新疆典型蝗虫适生区分布预测

Prediction of potential living area of typical locusts in Xinjiang based on species distribution model

新疆草原面积广阔,农牧业地位突出,蝗灾对当地经济、生态威胁很大,近年新疆极端天气日渐频发,蝗灾监测与防治任务艰巨。以意大利蝗、西伯利亚蝗和亚洲飞蝗为代表的蝗虫数据为基础,综合考虑对蝗虫各生命周期有重要影响的环境因素,运用BIOCLIM模型、领域模型(DOMAIN)、马氏距离模型(MAHAL)、广义线性模型(GLM)、随机森林模型(RF)、提升回归树模型(BRT)、支持向量机模型(SVM)、最大熵模型(MaxEnt)等八种经典物种分布模型及集成模型对新疆典型蝗虫适生区展开了预测。结果表明:(1)不同模型对新疆典型蝗虫适生区预测存在差异,其中DOMAIN最差(曲线下面积(AUC)=0.688,真实技巧统计(TSS)=0.301),而提升回归树(BRT)最佳(AUC=0.920,TSS=0.910),基于BRT、SVM和MaxEnt 3个集成模型预测的新疆蝗虫适生区可靠性更高;(2)新疆典型蝗虫不同等级适生区面积约56.844万km^(2),占新疆总面积的36%,其中高适生区面积16.568万km^(2);(3)新疆典型蝗虫适生区主要集中于北疆阿勒泰、塔城地区,此外东部哈密地区及南疆绿洲边缘地带亦有分布。研究可为新疆草原工作部门推进蝗虫监测防治工作提供支持。

Xinjiang grassland covers a vast area and occupies a prominent position in agriculture and animal husbandry. Locust plague is a great threat to the local economy and ecology. Furthermore, extreme weather in Xinjiang has become more frequent, the task of monitoring and preventing locust plagues is difficult. Based on the data of locusts represented by Italian locust, Siberian locust and Asian migratory locust, we comprehensively considered the environmental factors that had important influence on the different life cycles of locusts, and used eight typical species distribution models of BIOCLIM, DOMAIN, MAHAL, GLM, RF, BRT, SVM, MAXENT, and ensembled models to predict typical locust suitable areas in Xinjiang. The results showed that:(1) different models had differences in the prediction of typical locust suitable areas in Xinjiang, in which DOMAIN was the worst(AUC=0.688, TSS=0.301), and the BRT was the best(AUC=0.920, TSS=0.910). The prediction of the suitable area based on the three ensembled models of the BRT, SVM and MaxEnt was more reliable.(2) The suitable area of typical locusts of different grades in Xinjiang was about 568440 km^(2), accounting for 36% of the total area of Xinjiang, among which the high adaptable area was 165680 km^(2).(3) The typical locust breeding areas were mainly concentrated in the Altay and Tacheng areas of northern Xinjian. In addition, they were also distributed in Hami and the edge of oasis in southern Xinjiang. This study may provide support to the Xinjiang grassland authorities in their efforts to monitor and control locusts.