Potential Distribution Predictions of Three Precious Hard Broad-leaved Tree Species Based on MaxEnt and RF Models
Fraxinus mandshurica,Juglans madshurica and Phellodendron amurense are three precious hard broad-leaved tree species in northeast China.Analyzing relationships among distributions and environmental factors and predicting potential distribution areas of three precious hard broad-leaved tree species could pro-vide references for regeneration and replanting of three precious hard broad-leaved tree species.Taking Maoer-shan Experimental Forest Farm as study area,twelve environmental factors,including A1 soil layer thickness,AB soil layer thickness,soil type,slope gradient,slope position,slope aspect,altitude,canopy density,number of plants per hectare,accumulation per hectare,site type and total coverage of ground were selected to predict potential distributions of three precious hard broad-leaved tree species by maximum entropy(MaxEnt)model and random forest(RF)model,and accuracies of models were evaluated by area under receiver operating char-acteristic curve(AUC).Results showed that AUC means of maximum entropy model and random forest model were 0.854 and 0.920,respectively.Main environmental factors output by maximum entropy model were site type and slope position,which output by random forest model was altitude.Accuracy of random forest model was better than that of maximum entropy model,and output results of two models showed that main factor limit-ed distributions of three precious hard broad-leaved tree species was terrain factor.Western and northern parts of Maoershan Experimental Forest Farm were suitable for growths of three precious hard broad-leaved tree spe-cies,and all of distribution points of three tree species were within range of suitable and more suitable areas.
maximum entropy modelrandom forest modelpotential distribution prediction
国家科技期刊平台
NSTL
万方数据
