基于物种预测分布模型对桔小实蝇在全球适生区的评估

Assessment of potential distribution of Bactrocera dorsalis in the world based on predictive species distribution models

采集了桔小实蝇的发源地分布点数据,并分别用最大熵值法(MAXENT)模型和基因遗传算法(GARP)模型对桔小实蝇在全球的适生区进行了预测.桔小实蝇的发源地主要是在中非和南非,根据模型分析可知:2种模型预测的桔小实蝇的适生区分布范围相似,主要为北美洲的南部,南美洲的北部、中部和东部,中非和南非,亚洲南部,以及大洋洲的沿海地区.在实验中,用受试者工作特征曲线(ROC曲线)评估模型的准确性,MAXENT模型的曲线下面积(AUC)为0.82,而GARP模型的AUC为0.92,说明GARP模型在此研究中的准确度高于MAXENT模型.

Bactrocera dorsalis,a worldwide quarantine insect pest,had great danger to more than 250 kinds of fruits and vegetables.The original distribution locations were collected in this investigation,then the MAXENT and GARP models were applied to predict the potential distribution of Bactrocera dorsalis around the world.Central and south parts of Africa were the original area of the pest.According to the results of the two models,the main areas of the pest potential distribution were similar,including south part of North America,north,central and east parts of South America,south part of Asia and the coastal region of Oceania.ROC curve was used to evaluate the accuracy of the models,the AUC value of MAXENT was 0.82,while that of GARP was 0.92,which indicated that the GARP model was more accurate than MAXENT.The results suggested that the models could be useful in determination and prevention of Bactrocera dorsalis.