average [13].

In order to grasp the current status of forest resources more comprehensively, China has established a continuous forest inventory system since the 1970s, and the concept of the forest ecological function index (FEFI) was introduced in the Technical Provisions on the Continuous National Forest Inventory issued in 2004 [14]. FEFI includes factors such as forest productivity, naturalness, health, and the ratio of forest stand area to national land area, providing a comprehensive evaluation of FEF [2,15]. Modern technologies such as GIS and WEBGIS are used to construct a system allowing for a swift and comprehensive evaluation of FEF. More specifically, ecosystem service assessment models, represented by the InVEST model, have been widely used to quantify different functions of forest ecosystems, such as carbon storage, water retention, and erosion control [16]. Inspired by the Gross Ecosystem Product (GEP), the Gross Forest Ecological Economic Product (GFEEP) framework was constructed to evaluate the economic value of forest ecological functions [17].

Researchers have long been concerned with the impacts of natural factors and human activities on forest ecosystems. Climatic changes such as temperature, precipitation, and duration of sunshine can strongly affect the productivity and resilience of the global forest ecosystem [18,19,20]. Anthropogenic activities have often wielded greater influence on forest ecosystems than climate change in recent years [21]. Accelerated economic growth, urban land expansion, and population explosion have caused severe damage to surface vegetation [22]. The huge demand for agroforestry products has led to the conversion of forests into arable land or alternative uses, thereby diminishing ecological functions [23]. Despite this, it is undeniable that economic development helps to optimize the forestry production structure and minimize the waste of forest resources in certain areas [24]. Additionally, forest policies and laws worldwide have led to significant expansions of forest coverage. In China and India, large-scale ecological restoration policies contributed significantly to maintaining and improving forest ecological functions [11].

Despite obvious progress in the understanding of FEF, several limitations persist within the existing literature. Firstly, assessments of integrated forest ecological services often rely on data from statistical reports, lacking a more intuitive evaluation grounded in sample plot data. Secondly, previous studies primarily concentrated on examining forests in larger regions or water basins rather than at the county level, where the major policies were implemented and human activities were carried out. Arguably, an accurate evaluation of FEF at the county level should be a prerequisite for the formulation of effective forestry management measures and policies. Finally, more attention has been paid to the impact of natural factors on human activities. Given the difficulty in changing natural factors in the short term, our focus is on investigating how to better regulate human activities to improve FEF through more practical ways.

In this study, we take the Yellow River Basin (YRB) in China as a typical study area, and assess the county-level FEF for the first time using the data from the Ninth National Forest Inventory. We further analyze the characteristics of its spatial distribution and agglomeration. Using the multiple linear regression and geographically weighted regression modeling, we test how human activities impact the FEF and investigate their spatial variations. After controlling the interference from natural factors, we identify the various core human activity factors affecting FEF by dividing physical geographic areas. In the end, we put forward targeted recommendations to improve the county-level forest ecological functions.

2. Materials and Methods

2.1. Study Area

The Yellow River (located between 95°53′~119°05′ E and 32°10′~41°50′ N) originates in the northern foothills of Bayan Har Mountain. It traverses the Qinghai-Tibet Plateau, the Inner Mongolia Plateau, the Loess Plateau and the North China Plain form west to east, with a total length of 5464 km [25]. The Yellow River basin (YRB) covers a total area of 795,000 km², which exhibits almost all the typical climatic zones and topographic landscapes of northern China. Based on the vegetation hierarchy system formulated by the National Climate Division and Vegetation Map of China, and referring to the multi-year average temperature and precipitation and information of soil zoning, the YRB is divided into eight physical geographic regions [26]. As shown in Figure 1, I is the sub-frigid semi-arid meadow steppe region in Qinghai Plateau; II is the sub-frigid semi-humid scrub and meadow region in the Zoige Plateau and surrounding mountains; III is the temperate humid and sub-humid alpine deep valley coniferous forest region in West Sichuan and East Tibet Plateau; IV is coniferous forest region and grassland region in the temperate semi-arid alpine basin in Qilian Qingdong Plateau; V is the warm temperate steppe and desert-grassland region in the Yellow River loop and Ordos Plateau; VI is the warm temperate semi-arid grassland region in the northwest of Loess Plateau and semi-humid forest region in the southeast; VII is the warm temperate humid and sub-humid forest region in the mountains and plains of lower Yellow River; and VIII is the region of warm temperate semi-humid saline meadow of the Yellow River delta [26].

The YRB is taken as the study area for this paper based on the following considerations: first, the YRB is an ecological corridor connecting the east and west of China, and it holds paramount importance in wind-breaking, sand-fixing, and ecological protection. There has been a consistent concern about its direct impact on the evolution trend in ecological security and environmental quality in the medium to long term [27,28]. And the Chinese government positioned the ecological protection and high-quality development of the YRB as a prime national strategy since 2019 [29]. Second, the YRB has been one of the regions most seriously affected by human activities in the world [30]. Covering 448 counties in nine provinces, the YRB plays a crucial role in China’s economic development and food security [31]. And considering that stepped social and economic development levels are observed across the upper, middle, and lower reaches of the YRB, the findings of this study will be potentially universal to other countries and regions.

2.2. Variables and Data

2.2.1. Forest Ecological Function

Forest ecological function (FEF) is the core variable we would like to investigate in this study, which is commonly quantified by forest ecological function index (FEFI) [32]. We list the eight factors and their weights to calculate FEFI in Table 1, including forest volume, naturalness, community structure, tree species structure, total vegetation coverage, canopy density, mean tree height, and litter thickness.

The calculating equation for county-level FEFI can be expressed as [32]:

(1)$FEFI={\displaystyle {\sum }_k^n\frac{s_k}{S{\displaystyle \sum _{i=1}^8{\omega }_i{x}_{ki}}}}$

where the county-level FEFI is obtained by the weighted average method. Suppose there is a total number of $n$ sample plots surveyed in a certain county, $s_k$ represents the area of sample plot $k$ and $S$ is the sum area of $n$ sample plots; $x_{ki}$ is the value of factor $i$ of sample plot $k$ and ${\omega }_i$ is the weight shown in Table 1. Within the range of [0, 1], a higher FEFI means a better forest ecological function.

2.2.2. Influencing Factors of Forest Ecological Function

Accelerated economic development, urban expansion, and population booms can directly or indirectly damage the surface vegetation [22]. Population pressure stimulates the demand for agricultural and forestry products, which may initiate changes in land-use types, such as the conversion from woodland or grassland to farmland or other use, leading to a variation in vegetation cover and forest community structure [23,33]. Ecological projects have contributed substantially to maintaining the forest ecological functions [34,35]. Additionally, the long-term variations in FEF are interlinked with natural factors such as climate change that exert direct impacts on forest growth, cover and types [36,37]. The influence of temperature and precipitation is also determined by altitude, which varies across the study area. Therefore, we selected eight indicators that effectively capture human activities and their impacts on FEF. We also added control variables to alleviate the problems of missing variables of natural factors (Table 2).

2.2.3. Data Sources

In this study, we initially obtain sample plot-level data for calculating the forest ecological function index from the 9th National Forest Resources Inventory of China. Data pertaining to elevation, slope, soil and forest land protection at the sample plot scale are also sourced from this inventory. The sample plot is arranged in a 1 km × 1 km grid system (aligned with the spatial resolution of Landsat imagery). Each sample plot generally covers 0.0667 hm², with the fixed number of plots per county determined based on existing forest area, forest stock volume data, and precision requirements. There are a total of 114,887 sample plots in the 448 counties of the YRB, and they are aggregated to the county level through an area-weighted averaging approach. The land use data are acquired from the Chinese Academy of Sciences (CAS) Resource and Environmental Sciences and Data (RESD). We process the land use remote sensing monitoring data with a resolution of 1 km for 2018 using ArcGIS software. The data concerning human activities are obtained from the China County Statistical Yearbook, statistical bulletins of economic and social development for county-level areas, and statistical yearbooks of prefecture-level cities. Climate data are retrieved from the World Climate Database (https://worldclim.org, accessed on 10 December 2024).

2.3. Econometric Methods

2.3.1. Spatial Correlation Analysis

The global Moran’s I index and local Moran’s I index describe the global and local cluster characteristics of the FEF, respectively. They are expressed as follows [39]:

(2)$Moran’sI={\displaystyle \frac{n{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n(x_i-\overline{x})(x_j-\overline{x})}}}{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{W}_{ij}{\displaystyle {\sum }_{i=1}^n(x_i-\overline{x})}}}}}$

(3)$LocalMoran’sI=({\displaystyle \frac{x_i-\overline{x}}{s^2}}){\displaystyle {\sum }_{i=1}^n{W}_{ij}}(x_j-\overline{x})$

where $x$ represents the FEFI; $n$ is the number of counties; $\overline{x}$ is the average value of all counties; $W_{ij}$ is the spatial weight matrix; and $s^2$ is the sample variance.

Moran’s I index ranges from −1 to 1. The index approaching 1 signifies a positive autocorrelation and FEF in adjacent areas are similar, while a Moran’s I value nearing −1 denotes negative spatial autocorrelation. If the spatial distribution of FEF is absolutely random, the Moran’s I index approximates 0. As for the local Moran’s I index, a positive value indicates that counties with high (low) FEF are also surrounded by adjacent counties with high (low) values. On the contrary, a negative value indicates that counties with high (low) FEF are encircled by neighboring counties with low (high) values. Based on the local Moran’s I indices, the Local Indicators of Spatial Association (LISA) map is constructed and shows four clustering patterns: high–high (H-H), low–low (L-L), high–low (H-L), and low–high (L-H) [40]. In this study, a significant spatial correlation of FEF is the prerequisite for applying the geographically weighted regression (GWR) model.

2.3.2. Multiple Linear Regression Model

The multiple linear regression (MLR) model is used to quantify the impacts of human activities on FEF; the formula is as follows:

(4)$y_i={\beta }_0+{\displaystyle {\sum }_{j=1}^n{\beta }_j}{x}_{ij}+{\epsilon }_i$

where $y_i$ represents the FEFI of county $i$; $x_{ij}$ is the value of dependent variable $j$ in county $i$; ${\epsilon }_i$ is the error term; and $\beta$ is the coefficient to be estimated. The ordinary least square method is most commonly used to estimate the parameters in MLR.

2.3.3. Geographically Weighted Regression Model

The GWR model is unique in that it considers spatial non-stationarity, which leads to variable relationships that differ by location [41]. The formula is as follows:

(5)$y_i={\beta }_0(u_i,v_i)+{\displaystyle {\sum }_{i=1}^k{\beta }_j(u_i,v_i)}{x}_{ji}+e_i$

where $y_i$ denotes the FEFI of county $i$; $x_{ji}$ represents the human activity indicator; $k$ is the sample size; $e_i$ is the random error term; $(u_i,v_i)$ is the spatial location of county $i$; ${\beta }_0(u_i,v_i)$ is the intercept at location $i$; and ${\beta }_j(u_i,v_i)$ is the local estimated coefficient of variable $x_{ij}$.

We employ the Gaussian function to determine the weight and the Akaike information criterion (AIC) method to determine the optimal bandwidth [42]. Moreover, the adjusted coefficient of determination ${\mathrm{R}}_{\mathrm{adj}}^2$ and AIC are used to compare and confirm the accuracy of MLR and GWR models in this study [42]. Generally speaking, the model with a higher ${\mathrm{R}}_{\mathrm{adj}}^2$ and a lower AIC has a better fit for the data, and if the difference between AIC values is greater than 3, there is a significant difference between the model and the actual data.

3. Results

3.1. FEF Evaluation and Its Spatial Variation

3.1.1. Assessment of Forest Ecological Functions

In order to analyze the spatial distribution characteristics of the county-level forest ecological function in the study area, ArcGIS 10.8 software is used to perform visualization. According to the calculated indices, they are classified into five grades with natural breaks classification at first, as is shown in Figure 2a. Meanwhile, with reference to the classification criteria developed by the Technical Regulations for Continuous Forest Inventory (GB/T38590−2020) [32], FEF can be categorized into three grades: “Good” [0.667,1], “Moderate” (0.4,0.667), and “Poor” [0–0.4], which is demonstrated in Figure 2b.

We find that the overall ecological function of county-level forests in the YRB registers at a moderately low level, with an average score of only 0.438, and there are significant disparities in their spatial distributions. Overall, the FEF in the southern mountains is better than that in the northern plateau and desert, while the FEF in the central areas is superior to that in the western alpine zone and eastern plains. Counties with relatively better FEF occupy an extremely limited portion, and they are distributed mainly in the semi-moist areas in the middle reaches, especially along the Qinling, Ziwuling and Lvliang mountains. The counties with “Moderate” FEF fall into two subcategories: those abundant in natural forest resources, such as counties along the mountains, and those in river plain areas with rich arable land resources. The low-value area mainly clusters in the north-central YRB, such as the Loess Plateau and the Ordos Plateau, where there is an arid climate, limited precipitation, and, consequently, low vegetation coverage. There are also some counties with poor FEF scattered in the Guanzhong Plain and North China Plain, which are major suppliers of agricultural products in China.

3.1.2. Spatial Agglomeration Characteristics of FEF

By comparing the Moran’s I values and their statistical significance (p-values) of different spatial weight matrices, we ultimately selected the Rook contiguity matrix as the fundamental weight matrix for spatial autocorrelation analysis. It turns out that the global Moran’s I value is 0.561 with a robust statistical significance (p < 0.001), revealing a positive correlation of forest ecological functions in adjacent counties—those with high or low FEFIs tend to cluster together. Figure 2c depicts the spatial clustering patterns of forest ecological functions among counties. H-H areas, characterized by an aggregation of good FEF, occupy 13.62% of the total and are primarily situated in the south and middle of the YRB along the Qinling Mountains. In contrast, L-L areas are clustered in the upstream warm temperate steppe and desert-grassland regions of Ningxia Plain, Hetao Plain, and Ordos Plateau. H-L areas and L-H areas are scattered randomly throughout regions with no significant spatial patterns. It is reasonable to assume that the forest’s ecological function has formed a “point-to-surface” development pattern, manifested in homogeneous spillover. That is, counties with high values benefit from the growth of forest ecological function in the surrounding counties, while those with low values inhibit forest ecological function improvement in the surrounding counties.

3.2. Impact of Human Activities on FEF

A multiple covariance test is conducted with the help of IBM SPSS Statistics version 27.0 to avoid pseudo-regression. The results show that the variance inflation factor (VIF) of each variable is less than 5 (Table 3), indicating that the variables are independent of each other and meet the regression requirements. We also standardize all the variables to ensure the robustness of the model and the credibility of the results. After that, we apply MLR and GWR models to regress the influencing factors of forest ecological functions, respectively. The results of regression and the model fitting are shown in Table 3.

The results of the MLR regression show the overall impact of human activities on the FEF. Figure 3 exhibits that residual values are located around the diagonal line, indicating that residual values in this model follow a normal distribution, which displays the correctness of the performed regression. It can be seen that three of the eight human activity factors we selected have significant effects on FEF. Economic and agricultural development, growth in population density, and accelerated urbanization mostly inhibit FEF, whereas the development of secondary industry, the increasing proportion of productive land and living land and forest protection contribute to FEF. Since the GWR regression shows a higher value of ${\mathrm{R}}_{\mathrm{adj}}^2$ (0.692) and a smaller AIC (−797.521) than MLR regression, we believe that the GWR model better contributes to elucidating spatial variations in the influencing factors we analyzed. The local R² values can be seen in Appendix A Figure A1, which shows the GWR model fits better than OLS model in most of the counties. Therefore, the results of GWR regression will act as the basis for the following discussion.

3.3. Spatial Heterogeneity of Human Impacts on FEF

In order to compare the spatial differences between the driving factors, we calculate the mean regression coefficients in different physical geographic regions and visualize them using ArcGIS 10.8. Similarly, the coefficients are all classified into five grades with natural breaks classification, as is shown in Figure 4.

We find that human activities exert a substantial impact on FEF, and the coefficients and significance levels of each factor show marked spatial heterogeneity. After controlling the interference from natural factors, the core human activity factors affecting FEF are caught in five of the eight natural geographic regions, as shown in Table 4. The results and discussion are carried out according to physical geographic subdivisions in the following section.

The combined analysis of Figure 4 and Table 4 shows that in the source area of YRB (Region I), the expansion of productive land (PDUC) is the major inhibitor of FEF. Economic development (EDS) exhibits both positive and negative impacts on FEF, while other human activities mentioned contribute positively to the FEF. In Region II, occupied mainly by the Qinghai Lake and the Gonghe Basins, economic development, especially agricultural growth (AGRI), and encroachments on ecological land are detrimental to the FEF, which can be improved by population agglomeration (PDS) and forest protection (PROT). The development of the secondary sector (INDS) and the urbanization of permanent residents (URB) reveal both positive and negative impacts on FEF.

Regarding Region III, the wetlands of West Sichuan and East Tibet Plateau, EDS, AGRI, PDUC and LIVE are significantly constraining the FEF. We find that changes in FEF are very sensitive to the above activities compared to others. Region IV is the most livable in the upper reaches of the Yellow River, where the more densely populated counties have better FEF. The protection of forests also greatly benefits the improvement of FEF. All the activities aimed at economic development hinder the enhancement of FEF, and the negative effect of AGRI is most pronounced in the main agricultural and pastoral areas.

In Region V, an arid and semi-arid area with scarce water resources, where vegetation growth depends on the groundwater, economic growth driven by primary and secondary sector development (AGRI and INDS), PDS, and URB significantly increase the amount of water consumption for production and domestic use, making water resources scarcer, which in turn has a negative impact on FEF. Conversely, PDUC, LIVE, and PROT mainly contribute to FEF but not very much, with more significant effects of PDUC and LIVE. We also find heterogeneity in the effects of PDUC and PROT in this region. Region VI can be subdivided into semi-humid forest in the southeast and semi-arid grassland in the northwest of the Loess Plateau, and there are huge differences in vegetation cover types. This is why the impact of the same human activity also varies considerably from counties. Overall, EDS, AGRI, PDS, URB, and PDUC have significant negative influence on FEF, while INDS, LIVE, and PROT positively affect FEF remarkably. Increasing the proportion of living land and strengthening the protection of forest land can lead to better FEF in the area.

Region VII is the most important agricultural producing area in China, where densely inhabited hilly plain areas witnesses adverse impacts from EDS and AGRI on FEF. INDS, URB, PDUC, and LIVE appear to be positively correlated with FEF, even though we have not found a significant direct correlation between them. The alluvial plains of the Yellow River Delta (Region VIII) have experienced improved FEF through population agglomeration and urbanization, as well as an increase in the proportion of productive land. Meanwhile, FEF may deteriorate with the ecological land being encroached on by the expansion of living space.

3.4. Spatial Trends in the Human Impact on FEF

In order to explore the spatial evolution of the factors influencing forest ecological function and to obtain their global trends, we used the trend analysis tool in Arc GIS 10.8 to draw a three-dimensional perspective map. As is shown in Figure 5, the X-axis curve indicates the longitudinal trend, and the Y-axis curve indicates the latitudinal trend.

In the east–west direction (X direction), the influence of LIVE and PROT increases linearly from west to east. FEF has the highest sensitivity to changes in LIVE, whose extent is much higher in the east than in the west. The influence of AGRI, PDS, URB, and PDUC on FEF is distributed in a U-shape. There is significant spatial differentiation in the impacts of AGRI, URB, and PDS, with AGRI and URB having greater impacts on the central region and PDS on the western counties. Additionally, the effects of PDUC are not significantly different from west to east; neither are those of EDS. In contrast, INDS has an inverted U-shaped influence on FEF. We also find that the impacts of the natural factors are all inverted U-shaped with significant spatial variations. PREC plays an important role in the central region, while TEMP, ELVE, and SOIL have higher impacts in the east of YRB.

In the north–south direction (Y direction), the influence of EDS, AGRI, URB PDUC, and PROT on FEF shows an inverted U-shape. There is a significant spatial differentiation in the effect of LIVE. EDS, PDUC, and LIVE have a more remarkable influence in the north. Factors such as INDS, PDS, and PROT show a U-shaped impact on FEF, with PDS and PROT having greater influence in the north and INDS in the south. The degree of influence of AGRI on FEF decreases from north to south and is more sensitive to differences between north and south regions. As for natural factors, the effects of TEMP and ELEV are declining from north to south, both of which are much more pronounced in the north. The opposite is true for PREC and SLOP. The effect of SOIL is inverted U-shaped and increases from north to south.

In general, the influence of human activity factors on FEF is more evident in the east–west direction than in the north–south direction, with LIVE, PDS, and PROT on FEF being the most sensitive in the east–west direction and AGRI, PDS, and URB being the most sensitive in the north–south direction. The degree of influence of natural factors varies considerably in both directions.

4. Discussion

4.1. Comparative Analysis of Core Human Activities Affecting FEF

The long-term variations in forest ecological functions were influenced by both natural and human activity factors, as well as their interactions. The direct impact of natural factors on forest ecological functions was evident in the wide variability of vegetation types, forest biomass, community structure, and naturalness concerning temperature, precipitation, and elevation. Additionally, natural factors also significantly shaped the scope and intensity of human activities. For example, while afforestation was commonly used to rehabilitate and expand forest cover, the regeneration and maintenance of artificial forest vegetation depended on adequate water resources. In China, greening initiatives had not surpassed the 400 mm precipitation threshold [43], and the lack of water resources post-afforestation posed a risk to the existing vegetation.

We find the positive effects of temperature, precipitation, and elevation on FEF in the entire YRB. Among them, the improvement of temperature is most significant in regions VI and VII, and precipitation is crucial in the middle and upper reaches. This is consistent with the findings that precipitation has the greatest effect on vegetation cover in semi-arid regions, while air temperature in semi-humid regions [44]. The higher the altitude of a region, the less interference it receives from human activities and, therefore, the better the FEF.

We also observe that natural factors have greater impacts on FEF than human activities in regions I, II, and VII. The fragility of forests tends to reduce with increasing elevation [45]. Since the challenging environmental conditions at high altitudes do not readily support a large human population, the overall impact of anthropogenic activities is greatly diminished [46]. That is why changes in FEF here are more attributable to climate rather than human factors [47]. We are sure that maintaining the natural state of vegetation and minimizing human activities to the greatest extent are beneficial to the FEF in Region Ⅰ. Due to the economic backwardness and sparse population, primitive economic activities such as grazing and farming have led to the degradation of alpine meadows [48]. Excessive nibbling also slowed the regeneration and renewal of pasture grasses and reduced vegetation cover [49]. Region II has become one of the areas of serious land degradation in the semi-arid steppe region of northern China. Considering that the traditional economic development mode through agricultural sprawl is unsuitable for this region, we insist that the first imperative is to promote economic growth through new urbanization and leave more space for nature to achieve self-renewal.

When it comes to Region VII, the complex terrain of the western mountainous area protects it from disturbance by human activities [33]. The abundant resource of forests, shrubs and herbs facilitates better FEF [50]. By contrast, economic activities such as farming and mining occur mainly in the densely inhabited hilly plain areas in the central and eastern part of this region. Meanwhile, ecological projects led by the government, for example, conceding farmland to forests and afforestation in barren mountains, compensates for some of the damages to FEF caused by human activities to some extent [51]. We find that as the impacts of different human activities canceled each other out, we can only capture the significant influence of natural factors on FEF, although human activities are very frequent in this area. Moreover, despite the severe uneven distribution as well as reduced precipitation in recent years, the effect of precipitation on FEF is not significant due to improved irrigation conditions. This seems to be evidence that human activities can indirectly affect FEF by influencing natural factors.

In addition to the three physical geographic regions mentioned above, we identify core human factors affecting FEF in all other regions, and each will be discussed next.

As one of the national key ecological function areas with restricted development, the area of red lines for ecological protection in Region III is close to 50% [52]. Although counties attempt to better balance economic and ecological benefits, the chronically negative impacts of overloaded grazing and indiscriminate mining on the ecological environment caused by rapid population growth have not diminished [53]. A series of key ecological projects such as “Natural forest protection”, “Sand control in northwest Sichuan”, “Subsidy incentives for grassland ecological protection”, and “Ecological compensation for wetlands” have not yet fulfilled their positive roles in improving FEF, indicating that protection is more important than treatment in this region. It is wise to severely restrict or prohibit deforestation, reclamation of wetlands and grasslands, and over-grazing for a considerable period of time.

The obvious location and resource superiority make Region IV an important center of agricultural and husbandry production in the upper reaches. It occupies 60% of the arable land and is inhabited by more than two thirds of the population in Qinghai province. The growth of agriculture has always exerted pressure on ecological water use and inhabited the FEF in the long term. Six national parks and three national nature reserves have been designated in Hehuang Valley, playing a powerful role in maintaining the local ecosystem. Moreover, we should still be alert to the ecological threats posed by economic development. The vegetation cover and habitat quality of Lanzhou and Xining have deteriorated and radiated to the periphery [54], and greater challenges on FEF may come with the construction of urban clusters centering these two cities.

Although the vegetation cover increases through the extension of the cultivated area and afforestation in the irrigation district and desert [55], water-intensive agriculture may also be detrimental to FEF by intensifying conflict for water and land with existing forest and grass, as demonstrated in Region V. The abundant and various mineral deposits provide a foundation for mining industries, but the increase in industrial and mining land has led to the loss of natural ecosystems [56]. Recent urbanization has increasingly been dedicated to greening and livability with the labor force and population agglomeration [9,57]. Expanding artificial vegetation through construction emerged as an effective way to improve forest ecological functions in the region. Expanding artificial vegetation through construction emerges as an effective approach to ecological function restoration.

The vegetation types in Region VI transition from warm–temperate deciduous broad-leaved forests, forest-steppe, typical grassland to desert-steppe from south to north [35]. The south plain is the most populated and industrially concentrated area, and its rapid development requires more construction lands converted from arable land or other natural land, which directly leads to vegetation degradation [58]. The Chinese government has realized early that the Loess Plateau is critical in biodiversity protection and soil conservation. Consequently, since 2000, this region has become the most remarkable ecological restoration area and benefited the most from a series of ecological projects in China.

The saline–alkali Yellow River alluvial plains have experienced increased vegetation coverage through saline–alkaline land improvement and seed industry innovation [59]. Industrial mining development, coupled with population growth and urban agglomeration, positively impacted economic growth and forest ecological functions [60]. In this process, a vast expanse of natural lands have been occupied or transformed into artificial ones, resulting in continuous degradation of ecological functions [61].

4.2. Robust Test

In order to verify the robustness of the baseline regression results above, and considering that the impact of human activities on forest ecological functions may exhibit a time lag, this study adopts the average values of socioeconomic indicators from 2013 to 2017 as independent variables for re-estimation. The results in Table 5 demonstrate that the two sets of human activity data exhibit highly consistent directionality and magnitude in their impact on FEF. Table 6 further confirms that the regression results remain largely consistent across different natural geographical regions. This provides strong evidence for the robustness and reliability of our findings.

4.3. Limitations

This study proves that it is a worthwhile endeavor to evaluate the forest ecological function on the basis of administrative regions. This work contributes to a more systematic understanding of interactions between human and nature, and it should be an indispensable foundation for all the forest-related discussions.

We also gain some interesting findings through exploring the influence of human activities on forest ecological function. It is well-known that natural factors have greater impacts on forest ecological function than human activities in untrod areas of high altitude. Surprisingly, this study confirms that in such regions with considerable human activities, the level of forest ecological function may still ultimately depend on natural factors. This is due to the fact that the positive or negative effects of different human activities on forest systems constrain or counteract each other, but the final effect of the counteracting does not spatially flip the natural endowment. This is true at least in the Yellow River basin of China. It is regrettable that this study does not capture changes in forest ecological functions over time, and thus cannot clarify whether the human activities facilitate or inhibit forest ecological function in the end. For the time being, we can only suggest that local governments encourage those human activities that can improve local forest ecological functions. We need to supplement relevant studies on time series in the future.

Further refinements for future research can also be made from the following perspectives. First, the indicator selection for natural factors can be more detailed. In addition to temperature, precipitation, and elevation, the soil type, slope direction, and slope gradient may also affect forest ecological function. Second, the portrayal of forest ecological function in this paper is rather general and does not focus on certain ecological service functions, such as carbon sequestration and water conservation, even if these are hot topics in current research. We will also expand the related contents in our future research.

5. Conclusions and Implications

Taking the Yellow River Basin (YRB) as a typical area, we use multiple linear regression (MLR) and geographically weighted regression (GWR) modeling to compare and analyze the impacts of human activities on forest ecological function (FEF) in county-level administrative areas in China. The main findings are as follows:

First, the county-level FEF in the YRB is poor overall but with significant regional differences, showing an obvious positive spatial agglomeration. The agglomeration manifests predominantly as high–high and low–low clusters, with high–high areas primarily clustered in the southern and central mountainous areas, whereas low–low areas are distributed in the upstream warm temperate steppe and desert-grassland regions.

Second, human activities exert a substantial impact on forest ecological function, and the coefficient and significance level of each factor show marked spatial heterogeneity. Trend analysis indicates that FEF is more sensitive to the increase in living land, population density, and forest protection in the east–west direction. And in the north–south direction, FEF is more easily affected by agricultural development, population growth and urbanization.

Third, it is recognized that FEF is more dependent on natural factors in areas where human activities are scarce. We find that in some regions with considerable human activities, FEF may still ultimately depend on the natural endowments, which is due to the inter-constraint or counteract effects among different human activities. In regions with more natural forest resources, economic development, especially agricultural production, has significant dampening effects on FEF, as does the encroachment on ecological land by irrational expansion of productive and living lands. In the water-scarce counties, population clustering and forest protection are the most efficient for enhancing FEF. As for the Yellow River Delta with salinized soil, FEF can be improved through accelerated urbanization.

Accordingly, we propose the following recommendations for bettering the forest ecological functions:

First, we should take full account of the limitations of climatic conditions. We should reduce human disturbances and promote natural ecological recovery as much as possible in areas that have not been that developed.

Second, counties should formulate measures to improve FEF in line with local conditions. More efforts could be put into raising the efficiency of available resources in counties with abundant natural endowments. And in less endowed counties, the improvement of FEF should be on the premise of increasing vegetation cover through ecological projects.

Third, counties should articulate their positions in regional ecological protection, considering the possible repercussions of ecological measures on neighboring regions. Leveraging the comprehensive effects of combined ecological measures is imperative, avoiding reliance on single ecological engineering or restoration measures.

Author Contributions

Conceptualization, X.Z.; data curation, L.F.; formal analysis, Q.W. and X.Z.; methodology, Q.W.; software, Q.W.; funding acquisition, X.Z.; writing—original draft, Q.W.; writing—review and editing, Y.D., R.P.S. and X.Z. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data will be made available on request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1 (a) Location of Yellow River basin in China; (b) land-use types in different physical geographic regions. Note: Figure 1b is mapped on the basis of the Globe Land30 land cover remote sensing data for 2020, where land uses are classified into10 types: cropland, forest, shrub, grassland, tundra, artificial surfaces, wetland, water bodies, barren, glacier and perennial snowfield. In order to be more relevant to the topic of this study, the last four types are combined into “others”.

Figure 2 Spatial characteristics of forest ecological functions: (a) distribution of forest ecological function index; (b) classification results; (c) spatial agglomeration patterns.

Figure 3 Regression error probability distribution.

Figure 4 Spatial distribution of mean regression coefficients for each influencing factor. Refer to Table 2 for definitions of the acronyms.

Figure 5 Trend analysis of impact extent of all influencing factors on FEF.

Table 1

Evaluation factors and classification standards of forest ecological function.

Factor	Classification Standard			Weight
	I	II	III
Forest volume	$\ge 150 {\mathrm{t}/\mathrm{hm}}^2$	${50 \mathrm{t}/\mathrm{hm}}^2{~ 149 \mathrm{t}/\mathrm{hm}}^2$	$<{50 \mathrm{t}/\mathrm{hm}}^2$	$0.20$
Forest naturalness	$1,2$	$3,4$	$5$	$0.15$
Forest community structure	$1$	$2$	$3$	$0.15$
Tree species structure	$6,7$	$3,4,5$	$1,2$	$0.15$
Total vegetation coverage	$\ge 70\%$	$50\%~69\%$	$<50\%$	$0.10$
Canopy density	$\ge 0.70$	$0.40~0.69$	$0.20~0.39$	$0.10$
Mean tree height	$\ge 15.0 \mathrm{m}$	$5.0 \mathrm{m}~14.9 \mathrm{m}$	$<5.0 \mathrm{m}$	$0.10$
Litter thickness grade	$1$	$2$	$3$	$0.05$

Note: I, II and III refer to the categories in which each indicator is classified in the raw data. They are assigned with values 3, 2, and 1 when calculating the FEFI of sample plots for ease.

Table 2

Description of variables.

Variable	Symbol	Description	Unit
The level of economic development	EDS	GDP per capita	ten thousand yuan/km2
Development of the primary sector	AGRI	the share of primary sector in the GDP	%
Development of the secondary sector	INDS	the share of secondary sector in the GDP	%
Population density	PDS	the ratio of resident population to county area	thousand people/km2
Rate of urbanization	URB	the urbanization ratio of resident population	%
Production space	PDUC	the ratio of the area of productive land to county area	%
Living space	LIVE	the proportion of the area for living land to county area	%
Forest protection	PROT	level of forest land protection	-
Temperature	TEMP	annual average temperature	°C
Precipitation	PREC	annual average precipitation	mm
Elevation	ELEV	average elevation of the county	km
Slope	SLOP	average slope of the county	°
Soil	SOIL	average soil condition of the county	-

Note: The land use is merged into three types, which are production space, living space, and ecological space. Among them, production space includes the land used for agricultural production as well as industrial and mine production, and living space includes urban and rural living land. The others are summarized as ecological space [38]. There are three levels of forest land protection in China. Grade I areas are specially protected and strictly regulated within critical ecological function zones, designated to safeguard biodiversity and unique natural landscapes; Grade II focuses on ecological restoration, management, and the construction of ecological barriers; and Grade III is dedicated to maintaining regional ecological balance and ensuring the supplies of forest products [32]. Evidently, Grade I represents the highest level of forest land protection. For clarity, this study employs the reciprocal of the protection grade for each sample plot to calculate the PROT index.

Table 3

Results of MLR and GWR models and comparison of their performance.

Variable	MLR	GWR						VIF
		Min	Max	Lwr Quartile	Medium	Upr Quartile	Mean
ECO	−0.103 *	−0.737	0.171	−0.117	−0.073	−0.026	−0.071	3.047
AGRI	−0.174 ***	−0.354	0.119	−0.166	−0.118	−0.073	−0.123	2.903
INDS	0.021	−0.139	0.233	−0.014	0.028	0.055	0.015	2.251
PDS	−0.035	−0.754	1.491	−0.117	−0.079	0.003	−0.063	2.265
URB	−0.012	−0.201	0.314	−0.047	−0.001	0.030	−0.003	4.657
PDUC	0.011 *	−0.862	0.317	−0.038	−0.005	0.073	0.006	2.423
LIVE	0.018	−0.595	0.263	0.043	0.094	0.126	0.075	4.142
PROT	0.031	−0.087	0.355	−0.038	−0.005	0.123	0.047	1.117
TEMP	0.504 ***	−0.801	0.892	0.224	0.368	0.571	0.378	2.029
PREC	0.299 ***	−0.089	0.618	0.161	0.296	0.374	0.270	1.343
ELEV	0.386 *	−1.257	1.777	0.249	0.403	0.932	0.587	1.139
SLOP	0.438 **	−0.037	0.313	0.096	0.194	0.258	0.175	2.313
SOIL	0.000	−0.425	0.145	−0.048	0.021	0.083	0.007	1.328
AIC	−749.032	−797.521						-
${\mathrm{R}}_{\mathrm{adj}}^2$	0.467	0.692						-

Note: ***, **, and * indicate significance at 0.01, 0.05, and 0.1 levels, respectively. Acronyms are defined in Table 2.

Table 4

The core influencing factors on FEF in different physical geographic regions.

Variable	Coefficient
	I	II	III	IV	V	VI	VII	VIII
ECO	0.047	−0.175	−0.631 ***	−0.021	−0.079	−0.068	−0.073	−0.059
AGRI	0.051	−0.133	−0.327 ***	−0.170 **	−0.179 *	−0.108 *	−0.097	−0.021
INDS	0.116	0.008	0.162	−0.094	−0.037	0.044	0.030	0.105
PDS	1.138	0.136	−0.212	−0.282	−0.082	−0.085	0.075	0.579
URB	0.067	0.100	0.246	0.010	−0.083	−0.021	0.056	0.273 **
PDUC	−0.198	−0.086	−0.730 ***	0.050	0.101	−0.033	0.065	0.166 **
LIVE	0.281	−0.186	−0.343	0.026	0.157	0.098 *	0.072	−0.108
PROT	0.122	0.090	0.024	0.211 *	0.058	0.022 *	−0.014	−0.016
TEMP	0.043	−0.016	−0.652 ***	0.246	0.511 **	0.516 ***	0.092	−0.159
PREC	0.315	0.308 **	0.508 ***	0.328 **	0.174	0.347 ***	0.030	−0.015
ELEV	0.020	−0.097	−0.943 ***	0.304 **	0.694 **	0.825 ***	0.204	0.001
SLOP	0.005	0.124	0.231 **	0.165 **	0.107	0.180 *	0.236 **	0.193
SOIL	−0.041	−0.202	−0.397 ***	−0.125	−0.012	0.055	0.034	−0.007

Note: ***, **, and * indicate significance at 0.01, 0.05, and 0.1 levels, respectively (two-tailed). Refer to Table 2 for definitions of the acronyms.

Table 5

Results of robustness test.

Variable	MLR	GWR
		Min	Max	Lwr Quartile	Medium	Upr Quartile	Mean
EDS	−0.102 **	−0.330	0.517	−0.218	−0.138	−0.021	−0.109
AGRI	−0.135 **	−0.484	0.213	−0.147	−0.100	−0.075	−0.114
INDS	0.034	−0.143	0.318	−0.025	0.030	0.062	0.017
PDS	0.046	−2.720	4.899	−0.109	0.071	0.169	−0.072
URB	0.039	−0.481	0.321	−0.045	−0.018	0.019	−0.015
PDUC	0.030	−1.004	0.351	−0.052	−0.012	0.069	0.002
LIVE	−0.028	−0.736	0.301	0.000	0.088	0.145	0.071
PROT	0.006	−0.094	0.336	−0.046	0.027	0.115	0.046
TEMP	0.607 ***	−0.682	0.944	0.296	0.457	0.628	0.440
PREC	0.376 ***	−0.095	0.624	0.192	0.398	0.506	0.345
ELEV	0.612 ***	−1.604	1.799	0.343	0.504	0.951	0.658
SLOP	0.356 ***	−0.107	0.315	0.050	0.158	0.236	0.136
SOIL	0.032	−0.270	0.153	−0.027	0.048	0.104	0.027
AIC	−766.555	−811.911
${\mathrm{R}}_{\mathrm{adj}}^2$	0.488	0.698

Note: *** and ** indicate significance at 0.01 and 0.05 levels, respectively (two-tailed). Refer to Table 2 for definitions of acronyms.

Table 6

Robustness test results for different physical geographic regions.

Variable	Coefficient
	I	II	III	IV	V	VI	VII	VIII
ECO	0.029	0.208	0.382	−0.001	−0.030	−0.196 *	−0.077	0.093
AGRI	0.062	−0.143	−0.445 ***	−0.184 **	−0.127	−0.107 **	−0.065	0.001
INDS	0.102	−0.091	−0.063	−0.087	−0.035	0.056 *	0.031	0.088
PDS	1.686	−1.382	−1.727	−0.787	0.120	0.073	0.122	0.636
URB	0.105	−0.083	−0.386 **	−0.008	−0.070	−0.015	0.039	0.233 **
PDUC	0.193	−0.099	−0.790 ***	0.020	0.129	−0.036	0.055	0.206 **
LIVE	−0.420	−0.017	−0.024	0.035	0.131	0.075	0.092	−0.010
PROT	0.124	0.142	0.258 *	0.130	0.034	0.030 *	0.008	0.032
TEMP	0.006	0.091	−0.557 **	0.285	0.479 **	0.602 ***	0.160	−0.116
PREC	0.363	0.319 **	0.487 ***	0.332 **	0.239 *	0.463 ***	0.059	−0.005
ELEV	−0.233	−0.141	−1.218 ***	0.406 **	0.791 **	0.889 ***	0.302	0.179
SLOP	−0.006	0.090	0.155	0.115 *	0.081	0.138 *	0.202 **	0.199
SOIL	−0.039	−0.126	−0.248 *	−0.113	−0.006	0.074 *	0.058	0.052

Note: ***, **, and * indicate significance at 0.01, 0.05, and 0.1 levels, respectively (two-tailed). Refer to Table 2 for definitions of the acronyms.

Appendix A

Figure A1 A map of the local R² values for each county. Note: Using the OLS global goodness of fit (0.467) as the cutoff point, the model fit goodness of GWR was divided into two categories. It is found that the GWR model fits better in 61.6% of the counties, especially in the entire Region V and counties along the mountain in region VI, where the GWR successfully captures the spatial heterogeneity of human impacts on FEF.

References

1. Keenan, R.J.; Reams, G.A.; Achard, F.; de Freitas, J.V.; Grainger, A.; Lindquist, E. Dynamics of global forest area: Results from the FAO global forest resources assessment. For. Ecol. Manag.; 2015; 352, pp. 9-20. [DOI: https://dx.doi.org/10.1016/j.foreco.2015.06.014]

2. Yuan, Y.; Liu, Z.G.; Dong, L.B. Evaluation and analysis of forest ecological function level in Greater Khingan Mountains Pangu forest farm based on GIS. J. Cent. South Univ. For. Technol.; 2016; 36, pp. 108-114.

3. Zhang, M.N.; Liu, S.R.; Luo, X.Z.; Keenan, T.F.; Fu, L.Y.; Xiao, C.W.; Zhang, Y.; Gong, P. Incorporating site suitability and carbon sequestration of tree species into China’s climate-adaptive forestation. Sci. Bull.; 2025; in press [DOI: https://dx.doi.org/10.1016/j.scib.2025.03.035] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/40190000]

4. Wang, N.; Bi, H.X.; Peng, R.D.; Zhao, D.Y.; Liu, Z.H. Disparities in soil and water conservation functions among different forest types and implications for afforestation on the Loess Plateau. Ecol. Indic.; 2023; 155, 110935. [DOI: https://dx.doi.org/10.1016/j.ecolind.2023.110935]

5. Pandey, R.; Mehta, D.; Kumar, V.; Pradhan, R.P. Quantifying soil erosion and soil organic carbon conservation services in indian forests: A RUSLE-SDR and GIS-based assessment. Ecol. Indic.; 2024; 163, 112086. [DOI: https://dx.doi.org/10.1016/j.ecolind.2024.112086]

6. Zhao, J.Y.; Fu, X.; Sa, N.; Kou, X.Y.; He, X.J.; Zheng, S.N.; Lu, Z.H.; Wu, G.; Sang, W.G. Forest eco-function restoration and its positive effects on biodiversity improvement in China’s ecological conservation programs. Ecol. Eng.; 2025; 212, 107530. [DOI: https://dx.doi.org/10.1016/j.ecoleng.2025.107530]

7. Lecina-Diaz, J.; Senf, C.; Grünig, M.; Rupert Seidl, R. Ecosystem services at risk from disturbance in Europe’s forests. Glob. Change Biol.; 2024; 30, e17242. [DOI: https://dx.doi.org/10.1111/gcb.17242]

8. Repo, A.; Albrich, K.; Jantunen, A.; Aalto, J.; Lehtonen, I.; Honkaniemi, J. Contrasting forest management strategies: Impacts on biodiversity and ecosystem services under changing climate and disturbance regimes. J. Environ. Manag.; 2024; 371, 123124. [DOI: https://dx.doi.org/10.1016/j.jenvman.2024.123124]

9. Zhang, S.Y.; Zhu, H.K.; Zeng, K.; Zhang, Y.W.; Jin, Z.W.; Wang, Y.; Zhang, R.; Jürgen, B.; Liu, M. From city to countryside: Unraveling the long-term complex effects of urbanization on vegetation growth in China. J. Environ. Manag.; 2025; 380, 124975. [DOI: https://dx.doi.org/10.1016/j.jenvman.2025.124975]

10. Shen, Z.; Zhang, Q.; Chen, D.; Singh, V.P. Varying effects of mining development on ecological conditions and groundwater storage in dry region in Inner Mongolia of China. J. Hydrol.; 2021; 597, 125759. [DOI: https://dx.doi.org/10.1016/j.jhydrol.2020.125759]

11. Chen, C.; Park, T.; Wang, X.; Piao, S.; Xu, B.; Chaturvedi, R.K.; Fuchs, R.; Brovkin, V.; Ciais, P.; Fensholt, R. . China and India lead in greening of the world through land-use management. Nat. Sustain.; 2019; 2, pp. 122-129. [DOI: https://dx.doi.org/10.1038/s41893-019-0220-7] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30778399]

12. Zhang, H.R.; Lei, X.D.; Fu, L.Y.; Liu, X.Z.; Zhang, X.H.; Yang, L. The current situation and improvement strategies of forest resource quality in China. Terr. Ecosyst. Conserv.; 2024; 4, pp. 46-54.

13. Jian, N.; Zhao, H.; Liu, M. Driving force and potential of forest resource change: Based on the results of the Ninth National Forest Resources Inventory. J. For. Grassl. Policy; 2022; 2, pp. 64-71.

14. Wang, B. Construction and utilization of continuous inventory system for forest ecology. J. Beijing For. Univ.; 2015; 37, pp. 1-8.

15. Qian, C.; Zhao, Y.; Li, M.; Li, T. Dynamic analyses of forest ecological functions based on fixed plot data of continuous forest resources inventories from 1979 to 2012 in Guangzhou. J. Nanjing For. Univ. (Nat. Sci. Ed.); 2022; 46, pp. 205-212.

16. Zou, W.T.; He, Y.J.; Ye, B.; Zhao, X.D.; Xu, D.Y.; Xiao, R.Q.; Duan, Y.X. Research advances in forest ecosystem services evaluations based on the InVEST model. World For. Res.; 2020; 33, pp. 19-24.

17. Xu, P.Z.; Zhang, Y.D.; Li, Q.W.; Sun, C.W. Evaluation of forest eco-efficiency: A transformation of ecological value quantity perspective. Sci. Total Environ.; 2025; 965, 178612. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2025.178612]

18. Boulanger, Y.; Pascual, P.J. Boreal forests will be more severely affected by projected anthropogenic climate forcing than mixedwood and northern hardwood forests in eastern Canada. J. Landsc. Ecol.; 2021; 36, pp. 1725-1740. [DOI: https://dx.doi.org/10.1007/s10980-021-01241-7]

19. Letelier, L.; Valderrama, A.; Stoll, A.; Garcia-Gonzales, R.; Gonzalez-Rodriguez, A. Patterns of composition, richness and phylogenetic diversity of woody plant communities of Quillaja saponaria Molina (Quillajaceae) in the Chilean sclerophyllous forest. Gayana Bot.; 2017; 74, pp. 57-72. [DOI: https://dx.doi.org/10.4067/S0717-66432017005000105]

20. Vefago, M.B.; da Silva, A.C.; Cuchi, T.; dos Santos, G.N.; Nunes, A.D.; Rodrigues, L.C.; Lima, C.L.; Gross, A.; Kilca, R.D.; Higuchi, P. What explains the variation on the regenerative component dynamics of Araucaria forests in southern Brazil?. Sci. Agric.; 2019; 76, pp. 405-414. [DOI: https://dx.doi.org/10.1590/1678-992x-2017-0304]

21. Danneyrolles, V.; Dupuis, S.; Fortin, G.; Leroyer, M.; de Romer, A.; Terrail, R.; Vellend, M.; Boucher, Y.; Laflamme, J.; Bergeron, Y. . Stronger influence of anthropogenic disturbance than climate change on century-scale compositional changes in northern forests. Nat. Commun.; 2019; 10, 1265. [DOI: https://dx.doi.org/10.1038/s41467-019-09265-z] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30894543]

22. Bryan, B.A.; Gao, L.; Ye, Y.; Sun, X.; Connor, J.D.; Crossman, N.D.; Stafford-Smith, M.; Wu, J.; He, C.; Yu, D. . China’s response to a national land-system sustainability emergency. Nature; 2018; 559, pp. 193-204. [DOI: https://dx.doi.org/10.1038/s41586-018-0280-2] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29995865]

23. Ma, J.; Wang, X. Differences in industrial environmental efficiency of the Yangtze River economic belt and their influencing factors: Based on the super efficiency DEA-Malmquist-Tobit model. J. Hohai Univ.; 2017; 19, pp. 49-54.

24. Xie, H.; Chen, Q.; Lu, F.; Wang, W.; Yao, G.; Yu, J. Spatial-temporal disparities and influencing factors of total-factor green use efficiency of industrial land in China. J. Clean. Prod.; 2019; 207, pp. 1047-1058. [DOI: https://dx.doi.org/10.1016/j.jclepro.2018.10.087]

25. Ding, Y.; Zhang, L.; He, Y.; Cao, S.; Wei, X.; Guo, Y.; Ran, L.; Filonchyk, M. Spatiotemporal evolution of agricultural drought and its attribution under different climate zones and vegetation types in the Yellow River Basin of China. Sci. Total Environ.; 2024; 914, 169687. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2023.169687]

26. Wu, Q.; Zhu, J.; Zhao, X. Effects of human social-economic activities on vegetation suitability in the Yellow River Basin, China. Forests; 2023; 14, 234. [DOI: https://dx.doi.org/10.3390/f14020234]

27. Zhang, W.; Wang, Y.; Li, J.; Hao, Z. Coupling coordination network analysis of ecological protection and high-quality economic development in the Yellow River basin. Ecol. Econ.; 2022; 38, pp. 179-189.

28. Jin, F.; Ma, L.; Xu, D. Environmental stress and optimized path of industrial development in the Yellow River basin. Resour. Sci.; 2020; 42, pp. 127-136. [DOI: https://dx.doi.org/10.18402/resci.2020.01.13]

29. Guo, Z.; Tang, S.; Wang, D.; Zhang, J. Spatial distribution and factors influencing ecological efficiency of the Yellow River basin in China. Front. Environ. Sci.; 2022; 10, 815898. [DOI: https://dx.doi.org/10.3389/fenvs.2022.815898]

30. Fan, J.; Wang, Y.; Wang, Y. High quality regional development research based on geographical units: Discuss on the difference in development conditions and priorities of the Yellow River basin compared to the Yangtze River basin. Econ. Geogr.; 2020; 40, pp. 1-11.

31. Li, T.; Zhang, Q.; Singh, V.P.; Zhao, J.; Song, J.; Sun, S.; Wang, G.; Shen, Z.; Wu, W. Identification of degradation areas of ecological environment and degradation intensity assessment in the Yellow River basin. Front. Earth Sci.; 2022; 10, 922013. [DOI: https://dx.doi.org/10.3389/feart.2022.922013]

32.GB/T 38590−2020 Technical Regulations for Continuous Forest Inventory; Standardization Administration of China, Standards Press of China: Beijing, China, 2020.

33. Wang, L.; Li, N.; Wen, G.; Yang, Y. Vegetation coverage changes and driving forces in the Henan section of the Yellow River basin. Bull. Soil Water Conserv.; 2022; 42, pp. 393-399.

34. Lu, Y.; Zhang, Q.; Li, X.; Yang, S.; Yang, Q. Temporal and spatial variations of vegetation coverage and its response to climate factors in Gansu section of Yellow River basin. Bull. Soil Water Conversat.; 2020; 40, pp. 232-238.

35. Yang, Y.; Zhang, P.; Wu, F.; Zhou, Y.; Song, Y.; Wang, Y.; An, S. The significance and countermeasures of vegetation construction on the Loess Plateau to carbon neutrality. Acta Ecol. Sin.; 2023; 43, pp. 9071-9081.

36. Shi, Y.; Jin, N.; Ma, X.; Wu, B.; He, Q.; Yue, C.; Yu, Q. Attribution of climate and human activities to vegetation change in China using machine learning techniques. Agric. For. Meteorol.; 2020; 294, 108146. [DOI: https://dx.doi.org/10.1016/j.agrformet.2020.108146]

37. Zheng, K.; Wei, J.; Pei, J.; Cheng, H.; Zhang, X.; Huang, F.; Li, F.; Ye, J. Impacts of climate change and human activities on grassland vegetation variation in the Chinese Loess Plateau. Sci. Total Environ.; 2019; 660, pp. 236-244. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2019.01.022]

38. Feng, Y.; Zhai, S.; Jiang, X.; Luo, J.; Yan, P.; Dong, C.; Han, J. Simulating spatio-temporal pattern of producing-living-ecological space and evaluating eco-environmental effects in Zhongyuan urban agglomeration based on FLUS model and shared socioeconomic pathways. Acta Ecol. Sin.; 2023; 43, pp. 8292-8308.

39. Guo, S.; Wang, X.; Wang, N.; Dong, X.; Fu, Y. Evaluation and spatial-temporal variations of ecological comprehensive index along the Yellow River of Shanxi province, China. Chin. J. Applies Ecol.; 2023; 34, pp. 3385-3392.

40. Anselin, L. Local indicators of spatial association—LISA. Geogr. Anal.; 1995; 27, pp. 93-115. [DOI: https://dx.doi.org/10.1111/j.1538-4632.1995.tb00338.x]

41. Brunsdon, C.; Fotheringham, A.S.; Charlton, M.E. Geographically weighted regression: A method for exploring spatial nonstationarity. Geogr. Anal.; 1996; 28, pp. 281-298. [DOI: https://dx.doi.org/10.1111/j.1538-4632.1996.tb00936.x]

42. Zheng, Y.; Zhao, X.; Li, X.; Chen, H.; Li, C.; Zhang, C. Mapping soil organic carbon density via geographically weighted regression with smooth terms: A case study in Shanxi Province. Ecol. Indic.; 2024; 166, 112588. [DOI: https://dx.doi.org/10.1016/j.ecolind.2024.112588]

43. Gao, X.; Sun, M.; Luan, Q.; Zhao, X.; Wang, J.; He, G.; Zhao, Y. The spatial and temporal evolution of the actual evapotranspiration based on the remote sensing method in the Loess Plateau. Sci. Total Environ.; 2020; 708, 135111. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2019.135111] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31810704]

44. Chen, C.; Wang, Y.; Li, Y.; Zhou, S. Vegetation changes and influencing factors in different regions of Yellow River Basin from 1982 to 2015. J. Yangtze River Sci. Res. Inst.; 2020; 39, pp. 56–62+81.

45. Thakur, S.; Negi, V.S.; Pathak, R.; Dhyani, R.; Durgapal, K.; Rawal, R.S. Indicator based integrated vulnerability assessment of community forests in Indian west Himalaya. For. Ecol. Manag.; 2020; 457, 117674. [DOI: https://dx.doi.org/10.1016/j.foreco.2019.117674]

46. Chakraborty, A.; Ghosh, A.; Sachdeva, K.; Joshi, P.K. Characterizing fragmentation trends of the Himalayan forests in the Kumaon region of Uttarakhand, India. Ecol. Inform.; 2017; 38, pp. 95-109. [DOI: https://dx.doi.org/10.1016/j.ecoinf.2016.12.006]

47. Zhang, Y.; Sun, J.; Song, X.; Lu, Y. Revealing the main factors affecting global forest change at distinct altitude gradients. Ecol. Indic.; 2023; 148, 110131. [DOI: https://dx.doi.org/10.1016/j.ecolind.2023.110131]

48. Zhang, T.; Wang, J.; Ye, H.; Lai, W.; Zhang, X. Vulnerability of alpine ecosystems and its response to climate change and human activities. Acta Ecol. Sinica.; 2024; 44, pp. 154-170.

49. Tang, D.; Yang, K.; Zeng, L.; Liu, X.; Xin, C.; Xu, Y. Spatio-temporal differences of enhanced aeolian sand activity in the northeastern Tibetan Plateau over the past 1500 years. Acta Geogr. Sin.; 2023; 78, pp. 2284-2298.

50. Jin, B.; Zeng, Z.; Peng, P.; Wang, S.; Zhang, C.; Yang, R. Tree biomass of evergreen broad-leaf forest in mid-subtropical region of China at three succession stages. Hunan For. Sci. Technol.; 2017; 44, pp. 42-45.

51. Yang, Y.; She, Z.; Song, J.; Zhu, D. Study on vegetation variation characteristics and influencing factors in different geomorphic zones in the Chanbahe River basin since 2000. Ecol. Environ. Sci.; 2022; 31, pp. 224-230.

52. Wang, Y.; Gao, L.; Hu, S.; Bao, L. Countermeasures and recommendations for implementing ecological compensation in Sichuan province of the Yellow River basin. Water Resour. Dev. Res.; 2022; 22, pp. 60-62.

53. Wang, Y.; Chun, W.; Ding, W.; Zhang, Y. Study on the high-quality development of Yellow River basin in Sichuan based on the social-ecological system framework. J. Chengdu Univ. Technol. (Soc. Sci.); 2022; 30, pp. 32-39.

54. Hou, G.; Xu, C.; Li, X.; Qi, B. Spatial pattern of water conservation function in Hehuang Valley. Ecol. Sci.; 2021; 40, pp. 169-176.

55. Zhang, H.; Zhan, C.; Xia, J.; Hu, S.; Ning, L.; Deng, X. Spatial-temporal variations and strip patterns of vegetation in the Inner Mongolia of Yellow River basin. Acta Ecol. Sin.; 2022; 42, pp. 8818-8829.

56. Zeng, X.; Liu, Z.; He, C. Quantifying surface coal-mining patterns to promote regional sustainability in Ordos, Inner Mongolia. Sustainability; 2018; 10, 1135. [DOI: https://dx.doi.org/10.3390/su10041135]

57. Ding, M.; Sun, Z.; Liu, Z.; He, C. Impacts of urban expansion on net primary productivity in the agro-pastoral ecotone in northern China: A case of Hohhot-Baotou-Ordos region. Arid Land Geogr.; 2017; 40, pp. 614-621.

58. Niu, L.; Shao, Q.; Ning, J.; Yang, X.; Liu, S.; Liu, G.; Zhang, X.; Huang, H. Evaluation on the degree and potential of ecological restoration in Loess Plateau. J. Nat. Resour.; 2023; 38, pp. 779-794. [DOI: https://dx.doi.org/10.31497/zrzyxb.20230314]

59. Zuo, Q.; Wu, X.; Shi, J.; Wang, Q.; Liu, Z.; Zhu, A.; Yin, D.; Feng, Q.; Ji, W.; Kang, S. Constraints and coordinated allocation strategies of water and land resources for sustainable use of coastal saline-alkali land in the Yellow River delta. Eng. Sci.; 2023; 25, pp. 169-179. [DOI: https://dx.doi.org/10.15302/J-SSCAE-2023.07.028]

60. Ren, L.; Li, Y.; Yu, M.; Yang, J.; Zhan, C.; Zhou, D. Evolution of artificial wetlands in the Yellow River Delta and related driving factors during 1984–2015. J. Agric. Resour. Environ.; 2020; 37, pp. 493-502.

61. Song, Y.; Zhang, Z.; Niu, B.; Li, X. Temporal and spatial variations of landscape pattern vulnerability in Yellow River Delta during 2005–2018. Bull. Soil Water Conversat.; 2021; 41, pp. 258-266.