基于MaxEnt模型的科尔沁沙地刺萼龙葵适宜生长区预测

Prediction of suitable Solanum rostratum growth areas in Horqin Sandy Land based on the Maximum Entropy

为明确外来入侵植物刺萼龙葵(Solanum rostratum)在科尔沁沙地的适生区分布,利用刺萼龙葵的记录点和19个环境变量,用最大熵模型(MaxEnt)和地理信息系统(ArcGIS)相结合对分布在科尔沁沙地的刺萼龙葵的地理数据进行分析,预测其适生区,并采用受试者操作特征曲线(ROC)分析法对预测结果进行验证,训练数据和验证数据在特性曲线下的面积值(AUC)分别为0.849和0.950,预测结果较为准确。结果表明,气象因子中最暖月高温、最冷季平均温度、最干月降水量、昼夜温差日均温是影响刺萼龙葵分布的主要环境变量;刺萼龙葵在科尔沁沙地的高适生区约占3.9万km^(2),主要集中分布于科尔沁沙地中部地区。

To determine the characteristics of suitable habitats for Solanum rostratum,an alien invasive plant,in Horqin Sandy Land,geographic data matching S.rostratum distribution were analyzed.Suitable habitat requirements were predicted based on recorded S.rostratum locations and 19 environmental variables,using the Maximum Entropy Model and Geographic Information System(ArcGIS).Receiver operating characteristic curve analysis validated the predictions.The areas under the characteristic curve values of the training and validation data were 0.849 and 0.950,respectively,indicating that the prediction results were relatively accurate.The main environmental variables affecting S.rostratum distribution included the highest temperature in the warmest month,the average temperature in the coldest season,the amount of precipitation in the driest month,and the daily average temperature difference between day and night.The most adaptable area for S.rostratum in Horqin Sandy Land spans approximately 39000 km^(2),primarily distributed in the central region.