Spatiotemporal evolution effects of habitat quality with the conservation policies in the Upper Yangtze River, China

Abstract

As a critical ecological barrier, the upper Yangtze River (UYR) holds strategic importance for national ecological security. Understanding its habitat quality dynamics is essential for evaluating conservation efforts. However, there is a relative lack of long-term monitoring studies on habitat quality in this region, and the influencing factors remain insufficiently explored. Using the InVEST model, this study quantified the spatiotemporal evolution of habitat quality in the UYR from 1990 to 2020. Spatial autocorrelation analysis revealed distinct clustering patterns, and spatial regression models identified driving mechanisms. Results showed that habitat quality experienced a sharp decline (1990–2000), especially in the Jinsha River basin, followed by recovery due to the Natural Forest Protection (1998) and Grain-for-Green (2000) programs. High-quality areas clustered in the upper Jinsha and Min-Tuo basins, while low-quality areas were concentrated in urbanized regions of the Sichuan Basin and Jialing River basin. Elevation and slope indirectly improved habitat quality by promoting vegetation, whereas temperature, PM2.5, population density, and GDP had negative effects. Although ecological policies alleviated pressures, urbanized areas require further restoration. This study provides critical insights into conservation policy effectiveness and supports zonal ecological management in the UYR.

Introduction

Habitat quality is one of the key indicators for assessing ecosystem health and a critical component of ecosystem services¹. It reflects not only an ecosystem’s ability to provide suitable habitats for species but also its resilience and stability in response to external pressures^2,3. With the intensification of climate change and human activities, the degradation of habitat quality and subsequent loss of biodiversity have become global challenges^4,5,6. As a vital ecological security barrier and biodiversity hotspot in China, maintaining healthy and stable habitat quality in the UYR is essential for ensuring ecological security and promoting sustainable development in the region⁷. Since the Reform and Opening-up policy, China has implemented numerous ecological protection initiatives, including the Natural Forest Protection Project, Yangtze River Shelterbelt Program, and the Grain for Green Program^8,9,10. However, the effectiveness of these policies and restoration projects in improving regional habitat quality has not been fully quantified and analyzed. Therefore, scientifically evaluating habitat quality in the UYR region, analyzing its spatiotemporal evolution, and identifying the driving mechanisms are of great importance for biodiversity conservation and the construction of a national ecological security framework.

In recent years, with the deepening of ecosystem service research, habitat quality assessment models have become increasingly diversified. Commonly used models include GLOBIO¹¹, HSI (Habitat Suitability Index)¹², MaxEnt (Maximum Entropy Model)¹³, ARIES (Artificial Intelligence for Ecosystem Services)¹⁴, and InVEST (Integrated Valuation of Ecosystem Services and Trade-offs)¹⁵. While MaxEnt is widely applied for species distribution modeling, it is less effective at comprehensively capturing the impact of land use changes on habitat quality¹⁶. ARIES, although advantageous in modeling multiple ecosystem services, is complex and requires extensive data¹⁷. GLOBIO is more suitable for global or regional biodiversity trend assessments, focusing on macro-level impacts but lacking the granularity needed for local habitat analysis¹¹. HSI, on the other hand, focuses more on the suitability of habitats for specific species¹⁸. In contrast, the InVEST model integrates land use changes, threat factors, and habitat sensitivity, offering a modular design, moderate data requirements, and strong spatial analysis capabilities^2,19. These attributes make it particularly effective for large-scale and long-term habitat quality assessments, especially in heterogeneous and complex geographical environments^20,21. As a result, InVEST has been widely adopted for habitat quality assessments in large river basins. For example, Berta Aneseyee et al. demonstrated its utility in southwestern Ethiopia’s Winike River Basin, revealing significant correlations between land use intensity, slope, and habitat quality²². In China, Sun et al. applied InVEST to monitor the spatiotemporal dynamics of habitat quality in the South Four Lakes Basin, discovering severe habitat degradation in lake areas and the limitations of existing conservation strategies²³. Similarly, Bao et al. assessed habitat quality in the Yellow River Wetland Nature Reserve, linking habitat degradation to wetland reduction and the expansion of construction land²⁴. Within the Yangtze River Basin, research on habitat quality has primarily focused on specific provinces, cities, or reservoir areas, with a lack of comprehensive, region-wide habitat quality monitoring. For example, Zhang et al. investigated habitat quality in the Wuding River Basin²⁵, while Jin et al. explored changes in habitat quality across the Sanjiang Plain over several decades²⁶. Additionally, Zhou et al. examined the impact of topographic gradients on habitat quality in the Hubei section of the Three Gorges Reservoir, highlighting how land use changes influence biodiversity²⁷. While these studies provide valuable insights into localized habitat dynamics, they are geographically constrained and fail to capture the broader spatiotemporal patterns of habitat quality across the Yangtze River Basin. This limitation is particularly pronounced in the Upper Yangtze River, underscoring the need for a holistic, basin-wide assessment.

In summary, this study adopts a multi-model, multi-scale evaluation framework to comprehensively investigate the spatiotemporal evolution of habitat quality in the UYR region from 1990 to 2020. By examining habitat quality dynamics and driving mechanisms across different river basins, we explore the ecological restoration effects and regional variations in response to policy interventions in major basins such as the Jinsha, Min-Tuo, and Jialing Rivers. The findings aim to provide a scientific basis for assessing the effectiveness of large-scale ecological protection policies in the UYR and offer targeted recommendations for region-specific ecological governance and future conservation efforts in this ecologically critical area.

Methodology

Study area overview

The upper Yangtze River extends from the river’s source to the Yichang section in Hubei Province, covering a geographical range between 24°27′40″~35°45′10″N and 90°32′4″~35°45′10″E, covering an area of approximately 1 million km². As shown in Fig. 1: this region lies within China’s transition zone from the first to the second topographic step, spanning various geomorphological units, including the Qinghai-Tibet Plateau, Hengduan Mountains, Yunnan-Guizhou Plateau, Qinling-Daba Mountains, and the Sichuan Basin. Influenced by monsoon climates and plateau effects, the upper Yangtze River area is highly sensitive to global climate warming. The terrain slopes from northwest to southeast, with significant elevation variation and diverse landscapes; mountains and hills account for 98% of the basin’s area. The river system is complex, including major tributaries such as the Jinsha, Min-Tuo, Jialing, and Wu Rivers.

The Jinsha River basin primarily flows through Yushu Tibetan Autonomous Prefecture in Qinghai Province; Chamdo in the Tibet Autonomous Region; Ganzi Tibetan Autonomous Prefecture and Liangshan Yi Autonomous Prefecture in Sichuan Province; and Lijiang, Diqing Tibetan Autonomous Prefecture, Dali Bai Autonomous Prefecture, and Chuxiong Yi Autonomous Prefecture in Yunnan Province. The Min-Tuo River basin primarily flows through Chengdu, Yibin, Mianyang, Deyang, and Leshan in Sichuan Province. The Jialing River basin primarily flows through Longnan in Gansu Province; Hanzhong in Shaanxi Province; Guangyuan and Nanchong in Sichuan Province; and Hechuan District in Chongqing. The Wu River basin primarily flows through Zunyi, Bijie, and Tongren in Guizhou Province; and Fuling District and Pengshui Miao and Tujia Autonomous County in Chongqing. The Yangtze mainstem, from Yichang to the headwaters, primarily flows through Yushu Tibetan Autonomous Prefecture in Qinghai Province; Chamdo in the Tibet Autonomous Region; Yibin in Sichuan Province; Zhaotong in Yunnan Province; and Yichang in Hubei Province.

Fig. 1

Location map of the Upper Yangtze River Basin. This map is based on the standard map from the Ministry of Natural Resources of China (Approval Number: GS (2020) 4691), downloaded from the National Standard Map Service website (http://bzdt.ch.mnr.gov.cn/index.html) and rendered using ArcGIS 10.8 (https://www.esri.com/en-us/arcgis/about-arcgis). The base map has not been modified.

Data sources

The primary data used in this study include land use data of the UYR from 1990 to 2020. The accuracy and sources of the various baseline datasets are listed in Table 1.

Table 1 Data sources.

Habitat quality calculation based on the InVEST model

The habitat quality module within the InVEST model was applied to analyze habitat quality in the study area. This module evaluates habitat quality by incorporating the sensitivity of different land cover types to threat factors, the intensity of external threats, the impact distance of threats, spatial weights, and the degree of legal protection for the land. Habitat quality is treated as a continuous variable²⁸. The model assumes that areas with high habitat quality also exhibit high biodiversity. By calculating the negative impacts of threat sources on habitats, the model assesses habitat degradation, which is then used to calculate habitat quality based on habitat suitability and degradation levels²⁹. Habitat degradation represents the extent of disturbance caused by various threats, and the model requires the identification of both threat sources and habitat types.

For the selection of threat sources, this study considered the actual conditions of the UYR and referred to relevant studies and expert opinions^30,31,32. Six main threat factors were selected: paddy fields, dryland, urban land, rural settlements, other construction land, and bare land. Cultivated land and construction land were identified as areas of intense human activity, which exert a considerable impact on habitat quality in the UYR. Additionally, bare land and unused land, such as saline-alkali land, were considered ecologically fragile areas. Due to the varying impacts of different farming practices on habitat quality, cultivated land was divided into paddy fields and dryland for separate calculations. Given the highly variable topography of the UYR region and the scattered distribution of construction land, construction land was further classified into urban land, rural settlements, and other construction land. The specific values assigned to these factors are shown in Table 2. The calculation formula is as follows:

$$\:{\text{D}}_{\text{x}\text{j}}=\sum\:\begin{array}{c}r\\\:l\:\end{array}\sum\:\begin{array}{c}y\\\:l\end{array}\left(\frac{{\omega\:}_{r}}{\sum\:\begin{array}{c}n\\\:r=1\end{array}{\omega\:}_{r}}\right)\times\:{r}_{y}\times\:{i}_{rxy}\times\:{\beta\:}_{x}\times\:{S}_{jr}$$

(1)

$$\:i_{{rxy}} = \left\{ {\begin{array}{*{20}c} {1 - \left( {\frac{{d_{{xy}} }}{{d_{{r\:\max }} }}} \right)\:\:\:\left( {Linear\:Decay\:} \right)} \\ {\:\exp \left[ { - \left( {\frac{{2.99}}{{d_{{r\:\max }} }}} \right) \times \:d_{{xy}} } \right](Exponential\:Decay)} \\ \end{array} } \right.$$

(2)

$$\:{Q}_{xj}={H}_{xj}\times\:\left[1-\left(\frac{{D}_{xj}^{2}}{{D}_{xj}^{2}+{k}^{2}}\right)\right]$$

(3)

In the formula: D_xj represents habitat degradation; $\:{{\upomega\:}}_{\text{r}}$ is the weight assigned to each threat factor; $\:{\text{r}}_{\text{y}}$denotes the intensity of the threat factor; $\:{{\upbeta\:}}_{\text{x}}$indicates the resistance of the habitat to disturbance; $\:{\text{S}}_{\text{j}\text{r}}$is the relative sensitivity of different habitats to various threat factors; $\:{\text{i}}_{\text{r}\text{x}\text{y}}$represents the impact of threat factor r in grid cell y on grid cell x, and it can follow either an exponential or linear decay; r is the habitat threat factor; y represents the grid cell containing threat factor r; d_xy is the distance between grid cells x and y; d_rmax denotes the maximum impact range of threat factor r; The variable Q_xj refers to the habitat quality of grid cell x within land use type j; while H_xj represents the habitat suitability of grid cell x within land use type j, ranging from [0,1]. Higher values indicate better habitat suitability, with artificial environments typically having lower suitability. The sensitivity S_jr refers to the relative vulnerability of each habitat type to threats. In this study, sensitivity values for ecological threat factors were assigned between 0 and 1 based on the ecological importance of each habitat type. Specific values for habitat suitability and sensitivity to ecological threats are shown in Table 3. The half-saturation constant k is set to 0.05. Habitat quality values range from 0 to 1, with higher Q_xj values indicating better habitat quality. To further analyze habitat quality changes in the UYR, habitat quality is classified into five levels: low (0−0.25), moderate (0.25–0.5), good (0.5–0.75), and high (0.75−1).

Table 2 Threat factors.

Table 3 Habitat suitability and sensitivity of different land use types.

Spatial autocorrelation analysis

According to Tobler’s First Law of Geography, habitat quality in the UYR, as a continuous spatial variable, exhibits spatial autocorrelation or similarity³³. Detecting whether spatial entities are correlated is dependent on global spatial autocorrelation analysis, while the degree of correlation is measured using local spatial autocorrelation indices. The Global Moran’s I was used to evaluate the spatial dependence of habitat quality across the UYR region. Local spatial autocorrelation analysis provides more detailed insights into the clustering patterns and specific locations where individual units exhibit spatial similarity with neighboring units, offering a more intuitive understanding of the spatial differentiation of habitat quality³⁴. By visualizing the Moran scatter plot derived from the local spatial autocorrelation, the spatial pattern of habitat quality in the study area can be classified into four quadrants: positive spatial correlation represented by High-High (H-H) and Low-Low (L-L) clusters, and negative spatial correlation represented by High-Low (H-L) and Low-High (L-H) clusters. Other scattered patches are classified as “Not Significant.” The H-H and L-L clusters indicate spatially positive correlations, where the habitat quality of a given unit aligns with that of its neighboring units, showing consistent evolutionary trends. In contrast, the L-H and H-L clusters indicate spatially negative correlations, where the habitat quality of a given unit contrasts with that of its surrounding neighbors, forming patterns akin to “peaks” and “valleys.” According to the theory of spatial polarization, a target area will eventually be assimilated by surrounding areas, transforming into either H-H or L-L regions.

Spatial regression model

Traditional statistical methods, such as the classical linear regression model, typically assume uniform data distribution and independence across the study area, which often does not account for the spatial autocorrelation commonly present in geographical data³⁵. Spatial regression models, including the spatial lag model and the spatial error model, address this limitation by incorporating spatial dependency into the regression equation. In this study, we applied spatial regression models to analyze the driving mechanisms behind habitat quality in the UYR region. The model equations are as follows:

$$\begin{gathered} Y=\rho {W_1}+\beta X+\mu \hfill \\ \mu =\lambda {W_2}+\varepsilon \hfill \\ \end{gathered}$$

(4)

In the formula, Y represents the dependent variable, and X is the explanatory variable. β denotes the spatial regression coefficient of the explanatory variable, µ is the error term, and ε represents white noise. W₁ is the spatial weight matrix reflecting the spatial trend of the dependent variable itself, while W₂ is the spatial weight matrix reflecting the spatial trend of the residuals. Ρ is the coefficient of the spatial lag term, and λ is the spatial error coefficient. Both coefficients range from [0,1], with values closer to 1 indicating greater similarity in the dependent or explanatory variables across neighboring areas.

Result

Spatiotemporal evolution of habitat quality in the Upper Yangtze River from 1990 to 2020

The spatial distribution of habitat quality in the UYR exhibits significant variation, with a general trend of higher habitat quality in the west and lower quality in the east, as shown in Fig. 2. Areas with excellent habitat quality are primarily concentrated in the Jinsha River and western Min-Tuo River basins, including regions in Tibet (Changdu), western and central Sichuan (Ganzi, Liangshan, Yibin), and Yunnan (Diqing, Dali, Lijiang). Other high-quality habitat areas are scattered across the northwestern Jialing River basin, the Wujiang River basin, and the upper reaches of the Yangtze River, including parts of Gansu (Gannan), northwestern Guizhou (Zunyi), and areas in Hubei (Yichang and Shennongjia Forest District), which are predominantly covered by forests and grasslands. These high-quality areas are closely associated with adjacent regions of relatively high habitat quality. Moderate habitat quality areas are mainly distributed in the northern Jinsha River basin, southern areas of the Yangtze River mainstream, and the Wujiang River basin. These areas include the western part of Qinghai (Yushu), eastern Sichuan (Nanchong, Suining, Chengdu, Meishan, Ziyang, Neijiang, Zigong), Chongqing (Yongchuan, Dazu, Bishan, Hechuan), and southern Guizhou (Anshun, Guiyang).

Fig. 2

Spatial distribution of habitat quality classes in the upper reaches of the Yangtze River, 1990–2020. This map is based on the standard map from the Ministry of Natural Resources of China (Approval Number: GS (2020) 4691), downloaded from the National Standard Map Service website (http://bzdt.ch.mnr.gov.cn/index.html) and rendered using ArcGIS 10.8 (https://www.esri.com/en-us/arcgis/about-arcgis). The base map has not been modified.

Low-quality habitat areas are primarily located in economically developed regions and areas of intensive agricultural activity, where human disturbances are significant. These areas are concentrated in the southeastern Min-Tuo River basin, the western part of the Yangtze River mainstream, and the Jialing River basin, including Sichuan (Guangyuan, Chengdu, Nanchong, Deyang, Suining, Neijiang, Zigong), Gansu (Chengxian), Shaanxi (Hanzhong), and Chongqing. There are also small low-quality habitat areas in the northern and southeastern Jinsha River basin and the Wujiang River basin, including parts of Qinghai (southern Golmud, Zhiduo County), Yunnan (Chuxiong, Kunming), Sichuan (Panzhihua, Xichang), and Guizhou (Guiyang, Anshun, Tongren). The spatial distribution shows a clustered pattern around the Sichuan Basin, with concentrations in the eastern and southern parts of the UYR, which aligns with the natural resource endowments and regional characteristics of socioeconomic development in the UYR.

Over time, habitat quality in the UYR region exhibited an initial decline followed by gradual recovery. Between 1990 and 2000, the area where habitat quality remained stable totaled78.97 × 10⁴km², accounting for 80.57% of the study area. Areas where habitat quality improved covered 6.31 × 10⁴km² (6.43%), while those with declining habitat quality spanned 12.62 × 10⁴km² (12.90%). The most notable declines were concentrated in the western and southeastern Jinsha River basin, the Yangtze River mainstream, and most of the Wujiang River basin. These areas include Tibet (Changdu), Yunnan (Kunming, Zhaotong), Sichuan (Yibin, Luzhou), Chongqing, Hubei (Enshi, Yichang), and Guizhou (Bijie, Zunyi, Guiyang, Tongren). Smaller areas of habitat quality decline were also observed in the Min-Tuo River basin and the Jialing River basin, including parts of Sichuan (Barkam, Mianyang, Guangyuan, Dazhou). As shown in Fig. 3, the decline during this period can be attributed to extensive deforestation and land reclamation in the UYR during the 1990s, which had detrimental effects on biodiversity and led to a marked decrease in habitat quality.

Fig. 3

Spatial and temporal distribution of habitat quality changes in the upper reaches of the Yangtze River, 1990–2020. This map is based on the standard map from the Ministry of Natural Resources of China (Approval Number: GS (2020) 4691), downloaded from the National Standard Map Service website (http://bzdt.ch.mnr.gov.cn/index.html) and rendered using ArcGIS 10.8 (https://www.esri.com/en-us/arcgis/about-arcgis). The base map has not been modified.

From 2000 to 2010, the overall rate of habitat quality decline slowed. During this period, stable areas accounted for 86.47 × 10⁴km² (88.33%), areas of habitat improvement covered 7.39 × 10⁴km² (7.55%), and areas of decline totaled 4.04 × 10⁴km² (4.12%). Compared to the previous decade, habitat quality showed gradual improvement, especially in the Guizhou section of the Wujiang River basin (Bijie, Zunyi, Guiyang, Tongren), where recovery was particularly significant. From 2010 to 2020, habitat quality exhibited a mixed trend: while declines were observed in the northern and western parts of the UYR, improvements were noted in the southern and eastern regions. During this period, stable areas accounted for 86.74 × 10⁴km² (88.61%), while areas of habitat improvement covered 6.91 × 10⁴km² (7.06%). Notable improvements occurred in the Jinsha River basin, the Yangtze River mainstream, and other areas, including Tibet (Changdu), Yunnan (Zhaotong), Sichuan (Yibin), Chongqing, and Guizhou (Bijie, Zunyi). Areas of habitat quality decline totaled 4.24 × 10⁴km² (4.33%), with significant declines in the Qinghai section (Yushu) and the western Sichuan section (Aba, Ganzi) of the Jinsha River basin.

Overall, from 1990 to 2020, most areas experienced an initial decline in habitat quality followed by gradual recovery, with habitat quality eventually returning to levels comparable to those in 1990. During this period, 98.88% of the study area showed no change in habitat quality, 0.29% of the area saw slight improvements, and 0.83% experienced significant declines. While habitat quality in low-quality regions improved by 1.17%, areas with moderate, good, and excellent habitat quality decreased by 0.71%, 0.23%, and 0.88%, respectively.

Spatial autocorrelation analysis of habitat quality in the Upper Yangtze River

This study applied the Global Moran’s I index to analyze the spatial autocorrelation of habitat quality in the UYR and generated the corresponding Moran scatter plots (Fig. 4). The results of the global spatial autocorrelation analysis at the county level for the four periods (1990, 2000, 2010, and 2020) showed Global Moran’s I indices of 0.744, 0.743, 0.737, and 0.744, respectively, all of which passed the significance test. This indicates that habitat quality in the region exhibited strong spatial autocorrelation and clear clustering patterns.

Fig. 4

Scatterplot of global spatial autocorrelation Moran index for habitat quality.

The results of the local spatial autocorrelation analysis, as shown in Fig. 5, reveal significant spatial clustering of habitat quality in the UYR region. The High-High (H-H) clusters are primarily concentrated in the central and western parts of the UYR. In the west, they cover areas in Tibet, while in the east, the boundary lies along the transition zone between the western Sichuan Plateau and the Sichuan Basin. This includes regions such as Jiuzhaigou, Songpan, Mao County, Lixian, and Baoxing in Sichuan, extending southward to Yongsheng, Binchuan, and Dayao counties. These areas are critical for biodiversity conservation and water resource preservation, and they also host numerous nature reserves, giving them a strong ecological foundation. Between 1990 and 2020, the high-value clusters expanded eastward towards Wen County and Pingwu County, and westward towards Shiqu and Yushu. During this period, several areas transitioned from H-H clusters to non-significant, including Yiling District in Hubei, Zhaoyang District in Yunnan, and Shiqian County in Guizhou, which experienced significant urban and agricultural expansion, leading to greater disturbance of habitat quality.

Low-Low (L-L) clusters are mainly concentrated in eastern Sichuan, forming a ring around Chengdu, and expanded eastward between 1990 and 2020. Since 2010, new L-L clusters have emerged in southern areas, particularly in the urban centers of Kunming and Guiyang. Specific areas include Xishan, Wuhua, Panlong, Guandu, and Chongming districts in Kunming, and Huaxi District in Guiyang. These counties are located in the Dianchi Lake basin, which serves as the primary urban and economic center of Kunming. In recent years, these areas have been heavily impacted by urbanization, and the construction of multiple industrial parks since 2015 has contributed to significant habitat degradation. The distribution of Low-High (L-H) and High-Low (H-L) clusters is relatively sparse. The L-H cluster was only observed in Guiding County, which transitioned to non-significant by 2020, while the H-L cluster was only present in Wudang District, adjacent to the low-value habitat quality areas in southern Guizhou.

Fig. 5

Aggregation of local spatial correlations of habitat quality. This map is based on the standard map from the Ministry of Natural Resources of China (Approval Number: GS (2020) 4691), downloaded from the National Standard Map Service website (http://bzdt.ch.mnr.gov.cn/index.html) and rendered using ArcGIS 10.8 (https://www.esri.com/en-us/arcgis/about-arcgis). The base map has not been modified.

The results of the local spatial autocorrelation analysis, as shown in Fig. 5, reveal significant spatial clustering of habitat quality in the UYR region. The High-High (H-H) clusters are primarily concentrated in the central and western parts of the UYR. In the west, they cover areas in Tibet, while in the east, the boundary lies along the transition zone between the western Sichuan Plateau and the Sichuan Basin. This includes regions such as Jiuzhaigou, Songpan, Mao County, Lixian, and Baoxing in Sichuan, extending southward to Yongsheng, Binchuan, and Dayao counties. These areas are critical for biodiversity conservation and water resource preservation, and they also host numerous nature reserves, giving them a strong ecological foundation. Between 1990 and 2020, the high-value clusters expanded eastward towards Wen County and Pingwu County, and westward towards Shiqu and Yushu. During this period, several areas transitioned from H-H clusters to non-significant, including Yiling District in Hubei, Zhaoyang District in Yunnan, and Shiqian County in Guizhou, which experienced significant urban and agricultural expansion, leading to greater disturbance of habitat quality.

Low-Low (L-L) clusters are mainly concentrated in eastern Sichuan, forming a ring around Chengdu, and expanded eastward between 1990 and 2020. Since 2010, new L-L clusters have emerged in southern areas, particularly in the urban centers of Kunming and Guiyang. Specific areas include Xishan, Wuhua, Panlong, Guandu, and Chongming districts in Kunming, and Huaxi District in Guiyang. These counties are located in the Dianchi Lake basin, which serves as the primary urban and economic center of Kunming. In recent years, these areas have been heavily impacted by urbanization, and the construction of multiple industrial parks since 2015 has contributed to significant habitat degradation. The distribution of Low-High (L-H) and High-Low (H-L) clusters is relatively sparse. The L-H cluster was only observed in Guiding County, which transitioned to non-significant by 2020, while the H-L cluster was only present in Wudang District, adjacent to the low-value habitat quality areas in southern Guizhou.

Analysis of the driving mechanisms of spatiotemporal evolution of habitat quality in the Upper Yangtze River

Selection of driving factors

Changes in habitat quality are influenced by a combination of natural and anthropogenic factors. In this study, seven key factors affecting habitat quality in the UYR were selected: DEM, slope, NDVI, AMT, GDP, PM2.5 concentration, and population density. A bivariate spatial autocorrelation analysis was conducted between these indicators and the habitat quality of the UYR in 2020. The results indicated a significant spatial correlation between the Moran’s I values of the selected indicators and habitat quality (Fig. 6), suggesting that these factors exert varying degrees of influence on habitat quality across the study area.

Fig. 6

Bivariate spatial autocorrelation Moran index values.

Modeling of driving mechanisms

Classical linear regression analysis, based on the method of Ordinary Least Squares (OLS), is a necessary step before conducting spatial regression analysis. To further analyze the factors influencing habitat quality changes and their degree of impact, the selected driving factors were used as independent variables, and habitat quality was used as the dependent variable to construct a spatial regression model.

As shown in Table 4, the positive or negative coefficients indicate that habitat quality is positively correlated with DEM, slope, and NDVI, while it is negatively correlated with AMT, PM2.5 concentration, GDP, and population density. Additionally, the OLS model results listed in the table below provide the parameters necessary for selecting the most appropriate spatial regression model.

Table 4 OLS modeling results (2020).

As shown in Table 5, the Moran’s I value for the residuals of the OLS model is 0.4735, and it passed the significance test, indicating that there is spatial autocorrelation in the residuals. Therefore, introducing a spatial regression model is necessary. First, the significance of the Lagrange Multiplier (lag) and Lagrange Multiplier (error) tests should be examined. If either test is significant, the corresponding spatial regression model should be used. If both tests are significant, the Robust LM (lag) and Robust LM (error) statistics should be compared. In this study, Robust LM (error) is more significant than Robust LM (lag), suggesting that the spatial error model is the more appropriate choice.

Table 5 Spatial correlation test.

Analysis of driving mechanisms based on the spatial error model

As shown in Table 6, the results from the spatial error model indicate that DEM, slope, NDVI, and annual mean temperature significantly impact habitat quality, with most of the factors passing the 1% significance test, except for PM2.5.

Table 6 Spatial error modeling results (2020).

From the regression coefficients, DEM, slope, and NDVI have a considerable positive impact on habitat quality. As DEM, slope, and NDVI increase, habitat quality also improves. Regions with higher DEM and steeper slopes tend to experience less human interference, indicating that areas in the UYR that are at high altitudes, have steep slopes, and are rich in vegetation generally have better ecological conditions. Specifically, for every 1-meter increase in DEM, habitat quality improves by 0.0001%, and for every 1-degree increase in slope, habitat quality improves by 0.01%. This suggests that slope has a greater impact than DEM, as regions with higher elevations, while less impacted by human activity, tend to exhibit less biodiversity due to harsh climate conditions. In contrast, regions with steeper slopes in the UYR are less disturbed by humans and less affected by adverse weather conditions, leading to better habitat quality.AMT shows a negative correlation with habitat quality. For every 1 °C increase in AMT, habitat quality decreases by 0.016%. The UYR is particularly sensitive to AMT changes, especially since 2005, when global warming has led to more frequent droughts in the region, significantly impacting habitat quality.

In general, socioeconomic factors tend to have a negative impact on habitat quality. As shown in Table 6, for every 10,000 yuan increase in GDP, habitat quality declines by 1.3605E-006% points. Similarly, for every 1 person/km² increase in population density, habitat quality decreases by 4.55692E-006% points. Low habitat quality areas are primarily clustered in the eastern part of the UYR, indicating that factors such as terrain, population density, and economic development negatively affect habitat quality.

Discussion

Spatial dynamics of habitat quality: patterns and policy impacts

The UYR region spans a vast area with significant altitudinal gradients, resulting in pronounced spatial variations in ecological conditions and economic development across different areas²⁹. This heterogeneity leads to notable spatial differences in habitat quality. The western parts, including the eastern edge of the Tibetan Plateau and the Sichuan-Tibet highlands, maintain high habitat quality due to limited human interference and well-preserved natural vegetation. In contrast, the eastern Sichuan Basin and the Jialing River basin exhibit lower habitat quality, driven by intensive land development, high population density, and urban expansion and agricultural activities that have significantly impacted the ecosystem. The spatial differentiation in habitat quality reflects not only ecological conditions but also underlying tensions between regional ecological patterns and socioeconomic development.

From 1990 to 2020, habitat quality in the UYR followed a “decline-recovery” trajectory. Between 1990 and 2000, habitat quality significantly decreased due to high-intensity land use and rapid urbanization, especially in the Sichuan Basin and Jialing River Basin^27,36. Since 2000, national ecological conservation policies, such as the Natural Forest Protection Program (1998) and the Grain-for-Green Program (2002), facilitated a gradual recovery of habitat quality, particularly in ecologically sensitive areas like the Jinsha and Min-Tuo River basins³⁷. These findings highlight the effectiveness of government-led interventions in mitigating habitat degradation. However, the rate of habitat recovery remains uneven across regions, differences in natural conditions and human activity intensity result in spatial variability in policy outcomes. Urbanized areas such as Chengdu and Kunming exhibit slower recovery, reflecting the persistent challenges posed by rapid economic growth and population increases. Additionally, certain high-altitude areas in Qinghai have shown further declines in habitat quality, highlighting persistent ecological fragility in parts of the UYR region. In highly sensitive areas with poor ecological baselines, such as the Yangtze River source region, sustaining the benefits of ecological projects remains challenging, underscoring the need for differentiated conservation strategies.

The “Go West” strategy (2010), designed to promote economic development in western China, has exacerbated the tension between ecological conservation and economic expansion. Without proactive interventions, high-quality habitats in the western UYR may face increasing developmental pressures. Addressing these challenges requires balancing economic growth with ecological protection through integrated regional policies and sustainable development strategies. Enhanced dynamic monitoring mechanisms, such as integrating remote sensing technologies with ground-based assessments, are essential not only for evaluating the long-term efficacy of current restoration projects but also for providing timely feedback to refine conservation strategies in the UYR.

Drivers of habitat quality: balancing nature protection and development

Habitat quality in the UYR is influenced by a complex interplay of natural and anthropogenic factors. The spatial regression analysis identifies altitude (DEM), slope, and vegetation cover (NDVI) as key positive factors influencing habitat quality, while average annual temperature (AMT), PM2.5 concentration, population density, and GDP are negatively correlated with habitat quality. The positive effects of DEM and slope reflect the ecological integrity of high-altitude, steep-slope areas, such as the Jinsha River and Min-Tuo River basins, which experience minimal human disturbance. However, these areas also face ecological vulnerabilities due to low vegetation productivity and poor ecological baselines, making them highly sensitive to climate change³⁹. According to the findings of Bao and Wang^24,38, the Hengduan Mountains in the Jinsha River basin and the Min River basin are characterized by steep slopes, complex terrain, relatively homogeneous vegetation types, and fragile ecosystems prone to degradation. Therefore, habitat quality in these areas is highly sensitive to climate change, making natural factors the main driving force behind habitat quality changes in the UYR. Rising temperatures, driven by global climate change, further exacerbate the vulnerability of these ecosystems. For example, extreme weather events and temperature increases may alter vegetation distribution and ecosystem stability in these regions, necessitating adaptive management approaches.

Conversely, GDP, population density, and PM2.5 concentrations negatively impact habitat quality, particularly in urbanized areas such as the Sichuan Basin and Dianchi Lake basin. Urban expansion not only intensifies negative impacts such as habitat fragmentation but also diminishes the buffering role of natural factors, such as slope and vegetation cover, amplifying regional habitat degradation. For instance, the expansion of industrial parks in Kunming has disrupted surrounding ecosystems, while urban sprawl in Chengdu has reduced habitat connectivity. To mitigate these impacts, integrated urban planning and green infrastructure investment are critical. Measures such as compact city designs, ecological zoning, and stricter pollution controls can help balance urban development with ecological preservation.

Regional protection measures and policy strategies

The spatial heterogeneity of habitat quality in the UYR necessitates tailored policy interventions. For High-High (H-H) clusters, current ecological conservation achievements should be further consolidated. Conservation efforts should focus on natural restoration and assisted rehabilitation to enhance ecosystem resilience. In areas such as the Jinsha and Min-Tuo River basins and regions like Yushu and Shiqu, continued ecological restoration projects (e.g., Grain-for-Green, Natural Forest Protection) and strict limitations on human activities are necessary to maintain ecological stability and safeguard this crucial ecological barrier. Additionally, establishing ecotourism projects in these regions may help achieve a balance between economic development and ecological preservation.

For Low-Low (L-L) clusters, particularly in the eastern Sichuan Basin and the urbanized areas of Kunming and Guiyang in the Dianchi Lake basin, stronger pollution control measures and stricter land-use planning are essential. Establishing an ecological compensation mechanism could help mitigate the adverse impacts of urban expansion⁴⁰. Measures could include defining urban growth boundaries to limit construction land expansion, adjusting industrial structures to prioritize low-carbon industries, and designating green industrial parks to concentrate renewable energy and new material industries, thereby promoting coordinated regional development. Increased investment in green infrastructure could also reduce the ecological pressures associated with economic growth. Given the competing priorities of economic growth and ecological preservation in these regions, ensuring adequate funding and cross-jurisdictional coordination remains a challenge. Policymakers should prioritize collaborative frameworks and fiscal incentives to encourage regional cooperation in addressing shared ecological pressures.

For areas with unstable habitat quality, such as Guiding and Wudang counties, enhanced cross-regional ecological linkages are recommended. Building ecological corridors to mitigate external ecological impacts and increase habitat connectivity can improve regional habitat functionality. Gradual habitat quality improvements in these transitional zones can be achieved through targeted ecological restoration measures, such as soil management and vegetation rehabilitation.

Innovation, challenges, and future directions for UYR conservation

This study offers a novel framework for assessing habitat quality dynamics in the UYR by integrating the InVEST model with spatial autocorrelation and regression analysis. By quantifying the impacts of both natural and anthropogenic drivers, this research provides a comprehensive evaluation of habitat quality trends and their underlying mechanisms. The findings not only highlight the spatially differentiated impacts of conservation policies but also offer region-specific insights to guide ecological governance. However, the study has several limitations. First, while the InVEST model effectively quantifies habitat quality, it does not fully capture ecosystem services trade-offs or the socioeconomic complexities of conservation efforts. Future studies should integrate models like ARIES or GLOBIO to provide a more holistic evaluation. Second, while this study evaluates the effectiveness of large-scale ecological conservation policies, future studies could benefit from integrating insights from local stakeholders or policymakers. This approach could provide a more nuanced understanding of the practical challenges and opportunities in implementing conservation strategies at local scales.

Looking ahead, future research should prioritize the integration of climate change scenarios into habitat quality assessments. Predictive modeling under Representative Concentration Pathways (RCPs) can help identify the potential impacts of rising temperatures, extreme weather events, and changing precipitation patterns on regional habitat quality. This would provide a basis for adaptive conservation strategies in the UYR. On the policy front, the establishment of a long-term, dynamic monitoring framework that incorporates real-time data collection and adaptive management will be crucial for sustaining conservation benefits. Future studies could incorporate participatory approaches, engaging local stakeholders to align conservation measures with community needs. This would enhance the practicality and acceptance of ecological policies while ensuring their long-term sustainability. These efforts would not only enhance our understanding of ecological dynamics but also ensure the long-term resilience and sustainability of conservation policies in the Upper Yangtze River region.

Conclusion

This study reveals the spatially uneven effectiveness of large-scale ecological conservation policies in improving habitat quality in the Upper Yangtze River (UYR). The findings highlight critical challenges in balancing ecological preservation with socioeconomic development, offering actionable insights for precision ecological governance. Key conclusions are as follows.

• (1) Spatial Variability of Habitat Quality: Habitat quality in the UYR exhibits significant spatial heterogeneity, with high-quality habitats concentrated in the Jinsha and Min-Tuo River basins due to favorable natural conditions and minimal human disturbance. Conversely, urbanized and agriculturally intensive regions, such as the Sichuan Basin and Jialing River basin, show lower habitat quality, reflecting the pressures of intensive human activities.

• (2) Conservation Policy Effectiveness with Temporal and Regional Variability: From 1990 to 2020, habitat quality exhibited a “decline-recovery” trajectory. The period of 1990–2000 saw significant habitat degradation driven by high-intensity land use and rapid urbanization. In contrast, the introduction of ecological conservation policies, such as the Natural Forest Protection Program (1998) and the Grain-for-Green Program (2002), effectively reversed this trend, leading to gradual habitat recovery, particularly in the Jinsha and Min-Tuo River basins. However, habitat recovery in urbanized areas such as Chengdu and Kunming remains limited, highlighting the persistent challenges of balancing ecological restoration with rapid economic growth.

• (3) Spatial Clustering and Habitat Quality Trends: High-value habitat clusters (H-H) expanded along the eastern edge of the Tibetan Plateau, demonstrating the success of conservation measures in low-disturbance areas. Conversely, low-value clusters (L-L) proliferated in urban centers, such as Chengdu and Kunming. These patterns highlight the necessity of region-specific conservation strategies to address diverse ecological challenges.

• (4) Drivers of Habitat Quality: Natural factors such as elevation, slope, and vegetation positively influence habitat quality, while anthropogenic pressures (population density, urbanization, GDP, and PM2.5 emissions) were negatively associated with it. Climate change, exemplified by rising temperatures, emerged as a critical challenge, particularly for high-altitude ecosystems, which remain ecologically fragile despite limited human disturbance, requiring adaptive management strategies to mitigate the impacts of rising temperatures.

• (5) Implications for Precision Ecological Governance: Effective ecological governance requires spatially tailored strategies. High-value clusters (e.g., Jinsha and Min-Tuo basins) demand ongoing restoration and strengthened ecological barriers, while low-value clusters (e.g., Chengdu Plain and Dianchi Lake basin) require stricter land-use regulations, pollution control, and ecological compensation. Enhancing ecological connectivity through corridors in transitional zones (e.g., Jialing and Wu basins) is essential for sustaining regional resilience. Globally, these insights can inform conservation strategies in other ecologically sensitive regions, offering a roadmap for balancing environmental and socioeconomic priorities.