Get reliable science

Abstract Invasive alien plants pose a threat to biodiversity worldwide, and the costs of control are ever-escalating. Early detection and prediction of areas potentially at risk is crucial to minimise ecological and socio-economic costs. Maxent was used to predict the area within which Ageratina adenophora can potentially naturalise and spread in South Africa. The model was set up with 1020 occurrence records (10 replicates, 70% of records for calibration:30% for validation), and four climatic predictor variables. Background data were selected using KĂśppen-Geiger (vegetation-based) climate classification zones. All model replicates performed better than random in both binomial tests of omission and ROC analysis. The model was statistically significant and its mean AUC was 94%. The modeled prevalence was 0.21 and the sensitivity was 0.99. The Eastern Cape, KwaZulu-Natal, Mpumalanga and Gauteng provinces have climatic conditions indicative of a high potential for invasion by A. adenophora, followed by parts of the Western Cape, North West and Limpopo provinces. The model predicted areas beyond the current distribution, suggesting that A. adenophora has potential for further spread, and that searches for it need to be made beyond its currently known distribution. On the other hand it appears not to have spread into some climatically suitable areas near its current occupancy sites, such as throughout the KwaZulu-Natal mist belt, suggesting that unknown biotic (including human) or abiotic factors are also limiting its naturalization and require further study to be identified.