Low persistence of a monocarpic invasive plant in historical sites biases our perception of its actual distribution

Aim As accurate and up-to-date distribution data for plant species are rarely available, cumulative records over long periods of time are frequently used for mapping distributions, without taking into account that species do not persist in their historical localities forever. However, persistence is highly relevant in changing modern landscapes, especially for invasive species that dynamically spread in unstable human-made habitats. We studied how an invasive species, Heracleum mantegazzianum, persists at sites once colonized and how its ability to persist affects its distribution. Location The Czech Republic. Methods We visited 521 localities of H. mantegazzianum occurrence reported in the literature and herbaria to determine whether the species still occurs at these sites. By using G-tests and classification trees, we explored the roles of various factors affecting its persistence at a site. Results Of the total number of 521 historical sites at which the species has occurred since the end of the 19th century, it persists at only 124 (23.8%). The persistence rate differs with respect to habitat type and is highest in meadows and forest margins. Analysis using classification trees indicated that the factors that best explain persistence are: type of habitat (with meadow and forest margins over-represented); urbanity (with a higher persistence outside urban areas); proximity to the place of the species introduction into the country; metapopulation connectivity; and distance to the nearest neighbouring population. Main conclusions The use of cumulative historical records as a measure of species distribution, which is common in invasion literature, can seriously overestimate the actual distribution of alien plant species with low persistence. In the case of alien species such as H. mantegazzanium, which is non-clonal and reproduces only by seed, estimates of distribution and spread based on historical data are informative about potentially suitable habitat but may be unreliable as indicators of current occurrence and invasion dynamics.