How do introduction characteristics influence the invasion success of Mediterranean alien plants?

Invasive plant species are becoming increasingly widespread following accelerated anthropogenic activity in the Mediterranean region. Humans have played a central role in the expansion process, and it is important to incorporate such considerations into management plans. Using generalized linear models, our first aim was to describe how the invasion success of 862 prominent alien plant species on Mediterranean islands is related to characteristics of the introduction process: introduction frequency, date and region of origin, range size and purpose of import. The importance of each was measured by the numbers of species present and their average invasiveness. The main findings were: (a) accidental imports and ornamentals accounted for a high proportion of all aliens, although neither group had particularly high average invasiveness; (b) introduction frequency had a comparatively modest influence, with the most commonly-introduced species naturalized only three times more widely than those rarely-introduced; (c) rates of species introduction appear to have increased dramatically in the last century, although aliens which have been present in the region for more than 200 years were most widespread, indicating that it may be centuries before some species fill their potential range; (d) there were small tendencies for successful invaders to originate in the Neotropics or in regions with Mediterranean climate biomes and to have large range sizes. Our second aim was to determine whether the number or average invasiveness of species introduced via a given pathway had the most influence on the overall probability of invasion on a given island. An elasticity analysis suggested that the number of species was substantially the best predictor of the two. This finding arises largely because invasion events are rare and remain unpredictable, and has significant implications for assessing invasion risk. We discuss how substantial sources of error and intrinsic variability in invasiveness within species groups limit the potential for developing accurate risk models.