Reconstructing the spread of invasive plants: taking into account biases associated with herbarium specimens

Aim To reconstruct the spread of invasive wetland species using herbarium specimens and to develop a method that accounts for the biases associated with this type of historical record. Location Southern Quebec. Methods The temporal and spatial distribution of herbarium specimens of vascular plants was examined. Six invasive species that are mainly found in wetlands were compared with five native, non-expanding hydrophytes. The cumulative number of locations was plotted against time to construct invasion curves. For native species, this 'invasion curve' indicates the spatiotemporal distribution of the sampling for herbarium specimens of wetland plants. It also represents the history of the state of knowledge of the distribution of native species. Deviation from the native species invasion curve can be used to describe the spread of invasive species. For every year on record, the cumulative number of locations of each exotic species was divided by the cumulative number of locations of all native species, and the proportions obtained were plotted against time. Periods of invasiveness were then delineated. Results During the twentieth century, two periods of intensive herbarium specimen collection in Quebec can be clearly distinguished, i.e. from 1930 to 1940 and from 1950 to 1985. Several periods of invasiveness were delineated for Butomus umbellatus L. (1922-35), Hydrocharis morsus-ranae L. (1957-96), Lycopus europaeus L. (1963-2000), Lythrum salicaria L. (1890-1905, 1923-46), Phragmites australis (Cav.) Trin. ex Steudel (1963-84) and Rorippa amphibia (L.) Bess. (1929-34, 1943-52, 1971-79). Main conclusions Accounting for sampling biases associated with herbarium specimens is important in order to objectively delineate periods of invasiveness for exotic species. The spread of wetland vascular plant species can be reconstructed using herbarium specimens, even when the phenomenon is rapid, but the reconstruction is certainly more accurate when the invasion occurred during an active sampling period. The highly variable sampling effort for herbarium specimens in Quebec and in other regions strongly cautions against using herbarium data without correction procedures.