摘要
【目的】红松是我国重要的经济树种和生态树种,预测红松的潜在分布范围和分析影响其生长的重要环境因子很有必要。在现有的建模工具中,最大熵(MaxEnt)模型是应用最为广泛的模拟物种分布模型。在建模过程中,大多数研究人员常使用默认参数来进行计算,然而,最近研究表明,基于默认参数构建的模型结果并非最优,这些参数的设置仍有待讨论。【方法】利用红松分布数据和环境变量,在不同随机测试百分比下计算训练数据和测试数据的建模精度,用最适宜的比例模拟了红松的潜在分布。【结果】训练数据的AUC在不同百分比下的范围为0.890~0.898,而测试数据的AUC在不同百分比下的范围为0.854~0.884。当百分比为5%时,两组数据的准确率都比较准确。在此比例下,Bio12(年平均降水),Bio14(最干月降水量)和Bio15(降水量变异系数)对红松分布最为重要。【结论】研究结果表明在不同随机测试百分比下,模型的精度有所差异,默认值25%不是最合理的百分比,而在5%时训练数据和测试数据的结果更准确。降水和温度是影响红松潜在分布的气象要素,但是在东北地区,由于温度空间差异较小,影响红松潜在分布的主要为降水。红松主要分布范围在东北地区的东北和西南地区,特别是在辽宁省和吉林省交界处。研究的结果表明模型参数的设置对于模型的精度十分重要,调整参数可以更合理地反映物种对环境因素的反应,准确地模拟物种的潜在分布。
[Objective]Pinus koraiensis is an important economic and ecological tree species in China.Therefore,it is necessary to predict the potential distribution area of Pinus koraiensis and analyze the important environmental factors affecting its growth.Among the existing modeling tools,MaxEnt is the most popular and widely used model in species distribution simulation.When modeling in Maxent,most researchers often use default parameters.However,recent studies have shown that the default parameters are not always the best results when building the model,and the setting of these parameters remains to be discussed.[Method]The distribution data of Pinus koraiensis and environmental variables were used to calculate the modeling accuracy of training and test data under different random test percentages,and simulate the potential distribution of Pinus koraiensis with the most suitable percentage[.Result]The AUC of the training data ranged from 0.890-0.898 at different percentages,while the AUC of the test data ranged from 0.854-0.884.When the percentage was 5%,the accuracy of both sets of data was higher.At the percentage of 5%,Bio12(Annual Precipitation),Bio14(PrecipitationofDriestMonth)andBio15(Precipitation Seasonality)werethemostimportantforPinus koraiensis.[Conclusion]The accuracy of the model varies in different random test percentages.The default value of 25%is not the most reasonable percentage,and the results of training data and test data are more accurate at 5%.Precipitation,instead of temperature,is the main climate factor that affects the potential distribution of Pinus koraiensis in Northeast China,which is due to the small spatial difference in temperature.Pinus koraiensis is mainly distributed in the northeast and southwest regions of the Northeast China,especially at the junction of Liaoning and Jilin provinces.The results of this study indicate that the setting of model parameters is very important for the accuracy of the model and the potential distribution simulation can be more accurate when the parameters are adjusted.
作者
张劳模
庞丽峰
许等平
唐小明
ZHANG Lao-mo;Pang Li-feng;XU Deng-ping;TANG Xiao-ming(The Research Institution of Forest Resource Information Technique,Chinese Academy of Forestry,Beijing 100091,China;College of Computer Science,Henan University of Engineering,Zhengzhou 451191,China;Forestry Industry Planning and Design Institute,National Forestry and Grassland Administration,Beijing 100010,China)
出处
《江西农业大学学报》
CAS
CSCD
北大核心
2020年第1期74-83,共10页
Acta Agriculturae Universitatis Jiangxiensis
基金
“十三五”国家重点研发计划项目(2017YFC0504106)
河南省高等学校重点科研项目资助计划(18A520024)~~
关键词
潜在分布模拟
最大熵模型
随机测试百分比
红松
potential distribution simulation
MaxEnt model
random test percentage
Pinus koraiensis