Abstract :
Global data sets of serious agricultural weeds (1348 species), widespread agricultural weeds (1041 species), and threatening natural area invaders (381 species) were assembled, and taxonomic patterns among these data sets were compared to gain insights into how these groups differ ecologically. Angiosperm taxonomic groups (families, orders and subclasses) were tested for over- and under-representation using resampling tests, and ecological characteristics of plant families were correlated with the prominence of each family in the data sets. As predicted by Bakerʹs previously proposed attributes of ‘ideal’ weeds, many over-represented families among agricultural weeds contained primarily herbaceous, rapidly reproducing, abiotically dispersed species. The natural area invaders, in contrast, were represented by a more ecologically diverse range of families and were over-represented by more largely woody families. Families with at least some abiotically pollinated species averaged significantly higher proportions of natural area invaders (p = 0·001) and agricultural weeds (p < 0·001) than biotically pollinated families. Families containing climbing species were more highly represented among natural area invaders than families without climbers (p < 0·009). While all non-native plants may carry some risk of becoming natural area invaders, this taxonomic analysis suggests that plants with amongst the highest risk of becoming natural area invaders worldwide include: species that are primarily aquatic or semi-aquatic, grasses, nitrogen-fixers, climbers, and clonal trees. Only 25% of natural area invaders were serious agricultural weeds, so quarantine legislation aimed at curbing the introduction of new agricultural weeds will be unlikely to prevent the introduction of most natural area invaders.
Keywords :
Ecological characteristics , statistical trends , plant invaders , Weeds , plant families