Predicting naturalization of southern African Iridaceae in other regions
Mark van Kleunen, Centre for Invasion Biology, School of Biological and Conservation Sciences, University of KwaZulu-Natal, P Bag X01, Scottsville, Pietermaritzburg 3209, South Africa (fax +27 33 2605105; e-mail email@example.com).
- 1 One of the major challenges in invasion biology is to predict the likelihood of naturalization, and ultimately invasiveness, of species from properties that can be assessed in the native range prior to a species’ introduction elsewhere. This is particularly relevant as intentional introduction for horticultural usage has been predicted to be the over-riding factor associated with naturalization.
- 2 We compiled a data set on 1036 species of Iridaceae native to southern Africa to test whether traits differ between horticulturally used and unused species, whether the likelihood of naturalization elsewhere is higher for horticulturally used species, and whether it differs according to species’ taxonomic affinities, geographical range size, altitudinal range, number of subtaxa and plant size.
- 3 Our results show that at least 306 southern African Iridaceae species are used in horticulture elsewhere. Of the 67 that have become naturalized elsewhere, no less than 62 are in horticulture, indicating horticultural trade as the main source of naturalized Iridaceae.
- 4 Global horticultural usage differs among the three subfamilies and among the genera. In addition, horticultural usage is more likely for species with a larger distributional range, a lower maximum altitude, more subtaxa and a taller height. This indicates that species introduced elsewhere for horticultural usage have a biased set of biogeographical and biological characteristics that should be corrected for in analyses of naturalization.
- 5 After correction for horticultural usage, naturalization differs between genera, and is more likely for species with lower maximum altitude, species with higher numbers of subtaxa and taller species.
- 6 Cross-validation of our predictive logistic regression model revealed a low kappa (0·279 ± 0·069, mean ± SE) that was significantly different from zero. This indicates that the estimates from our logistic regression model can be used to predict naturalization of Iridaceae but that the accuracy is relatively low.
- 7Synthesis and applications. Our study shows that screening protocols for potential invasiveness of species of Iridaceae should include international horticultural usage, and taxonomic, biogeographical and biological characteristics in the native range, as predictors. Moreover, for the development of more accurate predictive models, experimental assessment of other plant characteristics associated with naturalization is still required.