Climate warming drives invasion history of Ambrosia artemisiifolia in central Europe

Abstract

It is thought that future climate change is likely to foster biological invasions, but effects of climate warming on invasions in recent decades are little explored. In this paper, we analyse the history of the spread of Ambrosia artemisiifolia (common ragweed) in central Europe in order to determine the effects of climate warming. In addition, we infer the likely history of this species' spread and current distribution from incomplete documentation. The area studied in central Europe includes Austria, the Czech Republic, Germany, Hungary, Liechtenstein, Slovakia, Slovenia and Switzerland. These countries were represented by a lattice of grid cells of size 5' × 3' (∼ 6 × 6 km2), and we modelled this species' increase in range in this grid from 1900 to 2010 at an annual resolution. The spread was modelled to be driven by the spatio-temporal variation in environmental suitability of grid cells as determined by climatic conditions and land use, propagule production and dispersal from invaded cells, and 'background' introductions from unknown sources. A hierarchical Bayesian modelling approach accounted for lagged and incomplete records of occurrence and spatio-temporal variation in sampling intensity.We fitted models with different representations of climate variation over time, and further contrasted the hierarchical model to a simplified model, which assumed that records accurately reflected the species' actual spread. Climatic conditions were the most important determinants of environmental suitability for invasion, and suitability also increased with increase in the proportion of urban area and length of railways in grid cells. Temperature was, on average, 2.7°C higher in the environmentally more suitable cells than in the less suitable cells. The pattern in the spread was determined by local range expansion frommultiple, spatially scattered points of introduction. Assuming a linear trend in climate warming over the modelling period provided a better model fit than using annual weather conditions or the long-term average of the climate. The model estimated that by 2010 only about 60% of the actually invaded grid cells were recorded, and records lagged behind actual colonizations by years up to decades. Inferences of the hierarchical and simplified models differed quantitatively.We conclude that by using our modelling framework it is possible to separate spatial effects of climate on the spread of non-native species from temporal effects, and that climate warming has already promoted the spread of Ambrosia artemisiifolia in central Europe. The recorded distribution reflects only a part of this species' actual distribution. Properly accounting for incomplete records hence improves inferences about the dynamics of spread.