Predicting White-naped Crane Breeding Habitat Distribution And Testing By GPS-GSM Tracking Data

White-naped Crane?Grus vipio?is classified as vulnerable by the International Union for Conservation of Nature?IUCN?and is enlisted as a class ? protected species in China.As a method for predicting suitable habitats,Species Distribution Models has been widely used in the research and conservation of endangered wildlife.For model assessment,previous studies usually rely on part of samples from whole samples,but still the practical field survey is the most recommended approach for validation.Currently the application of GPS-GSM tracking devices provides researchers with abundant species distribution data.Consequently,this study first built species distribution models for the White-naped Crane,with 177 nests data collected from our fieldwork and literatures during 2006 and 2016 and 29 environmental variables from Open Access sites.The results showed indicated that the model showed a rather high performance on the training data?AUC=0.982?..In Northeast Asia,the White-naped Crane mainly breed in Dauria on the border of Russia,Mongolia and China;Songnen plain in China,Lesser Khingan Mountains hinterland in Heilongjiang province,China and southern Amur state in Russia;Sanjiang plain,the downstream of Heilongjiang in Khabarovsk,Russia and the east coast of Ussuri river;Sunitezuoqi in Xilin Gol League,Inner Mongolia,China.Variable importance rankings indicated that the most significant predictors are slope?<0.9°?,bio₄?14,300-15,300?,and disriver?<800m?.Subsequently,model predictions got further validated by GPS-GSM tracking data:using 53,990 distribution data collected from GPS-GSM tracking work conducted on 17 individuals during 2014 and 2016.The results showed that our model predictions showed rather high consistency with the practical breeding distribution of the White-naped Crane?median:0.903?.Also,this study additionally proved that when using distribution data collected from tracking individuals for model assessment,breeding individuals are better than non-breeding individuals.Specially,for non-breeding individuals,data from pre-nest period was better than incubation and post-nest period?breeding period delineated according to the breeding individuals?.This study showed an effective method for model validation,greatly extending the application of GPS-GSM tracking data.