Home away from home - objective mapping of high-risk source areas for plant introductions

Abstract

Prevention is the best way to slow the escalation of problems associated with biological invasions. Screening of potential introductions is widely applied for assessing the risk of species becoming invasive. Despite advances in the understanding of the determinants of invasiveness, screening still relies heavily on assessments of the potential of species to 'fit in' to the broad environmental conditions of a target region. Most screening systems ask whether species are native to, or are known to be naturalized or invasive in, regions with 'similar' climatic/environmental conditions to the target region. The level of similarity required to make the species a high-risk introduction is generally not specified. This paper describes a protocol for making such assessments more objective, using South Africa as a test case. Using nonparametric niche-based modelling (generalized additive model; GAM) calibrated on the current distribution of each South African biome, we mapped regions of the world that are climatically similar to South African biomes. Lists were produced of countries with the largest areas climatically similar to South Africa overall, and to each biome separately. Validation of the usefulness of the approach was sought by evaluating whether the main invasive plant species in South African biomes occur naturally, or have adventive ranges, in regions mapped as analogous to South African biomes. A very large part of the world is climatically similar to South Africa, with eight countries having larger areas of land classified as climatically similar to South African biomes than the total area of South Africa. Almost all the most prominent invasive species in South African biomes occur naturally or are invasive outside their natural range in areas with similar climates to those that occur in parts of South Africa. This confirms the value of objective climate matching in screening protocols. We examined climatic conditions for a representative sample