基于GARP和MaxEnt的云杉矮槲寄生分布区的预测
Predicting the distribution of dwarf mistletoe (Arceuthobium sichuanense) with GARP and MaxEnt models
-
摘要: 云杉矮槲寄生(Arceuthobium sichuanense)是一种寄生性种子植物,主要寄生于云杉属植物,近年来对我国“三江源”地区的云杉天然林和次生林造成严重危害。本研究利用GARP和MaxEnt生态位模型,基于已报道的云杉矮槲寄生分布点的数据,对其在中国的分布区进行了预测和分析,并采用ROC曲线对预测结果进行检验和评价。结果表明:GARP模型预测的分布范围较广,MaxEnt内部层次更细致,为得到最佳结果,将2个模型的预测结果赋予一定的权重(4:1)从而获得最佳结果;ROC曲线评价结果表明,GARP-MaxEnt模型的AUC值为0.937,达到了极高的精度。通过该模型预测可见,云杉矮槲寄生在中国的适生区主要集中在青海、甘肃、四川和西藏地区,其中青海、甘肃、四川交界区域为云杉矮槲寄生的极度适宜分布区。本研究结果有利于全面了解云杉矮槲寄生的分布范围,对云杉矮槲寄生害的调查与监测以及制定科学防治策略具有重要的理论指导意义。Abstract: Sichuan dwarf mistletoe (Arceuthobium sichuanense) is a plant that parasitizes Picea trees. Recently, this dwarf mistletoe has caused serious damage to spruce trees in Sanjiangyuan area, Qinghai Province. In this study, based on the data on the reported sites, we applied GARP and MaxEnt niche models to predict and analyze the distribution area of A. sichuanense in China. The predicted results were then tested by ROC curves. The results showed that the predicted area by GARP model was more extensive while the area by MaxEnt model was more detailed. Thus, GARP and MaxEnt models were combined with a weight ratio 4:1 to predict the distribution area of A. sichuanense. The AUC value of GARP-MaxEnt model was 0.937, indicating that the precision of prediction was high. Based on the GARP-MaxEnt prediction model, it is shown that the distribution area of A. sichuanense is mainly in Qinghai, Gansu, Sichuan and Tibet. The junction area of Qinghai, Gansu and Sichuan was the most suitable area for A. sichuanense parasitism. The results will benefit our understanding of the distribution area of A. sichuanense, and have an guiding significance for monitoring the disease and making scientific strategy to control this dwarf mistletoe.