Abstract
Objective methods are needed for identifying areas where control measures against invasive alien species should be focused. We investigated classification tree models of the distribution of invasive plant stands (Ambrosia trifida, Eragrostis curvula, Sicyos angulatus, and Robinia pseudoacacia) in riparian areas of the Chikuma River. For each taxon, we used Classification And Regression Tree (CART) analyses to relate the distribution of invasive stands to explanatory variables (relative height from calculated water level, vegetation type, distance to nearest agricultural fields, areas dominated by the target species around each grid cell, and areas dominated by the target species in upstream river regions of each grid cell) and to derive the potential distribution of the target species. The current distribution of invasive stands along the Chikuma River was largely explained by relative height. We compared model predictions by testing the agreement between observed and simulated distributions (using the area under the receiver operating characteristic curve [AUC], specificity, sensitivity, and correct classification rate) by assessing consistency. Our results demonstrated that this modeling approach, combined with the visual presentation of predictions as a map, provides useful information for making decisions with regard to how different conservation actions and flood protection measures should be spatially allocated.
