MaxEnt生态学模型在野生植物近自然林引种区划中的应用——以红姜花为例

Application of MaxEnt ecology model in near-nature forestry plant introduction regionalization with Hedychium coccineum as an example

关键词:  引种区划  红姜花  MaxEnt生态学模型  近自然林

Key words:  introduction regionalization  Hedychium coccineum  MaxEnt ecologic model  near-nature forestry

摘要:

红姜花的野生地理分布数据基础上,选择温度、降雨、海拔为环境因子,以75%的数据进行建模,25%的分布数据作为检验作ROC(Receiver operating characteristic)曲线对模型的有效性进行评价。结果表明:ROC曲线下面积AUC值为0.991,评价结果优秀,表明预测模型可靠性高; 进一步以全部分布数据在MaxEnt中制作区划图,将引种栽培数据的分布位置与区划预测图进行比对,划分适生性等级; 在适生性被划分为0～1的11级时,区划图中大于0.01的区域内红姜花即可成功引种。结果证明对于缺少引种栽培实践、拟采用近自然林模式栽培的野生植物,可采用MaxEnt生态学模型制作引种区划图。

Abstract:

The near-nature forestry construction is an important direction for landscape development in future. As species with potential landscape value in a large number but short of introduction practice, it has important significance to carry out introduction regionalization in only wild distribution data condition. The principle of MaxEnt ecological model is predicting potential distribution rang of species based on the wild species distribution data together with wild species distribution and climate adaptation on the premise, which is consistent with the need for identifying the introduction regionalization of potential landscape plants in near-nature forestry. In the study, Hedychium coccineum was used to be an example to evaluate the valid of introduction regionalization conducted by MaxEnt modeling framework according to the theoretical method together with introduction practice. Temperature,rainfall and altitude environmental factors were selected to be analyzed on the basis of collection for geography distribution data. And 75% data were used to construct model while the remaining were used to evaluate the valid of model by drawing receiver operating characteristic(ROC)plot. At last, suitable categories were determined by comparing the cultivated data with prediction map. The results showed that the area under the curve(AUC)was 0.991 diagnosed as excellent, which indicated that the model was highly reliable. Meanwhile, all the data were used to make a predicition map in MaxEnt. Further, categories were divided by comparing the location of introduction data with the map. The results indicated that H. coccineum could be cultivated successfully in the area where the above 0.01 when the suitability degrees between 0 and 1 were divided into eleven categories. In all, if the wild plant to be introduced in near-nature forestry, the highly reliable introduction regionization map could be made in MaxEnt.