Challenges in predicting invasive reservoir hosts of emerging pathogens: mapping Rhododendron ponticum as a foliar host for Phytophthora ramorum and Phytophthora kernoviae in the UK

Invasive species can increase the susceptibility of ecosystems to disease by acting as reservoir hosts for pathogens. Invasive hosts are often sparsely recorded and not in equilibrium, so predicting their spatial distributions and overlap with other hosts is problematic. We applied newly developed methods for modelling the distribution of invasive species to the invasive shrub Rhododendron ponticum-a foliar reservoir host for the Phytophthora oomycete plant pathogens, P. ramorum and P. kernoviae, that threaten woodland and heathland habitat in Scotland. We compiled eleven datasets of biological records for R. ponticum (1,691 points, 8,455 polygons) and developed Maximum Entropy (MaxEnt) models incorporating landscape, soil and climate predictors. Our models produced accurate predictions of current suitable R. ponticum habitat (training AUC = 0.838; test AUC = 0.838) that corresponded well with population performance (areal cover). Continuous broad-leaved woodland cover, low elevation (< 400 m a.s.l.) and intermediate levels of soil moisture (or Enhanced Vegetation Index) favoured presence of R. ponticum. The high coincidence of suitable habitat with both core native woodlands (54 % of woodlands) and plantations of another sporulation host, Larix kaempferi (64 % of plantations) suggests a high potential for spread of Phytophthora infection to woodland mediated by R. ponticum. Incorporating non-equilibrium modelling methods did not improve habitat suitability predictions of this invasive host, possibly because, as a long-standing invader, R. ponticum has filled more of its available habitat at this national scale than previously suspected.