Changes in uncertain factors such as parameter setting can lead to variations in the ecological distribution predicted by the same model.Therefore,quantifying the contributions of different uncertainty factors is crucial for reducing variability in ecological predictions.However,there is limited research analyzing the modeling uncertainty of specific models.This study,using Gentiana macrophylla as an example,explores the uncertainty in its distribution prediction,with a specific focus on parameter settings.Initially,principal component analysis(PCA)and ecological variable grouping method(EVGM)were employed to select environmental factors.Six sets of models were established using 25%of presence point data and two methods for handling missing point data as test methods.The study investigates the impact of the presence point test set proportion on model performance,conducting a comprehensive analysis of training,testing AUC values,and spatial distribution area.The optimal model parameters for species were determined,revealing that a 20%random testing sampling proportion was optimal.This model not only provides guidance for the conservation and ecological planning of Gentiana macrophylla and other medicinal herbs but also serves as a theoretical reference for determining the optimal model for species spatial distribution.
parameter settinguncertainty analysisMaxEntGentiana macrophyllapresence point data
国家科技期刊平台
NSTL
万方数据
