关键词: 葫芦寡鬃实蝇, 埃塞俄比亚寡鬃实蝇, 西瓜寡鬃实蝇, 生态位模型, 最大熵值, Maxent, BIOCLIM, DOMAIN

Abstract: In order to predict and analyse the potential geographical distributions of three important quarantine invasive pests, Dacus bivittatus, D. ciliatus and D. vertebratus, three ecological niche modeling techniques, BIOCLIM, DOMAIN and Maximum entropy nichebased modeling (Maxent) were implemented by using distribution records of the three fruit fly species and a set of environmental predictor variables. Differences in prediction performance of the three models with thresholds were observed. An evaluation using independent records of D. bivittatus showed that Maxent offers the most accurate predictions than two other models based on three values of ROC/AUC, Kappa, and TSS. Prediction outcomes made by Maxent revealed that the three fruit fly species have broadly similar potential ranges in Central American, South American, Southeast Asia, and Coastal areas of Australia in general. D. ciliatus has the comparatively widest potential range among the three species, including Coastal areas of Mediterranean Sea, Saudi Arabia, Yemen, Oman and South Iran, suggesting that it may be tolerant of the widest range of climatic conditions among the three species. In China, while large areas of Yunnan and Hainan are very habitable for all the three fruit fly species, southern part of Guangdong and Taiwan are also their habitable areas. D. ciliatus has the widest potential distribution area, with southern part of Sichuan, Guizhou and Tibet plus the coastal areas of southern China all being its suitable areas. The risk of the three fruit fly species permanently establishing in Guangdong if introduced exist but low. Jackknife analysis revealed that temperature and its variation have comparative significant influence on the distribution patterns of three fruit fly species both in global and restricted regions.