Can the invaded range of a species be predicted sufficiently using only native-range data?: Lehmann lovegrass (Eragrostis lehmanniana) in the southwestern United States

Predictions of species invasions are often made using information from their native ranges. Acquisition of native-range information can be very costly and time-consuming and in some cases may not reflect conditions in the invaded range. Using information from the invaded range can enable much faster modeling at finer geographic resolutions than using information from a species' native range. We used confirmed presence points from the native range, southern Africa, and the invaded range, the southwestern United States, to predict the potential distribution of the perennial bunchgrass Eragrostis lehmanniana Nees, (Lehmann lovegrass), in its invaded range in the United States. The two models showed strong agreement for the area encompassed by the presence points in the invaded range, and offered insight into the overlapping but slightly different ecological niche occupied by the introduced grass in the invaded range. Regions outside of the scope of inference showed less agreement between the two models. E. lehmanniana was selected via seeding trials before being planted in the United States and therefore represents an isolated genotype from the native-range population. Models built using confirmed presence points from the invaded range can provide insight into how the selected genotype is expressed on the landscape and considers influences not present in the native range. Models created from locations in both the invaded and native ranges can lead to a more complete understanding of an introduced species' potential for spread, especially in the case of anthropogenic selection. (C) 2005 Elsevier B.V. All rights reserved.