These spatial layers quantify the predicted habitat suitability for Rhododendron ponticum across Scotland. These layers were developed with reference to this species role as reservoir host for Phytophthora plant pathogens, but should have value for management of Rhododendron ponticum as a problematic invasive species. The models were developed by combining biological records of R. ponticum with climate, soil, elevation and woodland cover data. The dataset contains averaged estimates for R. ponticum presence, associated standard deviation for each estimate and locations where environmental conditions in the study region strayed too far from the training set data. This research was funded by the Scottish Government under research contract CR/2008/55, 'Study of the epidemiology of Phytophthora ramorum and Phytophthora kernoviae in managed gardens and heathlands in Scotland' and involved collaborators from St Andrews University, Science and Advice for Scottish Agriculture (SASA), Scottish Natural Heritage (SNH), Forest Research, Forestry Commission and Centre for Ecology & Hydrology (CEH).
Publication date: 2016-07-08
This dataset is part of the following
Alexandra Schlenzig at SASA provided advice on premise types. Jim McCleod and Steve Albon from the Macaulary Land Use Research Institute provided the soil data. Prof. David Rogers and the Spatial Ecology and Epidemiology Group at Oxford University provided the MODIS data. Adam Butler at Biomathematics and Statistics Scotland (BioSS) provided statistical advice. Local Record Centres and their contributing biological recorders provided the R. ponticum data. Based on previous empirical studies of the habitat associations of R. ponticum, seven environmental predictors were selected apriori for modelling. R. ponticum data and environmental data were to run the Maximum Entropy (MaxEnt; version 3.3.3e) model. MaxEnt was selected to create the suitability maps from amongst Species Distribution Model techniques because it is specialised for presence-only biological records and is consistently competitive with the highest performing methods. More details on the data and model can be found in the supporting documentation.