UCLA

Species distribution models (SDMs) are valuable risk assessment tools in the management of invasive species, for which early detection and containment are critical. Few studies have evaluated the utility of invasive SDMs trained on temporally-biased presence data. While the abundance and range of occurrence records may increase with time after invasion, management objectives become more difficult to achieve as a destructive species nears establishment. This research assesses the relative predictive ability of models for the invasive shot hole borer (Coleoptera: Curculionidae: Scolytinae: Euwallacea sp. nr. fornicatus), which was first detected in Southern California in 2003. A series of 100-meter resolution models were developed in Maxent, selected for its ability to produce reliable models with relatively few occurrence records. Models were trained using data from five chronologically cumulative sampling periods, which simulate stages of invasion. The effects of spatial extent and spatial filtering were also examined. All models achieved high AUC (area under the receiver operating curve) values > 0.93 and correctly classified 87.7 � 18.8% of independent test records, indicating high model performance regardless of the degree of temporal bias. The leading contributing variables were minimum temperature of the coldest month (for sixteen models) or percent impervious surface (for four models). Sensitivity was consistently higher for models that used the larger spatial extent, which suggests that for an emerging species, larger backgrounds may be less restrictive on model outcomes. Spatial filtering produced more discriminating results without compromising model sensitivity. The study finds that invasive SDMs can be useful in identifying areas vulnerable to invasion, particularly if they are integrated into adaptive management strategies.