基于MaxEnt模型的突颊侧琵甲新疆潜在地理分布预测和主要影响因子分析

为有效预警和防控突颊侧琵甲Prosodes(Lioprosodes)dilaticollis Motschulsky发生,基于该物种分布数据、1950―2000年环境下生物气候数据和典型受害区土壤因子数据、气象数据等,采用MaxEnt模型对其在新疆的潜在地理分布及其主要影响因子进行预测和分析验证。结果表明:受试者工作特征曲线下面积(area under the ROC curve,AUC)都大于0.990,预测结果可靠。突颊侧琵甲的潜在分布区位于新疆天山一带,高、中度适生区分布在伊犁河谷草原和昌吉州玛纳斯县、呼图壁县草原及塔城地区沙湾县等地;刀切法得出降水量是影响其潜在地理分布的主要因素,特别是5、6、11、12月降水量的影响最大。该虫典型受害区的降水数据验证了分布概率响应曲线的准确性;土壤湿度在15%~30%范围时,30 cm处土壤湿度与虫口密度呈极显著弱正相关。表明降水作为影响草原土壤湿度、土壤含氧量的主要和直接因素,对不同生长期突颊侧琵甲生存起到重要作用。

Prediction of potential geographic distribution and analysis of the major driving factors for Prosodes(Lioprosodes)dilaticollis Motschulsky in Xinjiang by using MaxEnt model

Prosodes(Lioprosodes)dilaticollis Motschulsky is the main grassland pest in Xinjiang.In order to carry out early-warning,prevention and control effectively,the potential geographic distribution of P.dilaticollis was predicted and the major environmental effect factors were analyzed using GPS data of actual distribution and biological climatic data (1950-2000) in Xinjiang based on MaxEnt model.The soil factor and meteorological data of the typical hazard zone were used to verify the results of effect factors,and the results were also validated by ROC (receiver operating characteristics curve) analytical method.The results showed that AUC (area under the ROC curve) of the training data and testing data with MaxEnt model were all greater than 0.990,suggesting that the predicted results had achieved the level of excellence.The potential geographic distribution of P.dilaticollis in Xinjiang was mainly in Tianshan Mountains region of Xinjiang.The high-level suitable distribution region and moderate suitable distribution region of P.dilaticollis were in the grassland of Ili River valley and the grassland of Manas County,Hutubi County in Changji Prefecture,and Shawan County in Tachen Prefecture.Precipitation was the main factor for the potential distribution of hazard zones.Particularly the rainfall of May,June,November and December had the greatest impact on the geographic potential distribution of P.dilaticollis.Historical precipitation of the typical hazard zones verified the accuracy of response curves about environmental variables.When soil moisture was in the range of 15%-30%,soil humidity of field sampling and population densities showed the most significantly weak positive correlation.Moreover,precipitation was the main and direct factor for the change of grassland soil humidity and soil oxygen content,which played an important role in the survival of P.dilaticollis at different developmental stages.