基于MaxEnt模型的芦苇中国潜在分布区预测

Prediction of Potential Distribution of Phragmites australis in China Based on MaxEnt

研究基于338条芦苇分布数据和6个环境变量,利用MaxEnt模型对芦苇在我国的潜在分布区进行模拟,结合环境变量的贡献率、置换重要性、刀切法检验结果以及单变量响应曲线综合分析影响芦苇分布的主要环境变量及其适宜范围。结果表明:MaxEnt模型在RM=1、FC=LQHP组合参数下的AUC均值为0.909,表明预测结果准确度极高;芦苇在我国的潜在分布区主要位于甘肃、山西、陕西等地区,其中高、中适生区面积分别为1.25×10^(9)、1.88×10^(9) hm^(2);影响芦苇分布的主要环境变量是年平均气温、最湿月降水量、等温性、最冷季度降水量和年平均气温变化范围;当年平均气温为14~22℃、最湿月降水量为108~284 mm、等温性低于24、最冷季度降水量小于50 mm、年平均气温变化范围低于28.5℃时,芦苇的分布概率较高。

With 338 data records and 6 environmental variables,this study adopted MaxEnt to predict the potential distribution of Phragmites australis in China.The contribution rate,permutation importance,jackknife test,and univariate response curve were employed to analyze the environmental variables affecting the distribution and the suitable distribution zones of P.australis.With the parameters of RM=1 and FC=LQHP,the MaxEnt model showed the area under the receiver operating characteristic curve of 0.909,which indicated high prediction accuracy.The potential distribution zones of P.australis in China mainly included Gansu,Shanxi,and Shaanxi,and the highly and moderately suitable distribution areas were 1.25×10^(9) and 1.88×10^(9) hm^(2),respectively.The main environmental variables affecting the geographical distribution of P.australis were annual mean temperature,precipitation of the wettest month,isothermality,precipitation of the coldest quarter,and annual mean temperature range.The distribution probability of P.australis was high when the annual mean temperature,precipitation of the wettest month,isothermality,precipitation of the coldest quarter,and annual mean temperature range were 14-22°C,108-284 mm,below 24,less than 50 mm,and within mean 28.5°C,respectively.