Predicting and explaining plant invasions through analysis of source area floras: some critical considerations

Comparing species that become invasive with others from the same regional species pool that do not invade raises several issues about the accuracy of analyses attempting to define the determinants of invasiveness. The delimitation of the source area and deciding which species group(s) to include are especially relevant in analyses focusing on species originating in Europe. Historical patterns of immigration of alien species into Europe must be considered since European floras comprise a complex mix of native species, historical introductions (archaeophytes) and relative newcomers (neophytes). We make three main points: (1) Archaeophytes (species introduced to Europe before the discovery of America) differ from natives in a number of traits and in historical association with people; it is misleading to lump archaeophytes with native taxa. (2) Taxa from climatically and geographically different regions, representing distinct floristic geoelements, need to be treated separately, and not as a homogenous pool of potential invaders. Restricting the source species pool to native taxa with comparable phytogeographical characteristics reduces the variation associated with chance of dispersal by humans from the source area. (3) For prediction, a clear distinction should be made between accuracy (the proportion of those found to be alien that were also predicted to be there) and reliability (or predictive value, the proportion of those predicted to become aliens that do so). Information accumulated over centuries by botanists in Central Europe provides an excellent opportunity to deal with these issues and avoid spurious results. To illustrate these issues, we revisit a recently published study of Central‐European plant species as invaders in two Argentinean provinces ( Prinzing et al., 2002 ) to explore and demonstrate the implications of the above points. We hope that future studies will build on these points to achieve more reliable predictions.