Modelling the introduction and spread of non-native species: international trade and climate change drive ragweed invasion

Abstract

Biological invasions are a major driver of global change, for which models can attribute causes, assess impacts and guide management. However, invasion models typically focus on spread from known introduction points or non-native distributions and ignore the transport processes by which species arrive. Here, we developed a simulation model to understand and describe plant invasion at a continental scale, integrating repeated transport through trade pathways, unintentional release events and the population dynamics and local anthropogenic dispersal that drive subsequent spread. We used the model to simulate the invasion of Europe by common ragweed (Ambrosia artemisiifolia), a globally invasive plant that causes serious harm as an aeroallergen and crop weed. Simulations starting in 1950 accurately reproduced ragweed's current distribution, including the presence of records in climatically unsuitable areas as a result of repeated introduction. Furthermore, the model outputs were strongly correlated with spatial and temporal patterns of ragweed pollen concentrations, which are fully independent of the calibration data. The model suggests that recent trends for warmer summers and increased volumes of international trade have accelerated the ragweed invasion. For the latter, long distance dispersal because of trade within the invaded continent is highlighted as a key invasion process, in addition to import from the native range. Biosecurity simulations, whereby transport through trade pathways is halted, showed that effective control is only achieved by early action targeting all relevant pathways. We conclude that invasion models would benefit from integrating introduction processes (transport and release) with spread dynamics, to better represent propagule pressure from native sources as well as mechanisms for long-distance dispersal within invaded continents. Ultimately, such integration may facilitate better prediction of spatial and temporal variation in invasion risk and provide useful guidance for management strategies to reduce the impacts of invasion.