• Carpobrotus acinaciformis;
  • Coastal dune vegetation;
  • Exotic plants;
  • Imputation of environmental variables;
  • Natural stress and disturbance;
  • Predictive modelling;
  • Xanthium orientale;
  • Zero inflation


Questions: On sandy coastal habitats, factors related to substrate and to wind action vary along the sea–inland ecotone, forming a marked directional disturbance and stress gradient. Further, input of propagules of alien plant species associated to touristic exploitation and development is intense. This has contributed to establishment and spread of aliens in coastal systems. Records of alien species in databases of such heterogeneous landscapes remain scarce, posing a challenge for statistical modelling. We address this issue and attempt to shed light on the role of environmental stress/disturbance gradients and propagule pressure on invasibility of plant communities in these typical model systems.

Location: Sandy coasts of Lazio (Central Italy).

Methods: We proposed an innovative methodology to deal with low prevalence of alien occurrence in a data set and high cost of field-based sampling by taking advantage, through predictive modelling, of the strong interrelation between vegetation and abiotic features in coastal dunes. We fitted generalized additive models to analyse (1) overall patterns of alien occurrence and spread and (2) specific patterns of the most common alien species recorded.

Conclusion: Even in the presence of strong propagule pressure, variation in local abiotic conditions can explain differences in invasibility within a local environment, and intermediate levels of natural disturbance and stress offer the best conditions for spread of alien species. However, in our model system, propagule pressure is actually the main determinant of alien species occurrence and spread. We demonstrated that extending the information of environmental features measured in a subsample of vegetation plots through predictive modelling allows complex questions in invasion biology to be addressed without requiring disproportionate funding and sampling effort.