Species Distribution Model Construction And Model Comparison Under Climate Change

Changes in species distribution are affected by climate change.Climate change will lead to changes in the range of species and the spontaneous migration of species,and more and more species appear in high latitudes or high altitudes,which are originally unsuitable.With the development of the statistical model and niche theory,the species distribution models(SDMs)have experienced the gradual development procession from the poor to the advanced.The problems of uncertainty and accuracy on SDMs become increasing important within the process of SDMs.The choice of SDMs is the pivotal element to lead to the uncertainty of SDMs.On the existing technology,ensemble model(EM)that can model by synthesizing the merit of single models is the generic framework that improves the accuracy of SDMs prediction and reduces the uncertainty of the model.Notopterygium incisum Ting ex H.T.Chang is an endangered plant.The marginal areas of the Qinghai Tibet Plateau were the core habitat of N.incisum,which is very sensitive to climate change,and the future climate change will have an impact on the distribution of the species.Therefore,it is of great significance to determine the suitable habitat area of the species for resource conservation and sustainable utilization.In this study,the potential suitable distribution of N.incisum in China was simulated using 12 single SDMs(RF,CTA,GLM,GAM,MARS,GBM,FDA,SVM,ANN,MaxEnt,Maxlike,and SRE)based on species data(105 presence point data and 1500 pseudo-existence point data)and 9 environmental variables(climatic and terrain).We evaluate the results of each single model through the average value of the true skill statistic(TSS)and the area under ROC curve(AUC).Then,according to the average value of TSS,10 models were chosen to build an EM,and based on EM,we modelling CHS and projected the future distributions of N.incisum for the 2050s(average for 2041-2060)and 2070s(average for 2061-2080).In the process of modelling CHS,we select soil and vegetation factors as limited factors.Finally,we compere modeling effects and the results of the 12 single models and CHS,quantity the environmental variables of the suitable habitat,and portray the distribution of N.incisum under the climatic changing.The main conclusions of this study are as follows:1.In the 12 model,the optimal model is RF,the average value of TSS and AUC is0.929 and 0.984 respectively.The batter model were GBM?ANN?FDA?the worst models were Maxlike,SRE.The average value of TSS and AUC were partly lower than 0.7 and 0.8.2.According to the results of each single model,we know that the results of simple algorithm model overestimated the area of potential distribution of N.incisum;the complex machine learning model has a good spatial distribution performance to a certain extent.But all the results show that the high suitable area distributes in the east margin of the Qinghai-Tibet plateau.3.We model EM base on 10 SDMs,its average value of TSS and AUC are 0.857 and 0.963,respectively,so the effect of EM is good.The results of CHS showed that the area of suitable habitat for N.incisum is 7.539×104 km2 in current,and the habitats are mainly located in the Tibet Autonomous Region,Gansu Province,Qinghai Province,and Sichuan Province.EM simulations indicated that the area of suitable habitats for N.incisum would significantly decrease and would be 6.267×104 km2 and 5.776×104 km2 in the 2050s and 2070s,respectively.However,the area of marginally suitable habitat would remain relatively stable.4.The 6 main environmental variables that are biol(Annual mean air temperature),bio4(Temperature seasonality),bio 10(Mean temperature of warmest quarter),bio 11(Mean temperature of coldest quarter),bio 12(Annual precipitation),and bio 15(Precipitation of the driest quarter)effect the distribution and growth of N.incisum,and they are the reaction of characteristics of suitable habitat.5.In this study,to a certain extent,many methods reduced the uncertainty of model,like,the selection of sampling points,environmental variables,and models,and modelling EM and CHS.Meanwhile,they makes the results of distribution is consistent with the actual results.Results provide valuable information for future ecological conservation and management of N.incisum,predict the expansion or contraction of medicinal plants habitat,and can be used as a reference for other medicinal plants.And it provides primary habitat protection for economically valuable medicinal species and endangered plant species.