Based On Species Distribution Model Analysis Suitable Distribution Area Of Stipa And The Correlation With Climate Factors In China

Stipa is the most representative species of grassland ecosystem in China.With the zonal distribution of water and heat conditions,different types of Stipa steppe have been formed.The species distribution model(SDMs)is an important method for conservation biology and climate change research,in which the distribution information of species is correlated with the climatic and environmental factors in the distribution area through certain algorithms,so as to predict the distribution of target species.In this paper,use SSDM package in R language,selects the MAXENT,GAM,GBM,GLM,CTA model and ensemble these 5 single models,using the fifth assessment report of intergovernmental panel on climate change(IPCC)climate biological data under the situation of 3 kinds of climate scenario(current,2070 s RCP2.6 and RCP8.5).12 mainly species of genus of stipa were selected as research subjects,potential distribution area of these species were projected and calculate the changing areas of suitibile habitat of all species under three climate scenerio.The main research conclusions of this paper are as follows:(1).In order to avoid overfitting,environmental variables with a cross-correlation coefficient <± 0.9 and the contribution rate > 5% are chosen for model construction.After selecting,the standard deviation of temperature seasonal changes(bio04)is the best climatic factor that affects the distribution of Stipa baicalensis in the meadow steppe and the Stipa grandis in the typical steppe;Under the current climate scenario,the september precipitation(prec09)has the main effect on the distribution of stipa bungeana.The precipitation in June(prec06)is the best climatic factor affecting the distribution of Stipa krylovii.The solar radiation in March(srad03)affects the distribution of Stipa sareptana.The best climatic factors affecting the distribution of Stipa breviflora,Stipa klmenzii,Stipa caucasica and Stipa glareosa are August mean temperature(tavg08),June solar radiation(srad06),February wind speed(wind02),and annual Average temperature variation range(bio07).Under the 2070 RCP 2.6 scenario,the best climatic factors that affect the distribution of Stipa bungeana,Stipa krylovii,and Stipa lancea are the average annual precipitation(bio12),June precipitation(prec06),and November minimum temperature(tmin11).The best climatic factors that affect the distribution of species in desert grasslands are June precipitation(prec06),standard deviation of temperature seasonal changes(bio04),precipitation variation coefficient(bio15),and June precipitation(prec06).Under the 2070 s RCP 8.5 scenario,the best climatic factors for the distribution of Stipa bungeana,Stipa krylovii,and Stipa sareptana are the lowest temperature in March(tmin03),the standard deviation of temperature seasonal changes(bio04),and the dryest monthly precipitation(bio14).The best climatic factors affecting the distribution of species in desert grasslands are July precipitation(prec07),standard deviation of temperature seasonal changes(bio04),precipitation variation coefficient(bio15),and precipitation variation coefficient(bio15).Solar radiation and temperature are the best climatic factors that affect the distribution of species in the alpine steppe(Stipa roborowskyi,Stipa purpurea,Stipa subsessiliflora).(2).From the results of SDMs,we can see that the potential distribution area of Stipa baicalensis in meadow steppe is located in the eastern part of the Qinghai-Tibet Plateau,Loess Plateau and Mongolian Plateau,mainly in the eastern part of the steppe.As the climate changes,the highly suitable potential distribution area of Stipa baicalensis gradually increases.In typical steppe,the Stipa grandis is potentially located in the Songnen Plain,Mogolian Plateau,and northern Loess Plateau,and can extend to the eastern Qinghai-Tibet Plateau.Stipa sareptana is potentially concentrated in the Tianshan mountain;Stipa krylovii is potentially distributed in the mountains of Hulunbeier Plateau,Xilinguole Plateau,the eastern margin of the Qinghai-Tibet Plateau of Ulanqab Plateau,and desert areas of Qilian Mountain and Tianshan Mountain.Stipa krylovii is potentially located in the north-central part of the Loess Plateau,the Qilian Mountains,the west of Qinghai Lake,and the Xiliao River Plain.As the climate changes,its suitable habitat gradually increases.In desert steppe,Stipa breviflora is potentially distributed on the sunny slopes of the Loess Hilly Region,with sporadic distribution in Helan Mountains,Qilian Mountains,and Tianshan Mountains;the Stipa klemenzii potentially growing on the wavy plains of the Ulanqab Plateau and the western part of the Xilinguole Plateau Plains;Stipa caucasica is potentially distributed in areas with higher elevations in Xinjiang;the potential distribution area of Stipa glareosa is located in the desert steppe zone of the western Mongolian plateau,northern Qinghai-Tibet Plateau,and mountains of the desert area;The suitable habitat has a tendency to increase gradually with climate change.Among the alpine steppe,the Stipa roborowskyi is potentially located in the high mountain areas of Tibet and Xinjiang;Stipa purpurea is potentially distributed in the Qiangtang grassland,the Pamirs,the South Tibetan Lake Basin,the east and south of the Kekexili region,and the Qilian Mountains;the potential distribution areas of Stipa subsessiliflora are located in the Tianshan,Pamir Plateau,Kunlun Mountains and Altun Mountains;In the alpine steppe with the change of climate,the potential suitable habitat area of stipa reduced.(3).The accuracy of the model differs depending on the species and climate scenario.This paper selects the widely used AUC and Kappa statistic values to evaluate the accuracy of the model.It is found that the accuracy of the MAXENT and GBM models based on machine learning algorithms in the five models selected in this paper is generally higher than that of other models.(4).The Ensemble model is built by weighted average of the prediction results of single algorithms.In the meadow steppe,the potential distribution area of Stipa baicalensis increased from the current to 2070 s RCP2.6,and from the current to 2070 s RCP 8.5,its potential suitable distribution area increased;In the typical steppe,from the current to 2070 s RCP 2.6 climate scenario,the areas of potential suitable distribution areas for Stipa grandis,Stipa krylovii,and Stipa sareptana decreased.While the suitable potential distribution areas of Stipa bungeana increased.From the current to the 2070 s RCP 8.5 climate scenario,the potential distribution area of Stipa grandis and Stipa krylovii increased,while the area of potential distribution of Stipa bungeana and Stipa sareptana decreased;In the desert steppe,from the current to the 2070 s RCP 2.6 climate scenario,the suitable distribution area of Stipa breviflora,Stipa klemenzii and Stipa caucasica increased,the area of suitable distribution areas for Stipa glareosa decreased.From the current to the 2070 s RCP 8.5 climate scenario,the area of suitable distribution areas for Stipa breviflora,Stipa klemenzii and Stipa caucasica increased.the suitable distribution area of Stipa glareosa decreased;In the alpine grassland,from the current to 2070 s RCP 2.6,the potential distribution area of Stipa roborowskyi decreased,The potential distribution area of Stipa purpurea and Stipa subsessiliflora increased.From the current to the 2070 s RCP 8.5 climate scenario,the area of the potential distribution area of Stipa roborowskyi and Stipa subsessiliflora decreased,the potential distribution area of Stipa purpurea increased.