Ling Mao
Chunling Wang
Feng Deng
Xiaying Ye- Agronomy and Life Science Department, Zhaotong University, Zhaotong, Yunnan, China
Fargesia nitida (Mitford) P. C. Keng ex T. P. Yi is an alpine bamboo species endemic to the Hengduan Mountains (HDM). As the primary food source for giant pandas and a key component of subalpine ecosystems, it plays an irreplaceable ecological role. However, its exceptionally long flowering cycle (up to 109 years) and limited dispersal ability render it highly vulnerable to climate change impacts, while the potential shifts in its distribution under future climate scenarios remain inadequately explored. In this study, we employed ensemble modeling (Biomod2) to project the suitable habitats under current and future climate scenarios (SSP126, SSP245, SSP585) for the 2050s and 2090s, based on 78 occurrence records and seven environmental variables. Results showed that the ensemble model exhibits superior predictive performance (AUC = 0.995, TSS = 0.957, Kappa = 0.753) compared to optimized MaxEnt and other individual models. Temperature seasonality (bio4) and minimum temperature of the coldest month (bio6) were identified as the most influential factors. Currently, the total suitable habitat area is estimated at 83.10 × 104 km², with a central distribution located primarily in the HDM region and its adjacent areas. Although future projections suggest an overall expansion of the total suitable area, the highly suitable habitats—corresponding to the species’ current occurrence area—show a persistent contraction. By the 2090s, this contraction will exceed 56% under the SSP585 scenario. Distribution centroid analyses revealed that the current center of F. nitida is located in northern Sichuan. It will shift northward to higher latitudes under the low emission scenario, but southwestward to higher altitudes under the high emission scenario. These findings underscore the vulnerability of F. nitida to climate change and provide critical scientific insights for the development of targeted conservation strategies, as well as for the effective management of giant panda habitats and subalpine ecosystems under future climatic conditions.
1 Introduction
Climate change, characterized by rising temperatures and altered precipitation patterns, poses a severe threat to global biodiversity (Arneth et al., 2020; Wudu et al., 2023), especially for mountainous biodiversity hotspots like the Hengduan Mountains (HDM). As a core hotspot in southwestern China, the HDM harbors over 3500 endemic seed plant species, accounting for approximately 29% of China’s total endemic seed flora (Mittermeier et al., 2004). However, these specialized species are confronting heightened risks from rapid climate shifts (He et al., 2019). Projections have indicated that under a warming scenario of 3.2 °C above the pre-industrial levels by 2100, about 49% of insects, 44% of plants, and 26% of vertebrates are expected to lose more than half of their historical geographic ranges (Warren et al., 2018). This loss will be further compounded by cascading extinctions and reduced ecosystem stability, exacerbating overall vulnerability (Koh et al., 2004; Correia and Lopes, 2023). Thus, predicting the future climate suitability for these endemic species in the HDM is critical for guiding targeted conservation efforts.
Climate change reduces the fitness of species by altering phenological patterns, disrupting interspecific interactions, and shifting climatic niches, thereby driving population declines and ultimately increasing extinction risks (Bellard et al., 2012). Species with restricted distributions and slower migration rates are especially susceptible to these impacts (Zhang et al., 2024; Alves-Ferreira et al., 2025), and bamboos (Poaceae: Bambusoideae) are a typical example. As the only major grass lineage diversified in forests, bamboos offer important environmental, social and economic benefits, but their exceptionally long reproduction intervals (up to 120 years) and short rhizomes (as short as 1 cm) limited dispersal capacity, leading to geographically constrained distributions—particularly among sympodial bamboo species (Janzen, 1976; Li et al., 2006). These characteristics prevent bamboos from keeping pace with the rapidly shifting climate envelopes, leaving them exposed to threats posed by accelerating climate change (Li et al., 2019; Zhao et al., 2023; Wang et al., 2024). Therefore, accurately evaluating climate impacts on bamboo species is critical for formulating effective conservation and development strategies— especially for the alpine bamboo genus Fargesia Franchet. Globally, there are ca. 90 Fargesia species, 86% (77 species) of which are endemic to China, with 67% (60 species) concentrated in the HDM (Li et al., 2006; Ye et al., 2019).
Fargesia nitida (Mitford) P. C. Keng ex T. P. Yi is an HDM-endemic understory bamboo species at an altitude of 2450 m to 3200 m. It possesses a typical caespitose sympodial rhizome (rhizome neck 10–13 cm, Figure 1) and an unusual extended flowering cycle, with intervals reaching up to 109 years (Ma et al., 2017). Furthermore, this species constitutes approximately 90% of the giant panda’s (Ailuropoda melanoleuca, IUCN Vulnerable) diet (Yi and Jiang, 2010). Its decline—driven by altered precipitation and temperature—threatens both pandas and subalpine forests. For example, historical records show that more than 200 giant pandas perished due to food scarcity during its flowering-induced mortality event in the 1980s (Yi et al., 2008). Molecular dating analyses of alpine bamboos indicate that F. nitida originated during the late Pleistocene and the Fargesia genus has likely suffered a decline in diversification since approximately 3 million years ago (Ye et al., 2019). This pattern suggests that climate change has long exerted negative effects on the species’ growth and evolutionary trajectory. While existing studies on F. nitida have primarily focused on its ecological characteristics and its relationship with giant panda (Wang et al., 2009; Guo et al., 2023; Shi et al., 2025), systematic analyses of its distribution pattern under future climate change scenarios remain limited.
Figure 1. Fargesia nitida. (A) Sampling site; (B) Habitat; (C) Rhizome.
Species distribution models (SDMs) have become essential tools for evaluating the impacts of climate change on species distribution and are widely applied in the field of conservation biology (Guisan and Zimmermann, 2000; Chen et al., 2025). These models project the potential species distributions and the associated ecological needs based on existing distribution data and relevant environmental variables (Srivastava et al., 2019). Currently, more than ten models have been developed, such as random forest (RF), generalized additive model (GAM), and the maximum entropy model (MaxEnt). Most studies employed individual model to forecast species distribution pattern under future climate scenarios (Santini et al., 2021). However, different SDMs are established based on distinct algorithmic principles and thus exhibit unique strengths and limitations when applied to the same species (Bellard et al., 2012; Qiao et al., 2019). The Biomod2 package, an ensemble modeling technique that integrates multiple models, is regarded as an effective approach to enhance accuracy and optimize predictive results (Yang et al., 2024; Kang et al., 2025). Meantime, area under the curve (AUC) of the receiver operating characteristic (ROC) curve, true skill statistics (TSS), and Kappa values can be used to assign weights to each model, yielding the best simulation outcomes.
In this study, we focus on F. nitida, the primarily food source of the giant panda, to assess the impacts of climate change on its distribution pattern using ensemble modeling. Our objectives are as follows: (1) identify the critical environmental variables influencing its distribution; (2) determine its potential distribution range in China under current conditions and two future periods (the 2050s and the 2090s) with three different climate scenarios (SSP126, SSP245, SSP585); (3) predict the migration trends of its habitat centroid from present to the future. This study will provide a theoretical basis for the scientific utilization, management and conservation of F. nitida, and hold practical significance for the protection of giant pandas and the maintenance of ecosystem stability.
2 Materials and method
2.1 Collection of distribution data
To obtain the occurrence data of F. nitida in China, we carried out the collection through the following pathways: (1) extensive field surveys conducted from 2019 to 2022; (2) comprehensive database search, including Chinese Virtual Herbarium Database (CVH: https://www.cvh.ac.cn/), Global Biodiversity Information Facility (GBIF; https://www.gbif.org), and National Specimen Information Infrastructure (NSII: https://www.nsii.org.cn/); (3) relevant published literatures (Zhang et al., 2019; Zhou et al., 2020; Huang et al., 2021). A total of 145 records were initially obtained, of which 51 were collected from our field surveys, 44 were retrieved from online database, and 50 were extracted from published literatures. For five records with specific locations but lacking GPS data, Baidu Maps was used to estimate their coordinate information. Data with ambiguous geographic details, duplicates, and outliers were excluded. Meanwhile, to migrate the sampling bias, we further processed the geographic distribution data using the ENMTools approach (Warren et al., 2010), with only one point retained per 30-second grid cell (1 km × 1 km). Finally, 78 reliable distribution points of F. nitida were confirmed (Figure 1; Supplementary Table S1).
2.2 Selection and processing of environmental variables
Initially, 56 environmental variables were selected in this study, including 19 bioclimate variables, 34 soil variables, and three topographic variables. Bioclimatic data were obtained from WorldClim v2.1 (https://worldclim.org/), covering the current period (1970-2000) and two future period: the 2050s (2041-2060) and 2090s (2081-2100). For future climate projections, the Beijing Climate Center Climate System Model (BCC-CSM2-MR) from the Coupled Model Intercomparison Project 6 (CMIP6) was employed, as it is well-suited for simulating bioclimate changes in China. Three Shared Socioeconomic Pathways (SSP) were selected to provide a more comprehensive understanding of the distribution pattern of F. nitida: SSP126, SSP245 and SSP585. These scenarios correspond to a sustainable development pathway (with global warming of 1.8°C and radiative forcing reaching 2.6 W/m² by 2100), a moderate development pathway (2.7°C and 4.5 W/m²), and a traditional high-emission pathway (4.4°C and 8.5 W/m²), respectively (IPCC, 2023). Soil variables were sourced from the Harmonized World Soil Database v2.0 (HWSD2, https://www.fao.org/soils-portal/), with D1 data extracted therefrom. Elevation data was obtained from the Geospatial Information Authority of Japan (https://globalmaps.github.io/el.html), while slope and aspect were derived from it using ArcGIS 10.7. Soil and topographic data were assumed to remain unchanged in the future. All environmental variable layers were resampled to a uniform 30 arc-second resolution and rasterized to align with the occurrence data layer in terms of boundaries, cell size, and coordinate system.
To reduce the impact of multicollinearity on prediction accuracy, we conducted a correlation analysis on the 56 environmental variables (Supplementary Table S2). First, we performed pre-simulation of all environmental factors in MaxEnt v.3.4.4 and excluded variables with a contribution percentage below 1% (Qiu et al., 2023b). Second, For the remaining variables, we used the Pearson method in SPSS v.23 to test correlations, when two variables exhibited a strong correlation (|r| > 0.8), the one with a higher contribution was retained. Finally, seven environmental variables were selected for subsequent ensemble modeling (Table 1).
Table 1. Environmental variables retained after processing.
2.3 Construction and evaluation of the ensemble model
Firstly, we optimized two key parameters of MaxEnt—regularization multiplier (RM) and feature class (FC)—using the R package Kuenm to enhance the model’s accuracy before ensemble modeling (Cobos et al., 2019). For the RM parameter, we set a range from 0.5 to 6 with an interval of 0.5. For FC, we combined the five features of MaxEnt (linear (L), quadratic (Q), product (P), threshold (T), and hinge (H)) and generated 31 combinations. A total of 1209 candidate models were tested, and their performance was evaluated based on the Akaike Information criterion correction (AICc) and 5% Test Omission Rate (OR). Parameters with lower delta AICc and OR values were selected (Gao et al., 2024).
After parameter optimization, we first evaluated the accuracy of ten individual models—Generalized Linear Model (GLM), Generalized Boosting Model (GBM), GAM, Classification Tree Analysis (CTA), Artificial Neural Network (ANN), Surface Range Envelop (SRE), Flexible Discriminant Analysis (FDA), Multivariate Adaptive Regression Splines (MARS), RF, and optimized MaxEnt—using kappa, TSS, and AUC values. Then an optimal ensemble model was constructed from models with kappa > 0.70, TSS > 0.85, and AUC > 0.95. Through comparing the predictive accuracy of individual models and the ensemble model, we selected the latter one for subsequent prediction due to its better performance. All model construction processes were implemented in the R package Biomod2 with 10 repetitions. Key parameters were set as follows: 75% of the 78 distribution points were randomly selected as the training data, with the remaining serving as the test data; all models employed default parameters in Biomod2 except MaxEnt, which used RM and FC combinations defined by the Kuenm package. To reduce sampling bias, 1000 pseudo-absence points with 2 replications were randomly generated and included in the modeling process.
The performance of the ensemble model was evaluated using AUC, kappa and TSS. The AUC value ranges from 0 to 1, while kappa and TSS values range from -1 to 1, with higher values indicating more reliable prediction results. Common evaluation criteria are as follows: for AUC, values between 0.50 and 0.70 indicate poor performance, 0.70 < AUC ≤ 0.80 indicate fair performance, 0.80 < AUC ≤ 0.90 indicate good performance, and values above 0.90 represent excellent performance; For kappa and TSS, values between 0.40 to 0.50 indicate poor performance, 0.50 < kappa/TSS ≤ 0.70 indicate fair performance, 0.70 < kappa/TSS ≤ 0.85 indicate good performance, and values above 0.85 represent excellent performance.
2.4 Analysis of key environmental factors
To identify the critical environmental variables influenced the distribution of F. nitida, we calculated the weights of the selected environmental variables in each model and selected the top three ones. Response curves of the significant environmental factors were plotted using “jackknife” method in the optimized MaxEnt model.
2.5 Analysis of suitable habitat patterns
The suitable areas for F. nitida under both current and future climate scenarios were simulated using the ensemble model. Projection outputs were imported into ArcGIS 10.7 and reclassified based on the ecological criteria of Fargesia denudate, a congeneric species distributed in a similar region (Hu and He, 2020). These areas were divided into four categories: unsuitable areas (< 0.25), low suitable areas (0.25-0.50), moderately suitable areas (0.50-0.75), and highly suitable areas (> 0.75). The area of suitable habitats for F. nitida under each climate scenario was calculated using ArcGIS 10.7. Binary maps were then generated to determine the changing pattern of habitats under different emission scenarios, with contraction, expansion, and unchanged areas calculated using the “distribution changes between binary SDMs” tool from the SDM Toolbox v2.5 (Brown, 2014). Finally, the “centroid changes clines” tool in ArcGIS was used to identify the shift direction of the distribution centroid.
3 Results
3.1 Model evaluation and key environmental factors
Using the Kunem package, we optimized the two key parameters of MaxEnt based on 78 distribution points of F. nitida and seven environmental factors. The delta AICc and OR values of the default parameters (RM = 1, FC = LQHPT) were 103.10 and 0.12, respectively, whereas these values decreased substantially to 0 and 0.04 under the parameter configuration of RM = 0.1 and FC = LQ. Hence, this optimized configuration was selected for subsequent MaxEnt analyses. Meanwhile, comparison of the evaluation metrics across all models revealed that the ensemble model achieved the highest kappa, TSS and AUC values, reaching 0.753, 0.957 and 0.995, respectively (Figure 2), indicating optimal prediction accuracy. Consequently, the ensemble model was chosen for projecting the distribution pattern of F. nitida.
Figure 2. Model evaluation of the ensemble model and ten individual models based on kappa, TSS and AUC.
The importance values of bioclimatic variables across all individual models revealed that temperature seasonality (bio4) and min temperature of coldest month (bio6) had the greatest impact on the geographic distribution pattern of F. nitida, followed by elevation, precipitation of driest month (bio14), and annual precipitation (bio12) (Table 2). Notably, in the ANN model, bio12 emerges as the most important factor, indicating that precipitation may also play an important role in shaping the distribution pattern of F. nitida. In contrast, the effects of soil type (WRB4) and slope were negligible.
Table 2. Importance values of environmental variables influencing the distribution of F. nitida.
Response curves can illustrate the quantitative relationship between the logistic occurrence probability and the appropriate range of environmental variables. Generally, a probability exceeding 0.5 indicates that such an environment is conducive to the growth of bamboos. As shown in Figure 3, F. nitida exhibits favorable growth and occurrence when temperature seasonality ranges from 613% to 834%, min temperature of coldest month from -11.43 °C to -2.79 °C, elevation from 1324 m to 3457 m, annual precipitation from 630 mm to 991 mm, and precipitation of driest month from 3.20 mm to 11.65 mm.
Figure 3. Response curve for the main environmental variables, including temperature seasonality bio4), min. temperature of coldest month (bio6), annual precipitation (bio12), precipitation of driest month (bio14), and elevation.
3.2 Current potential suitable habitat of F. nitida
As determined by the ensemble model, the current suitable habitat for F. nitida is mainly located in the HDM region and its adjacent areas, ranging from 102° E – 113° E, 29° N – 36° N. The total suitable area is 83.10 × 104 km2, accounting for 8.69% of China’s land area. However, within this total area, the highly suitable area constitutes a relatively small proportion (only 39.51%), covering 32.84 × 104 km². Highly suitable habitats are mainly concentrated in Sichuan, Gansu, Shaanxi, and Henan, and almost completely overlap with the actual distribution points (Figure 4). Moderately suitable areas are mainly distributed around the highly suitable habitats, with sporadic distribution in Hubei, Yunnan and Xizang, covering 16.18 × 104 km², which accounts for 19.47% of the total suitable area. Low suitable areas, where no sampling points have been recorded, are primarily located in southern Sichuan, Henan, Shandong, and Shanxi, with scattered occurrences in Yunnan, Guizhou, Xizang and Xinjiang.
Figure 4. Potential geographical distribution of Fargesia nitida at current condition.
3.3 Dynamic changes in suitable habitats under different climate scenarios
With ongoing environmental changes, the suitable habitat of F. nitida remains concentrated in the HDM region and its adjacent areas, but the total suitable area exhibits an expanding trend across climate scenarios (Figures 5, 6; Supplementary Figure S1). Under the SSP126 scenario, the total suitable area will increase to 97.07 × 104 km² by the 2050s period and then decrease to 85.18× 104 km² by the 2090s, representing a 17.89% and 2.5% increase compared to the current scenario, respectively. Under the SSP245 scenario, the total suitable area will continue to increase, reaching 116.05 × 104 km² by the 2050s and 120.01 × 104 km² by the 2090s, with respective growth ratio of 36.95% and 44.41%. Under the SSP585 scenario, an even greater increase in the total suitable area is projected, reaching 112.73 × 104 km² by the 2050s and 135.02 × 104 km² by the 2090s, with the growth rate reaching 62.48% in the 2090s.
Figure 5. Potential suitable areas for Fargesia nitida under future climate scenarios. 2050s represents 2040–2060s period and 2090s represents 2080–2100 s period. SSP126, SSP245, and SSP585 represents different shared socioeconomic pathways.
Figure 6. Variation in suitable area for Fargesia nitida under different climate scenarios. SSP126, SSP245, and SSP585 represents different shared socioeconomic pathways.
Comparison suitability zones (Figure 6; Supplementary Figure S1) revealed that the total suitable area expansion is primarily driven by increased low suitable area, supplemented by a slight growth in the moderate suitable areas. However, the highly suitable areas—the primarily distribution zone of F.nitida—decline consistent across all scenarios, indicating that the actually suitable area for this bamboo species is decreasing. While the expanded area exceeded the contracted area overall, the extent of expansion and contraction varies across different climate scenarios, which could lead to changes in the distribution pattern of F. nitida. For example, both the greatest expansion and contraction occurred at the SSP585 scenario in the 2090s, whereas the SSP245 scenario in the 2090s experiences the largest increase in the moderately suitable area.
Along with changes in suitable habitats, the spatial distribution centroid of F. nitida is projected to shift under future climate scenarios, as shown in Figure 7. Currently, the distribution centroid of F. nitida is situated at the northern border of Sichuan province (105°51′32.02″ E, 32°28′32.35″ N), and it will migrate to different directions under various emission scenarios. For instance, under the sustainable development scenario (SSP126), the species shift northward by 56 km in the 2050s and 61 km in the 2090s. Under the moderate development scenario (SSP245), F. nitida first move 90 km northwestward in the 2050s, then shift slightly southwestward by 21 km in the 2090s. Such northward or northwestward shift in the low-emission scenarios (SSP126 and SSP245) is primarily driven by the expansion of low-moderate suitable habitats in northern Gansu, Shaanxi, and central Shanxi (expansion ca. 50%) and the contraction of highly suitable area in northern Sichuan (20%-30% loss) (Figure 5; Supplementary Figure S1). Under the traditional high-emission scenario (SSP585), the species will continuously migrate southwestward, with distances of 191 km in the 2050s and 160 km in the 2090s. This southwestward shift is dominated by the contraction of core highly suitable area in Sichuan and Shaanxi (with a loss of up to 56%) and the expansion of marginal habitats toward Xizang, Qinghai and western Sichuan. Concurrently, the altitude of the distribution centroid will gradually increases (Figure 7), particularly under the SSP585 scenario. As the species’ range shift to the Qinghai-Tibet Plateau, the altitude of its distribution centroid will rise to 3343 m by the 2050s and reach 4396 m by the 2090s. These findings demonstrate that different climate scenarios exert a significant influence on the distribution pattern of F. nitida.
Figure 7. The migration trajectory of distribution centroids of Fargesia nitida under different climate scenarios. SSP126, SSP245, and SSP585 represents different shared socioeconomic pathways.
4 Discussion
4.1 Optimization and limitation of SDMs
In the last thirty years, SDMs have been widely applied in conservation biology, ecology, and evolutionary biology to guide species conservation and management efforts (Wudu et al., 2023; Zhang et al., 2023). However, challenges and caveats have been pointed out regarding their predictions of species’ future distribution (Peterson et al., 2018), and parameter setting and model selection are critical to improve the accuracy of species distribution modeling (Bellard et al., 2012; Gao et al., 2024). Among SDMs, MaxEnt has been extensively employed to predict the distributions of endangered species, alien species and pests for its better performance than most other models (Elith et al., 2006). Furthermore, it yield more accurate predictions when using optimized configurations than those using default settings (Fang et al., 2022; Zhao et al., 2024). However, different models may produce biased predictions for their varied algorithm principles and assumptions. To address this limitation and improve prediction accuracy, we integrated the optimized MaxEnt model with biomod2 to construct an ensemble model. By combing models with kappa > 0.70, TSS > 0.85, and AUC > 0.95, the ensemble model achieved higher accuracy than both the optimized MaxEnt model and other individual model across all evaluation metrics (Figure 2). Both TSS and AUC values of the ensemble model reached the excellent level (TSS: 0.907; AUC: 0.988), indicating that our ensemble model produces credible simulation results and can reliably predict shifts in the distribution pattern of F. nitida.
Notably, several limitations may affect the interpretation of results in our study. First, the spatial scale may introduce uncertainties into SDMs outputs. Researches on geographic distribution of alien ornamental plants have revealed that predictions at the regional scale can yield a broader suitable range compared to those at the national scale (Oduor et al., 2023). Therefore, our simulation of the potential distribution area at China’s national scale may underestimate the suitable range in localized microhabitats according to the geographical scale-dependent perspective (Levin, 1992). Second, although we used ENMTools to reduce sampling bias (retaining one point per 1 km × 1 km grid), the 78 final occurrence records still have limited coverage of remote areas in the western HDMs. Such under-sampling and uneven sampling may lead to biased estimation of suitable habitats in these regions, which is consistent with the conclusions summarized by Santini et al. (2021). Third, physiological traits usually affect the climate niche breath of plants (Yan et al., 2025). Specifically, F. nitida has unique life history traits—short rhizome necks (10–13 cm) and long flowering cycles (up to 109 years)—which greatly restrict its dispersal capacity under rapid climate change. This is contradictory with the model assumption in our study that the species can migrate freely to newly suitable habitat. Hence, the actual distribution area may be narrower than the predicted one. Future studies could address these limitations by adjusting research scale, expanding field surveys, and incorporating physiological factors into model simulations.
4.2 Key environmental factors influencing the distribution of F. nitida
The geographic distribution pattern of F. nitida is primarily determined by temperature-related factors (bio4 and bio6), followed by elevation and precipitation-related factors (bio14 and bio12). This finding contrasts with several studies predicting bamboos’ distribution, which identified precipitation as the primary determinant of their distribution, with temperature playing a secondary role (Li et al., 2019; Zhao et al., 2023). Our results confirm that temperature is the dominant factor shaping the distribution of F. nitida—a pattern consistent with the modeling results of Fargesia denudata, a congeneric species also endemic to the HDM region (Hu and He, 2020). The dominance of temperature in shaping F. nitida’s distribution is closely tied to the species’ adaptation to alpine environments and its unique life-history traits. Specially, suitable seasonality (bio4: 613%-834%) is conducive to the emergence of bamboo shoots and the formation of lignification, which are the key processes for bamboos’ growth and cold resistance. Meanwhile, optimal minimum temperature in the coldest month (bio6: -11.43 °C to -2.97 °C) ensure winter survival by preventing freeze damage to rhizomes, the core organ for bamboos’ clonal reproduction (Li et al., 2019). This temperature-driven distribution pattern is not unique to F. nitida, researches on Incarvillea in the Himalayan region (Rana et al., 2021)and major tree species in Southwest China (Ma et al., 2022) have similarly found that temperature-dependent variables exert a more dominant influence on species distribution than precipitation-related factors. This suggests that temperature may play a more critical role in determining the habitat suitability for plants distributed in alpine and southwestern China mountains. However, additional studies involving more taxa are needed to validate this hypothesis.
Additionally, elevation plays an important role in shaping the richness pattern of montane plants (Hu and He, 2020; Qiu et al., 2023a). F. nitida, which is mainly distributed in the subalpine zone at an altitude of 2450–3200 m, is also influenced by elevation in terms of its distribution. Our study revealed that, in response to climate change, this bamboo species will migrate to northwest with higher altitude and may reach an upper limit of 3457 m. This upward shift could facilitate its survival under future climatic conditions, just as has been observed in Bergenia, Primula and plants distributed in Mt. Gongga etc (He et al., 2019; Zu et al., 2021; Qiu et al., 2023a).
Notably, soil parameters (WRB4) and slope exert negligible impacts on F. nitida’s distribution (Table 2). This species can adapt to both acidic and neutral soil types across the HDM region by regulating nutrient availability in its rhizosphere—an adaptive trait that alleviates soil-related constraints on its range (Guo et al., 2023). For slope, our field observations confirm that F. nitida’s dense caespitose rhizomes (Figure 1C) effectively stabilize its growth even on steep terrain (slope >30°), reducing its dependence on specific slope conditions and explaining why slope has a weak predictive power for its distribution.
4.3 Geographic distribution changes of F. nitida
Climate change significantly influences plant growth and distribution patterns. Many plant species have been observed migrating toward higher altitudes and latitudes to cope with increasing environmental pressures (Yang et al., 2024; Ye et al., 2024). F. nitida adopts a similar adaptive strategy, but the migration trajectory of its distribution centroid varies across climate scenarios. This divergence is closely tied to the intensity of climate warming and the species’ niche requirement. Under the low-emission scenarios (SSP126 and SSP245), F. nitida gradually shifts northward to higher-latitude regions—where temperature conditions align with its niche requirements (Figure 3). This pattern is consistent with the migration behavior of bamboo species Chimonobambusa tumidissinoda (Wang et al., 2024). Compared to the current centroid location, the migration distance under these scenarios remains relatively limited (approximately 100 km). This suggests that the species’ core habitats still maintain partial suitability, reducing the urgency of large-scale range shifts, consistent with observations in Alpinia officinarum, which exhibit distribution area increase and modest northward migration under low-emission scenarios (Kang et al., 2025). In contrast, under the high-emission scenario (SSP585), F. nitida exhibits a distinct southwestward shift—with migration distances of 191 km in the 2050s and 160 km in the 2090s—and a continuous increase in centroid altitude. This direction change stems from the severe loss of core highly suitable areas (up to 56% contraction in Sichuan and Shaanxi) resulting from extreme temperature increase (4.4 °C by 2100; IPCC, 2023). Such thermal conditions beyond the tolerance of F. nitida, particularly threatening rhizome survival of bamboos (Li et al., 2019). As a result, the species is forced to migrate toward the Qinghai-Tibet Plateau, where higher altitudes (rising to 3343 m in the 2050s and 4396 m in the 2090s) provide cooler microclimates that match its temperature niche. This high-altitude migration mirrors trends in other alpine taxa of the Hengduan Mountains, such as Bergenia and Primula, which shift upward to avoid heat stress under extreme climate scenarios (Qiu et al., 2023a, b). However, F. nitida’s limited dispersal capacity raises concerns about its ability to colonized these new high-altitude habitats. Ye et al. (2019) noted that alpine bamboos with low dispersal rates struggle to track rapid climate shifts, and our projections suggest this risk is amplified under SSP585, where suitable habitat in the Plateau’s remote areas may remain underutilized.
From a conservation perspective, the divergent migration patterns highlight scenario-dependent vulnerability. Although the total suitable area expands across all scenarios, the consistent contraction of highly suitable habitats (the most representative of the species’ actual growth conditions; Figure 6) indicates underlying risks. Particularly under the high-emission scenario in the 2090s, the highly suitable area will decrease sharply to 14.45 × 104 km², a reduction of over 56%, and the distribution centroid will shift to an elevation potentially beyond the dispersal ability of bamboos. Given F. nitida’s limited diffusion capacity and relatively recent evolutionary origin (Ye et al., 2019), its future survival prospects are concerning. This may also impose high pressure on the survival of giant pandas, which rely on it for approximately 90% of their diet. Targeted conservation measures, such as the implementation of long-term monitoring of bamboo dynamics, particularly flowering periods, the establishment of habitat corridors to prevent habitat fragmentation, and the facilitation of assisted migration to high-elevation refugia under severe emission scenarios, should be implemented to safeguard F. nitida populations and maintain the stability of the giant panda habitats.
5 Conclusion
In this study, an ensemble modeling approach was employed to evaluate the impacts of climate change on the distribution pattern of F. nitida, a primary food source for giant pandas and a key understory alpine bamboo species. The results revealed that temperature-related variables, particularly temperature seasonality (bio4) and minimum temperature of the coldest month (bio6), are the most critical drivers determining its distribution, followed by elevation and precipitation factors. The current total suitable habitat of F. nitida is estimated at 83.10 × 104 km², primarily concentrated in the HDM region and its adjacent areas (102° E – 113° E, 29° N – 36° N). Highly suitable areas, accounting for 39.51% of the total suitable area, are predominantly located in Sichuan, Gansu, Shaanxi, and Henan. These highly suitable areas encompass the species’ observed distributions points and are more representative of the conditions suitable for F. nitida. Under future climate scenarios, while the total suitable area shows an expanding trend, the extent of highly suitable areas consistently declines. For instance, under the SSP585 scenario, the total suitable area increased by 62.48% by the 2090s, whereas highly suitable areas decrease by more than 56%. The current distribution centroid, located in northern Sichuan, exhibits consistent migration to higher elevations and scenario-specific directional changes: a northward shift (high latitude) under low emission scenarios (SSP126 and SSP245), and a southwestward shift (high altitude) under high-emission scenario. These findings highlight the vulnerability of F. nitida to climate change, especially under high-emission scenarios, and provide a scientific basis for its conservation, the protection of giant panda, and the management of subalpine ecosystem.
Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Author contributions
LM: Writing – original draft, Writing – review & editing, Methodology, Formal analysis. CW: Writing – original draft, Software, Methodology. SH: Data curation, Writing – original draft. FD: Writing – original draft, Formal analysis, Data curation. XY: Methodology, Writing – review & editing, Writing – original draft, Funding acquisition, Visualization, Project administration, Formal analysis.
Funding
The author(s) declare financial support was received for the research and/or publication of this article. This research was supported by the National Natural Science Foundation of China (NO. 3220307), Cultivating Plan Program for the Leader in Science and Technology of Yunnan Province (NO. 202305AC160056), the Applied Basic Research Foundation of Yunnan Province (NO. 202301AT070348).
Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declare that no Generative AI was used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2025.1683441/full#supplementary-material
Supplementary Table 1 | Distribution points of Fargesia nitida used in this study.
References
Alves-Ferreira, G., Vancine, M. H., Mota, F. M. M., Bello, C., Sobral-Souza, T., Percequillo, A. R., et al. (2025). From hot to cold spots: Climate Change is projected to modify diversity patterns of small mammals in a biodiversity hotspot. Diversity Distributions 31, e70026. doi: 10.1111/ddi.70026
Arneth, A., Shin, Y.-J., Leadley, P., Rondinini, C., Bukvareva, E., Kolb, M., et al. (2020). Post-2020 biodiversity targets need to embrace climate change. PNAS 117, 30882–30891. doi: 10.1073/pnas.2009584117
Bellard, C., Bertelsmeier, C., Leadley, P., Thuiller, W., and Courchamp, F. (2012). Impacts of climate change on the future of biodiversity. Ecol. Lett. 15, 365–377. doi: 10.1111/j.1461-0248.2011.01736.x
Brown, J. L. (2014). SDMtoolbox: a python-based GIS toolkit for landscape genetic, biogeographic and species distribution model analyses. Methods Ecol. Evol. 5, 694–700. doi: 10.1111/2041-210x.12200
Chen, X., Xiao, D., Qi, Y., Shi, Z., Bai, H., Lu, Y., et al. (2025). Projected future changes in extreme climate indices affecting rice production in China using a multi-model ensemble of CMIP6 projections. Front. Plant Sci. 16. doi: 10.3389/fpls.2025.1595367
Cobos, M. E., Peterson, A. T., Barve, N., and Osorio-Olvera, L. (2019). kuenm: an R package for detailed development of ecological niche models using Maxent. PeerJ 7, e6281. doi: 10.7717/peerj.6281
Correia, A. M. and Lopes, L. F. (2023). Revisiting biodiversity and ecosystem functioning through the lens of complex adaptive systems. Diversity 15, 895. doi: 10.3390/d15080895
Elith, J., Graham, H., C., P., Anderson, R., Dudík, M., Ferrier, S., et al. (2006). Novel methods improve prediction of species' distributions from occurrence data. Ecography 29, 129–151. doi: 10.1111/j.2006.0906-7590.04596.x
Fang, B., Zhao, Q., Qin, Q., and Yu, J. (2022). Prediction of potentially suitable distribution areas for Prunus tomentosa in China based on an optimized MaxEnt model. Forests 13, 381. doi: 10.3390/f13030381
Gao, H., Qian, Q., Deng, X., Peng, Y., and Xu, D. (2024). Predicting the distributions of Morus notabilis C. K. Schneid under climate change in China. Forests 15, 352. doi: 10.3390/f15020352
Guisan, A. and Zimmermann, N. E. (2000). Predictive habitat distribution models in ecology. Ecol. Model. 135, 147–186. doi: 10.1016/S0304-3800(00)00354-9
Guo, M., Luo, L., Liang, J., Wang, Y. J., and Zhao, C. Z. (2023). Effects of freeze-thaw changes on soil physicochemical properties and enzyme activities in root zone of Picea asperata and Fargesia nitida under subalpine forests of southwest China. Chin. J. Plant Ecol. 47, 882–894. doi: 10.17521/cjpe.2022.0031
He, X., Burgess, K. S., Gao, L. M., and Li, D. Z. (2019). Distributional responses to climate change for alpine species of Cyananthus and Primula endemic to the Himalaya-Hengduan Mountains. Plant Diversity 41, 26–32. doi: 10.1016/j.pld.2019.01.004
Hu, S. P. and He, L. W. (2020). Analysis of suitable distribution areass of Fargesia denudata in Baishuijiang National Nature Reserve using MaxEnt model and ArcGIS. Chin. J. Ecol. 39, 2115–2133. doi: 10.13292/j.1000-4890.202006.023
Huang, L., Xing, X. C., Li, W. W., Zhou, Y., Zhang, Y. Q., Xue, C., et al. (2021). Population genetic structure of the giant panda staple food bamboo (Fargesia spathacea complex) and its taxonomic implications. J. Systematics Evol. 59, 1051–1064. doi: 10.1111/jse.12594
IPCC (2023). Climate change 2023: Synthesis report. contribution of working groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on climate change (Geneva, Switzerland: IPCC).
Janzen, D. H. (1976). Why bamboos wait so long to flower. Annu. Rev. Ecol. Systematics 7, 347–391. doi: 10.1146/annurev.es.07.110176.002023
Kang, Y., Lin, F., Yin, J., Han, Y., Zhu, M., Guo, Y., et al. (2025). Projected distribution patterns of Alpinia officinarum in China under future climate scenarios: insights from optimized Maxent and Biomod2 models. Front. Plant Sci. 16. doi: 10.3389/fpls.2025.1517060
Koh, L. P., Dunn, R. R., Sodhi, N. S., Colwell, R. K., Proctor, H. C., and Smith, V. S. (2004). Species coextinctions and the biodiversity crisis. Science 305, 1632–1634. doi: 10.1126/science.1101101
Levin, S. A. (1992). The problem of pattern and scale in ecology: the Robert H. MacArthur award lecture. Ecology 73, 1943–1967. doi: 10.2307/1941447
Li, X., Mao, F., Du, H., Zhou, G., Xing, L., Liu, T., et al. (2019). Spatiotemporal evolution and impacts of climate change on bamboo distribution in China. J. Environ. Manage. 248, 109265. doi: 10.1016/j.jenvman.2019.109265
Li, D. Z., Wang, Z. P., Zhu, Z. D., Xia, N. H., Jia, L. Z., Guo, Z. H., et al. (2006). “Tribe bambuseae,” in Flora of China, vol. 22 . Eds. Wu, Z. Y., Raven, P. H., and Hong, D. Y. (Science Press, Beijing and Missouri Botanical Garden Press, St. Louis), 7–180.
Ma, P. F., Vorontsova, M. S., Nanjarisoa, O. P., Razanatsoa, J., Guo, Z. H., Haevermans, T., et al. (2017). Negative correlation between rates of molecular evolution and flowering cycles in temperate woody bamboos revealed by plastid phylogenomics. BMC Plant Biol. 17, 260. doi: 10.1186/s12870-017-1199-8
Ma, D. X., Yin, X. J., Zhou, S. Y., Lu, S. F., and Chen, Z. (2022). Pattern of main tree species richness in Southwest China based on Chinese vegetation map. J. Fujian Agric. Forestry Univ. (Natural Sci. Edition) 51, 249–257. doi: 10.13323/j.cnki.j.fafu(nat.sci.).2022.02.015
Mittermeier, R. A., Gil, P. R., Hoffmann, M., and Fonseca, G. A. B. D. (2004). Hotspots revisited: earth’s biologically richest and most endangered ecoregions (Mexico: CEMEX Books on Nature: CEMEX).
Oduor, A. M. O., Yang, B., and Li, J.-M. (2023). Alien ornamental plant species cultivated in Taizhou, southeastern China, may experience greater range expansions than native species under future climates. Global Ecol. Conserv. 41, e02371. doi: 10.1016/j.gecco.2023.e02371
Peterson, A. T., Cobos, M. E., and Jiménez-García, D. (2018). Major challenges for correlational ecological niche model projections to future climate conditions. Ann. New York Acad. Sci. 1429, 66–77. doi: 10.1111/nyas.13873
Qiao, H. J., Feng, X., Escobar, L. E., Peterson, A. T., Soberón, J., Zhu, G. P., et al. (2019). An evaluation of transferability of ecological niche models. Ecography 42, 521–534. doi: 10.1111/ecog.03986
Qiu, L., Fu, Q.-L., Jacquemyn, H., Burgess, K. S., Cheng, J.-J., Mo, Z.-Q., et al. (2023a). Contrasting range changes of Bergenia (Saxifragaceae) species under future climate change in the Himalaya and Hengduan Mountains Region. Theor. Appl. Climatology 155, 1927–1939. doi: 10.1007/s00704-023-04746-0
Qiu, L., Jacquemyn, H., Burgess, K. S., Zhang, L. G., Zhou, Y. D., Yang, B. Y., et al. (2023b). Contrasting range changes of terrestrial orchids under future climate change in China. Sci. Total Environ. 895, 165128. doi: 10.1016/j.scitotenv.2023.165128
Rana, S. K., Rana, H. K., Luo, D., and Sun, H. (2021). Estimating climate-induced ‘Nowhere to go’ range shifts of the Himalayan Incarvillea Juss. using multi-model median ensemble species distribution models. Ecol. Indic. 121, 107127. doi: 10.1016/j.ecolind.2020.107127
Santini, L., Benítez-López, A., Maiorano, L., Čengić, M., Huijbregts, M. A. J., and Fourcade, Y. (2021). Assessing the reliability of species distribution projections in climate change research. Diversity Distributions 27, 1035–1050. doi: 10.1111/ddi.13252
Shi, Y. L., Wang, S. Y., Lei, F. F., Shi, Y., Sun, Y. L., Sun, Y. H., et al. (2025). Analysis of nutritional composition and palatability of Fagersia nitida and F. denudata in Axia Provincila Nature Reserve. Yunnan Chem. Technol. 52, 88–92. doi: 10.3969/j.issn.1004-275X.2025.06.20
Srivastava, V., Lafond, V., and Griess, V. (2019). Species distribution models (SDM): applications, benefits and challenges in invasive species management. CAB Rev. 14, 1–13. doi: 10.1079/pavsnnr201914020
Wang, W. H., Peng, S. L., Shu, H., Fu, X., and Ye, X. Y. (2024). Potential adaptive habitats for the narrowly distributed and rare bamboo species Chimonobambusa tumidissinoda J. R. Xue & T. P. Yi ex Ohrnb. under future climate change in China. Ecol. Evol. 14, e70314. doi: 10.1002/ece3.70314
Wang, Y. J., Tao, J. P., and Zhong, Z. C. (2009). Factors influencing the distribution and growth of dwarf bamboo, Fargesia nitida, in a subalpine forest in Wolong Nature Reserve, southwest China. Ecol. Res. 24, 1013–1021. doi: 10.1007/s11284-008-0573-2
Warren, D. L., Glor, R. E., and Turelli, M. (2010). ENMTools: a toolbox for comparative studies of environmental niche models. Ecography 33, 607–611. doi: 10.1111/j.1600-0587.2009.06142.x
Warren, R., Price, J., Graham, E., Forstenhaeusler, N., and Van Der Wal, J. (2018). The projected effect on insects, vertebrates, and plants of limiting global warming to 1.5 °C rather than 2 °C. Science 360, 791–795. doi: 10.1126/science.aar3646
Wudu, K., Abegaz, A., Ayele, L., and Ybabe, M. (2023). The impacts of climate change on biodiversity loss and its remedial measures using nature based conservation approach: a global perspective. Biodiversity Conserv. 32, 3681–3701. doi: 10.1007/s10531-023-02656-1
Yan, F., Liu, L., Ding, J., Fan, K., Buggs, R. J. A., and Wang, N. (2025). Impacts of climatic niche breadth, phylogeny, traits and ploidy on geographical ranges of Betula species. J. Systematics Evol. 63, 939–951. doi: 10.1111/jse.13176
Yang, J., Huang, Y., Su, M., Liu, M., Yang, J., and Wu, Q. (2024). Spatial Distribution Patterns of the Key Afforestation Species Cupressus funebris: Insights from an Ensemble Model under Climate Change Scenarios. Forests 15, 1280. doi: 10.3390/f15081280
Ye, X. Y., Ma, P. F., Yang, G. Q., Guo, C., Zhang, Y. X., Chen, Y. M., et al. (2019). Rapid diversification of alpine bamboos associated with the uplift of the Hengduan Mountains. J. Biogeography 46, 2678–2689. doi: 10.1111/jbi.13723
Ye, X.-Y., Wang, W.-H., Wei, G.-R., Li, B., Li, Y., and Ma, P.-F. (2024). Genetic structure of a narrowly distributed species Chimonobambusa tumidissinoda in the Yunnan-Guizhou Plateau, and its implications for conservation. Global Ecol. Conserv. 53, e03028. doi: 10.1016/j.gecco.2024.e03028
Yi, T. P. and Jiang, X. L. (2010). Staple food bamboo species of the giant panda and their biodiversity. J. Sichuan Forestry Sci. Technol. 31, 1–20. doi: 10.16779/j.cnki.1003-5508.2010.04.001
Yi, T. P., Shi, J. Y., Ma, L. S., Wang, H. T., and Yang, L. (2008). Iconographia Bambusoidearum Sinicarum (Beijing: Science Press).
Zhang, Q., Shen, X., Jiang, X., Fan, T., Liang, X., and Yan, W. (2023). MaxEnt modeling for predicting suitable habitat for endangered tree Keteleeria davidiana (Pinaceae) in China. Forests 14, 394. doi: 10.3390/f14020394
Zhang, H., Sun, P., Zou, H., Ji, X., Wang, Z., and Liu, Z. (2024). Adaptive distribution and vulnerability assessment of endangered maple species on the Tibetan Plateau. Forests 15, 491. doi: 10.3390/f15030491
Zhang, Y.-Q., Zhou, Y., Hou, X.-Q., Huang, L., Kang, J.-Q., Zhang, J.-Q., et al. (2019). Phylogeny of Fargesia (Poaceae: Bambusoideae) and infrageneric adaptive divergence inferred from three cpDNA and nrITS sequence data. Plant Systematics Evol. 305, 61–75. doi: 10.1007/s00606-018-1551-y
Zhao, Z., Yang, L., and Chen, X. (2024). Globally suitable areas for Lycorma delicatula based on an optimized Maxent model. Ecol. Evol. 14, e70252. doi: 10.1002/ece3.70252
Zhao, H., Yang, X., Shi, S., Xu, Y., Yu, X., and Ye, X. (2023). Climate-driven distribution changes for Bashania fargesii in the Qinling Mountains and its implication for panda conservation. Global Ecol. Conserv. 46, e02610. doi: 10.1016/j.gecco.2023.e02610
Zhou, Y., Li, W. W., Zhang, Y. Q., Xing, X. C., Zhang, J. Q., and Ren, Y. (2020). Extensive reticulate evolution within Fargesia (s.l.) (Bambusoideae: Poaceae) and its allies: Evidence from multiple nuclear markers. Mol. Phylogenet. Evol. 149, 106842. doi: 10.1016/j.ympev.2020.106842
Keywords: alpine bamboo, Biomod2, climate change, conservation, distribution pattern, Fargesia nitida, giant panda
Citation: Mao L, Wang C, He S, Deng F and Ye X (2025) Ensemble modeling uncovers climate change-driven distribution shifts of Fargesia nitida (Mitford) P. C. Keng ex T. P. Yi, a primary food source for giant panda. Front. Plant Sci. 16:1683441. doi: 10.3389/fpls.2025.1683441
Received: 11 August 2025; Accepted: 11 November 2025; Revised: 28 October 2025;
Published: 01 December 2025.
Edited by:
Zhihang Zhuo, China West Normal University, ChinaReviewed by:
Mohamed M. El-Khalafy, Kafrelsheikh University, EgyptChi-Cheng Liao, Chinese Culture University, Taiwan
Copyright © 2025 Mao, Wang, He, Deng and Ye. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Xiaying Ye, MzQwMDdAenR1LmVkdS5jbg==