Sources of Species Data

In this study, the geographical coordinates of P. decomposita were confirmed and filtered by accessing the National Plant Specimen Resource Center (NPSRC) and conducting surveys on forest genetic resources. Non-specific or repetitive distribution records were eliminated. Concentrated distribution data in certain regions can lead to overfitting of the model, resulting in uncertainty. Buffer tools in ArcGIS 10.2 were employed to establish a buffer zone with a radius of 5 km around each distribution point, based on the resolution of environmental variables, ensuring that only one distribution point was retained within a 5 km radius. In total, 37 distribution points of P. decomposita were collected, and a distribution map was created (Fig. 1).

Sources of Environmental Variables

The climate and environmental data in this study were obtained by accessing the WorldClim database (http://www.world-clim.org/), which utilized a WGS84 coordinate system with a grid size of 2.5′. Human activity intensity (HAI) data grids were acquired from the “Human Footprint” dataset of 2009 [20]. Future climate data were generated using the BCC-CSM2-MR climate system model developed by the National Climate Center, incorporating three emission scenarios (SSP5- 8.5, SSP2-4.5, SSP1-2.6). These SSPs provide more scientifically descriptive predictions of future climate change outcomes [21]. Due to the intercorrelation between environmental variables, a correlation analysis was performed on the environmental factors before incorporating them into the MaxEnt model [22, 23]. In this study, Spearman correlation analysis was conducted on the environmental factors, and when the correlation coefficient between the two factors exceeded 0.8, the one with a higher contribution rate was retained. Ultimately, seven environmental factors were selected for the operation of the MaxEnt model (Table 1).

Model Construction and Evaluation

Regarding the research methods of Hao et al. and Yang et al. [22, 23], the distribution data of P. decomposita and environmental factor data were separately imported into the MaxEnt software for modeling calculations. The parameter settings included allocating 25% of the distribution points as the test set and 75% as the training set. The MaxEnt model requires users to specify a set of parameters, namely the percentage of training presence (i.e., the proportion of locations used for model development and internal testing), the number of background points, clamping (i.e., whether to limit predictions within the range of variability of the input predictor variables), and the regularization multiplier (i.e., to prevent excessive overfitting of the response curve). When using species distribution models to predict species distribution, there is often an issue of overestimation or underestimation, namely false positives and false negatives. Therefore, it is crucial to evaluate the accuracy of the model simulations using effective evaluation metrics as an important step in determining model accuracy and usability [24]. In this study, the prediction accuracy of the model was assessed using the area under the curve (AUC) of the receiver operating characteristic (ROC) curve [25]. A higher AUC value indicates better predictive performance of the model, and the evaluation criteria are detailed in Table 2 [3].

Fig. 1.

Field photographs and sample distribution records of P. decomposita.

Classification of Potentially Suitable Areas

Choose the average of 10 repetitions as the simulated result for this study in the output file. This result is generated based on the logical values (P) representing the probability of species presence, with P ranging from 0 to 1. A higher P value indicates a greater likelihood of species presence. Use ArcGIS 10.2 software to convert the predicted results into raster format and perform classification and visualization on suitable habitats. Based on the P values, employ the natural breakpoints method to divide the suitable habitats into four levels (Table 3)[2]. Calculate the number of cells in each level and determine the proportion of suitable habitat area for each level.

Determination of Dominant Environment Variables

The determination of the dominant factors remains a topic of debate among scholars. Most researchers identify dominant environmental factors based on the cumulative contribution rate, considering environmental factors with cumulative contribution rates exceeding a specific threshold as the primary environmental factors. This threshold is typically subjectively chosen based on the characteristics of the studied species and research findings, leading to varying standards. In this study, the top four contributing environmental factors were selected as dominant environmental factors [8]. The results were also examined by the Jackknife method to determine which environmental factor variables influenced the formalized training gains the most when only a single environmental factor variable was used [10].

Evaluation of Model Accuracy

Based on the distribution records of P. decomposita, the potentially suitable habitats of P. decomposita in China were simulated and predicted using the MaxEnt model. The AUC value for the test dataset of P. decomposita was 0.929, and the AUC value for the training dataset was 0.949, indicating a “good” performance of the P. decomposita simulation. The ROC curve, as shown in Fig. 2, illustrates that the model was appropriately designed and that the predicted results are highly reliable, making them suitable for subsequent analysis.

Fig. 2.

Receiver operating characteristic curve of P. decomposita Maxent model.

Potential Distribution and Suitability Evaluation of P. Decomposita

Fig. 3 and 4 display the results of the model’s prediction for the potentially suitable habitats of P. decomposita. The analysis reveals that the current suitable habitats for P. decomposita are primarily concentrated in a linear pattern within the southwestern region of Sichuan, covering a total area of 21.97×10⁴ km² (Fig. 3, 4 and 5a). The high suitability zone occupies 7.10% of the total suitable habitat area, spanning 1.56×10⁴ km², with a narrow distribution in Ganzi and Aba prefectures, along with scattered patches in Diqing prefecture and Changdu region (Fig. 3, 4 and 5a). The moderate suitability zone accounts for 26.17% of the total suitable habitat area, covering 5.75×10⁴ km². It mainly surrounds the high suitability zone in a linear pattern, with a concentrated distribution on the outskirts of Ganzi and Aba prefectures, and sporadic occurrences in Changdu region, Diqing prefectures, and Shaotong city (Fig. 3, 4, and 5a). The low suitable habitat area occupies 66.73% of the total suitable habitat area, amounting to 14.66×10⁴ km². It is primarily located along the outer edges of the moderate suitable habitat area in a linear pattern, with concentrated distribution in Ganzi, Aba, and Diqing, as well as sporadic or patchy occurrences in Liangshan, Shaotong, Longnan, and Changdu regions, constituting more than half of the total suitable habitat area (Fig. 3, 4 and 5a).

Fig. 3.

Potential geographical distribution of P. decomposita under future climate change scenarios.

Fig. 4.

Changes in the potential geographical distribution of P. decomposita under climate change scenarios.

Based on field surveys, literature reviews, and analysis of plant specimen databases, it is evident that P. decomposita is primarily distributed in Jinchuan County, Barkam City, Kangding City, and other areas. These regions are all within the predicted suitable zones for this study, which further confirms the reliability of the predictions for P. decomposita. The research conducted by Hong et al. [26] indicates the presence of P. decomposita in Jinchuan County, Danba County, Barkam City, and Kangding City, while Tan et al. [27] have found its distribution in Li County and Xiaojin County. Notably, all these findings align with the predicted suitable zones, providing further evidence of the accuracy of the predictions.

Overall, the area of highly and moderately suitable habitats for P. decomposita in the territory of Ganzi and Aba Prefectures was distributed in a narrow line, which corresponded well with the actual distribution. This further confirms the accuracy of the simulation results of this study. The extremely narrow range of highly suitable areas also indicates the fundamental reason why P. decomposita is an endangered plant species.

Fig. 5.

a) Suitable areas for P. decomposita under different climate change scenarios. b) Future changes in suitable habitat area of P. decomposita under climate change scenarios.

Impacts of Future Climate Change on the Geographical Distribution of P. Decomposita

In this study, the Maxent model was utilized to predict the potentially suitable habitats of P. decomposita in two time periods (2050s and 2070s) under three different emission scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5). As a result, the different levels of potentially suitable habitat areas for P. decomposita under various climate change backgrounds (Fig. 5a) and the potentially suitable habitats of P. decomposita under different climate change scenarios (Fig. 3) were obtained. In all three SSPs (SSP1-2.6, SSP2-4.5, and SSP5-8.5) scenarios, the areas of P. decomposita’s suitable habitats showed a significant increase in both the 2050s and 2070s. However, the high suitability zone and moderately suitability zone exhibited a more pronounced increase during these periods. In the 2070s, under the SSP5-8.5 scenario, the high suitability zone increased by 373.08%, which is more than four times its current size. Similarly, the moderate suitability zone also experienced the greatest increase in the 2070s under the SSP5-8.5 scenario, with an enlargement of 282.61%, which is more than three times its current size. These findings indicate that climate change has a substantial impact on the high and moderate suitability zones of P. decomposita. Moreover, the total suitable habitat area of P. decomposita exhibited a considerable increase in the 2050s and 2070s, with the emergence of numerous new suitable habitats (Fig. 3 and 5a). Compared to the current distribution, in the 2050s, the total suitable habitat area is projected to increase by 57.72%, 77.79%, and 90.26% under the three SSPs (SSP1-2.6, SSP2-4.5, and SSP5-8.5) scenarios, respectively. The high suitability zone is expected to increase by 124.36%, 112.82%, and 193.59%, while the moderate suitability zone is projected to increase by 94.43%, 121.74%, and 148.70%. The low suitability zone is anticipated to expand by 36.22%, 56.82%, and 56.34% (Fig. 5a). Similarly, in the 2070s, compared to the current distribution, the total suitable habitat area is forecasted to increase by 44.11%, 85.23%, and 150.98% under the three SSPs (SSP1-2.6, SSP2-4.5, and SSP5-8.5) scenarios, respectively. The high suitability zone is expected to increase by 59.62%, 145.51%, and 373.08%, while the moderate suitability zone is projected to increase by 65.22%, 149.04%, and 282.61%. The low suitability zone is anticipated to expand by 34.17%, 53.75%, and 75.72% (Fig. 5a). In summary, the future projections indicate a substantial overall increase in the various levels of suitable habitat areas for P. decomposita, with the high suitability zone and moderate suitability zone showing clear upward trends. These findings highlight the significant influence of climate change on P. decomposita.

Fig. 6 illustrates the trajectory of the centroid movement in the high suitability zone of P. decomposita under different periods and climate change scenarios. By analyzing the changes in the area of the high suitability zone and the displacement of its centroid, it is evident that from the reference climate to the 2050s and further to the 2070s, the centroid of the high suitability zone for P. decomposita migrates towards different latitudes and directions under the three emission scenarios (Fig. 6). Specifically, under the high emission scenario, the centroid of the high suitability zone shifts towards lower latitudes and the southwest direction, and then towards higher latitudes and the northeast direction, with a relatively significant magnitude of migration. Under the moderate emission scenario, it initially moves towards higher latitudes and a northwest direction, and then towards lower latitudes and a southwest direction. Under the low emission scenario, the centroid consistently moves towards higher latitudes and a northwest direction. In general, under the moderate and low emission scenarios, the centroid of the high suitability zone for P. decomposita tends to migrate towards higher latitudes and a northwest direction, with relatively smaller magnitudes of migration. Overall, in the high-emission scenario, the migration magnitude is larger, while in the moderate and low-emission scenarios, the migration magnitude is smaller. Thus, the increase in the size of the high suitability area is most pronounced in the context of high emissions.

Fig. 6.

Variations of the centroids of highly suitable areas of P. decomposita under climate change scenarios.

Based on the prediction results, the potential geographic distribution of P. decomposita is projected to increase at all levels of suitable habitats in both periods, regardless of emission concentration scenarios. In the 2050s and 2070s under high emission scenarios, the high suitability zone of P. decomposita exhibits a significant increase with magnitudes of 193.59% and 373.08%, respectively. The moderate suitability zone also experiences substantial growth with magnitudes of 148.70% and 282.61% (Fig. 4 and 5b). This could be attributed to the fact that in areas under high emission scenarios, temperature and precipitation reach ranges suitable for the growth of P. decomposita, combined with appropriate elevation and changes in climatic conditions, thereby increasing the survival probability in these regions. Research by Xia et al. [28] also suggests that soil moisture content is an important environmental factor influencing the distribution pattern of P. decomposita. Furthermore, studies conducted by Tan et al. [27] indicate that temperature and precipitation have a significant impact on the growth conditions of P. decomposita. The centroid of the high suitability zone for P. decomposita migrates towards different latitudes and directions under the three emission scenarios; overall, the migration magnitude is larger under high emission scenarios, while it is smaller under moderate and low emission scenarios. Thus, the increase in the area of the high suitability zone is most pronounced in high emission backgrounds (Fig. 6). These findings suggest that in the future, there will be many regions that meet the growth conditions of P. decomposita, becoming suitable for its growth and distribution. At the same time, certain regions will no longer be suitable for its growth and distribution. However, the number of newly added suitable habitats will exceed those disappearing. Thomas et al. [29] found that under a warming climate, some species may become extinct, but many others will experience varying degrees of growth and distribution. This indicates that climate warming has both positive and negative effects on the growth and distribution of species. P. decomposita has thrived in terms of growth and distribution under a warming climate.

In the 2050s, under the moderate emission scenario, P. decomposita experiences the largest loss of suitable habitat area, reaching 2.90×10⁴ km². Meanwhile, under the high emission scenario, the largest increase in suitable habitat area is observed, amounting to 21.66×10⁴ km². In the 2070s, under the high emission scenario, both the loss and addition of suitable habitat areas for P. decomposita are substantial, measuring 5.34×10⁴ km² and 30.25×110⁴ km², respectively (Fig. 5b). Despite the increasing trend of suitable habitat areas for P. decomposita under future climate change scenarios, the total area remains relatively small for this critically endangered plant species. Therefore, conservation efforts must be prioritized in policy-making and agricultural production. Under the low emission scenario, in the 2050s, the loss of suitable habitat for P. decomposita mainly occurs in the western border areas of Guizhou and the southern region of Gannan. Additionally, scattered patches of newly suitable habitats emerged in the western part of Sichuan, northwestern Sichuan, southeastern Tibet, and Gannan (Fig. 4). Comparatively, in the 2070s, there was an expansion of the lost habitat areas, with new losses occurring in the northern border region of Yunnan. The range of newly suitable habitats becomes slightly smaller (Fig. 4). Under the moderate emission scenario, in the 2050s, the loss of suitable habitat for P. decomposita is concentrated in the western border areas of Guizhou, the northern border areas of Yunnan, and the southern region of Gannan. Conversely, new suitable habitats develop in the western part of Sichuan, northwestern Sichuan, southeastern Tibet, and Gannan (Fig. 4). In the 2070s, similar to the 2050s, there is an expansion of lost habitat areas, accompanied by an enlargement of newly suitable habitats (Fig. 4). Under the high emission scenario, in the 2050s, the loss of suitable habitat for P. decomposita primarily occurs in the western border areas of Guizhou, the northern border areas of Yunnan, and the southern region of Gannan. Meanwhile, new suitable habitats emerged in the western part of Sichuan, northwestern Sichuan, southeastern Tibet, southern Ningxia, and Gannan (Fig. 4). Comparatively, in the 2070s, there was a significant increase in the area of lost habitats and newly suitable habitats (Fig. 4). The potential disappearance of certain regions in P. decomposita’s suitable habitats suggests the emergence of new regions where its growth conditions are met. Additionally, the number of newly suitable habitats exceeds the number of disappearing ones, indicating a minimal risk of extinction for P. decomposita under natural circumstances. However, precautions must be taken to prevent extensive harvesting, as these critically endangered plant species face immediate extinction risks due to excessive extraction. Research by Feng et al. [17] also demonstrates the significant impact of temperature changes on the survival and distribution of P. decomposita. Tan et al. [27] further highlight the substantial influence of temperature and precipitation on the growth conditions of P. decomposita. These findings indicate that P. decomposita exhibits a strong response to climate change and will migrate towards regions suitable for its growth and distribution.

Climate change indirectly affects the population and distribution characteristics of P. decomposita by directly impacting the ecosystem. In addition to the significant influence of climate change on the potential geographic distribution of P. decomposita, unsustainable human activities have led to a sharp decline in the wild population of P. decomposita. This study only considered environmental variables for the years 2050 and 2070. Therefore, future research on the response of species’ potential geographic distributions to climate change can include multiple periods to obtain a comprehensive understanding of the overall trends in the potential geographic distribution of the study subject.

Constraints of Environmental Variables on the Potential Geographic Distribution of P. Decomposita

Currently, there is a lack of consensus among researchers regarding the determination of the number of dominant factors. The majority of researchers choose to identify dominant factors based on their contribution rates. However, the selection criteria for determining the degree of dominance are subjective, leading to different standards. In this study, the four environmental factors with the highest contribution rates were selected as the dominant environmental factors. Through analysis and calculations, it was found that altitude, mean temperature of the coldest quarter (Bio11), annual precipitation (Bio12), and temperature seasonality standard deviation (Bio4) accounted for contribution rates of 32.50%, 25.40%, 23.00%, and 11.40%, respectively. The cumulative contribution rate of these four factors reached 92.30%, making them the dominant environmental factors chosen (Table 4).

The examination conducted through the Jackknife method reveals that, when using individual environmental factor variables, temperature seasonality standard deviation (Bio4), mean temperature of the coldest quarter (Bio11), annual precipitation (Bio12), and altitude (Altitude) have the greatest impact on the normalized training gain for P. decomposita. This indicates that these environmental factor variables play a dominant role in constraining the growth of P. decomposita (Fig. 7).

The response curves of species’ survival probabilities to environmental factors in the MaxEnt model output are commonly used to assess the relationship between species occurrence probability and environmental factors (Fig. 8). When the probability of P. decomposita’s occurrence exceeds the threshold for high suitability habitats, the corresponding range of environmental factors represents the suitable range for the survival of P. decomposita.

The results from the MaxEnt model reveal that when the mean temperature of the coldest quarter (Bio11) reaches -4.27ºC, the survival probability of P. decomposita reaches the threshold for high-suitability habitats. At a mean temperature of the coldest quarter (Bio11) of 0.46ºC, the occurrence probability of P. decomposita peaks, after which the survival probability decreases as the mean temperature of the coldest quarter (Bio11) increases. When the mean temperature of the coldest quarter (Bio11) reaches 5.2ºC, the survival probability of P. decomposita decreases to 0.6. Therefore, the suitable range of the mean temperature of the coldest quarter (Bio11) for P. decomposita’s survival is -4.27 to 5.2ºC. When the annual precipitation (Bio12) is 434.8 mm, the survival probability of P. decomposita reaches the threshold for high-suitability habitats. As the annual precipitation increases, the survival probability of P. decomposita shows an upward trend, peaking at an annual precipitation (Bio12) of 663.0mm. With a further increase in annual precipitation (Bio12), the survival probability of P. decomposita starts to decline, reaching 0.6 at annual precipitation (Bio12) of 900.3 mm. Hence, the suitable range of annual precipitation (Bio12) for P. decomposita’s survival is 434.8 to 900.3 mm. In conclusion, as the dominant environmental factors influencing the survival of P. decomposita, namely annual precipitation (Bio12) and the mean temperature of the coldest quarter (Bio11), increase, the survival probability of P. decomposita initially rises and then declines. The narrow range of suitable conditions indicates its high requirements for temperature and precipitation levels.

Fig. 7.

The jackknife test result of environmental factor for P. decomposita.

Fig. 8.

Response curves of existence probalitity of P. decomposita.

Climate factors are critical in determining the potential distribution of species at a macroscale and studying the interaction between plants and climate is an important aspect of ecology. The results from the MaxEnt model indicate that altitude, temperature seasonality standard deviation (Bio4), mean temperature of the coldest quarter (Bio11), and annual precipitation (Bio12) are the dominant environmental factors constraining the potential geographic distribution of P. decomposita (Table 1). This study highlights the significance of altitude as a key factor limiting the potential geographic distribution of P. decomposita. To some extent, the probability of P. decomposita’s occurrence increases with higher altitudes. Based on specimen records, P. decomposita is primarily distributed in the provinces of Sichuan, Gannan, and Aba, where its growth is constrained by altitude. In low-altitude areas, it struggles to grow and reproduce, indicating that regions with excessively low altitudes are unsuitable for its normal development. This finding aligns with the research conducted by Zhou et al. [28], confirming that altitude has the most prominent effect on the growth and reproduction of P. decomposita. Xia et al. [30] also demonstrated the significance of altitude as one of the crucial environmental factors affecting the distribution pattern of P. decomposita. Regarding temperature seasonality standard deviation, it also influences the probability of P. decomposita’s occurrence. As the seasonal temperature variation in regions suitable for P. decomposita’s natural distribution increases, the probability of its occurrence also rises. However, when the variation becomes too extreme or falls below a certain threshold, it exceeds the suitable range for P. decomposita’s survival. Therefore, regions with a high probability of P. decomposita’s occurrence exhibit noticeable temperature seasonality, but the variations should not be excessively drastic. Feng et al. [17] similarly demonstrated the significant impact of temperature seasonality standard deviation on the probability of P. decomposita’s occurrence. The mean temperature of the coldest quarter also restricts the distribution of P. decomposita to a considerable extent. With an increase in the mean temperature of the coldest quarter, the probability of P. decomposita’s occurrence gradually rises. However, as the mean temperature continues to rise, the probability declines. Similarly, annual precipitation moderately constrains the distribution of P. decomposita. As annual precipitation increases, the probability of P. decomposita’s occurrence gradually rises. However, with a sustained increase in annual precipitation, the probability decreases. These findings indicate that both temperature and precipitation have substantial influences on the distribution of P. decomposita. Wu et al. [31] found that P. decomposita has a higher probability of occurrence in areas with higher soil moisture. Xia et al. [30] emphasized the importance of soil moisture content as a significant environmental factor limiting the distribution pattern of P. decomposita, while Tan et al. [27] highlighted the substantial influence of temperature and precipitation on the growth conditions of P. decomposita. These studies provide further support for the accuracy of the results obtained in this research.

Conservation Strategies for P. Decomposita

Climate change has potentially accelerated species extinctions, reduced species diversity, and made regional ecosystems more vulnerable. However, some species may develop new physiological traits to adapt to climate change [32, 33]. Relevant studies indicate that changes in the spatial and temporal patterns of climate can alter the geographical distribution patterns of plants, posing a threat to their natural habitats [3, 34, 35]. To mitigate the impact of climate change on ecosystems, modeling species distributions can effectively inform conservation strategies. Based on the current growth status and predicted potential distribution of P. decomposita, the following recommendations are proposed for its protection and utilization:

(1) Enhancement of P. decomposita seedling trials to promote population renewal. The limited number of wild P. decomposita seedlings is one of the key reasons for its endangerment. Research on seed germination and seedling production can improve the age structure of P. decomposita populations, promote population renewal, and replenish population size.

(2) Reduce anthropogenic interference and protect wild P. decomposita communities. P. decomposita has a low resistance to external disturbances and is sensitive to environmental changes. Since P. decomposita inhabits the riverbanks of the Dadu River and Min River in the upstream basin, which are hotspots for hydropower development, the construction of hydroelectric stations or reservoirs can directly or indirectly alter the habitat conditions of P. decomposita, which may lead to a decrease in its population. Therefore, the impact of the construction project on the ecological environment of the P. decomposita habitat should be carefully considered before the project is implemented. Appropriate and effective conservation measures should be taken in advance to avoid irreversible consequences for P. decomposita.

(3) When releasing P. decomposita into the wild or implementing translocation conservation, the results of suitable area delineation for P. decomposita can serve as a scientific reference for site selection. If the habitat conditions for P. decomposita have been severely disrupted and on-site conservation is no longer feasible, measures such as translocation conservation should be taken. During translocation or wild release, efforts should be made to select sites within the most suitable areas for P. decomposita.

(4) Strengthen publicity and education for P. decomposita conservation. As an endemic plant species in China, P. decomposita represents a valuable natural asset. Local governments should recognize its significance as a valuable resource and utilize it responsibly to create economic benefits and promote poverty alleviation. At the same time, efforts should be intensified to protect P. decomposita, preventing further decline of the population due to unauthorized collection and illegal harvesting.

(5) To promote the cultivation of P. decomposita. Given its medicinal, oleaginous, and ornamental value, the promotion of the cultivation of P. decomposita is necessary. The results of ecological and geological suitability assessments should be taken into account when selecting planting sites for P. decomposita to ensure that appropriate sites are selected.