Prediction of Climate Change Impacts on the Distribution of an Umbrella Species in Western Sichuan Province, China: Insights from the MaxEnt Model and Circuit Theory

Abstract

Climate change poses a significant threat to biodiversity. Predicting the impacts of climate change on species distribution and dispersal through computational models and big data analysis can provide valuable insights. These predictions are crucial for developing effective strategies to mitigate the threats that climate change poses to biodiversity. Our study investigated the potential impact of climate change on an umbrella species (Ursus arctos pruinosus) in Western Sichuan Province, China. We employed the MaxEnt and Circuit Theory to assess both the current and potential future shifts in the distribution and migration corridors. The results indicated that climate and environmental factors had the greatest influence on species distribution, with bioclimatic variables bio12, bio3, and elevation contributing 22.1%, 21.5%, and 19.3%, respectively. Under current climatic conditions, the total suitable habitat area for the species was 70,969.78 km², with the largest suitable habitats located in Shiqu and Litang, accounting for 24.39% and 15.86% of the total area, respectively. However, under future climate scenarios, predictions for RCP 2.6, RCP 4.5, and RCP 8.5 showed a significant reduction in suitable habitat area, ranging from 7789.26 km² to 16,678.85 km². The Yajiang and Xinlong counties experienced the most severe habitat reductions, with declines exceeding 50%. Additionally, the altitudinal distribution of suitable habitats shifted, with suitable habitats gradually moving to higher elevations under future climate scenarios. Our study also analyzed the species’ dispersal paths. Under current climatic conditions, the dispersal paths predominantly followed a northwest-to-southeast orientation. However, by the 2070s, under all three RCPs, dispersal resistance is projected to significantly increase, the density of dispersal paths will decrease, and the connectivity of these paths will be reduced. In the most extreme RCP 8.5 scenario, southern dispersal paths nearly disappeared, and the dispersal paths contracted towards the northwest. These findings highlight potential threats posed by climate change to the species’ habitats and dispersal corridors, emphasizing the importance of considering both current and future climate change in conservation strategies to protect this vulnerable species and its ecosystem.

1. Introduction

Global climate change is widely regarded as one of the most pressing environmental challenges of our time, exerting profound and complex impacts on ecosystems and biodiversity [1]. While climate change is recognized as a significant driver of ecological transformation and has garnered considerable global attention for its potential to threaten biodiversity, current evidence suggests that the primary drivers of biodiversity loss and species extinction remain direct anthropogenic factors [2]. These include the direct exploitation of natural resources and environmental pollution [2]. However, climate change, as a global and progressively intensifying background stressor, exacerbates the effects of these direct drivers, further threatening biodiversity and accelerating ecosystem destabilization. This complex interaction not only amplifies ecosystem vulnerability but also presents greater challenges for future biodiversity conservation efforts [1].

One critical ecological consequence of climate change is its influence on species distribution patterns. Rising global temperatures and increasingly erratic precipitation regimes are inducing shifts in species’ geographic ranges, as many organisms are compelled to migrate to areas with more favorable climatic conditions [1]. These distributional changes disrupt long-established ecological relationships, as species that previously coexisted may no longer share the same habitats [3]. Such disruptions can alter interspecific interactions, including resource competition, predation dynamics, and mutualisms, potentially leading to population declines and local extinctions [4]. In some cases, these shifts result in habitat fragmentation, isolating populations into smaller, disconnected patches, which hampers gene flow, diminishes genetic diversity, and constrains adaptive potential [5]. Moreover, species with limited dispersal capabilities face heightened extinction risks due to their inability to track suitable habitats rapidly enough.

Although climate-induced range shifts have been documented across taxa, their ecological consequences are context-dependent and vary by region and species. Many species have migrated toward higher latitudes or altitudes in response to warming temperatures, seeking habitats that provide climatic conditions analogous to their historical ranges [6]. However, as species move into previously uninhabited or marginal environments, they encounter novel challenges, including altered resource availability, competition with native species, and exposure to different abiotic stressors. These changes affect not only individual species but also the composition and functioning of entire ecosystems [7]. For instance, the introduction of species into new areas can lead to the displacement of native species and the establishment of invasive populations, which may further threaten local biodiversity [8]. Additionally, the loss or decline of keystone species due to climate-driven habitat shifts can disrupt essential ecosystem processes such as pollination, seed dispersal, nutrient cycling, and water regulation, thereby undermining ecosystem resilience and the provision of critical ecosystem services [9].

Given the multifaceted threats posed by climate change, it is imperative to develop a comprehensive understanding of the spatial and ecological requirements of species. Identifying areas that are likely to function as climate refugia—regions that can support species persistence under changing climatic conditions—is a key priority for conservation planning [10]. By integrating ecological modeling with climate projections, we can predict potential range shifts and assess the future viability of protected areas. This approach enables the identification of priority regions for conservation interventions, such as the establishment of new protected areas or ecological corridors that facilitate species movement and connectivity [11]. Furthermore, incorporating dynamic habitat suitability assessments into ecosystem management strategies enhances the ability to mitigate biodiversity loss and ensure the long-term stability of ecosystem services [12].

In light of the accelerating pace of climate change, conservation strategies must be adaptive and forward-looking. Effective biodiversity conservation requires a shift from static, location-based protection toward dynamic approaches that account for ongoing and future climatic changes. This involves designing flexible management frameworks that incorporate species’ ecological thresholds and habitat requirements, as well as fostering cross-regional collaboration to address the transboundary nature of climate impacts [13]. By prioritizing both immediate actions to address current drivers of biodiversity loss and proactive measures to mitigate climate-induced risks, we can enhance species resilience, maintain ecosystem integrity, and safeguard the ecosystem services upon which human well-being depends.

In addition, 3S technologies (Remote Sensing, GIS, and GPS) and computational models play a crucial role in developing species climate suitability conservation strategies. The continuous development of Geographic Information Systems (GISs) and advancements in innovative statistical methods have significantly contributed to the widespread application of ecological niche models within species distribution models (SDMs). Among these models, the Genetic Algorithm for Rule Set Prediction, the Domain Distance Model, the Generalized Linear Model, Random Forest, the Maximum Entropy Model (MaxEnt), and the Generalized Additive Model are particularly prominent [14,15]. The MaxEnt model, based on the principle of maximum entropy, integrates species occurrence data with environmental variables from their habitats to simulate the optimal distribution of target species under ecological niche constraints [16]. Among various species distribution models (SDMs), MaxEnt has garnered widespread attention in recent years due to its superior predictive performance [17]. In addition, Circuit Theory, which simulates species’ dispersal paths, has also become a growing area of focus in ecological research [18]. The application of these models not only enhances the accuracy of species distribution predictions but also highlights the increasing importance of SDMs in biodiversity conservation and ecological studies.

An umbrella species is one whose protection indirectly benefits the conservation of a broader range of coexisting species and their habitats by ensuring the preservation of ecological processes and biodiversity across large landscapes. The Tibetan brown bear (Ursus arctos pruinosus) (Supplementary Materials), as a high-altitude apex predator with specialized habitat requirements and wide-ranging behavior, serves as an effective umbrella species in the fragile alpine ecosystems of the Qinghai–Tibet Plateau and its surrounding areas [7]. Primarily inhabiting high-altitude regions characterized by low temperatures, low oxygen levels, and limited human presence, the Tibetan brown bear is highly sensitive to climate change yet remains understudied due to its remote and extreme environment [7]. Listed as a second-class protected species in China’s National List of Key Protected Wildlife and in CITES Appendix II, it is particularly vulnerable to environmental changes because of its small population size and dependence on specialized alpine meadow ecosystems.

By focusing on the conservation of Tibetan brown bear habitats, we can simultaneously protect other sympatric species sharing the same ecosystem, such as the Asiatic golden cat (Pardofelis temminckii), snow leopard (Panthera uncia), Asiatic black bear (U. thibetanus), and blue sheep (Pseudois nayaur). Therefore, identifying suitable habitats and ecological corridors for this species can contribute to broader biodiversity conservation in the region. In our research, we employed the MaxEnt model, Circuit Theory, geographic environmental variables, and climate change data to assess both current and potential future shifts in the distribution and migration corridors. This approach aimed to evaluate the impact of climate change on habitat suitability and movement patterns, providing crucial scientific insights for its conservation. Furthermore, the findings contribute to understanding the broader ecological consequences for other sympatric species in the region, thereby informing a more comprehensive conservation strategy for this high-altitude ecosystem.

2. Materials and Methods

2.1. Study Area and Species Occurrences

Ganzi Tibetan Autonomous Prefecture (approximately 149,700 km²) is one of the regions with the richest landscape types, ecosystem types, and biodiversity globally [19]. It not only harbors a vast number of ancient biological groups, but numerous new species have also evolved there. It is considered one of the most pristine areas in China in terms of native ecosystems, with the most intact natural vertical zones, and it is the most representative region of temperate ecosystems worldwide. As an important part of China’s “Three Rivers Source”, it is known as the “Water Tower of China” [20]. This region is home to numerous rivers and lakes, with a well-developed water system. The upper reaches of the Yangtze River, including the Jinsha River, the Yalong River, and the Dadu River, pass through 18 counties within the prefecture. The area boasts a variety of natural protected areas, including 83 sites across six categories: 44 protected areas, 11 forest parks, 17 wetland parks, 6 scenic spots, 3 geoparks, and 2 natural heritage sites. Ganzi Prefecture is rich in wild ecological resources, with 652 species of wildlife (including 98 national key protected species) and 5223 species of plants, making it an important natural gene pool for global species conservation (The People’s Government of Ganzi Tibetan Autonomous Prefecture; https://www.gzz.gov.cn/; accessed on 17 May 2024).

A total of 382 GPS coordinates of the species were collected, of which 286 coordinates were obtained through ground surveys conducted between 2017 and 2023, recording the presence of bears, including coordinates of feces, footprints, hair, and foraging traces. Additionally, 65 coordinates were captured through infrared camera traps between 2022 and 2023, and 31 coordinates were sourced from the published literature [21] (Figure 1). To reduce spatial autocorrelation, we randomly selected one point from each 10 km² grid [22]. Given that the species hibernate from late October or early November until March or April of the following year, field surveys were conducted during periods outside their hibernation phase [22]. Consequently, the habitat predictions presented in this study do not encompass winter habitats.

2.2. Data Source

2.2.1. Bioclimatic Variables

We utilized the mean climate data from 1950 to 2000 provided by the WorldClim 1.4 database (https://www.worldclim.org/; accessed on 20 March 2020), downloading 19 bioclimatic variables (

Table 1. Description of each variable in the model.

No.	Variables	Abbreviation	Description	Resolution
1	Bioclimatic variables	Bio1	Mean Annual Temperature	1 km
2		Bio2	Mean Diurnal Range
3		Bio3	Temperature Constancy
4		Bio4	Temperature Seasonality
5		Bio5	Max Temperature of Warmest Month
6		Bio6	Min Temperature of Coldest Month
7		Bio7	Temperature Annual Range
8		Bio8	Mean Temperature of Wettest Quarter
9		Bio9	Mean Temperature of Driest Quarter
10		Bio10	Mean Temperature of Warmest Quarter
11		Bio11	Mean Temperature of Coldest Quarter
12		Bio12	Annual Precipitation
13		Bio13	Precipitation of Wettest Month
14		Bio14	Precipitation of Driest Month
15		Bio15	Precipitation Seasonality
16		Bio16	Precipitation of Wettest Quarter
17		Bio17	Precipitation of Driest Quarter
18		Bio18	Precipitation of Warmest Quarter
19		Bio19	Precipitation of Coldest Quarter
20	Environmental variables	LUCC	Current (2010) and Future (2070s) Land Use and Land Cover	30 m
21		ELE	Elevation	30 m
22		HII	Human Influence Index	1 km

). A subset of these variables, which were deemed significant based on the physiological requirements of the target species, was selected for use in predicting its distribution [7]. To investigate potential changes under future climate scenarios, we selected three widely applied General Circulation Models (GCMs) in China: the Beijing Climate Center Climate System Model (BCC-CSM1-1), the Community Climate System Model (CCSM4), and the Hadley Global Environment Model 2 (HadGEM2-AO) [23]. These models were used to simulate future species distribution for the period of 2061–2080 (referred to as the “2070s”). Additionally, each model incorporated three different emission scenarios (Representative Concentration Pathways: RCP 2.6, RCP 4.5, and RCP 8.5) to reflect possible variations in greenhouse gas emissions, ranging from the most optimistic to the most pessimistic projections for the coming decades [24].

2.2.2. Additional Variables

We utilized the Finer Resolution Observation and Monitoring of Global Land Cover (http://data.ess.tsinghua.edu.cn/data/Simulation/; accessed on 25 March 2021) for land use and land cover (LUCC) data for 2010 and the 2070s [25]. The spatial distribution of future LUCC was simulated using a downscaling approach based on the Cellular Automata Model, covering four major climate change scenarios (RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5). To maintain consistency with the climate data, we selected 2010 and the 2070s as the key time points for the analysis and RCP 2.6, RCP 4.5, and RCP 8.5 as representative climate scenarios for future LUCC changes. In the construction of the species distribution model, we additionally considered elevation and the Human Influence Index (HII). The elevation data were sourced from the ASTER GDEM V2 digital elevation model (http://www.gscloud.cn/; accessed on 15 January 2019), while the HII data were provided by NASA’s Socioeconomic Data and Applications Center, based on the “Last of the Wild” v2 version (http://sedac.ciesin.columbia.edu/; accessed on 5 July 2018) (Table 1). Due to the unavailability of elevation and HII data for the 2070s, these two variables were treated as static in the model predictions [26].

2.3. Data Processing

All spatial variables were resampled to a 1 km resolution using ArcGIS 10.6 software (ESRI Inc., Redlands, CA, USA), and the coordinate system was standardized to GCS_WGS_1984. To address potential multicollinearity among the variables, we calculated the correlation coefficients for each pair of variables using the Band Collection Statistics (BCS) tool in ArcGIS 10.6. Variables with correlation coefficients |r| > 0.7 were excluded from the analysis [20]. After this procedure, eight variables were selected for modeling the current and future distribution of the species. The retained variables included mean diurnal range, temperature constancy, annual precipitation, precipitation seasonality, precipitation of driest quarter, elevation, human influence index, and land use and land cover.

2.4. Species Distribution Model

The MaxEnt model, based on the maximum entropy principle, is a widely used niche modeling approach in conservation biology research [16,17]. MaxEnt utilizes species presence data and environmental variables to estimate the most probable distribution of a species, simulating its current suitable habitat range and predicting future distributions under varying environmental conditions. This model relies solely on species presence data and environmental variables to simulate the current suitable distribution range of a species and to predict its potential future distribution [16]. In our study, we employed the MaxEnt model to simulate both the current and future distributions of the target species. Model parameters were set as follows: (a) 75% of the species occurrence points were randomly selected as training data for model development, while the remaining 25% were used as testing data for model validation; (b) the importance of environmental variables was evaluated through the Jackknife test method; and (c) model performance was assessed using a resampling method, with 15 repeated experiments to generate response curves, while other parameters were left at their default settings [16,17,27]. The model’s performance was evaluated based on the receiver operating characteristic curve (AUC), where AUC values range from 0 to 1, with values closer to 1 indicating higher predictive accuracy [16]. The importance of each environmental variable in the model was quantified through percentage contribution. The logistic output from MaxEnt represents the probability of species occurrence, ranging from 0 to 1. To convert the continuous probability predictions of species presence into binary maps of suitable and unsuitable habitats, we calculated the mean threshold based on the maximum training sensitivity and specificity (MTSPS) [7]. Grids with probabilities exceeding this threshold were considered a suitable habitat. Furthermore, based on the species’ minimum home range threshold, unsuitable habitats with areas smaller than 10 km² were excluded from the suitable habitat classification [28].

2.5. Altitudinal Distribution and Dispersal Analyses

We performed overlay analyses of elevation and habitat suitability maps to examine changes in the distribution patterns of the target species across different elevation zones. Based on the habitat suitability index (HSI) calculated by the MaxEnt model, we employed Circuit Theory (CT) to simulate potential diffusion paths of the species in both the current and future scenarios [29]. The CT modeling method integrates the concept of “circuits” with random walk theory, using the landscape as a resistance layer to model the movement patterns of random walkers between source and target points. This method has been widely applied in modeling wildlife migration routes and gene flow. In the CT model, habitats that facilitate species migration and dispersal are assigned lower resistance values, while those that impede ecological processes are assigned higher resistance values, following a graph-theoretic data structure. Based on the connectivity of current and future habitats, we used Circuitscape 4.0.5 (https://circuitscape.org/; accessed on 10 August 2024) to simulate the potential diffusion paths of the species. Specifically, we linked suitable habitats to low-resistance diffusion paths while associating unsuitable habitats with high-resistance paths [30]. To ensure the accuracy of the resistance layer, we applied a negative exponential transformation to convert the HSI values into resistance values. The formula is as follows:

If HSI > Threshold → Suitable Areas → Resistance = 1

$$\mathrm{If} \mathrm{HSI}<\mathrm{Threshold}\to \mathrm{Non-suitable} \mathrm{Areas}/\mathrm{Matrix}\to \mathrm{Resistance}={\mathrm{e}}^{{\displaystyle \frac{\ln \left(0.001\right)}{\mathrm{threshold}}}\times \mathrm{HSI}}\times 1000$$

3. Results

3.1. Model Performance

In the MaxEnt model, we used 144 species presence points to construct the species distribution model. Based on multicollinearity tests, eight variables were selected as input variables for the model. Climate and environmental factors jointly influence the species’ distribution under the three RCP scenarios. Overall, climate and elevation variables had the greatest contributions, with annual precipitation, temperature constancy, and elevation making the most significant contributions to the species distribution, accounting for 22.1%, 21.5%, and 19.3%, respectively (

Table 2. Analysis of variable contributions.

Variable	Percent Contribution (%)	Permutation Importance (%)
bio12	22.1	24.3
bio3	21.5	32.8
ele	19.3	12
lucc	17.2	11.8
bio15	8.7	8.9
bio17	8.3	5.4
bio2	1.6	3.8
hii	1.2	1

). The cross-validation results indicate that the species distribution model performs well, with an average test AUC of 0.78 and a standard deviation of 0.05 for the replicate runs.

3.2. Climate-Induced Variations in Suitable Habitat Distributions

The presence probability distributions of the species under current and future climate scenarios and land use changes are shown in Figure 2a and Figure 3, respectively. The average logistic threshold value of MTSPS for the model output was 0.33. Based on this threshold, we obtained current (Figure 2b) and future (Figure 4) binary distribution maps for the species. Under current climatic conditions, the total suitable habitat area within the study region was 70,969.78 km², with Shiqu and Litang possessing the largest suitable habitat areas, accounting for 24.39% and 15.86% of the total, respectively. Under future climate scenarios, the total suitable habitat area was projected to decrease to 63,180.52 km² (RCP 2.6), 62,146.3 km² (RCP 4.5), and 54,290.93 km² (RCP 8.5), representing reductions of 7789.26 km², 8823.48 km², and 16,678.85 km², respectively, compared to current conditions. From a spatial perspective, the changes in suitable habitat areas were most pronounced in Yajiang and Xinlong. Under the RCP 8.5 scenario, Yajiang’s suitable habitat area decreased by 1365.65 km², a reduction of approximately 64.5%, the largest proportional decrease among all counties. Similarly, Xinlong experienced a reduction of 2394.78 km², equating to a 53.4% decrease (

Table 3. The potential distribution area under current and future climate change scenarios.

County	Current		RCP 2.6		RCP 4.5		RCP 8.5
	Area/km2	Proportion/%	Area/km2	Proportion/%	Area/km2	Proportion/%	Area/km2	Proportion/%
Batang	5027.66	7.08	3698.34	5.85	4931.10	7.93	3908.79	7.20
Bayu	5622.69	7.92	4471.74	7.08	4099.64	6.60	3035.34	5.59
Danba	233.50	0.33	228.92	0.36	190.30	0.31	169.70	0.31
Daocheng	4707.58	6.63	3873.22	6.13	2826.79	4.55	2780.28	5.12
Daofu	597.26	0.84	433.21	0.69	443.28	0.71	303.33	0.56
Dege	4829.94	6.81	4667.60	7.39	4862.12	7.82	4629.74	8.53
Derong	725.35	1.02	606.86	0.96	766.43	1.23	378.35	0.70
Ganzi	2741.77	3.86	2874.43	4.55	3216.21	5.18	2992.32	5.51
Jiulong	982.66	1.38	706.58	1.12	777.82	1.25	326.65	0.60
Kangding	1418.00	2.00	1436.57	2.27	965.27	1.55	1165.67	2.15
Litang	11,253.51	15.86	10,022.14	15.86	9422.72	15.16	7740.76	14.26
Luding	391.78	0.55	516.00	0.82	241.56	0.39	357.14	0.66
Luhuo	1041.55	1.47	914.97	1.45	958.36	1.54	476.78	0.88
Seda	3962.37	5.58	4040.19	6.39	4254.41	6.85	3800.01	7.00
Shiqu	17,307.07	24.39	16,455.19	26.04	16,609.27	26.73	16,268.06	29.96
Xiangcheng	3518.58	4.96	3343.86	5.29	2955.14	4.76	3109.93	5.73
Xinlong	4493.18	6.33	3547.38	5.61	3210.26	5.17	2098.40	3.87
Yajiang	2115.33	2.98	1343.32	2.13	1415.62	2.28	749.68	1.38

).

3.3. Changes in the Elevation Distribution of Suitable Habitats

Under current climate conditions, suitable habitats in the study area were primarily concentrated in the 4001–5000 m elevation range, with relatively smaller areas of suitable habitats in the low-elevation ranges (1112–2000 m and 2001–3000 m), especially in the 1112–2000 m range. Under future climate scenarios, the distribution of suitable habitats was projected to undergo significant changes. Regardless of whether the scenario was RCP 2.6, RCP 4.5, or RCP 8.5, high-elevation areas (4001–5000 ms) remained the primary suitable habitat zones. However, with the intensification of climate change, the suitable habitat area in high-elevation regions showed a gradually increasing trend. In the RCP 2.6 scenario, the number of suitable habitat grid cells in the 4001–5000 m range increased to 87,532; in the RCP 4.5 scenario, this further increased to 88,954; and in the RCP 8.5 scenario, the suitable habitat area further grew to 98,741, reaching its maximum value. In contrast, the suitable habitat area in the low-elevation range (especially in the 1112–2000 m range) showed a marked reduction under future climate scenarios. As climate warming progressed, the altitude distribution of suitable habitats for species shifted upward (Figure 5).

3.4. Distribution Characteristics of Dispersal Paths

Under current climate conditions, the dispersal paths of the bears exhibited a northwest-to-southeast orientation, primarily concentrated in areas such as Dege, Ganzi, Xinlong, and Luho. The southern dispersal paths were more concentrated in areas including southern Yajiang, southern Kangding, southern Litang, and eastern Daocheng. By 2070, under all three emission scenarios, the dispersal paths of this species were projected to show an increase in dispersal resistance, a decrease in the density of dispersal paths, and a reduction in the connectivity of these paths. Specifically, under the RCP 2.6 scenario, the dispersal path density in the central region of the study area significantly decreased and dispersal resistance increased, with the dispersal paths in the southern part of Litang being almost entirely lost. Under the RCP 4.5 scenario, the dispersal paths became narrower, the length of dispersal corridors decreased, and the paths extended northward, with the southern dispersal paths largely disappearing. In the RCP 8.5 scenario, the southern dispersal paths were completely lost, and the overall dispersal paths gradually contracted toward the northwest, with the paths being primarily concentrated in the counties of Ganzi, Baiyu, and Dege (Figure 6).

4. Discussion

Although the primary drivers of global biodiversity loss are anthropogenic activities that disrupt landscape patterns and exploit natural resources, it is precisely these prominent factors that often overshadow the more subtle yet equally serious threat of climate change [1,2]. While the potential impacts of climate change may not be as immediately visible as direct human activities, its long-term effects on ecosystems and species distribution could exacerbate biodiversity loss in ways that are less perceptible [1]. For instance, from a long-term perspective, climate change not only leads to habitat loss but also exacerbates habitat fragmentation and resistance to species dispersal pathways, further intensifying the pressure on species’ ability to adapt to environmental changes [31]. Our study revealed the potential impacts of climate change on the distribution and dispersal paths of the bears, particularly with respect to changes in habitat suitability, elevation distribution, and habitat connectivity. The results indicated that climatic and environmental factors, particularly annual precipitation, temperature constancy, and elevation, played a decisive role in shaping the species’ current distribution pattern. These variables contributed most significantly to the model, highlighting the bears’ high sensitivity to changes in climate conditions. As climate factors, annual precipitation and temperature constancy directly influenced vegetation coverage and water distribution in the region, thereby affecting the bears’ habitat suitability [7].

Elevation, as a critical factor in habitat selection, determined the spatial distribution of suitable habitats [32]. With the intensification of climate change, a noticeable reduction in suitable habitats for the bears was observed. In particular, under the RCP 8.5 scenario (the most extreme emission scenario), the suitable habitat area was projected to decrease by approximately 16,678.85 km², primarily occurring in low-elevation regions. Low-elevation areas such as Yajiang and Xinlong, which have historically served as important habitats for the bear, were expected to experience significant habitat reduction in future climate scenarios. As climate conditions in these areas became increasingly unsuitable, the bears were forced to migrate to higher elevations. This migration process not only increased pressure on high-altitude ecosystems but also led to overcrowding and resource competition in newly concentrated habitats. While the habitat suitability in high-elevation regions remained relatively favorable, the reduction in suitable habitats at lower elevations pushed bear populations into limited high-altitude habitats. This posed a severe challenge to the long-term survival of the species, as the available space and resources might not have been sufficient to support growing populations.

This reduction in suitable habitats in low-elevation areas not only directly impacts the living space of the bears but may also trigger a series of cascading effects [33]. Firstly, the species faces increased habitat pressure due to the reduction in high-quality habitats. [34]. Secondly, climate change affects the types, distribution, and availability of food resources for brown bears, thereby influencing their survival. The reduction and alteration of food resources will exacerbate survival pressures on brown bears, particularly in the context of habitat loss or deterioration in habitat quality [35]. Therefore, future climate change will significantly increase the vulnerability of the bears’ habitats, particularly in low-elevation areas that are already in a critical state under current climatic conditions. Our study further revealed that elevation plays a critical role in shaping the distribution of suitable habitats for the bears. Suitable habitats were primarily concentrated in the 4001–5000 m elevation range, which, under future climate scenarios, will remain a core area for the species. However, the warming climate is expected to lead to a significant reduction in suitable habitats in lower-elevation areas (1112–2000 m), forcing the bears to migrate to higher elevations. Although this upward migration may provide additional habitat space in the short term, it will, in the long run, exacerbate the pressure on high-altitude ecosystems. The concentration of brown bear populations in these areas will lead to several ecological challenges, including increased competition for limited food resources, heightened disturbance to fragile alpine vegetation, and potential disruptions to the balance of high-altitude predator–prey dynamics. Moreover, the influx of bears may also lead to overcrowding, further stressing local ecosystems and potentially accelerating habitat degradation, as high-altitude environments are generally more sensitive to environmental changes and have less capacity to recover from such pressures.

As habitats continue to shrink, particularly under future climate scenarios, the degree of habitat isolation will intensify, making species’ dispersal paths increasingly complex and difficult. Specifically, under the RCP 8.5 scenario, the most extreme emissions scenario, the connectivity of suitable habitats for the bears significantly decreases. Our study indicated that in southern regions, the dispersal pathways almost completely vanish, and the bears’ habitat is fragmented into several isolated patches. The remaining dispersal paths are largely concentrated in the northwest, such as in the regions of Ganzi, Baiyu, and Dege. Although these areas still retain some degree of habitat connectivity, spatial constraints and environmental pressures, such as fragmented landscapes and limited food resources, significantly impair the species’ dispersal capacity. These factors create barriers that hinder the bears’ ability to move across suitable habitats, ultimately reducing their genetic diversity and adaptability and intensifying risks to their long-term survival.

As climate change increasingly impacts the bears’ habitat, the species faces significant survival threats. The reduction in habitat area, the intensification of habitat fragmentation, and the obstruction of dispersal paths directly affect the species’ distribution, population dynamics, and gene flow. Climate change not only accelerates the degradation of habitats but also challenges the species’ ability to adapt to environmental changes. In particular, under future extreme climate scenarios, the species will face risks such as habitat loss, population isolation, and genetic bottlenecks, further increasing its extinction risk. Therefore, it is essential to implement effective conservation measures to mitigate the negative effects of climate change on the bears.

(1)

Strengthening habitat protection in key areas such as Dege, Ganzi, and Xinlong

The results indicated that areas like Dege, Ganzi, and Xinlong are core habitats for the bears, with large areas of suitable habitats. However, under future climate change scenarios, these regions are expected to experience significant habitat reductions, especially under the RCP 8.5 scenario, where suitable habitats are projected to shrink significantly. To reduce these impacts, it is crucial to enhance conservation efforts in these areas, including expanding the boundaries of existing protected areas, implementing strict protection measures, and reducing human activities that interfere with habitat quality, such as mining and overgrazing.

(2)

Protecting and restoring ecological corridors in southern regions like Yajiang, Kangding, and Litang

Research has shown that southern regions such as Yajiang, Kangding, and Litang are essential dispersal corridors for the bears. However, with the intensification of climate change, these dispersal paths are gradually disappearing, especially under the RCP 8.5 scenario, where southern dispersal routes will be almost completely lost. To ensure that the species can adapt to habitat changes brought about by climate change, priority should be given to the construction of ecological corridors in these areas to facilitate migration and gene flow. This can be achieved by restoring degraded ecological corridors and rebuilding vegetation, thereby enhancing the connectivity between habitat patches.

(3)

Enhancing habitat protection in high-elevation areas, particularly in Litang, Daocheng, and Baiyu

According to this study, with climate change, the bears’ habitat is gradually shifting to higher elevations, especially in the 4001–5000 m range. In the future, areas such as Litang, Daocheng, and Baiyu may see an increase in suitable habitats, but these regions also face the risk of fragmented dispersal paths. Therefore, it is important to strengthen protection in these high-elevation areas to ensure the species’ survival. This can be achieved by establishing new high-altitude protected areas, enhancing ecological restoration efforts in these regions, and conducting scientific monitoring of high-altitude ecosystems to assess their health and stability.

(4)

Implementing habitat restoration measures in low-elevation areas, especially in Yajiang and Xinlong

Our study found that future climate change will significantly reduce suitable habitats in low-elevation areas such as Yajiang and Xinlong. To mitigate habitat loss in these regions, habitat restoration projects should be implemented to restore degraded ecosystems and expand suitable habitats for the species. This could include measures such as vegetation restoration, the creation of artificial water sources, and improving soil quality to enhance the ecological functionality of low-elevation habitats and increase habitat suitability.

(5)

Strengthening regional cooperation

Our findings showed that the bears’ habitat and dispersal paths span multiple administrative regions, particularly in areas like Ganzi, Dege, and Shiqu, as well as the neighboring Xizang Autonomous Region and Qinghai Province. To address the challenges posed by climate change, it is crucial to enhance cross-regional cooperation, integrate resources, and jointly conduct conservation efforts. Such cooperation would facilitate information sharing, scientific collaboration, and data integration, providing a more comprehensive foundation for habitat protection. Joint monitoring and assessment across regions would help precisely understand changes in species distribution, habitat degradation, and environmental threats, which can inform more scientifically based conservation strategies. Furthermore, different regions should develop tailored management measures based on local ecological characteristics and conservation needs, implementing consistent protection standards across the region to avoid fragmented conservation efforts.

5. Conclusions

Our study thoroughly explored the potential impacts of climate change on the habitat suitability, distribution patterns, and dispersal paths of a bear species. The results indicated that climate variables, such as annual precipitation and temperature constancy, along with elevation, were critical factors in shaping the distribution of the species. Furthermore, the bears were found to be highly sensitive to climate change. Specifically, climate change led to a significant reduction in suitable habitats for the bears, especially in low-elevation areas, forcing the species to migrate to higher altitudes, thus further intensifying pressure on high-altitude ecosystems. Through an analysis of future climate scenarios, this study revealed a decrease in suitable habitats, particularly under the RCP 8.5 scenario, with an estimated reduction of approximately 16,678.85 km² in habitat area. This change will pose a major challenge to the species’ survival, particularly in the context of increased habitat fragmentation and reduced population connectivity. The fragmentation of habitats due to climate change and resistance to dispersal paths will further constrain the species’ ability to expand, leading to population isolation and a reduction in genetic diversity. This, in turn, will diminish the species’ ability to adapt to environmental changes. We emphasized the climate change challenges faced by the bears and suggested key conservation measures, such as enhancing habitat protection, establishing ecological corridors, and focusing on the conservation of high-altitude ecosystems. To address the impacts of climate change, future conservation efforts should prioritize maintaining habitat connectivity, restoring low-elevation habitats, and ensuring the sustainable management of high-altitude ecosystems. Additionally, as the effects of climate change on habitat suitability and distribution patterns become increasingly more significant, further research should integrate more refined ecological data to optimize model predictions and provide a solid theoretical foundation for formulating scientific conservation strategies.