基于投影寻踪模型的草原蝗虫栖境评价及风险评估

草原蝗虫发生与栖境存在紧密而复杂的关系,二者关系的研究是评估蝗灾发生风险的基础。本文分析了不同栖境内蝗虫种群密度与21个植被特征参数的相关关系,利用投影寻踪模型进行了栖境评价及风险评估,并进行模型验证。结果表明,低优参数植物生物量多样性对蝗虫种群密度影响最大,最佳投影向量a为0.6725;高优参数禾本科生态优势度对亚洲小车蝗密度影响最大,最佳投影向量a为0.6547;样点植被投影特征值Z_i与蝗虫种群密度线性相关关系极显著(y=48.861x-18.937,R=0.9509^**),Z_i越大,栖境内植被越适合蝗虫的发生,蝗灾发生的风险越高,根据Z_i值可预测不同栖境草原蝗虫的发生。投影寻踪模型评价不同植被条件下蝗虫的发生风险,可以排除与数据结构和特征无关或关系很小变量的干扰,是一种更稳健实用的方法,对于蝗虫的监测预警具有重要意义。

Comprehensive evaluation and risk assessment of grasshoppers’habitat based on a projection pursuit model

There are close and complex relationships between grasshopper occurrence and habitat vegetation. A comprehensive analysis of these relationships will provide a stable foundation for risk assessments of grasshopper infection. Grasshopper population density and 21 vegetation parameters were analyzed in the survey on which this paper is based. A projection pursuit model was developed and verified in order to evaluate the risks of grasshopper infection. Results showed that a low index for plant biomass diversity had the greatest influence on grasshopper density, with the best projection direction a at 0.6725. Poaceae dominance with a high index had the greatest influence on Oedaleus asiaticus density, with the best projection direction a at 0.6547. There was a significant linear relationship (P<0.01) between the projection eigenvalue (Z_i) and grasshopper density (y=48.861x-18.937, R=0.9509). The occurrence of grasshoppers can be predicted according to the projection eigenvalue (Z_i). The bigger the value of Z_i, the higher the risk of grasshopper occurrence. The projection pursuit model can be used to eliminate the effect of irrelevant variables. Its application will play an important role in monitoring and early warning for the ecological management of grasshoppers.