Although biological invasions are currently recognized as being especially severe on islands, little attention has been given to detecting parameters influencing this pattern. This study tested the common perceptions that uniparental reproductive modes are associated with invasive and/or island plant populations due to a lack of pollinator vectors and/or small initial population sizes, and that pollinator services and biparental seed production modes will subsequently be associated with adjacent mainland sites. Using controlled pollination and germination experiments on invasive Carpobrotus edulis and C. aff. acinaciformis populations in both island and mainland habitats in south-east France, we found no evidence to support these hypotheses. All significant mean differences found between locations for reproductive indices describing uniparental reproductive modes were significantly smaller in island populations. In contrast, seedling sizes issuing from manual outcrossing and manual hybridization experiments, regardless of taxon, as well as C. aff. acinaciformis hybridization indices, were larger in island populations. The presence of significant, though limited, pollinator service was detected for all populations, and stressed the importance of such mutualisms. We suggest that invasive plant reproductive traits could be highly contextual, and that islands may have a potential role in the acceleration of the invasion process through the production of highly variable populations. © 2005 The Linnean Society of London, Biological Journal of the Linnean Society, 2005, 85, 65–79.