外来物种入侵已经成为人类历史上最重大的生态事件之一,它直接威胁到了包括中国在内的世界上很多国家的经济、公众健康、农业生产力和生态的完整性。该研究利用遥感和GIS的方法进行了外来入侵物种的相对适生区分析,预测了外来物种-豚草在中国的潜在分布,并在信息理论的框架下建立了一种改进的加权平均逻辑回归模型。利用logit阈值和频率统计的方法对入侵物种的生境进行相对适应性划分,减小了生物多样性数据的不对称带来的影响。

Exotic species invasion has been one of themost dramatic ecological event in human history that threatens our economy, public health and ecological integrity. Explaining the nature of the species and species-environment relationship and predicting the spatial distribution of the invasive exotic plants is of great mi portance for invasive exotic plants prevention and early warning efforts. One approach to species-specific predictions involves the use of habitat-suitability or niche-based models whereby environmental conditions suitable formaintenance of populations of a species are identified and mapped onto geographic space. These approaches combine herbarium specmi en locations data with a suite ofGIS layers(e.g.clmi atic, topographic and land cover) to create the ecologicalmodels of the species’requirements. Coupledwith thesemodels, GIS can project the ecologymodel onto geographic space andmapping the habitat-suitabilitymaps in native ranges and exotic ranges.This paper proposes an mi proved logistic regression approach in an information theoretic framework to predict the suitability of ragweed in both native and invaded ranges. Information-theoretic approaches computed and assessed themodeling choice aswell as produced a weighed-averagemodel based on themultiple-models rather than using the solemodelwith the lowestAIC value or the highestAkaikeweight. Thismultiple-model inference is useful to reducemodel selection bias.Having obtained theweighted averagemode,l the resulting regression equationswere applied to the native samples including the presentpoints and pseudo-absence points to produce the outputof the logitvalue. Because of lack in true absence data, we didn’t transform the logit value back to regular space scaled from 0—1 representing probability of a pixel containing the speciesbut regard the logitvalue as the degree of the suitability for the species. Sowe proposed a new approach specifically to compartmentalize the habitat-suitability using logit value thresholds and frequency statistics. At last, we used this habitat-suitability model developed in native ranges to“project”onto the exotic ranges to predict the ragweed’s potential distribution in China.