Assessment On Habitat Suitability And Population Connectivity In Qionglai Mountain Under Climate Change Context

The giant panda(Ailuropoda melanoleuca)is a flagship species for global biodiversity conservation.It currently only lives in six isolated mountain systems in Sichuan,Shaanxi,and Gansu provinces in China,its conservation has attracted global attention.The study of habitat selection is the basis for formulating effective conservation and management strategies.The response of species to different environmental variables usually occurs at different scales.In previous studies,researchers usually ignored this point and constructed habitat models on a subjectively selected single scale,which may reduce the effectiveness of the habitat model and draw wrong conclusions.Climate change caused by human activities has led to the extinction of many species,and it has been known as one of the main factors threatening biodiversity in the 21st century.Giant pandas have a narrow distribution,single diet,weak migration ability,and sensitivity to human activities,all of which make giant pandas more affected by climate change.Predicting the changing trend of the giant panda’s habitat under future climate change scenarios is great significance to actively developing conservation and management strategies,it is also a research hotspot.However,in the process of forecasting the impact of climate change on species distribution,the selection of predictive variables,the selection of variable scales,the selection of niche models(also known as species distribution models or habitat suitability models),the selection of General Circulation Models,the setting of model parameters and the selection of thresholds when converting the continuous habitat suitability to binarized habitat and non-habitat all have a significant impact on the assessment results.At present,there is no research to reduce the uncertainty of the evaluation results in the study of the climate impact of giant pandas from the perspective of optimizing the modeling process.For this reason,species distribution models,geographic information systems,landscape ecology,and other technical means were used in the research,and on the basis of the data of the fourth giant panda survey of Qionglai Mountain,we conducted the following researches,and the corresponding results are:1.Multi-scale habitat suitability of giant pandaThe phenological index extracted from MODIS time-series remote sensing images,combined with the optimization of the maximum entropy(Maxent),successfully simulated the distribution of bamboo in the understory of the Qionglai mountain.Scale optimization analysis showed that the response of giant pandas to habitat factors on different spatial scales,such as altitude(250 m),road density(6000m),bamboo percentage(500m),the largest patch index of close coniferous forest(2000 m)and slope(2000 m).The performance of the multi-scale model is better than single-scale encounters.According to the AUC and CBI,regarding the habitat suitability prediction map,there is a difference between the multi-scale and single-scale models,their overlap index varies between 0.718 and 0.917.2.Comparison of machine learning and traditional statistics in giant panda multi-scale habitat modelingScale optimization based on random forest shows that environmental variables affect the occurring of giant pandas on relatively small scales(≤1000 m)or relatively large scales(≥5000 m).Different scale optimization methods show minor differences in the selected scale.The performance of the multi-scale random forest model is significantly better than the traditional statistical method-the logistic regression model.Prediction of random forest multi-scale model shows higher heterogeneity in the habitat suitability of the giant panda habitat,whereas the prediction of the logistic regression shows a smoother monotonous pattern.3.Impacts of climate change on habitat suitability of giant pandasBy 2041-2070,under different climate change scenarios,the suitable habitat area for giant panda’s staple food bamboo will be reduced by about 6.93%～12.95%.The reduction area occurs mainly in the low-altitude area of the eastern Qionglai mountain.Combining scale optimization and model parameter tuning,the performance of the ensemble model is significantly better than the single algorithm model and the Maxent model with default parameters which is the most commonly used method in former evaluation studies.Comparison of performance is based on three indicators of AUC,TSS,and CBI.Under three different greenhouse gas emissions(ssp126,ssp370,ssp585),the areas of giant panda habitat that will experience a suitability reduction are 981,1784,and 2274 km2,of which the areas with slightly suitability degraded account for the most significant proportion,818,1251,and 1580 km2.The habitat area with increased suitability is tiny(0-19 km2).Comparing the prediction results of the optimized ensemble model and the Maxent model with default parameters,it is found that the Maxent model predicts that the habitat areas with reduced suitability under the three scenarios are 750,1951,and 2908 km2,indicating the uncertainty is significantly higher than that of the ensemble model.4.Assessment of giant panda population connectivity in Qionglai Mountain under climate changeBased on the habitat suitability prediction of the optimized ensemble model,two connect approaches-resistant kernel and factorial least-cost analysis were used to simulate the population connectivity and long-distance dispersal corridors under the current and future climate change scenarios.The study found that under the 10000 units cost move ability threshold,the Qionglai giant panda population is currently well connected,and all population patches can be connected by long-distance dispersal(50000 units cost threshold).With climate change,the percentage of landscape and the largest patch index of population patch will decrease significantly with the percentages of decrease being 10.94～21.94%and 6.03～14.99%,respectively.The largest population patch is less affected by climate change than other small patches.In addition,climate change will lead population distributed in Tianquan to county loss connection with the large population patch distribution in Wenchuan,Dayi,Lushan,and Baoxing.Furthermore,there are six areas that can be used as corridors currently and predicted to become barriers under climate change.This study improves the effectiveness of the habitat model in conservation and management practices from the perspective of scale optimization,and reduces the uncertainty of the assessment results on the impact of climate change on the giant panda’s habitat from the perspective of optimizing the modeling process.Based on them,the impact of climate change on population connectivity was assessed.This study attempts to construct the methods and results can provide suggestions for the decision-making of giant panda protection and management departments.