Ancient trees are ecologically important keystone organisms and have tangible connections to folklore, history and sociocultural practices. As a result of a successful 15 year citizen science project, the UK has assembled the most extensive database of ancient and other noteworthy trees to date: the Ancient Tree Inventory (ATI). Despite its size, the ATI is far from complete, and suffers from common problems associated with many large citizen science species databases, including sampling bias. Our research focuses on the use of statistical and ecological modelling techniques to produce unbiased ancient tree distribution maps across the UK using the ATI. As well as being of high ecological and conservation value, the ATI is an excellent case study for this research as the sources of bias are well-known and can be individually investigated. We modelled known ancient tree locations in relation to 21 environmental, topographical and anthropogenic characteristics at both a local habitat scale (wood-pastures) and a national scale (UK) using a variety of models and bias correction methods. Predictions of ancient tree ‘hotspots’ across the UK were generated and evaluated using historical aerial maps spanning almost 200 years, and independent field surveys.
Results/Conclusions
Our initial findings suggest that ancient tree distributions are influenced strongly by soil type, land use, distance to a city and distance to historic features such as Royal forests or Tudor deer parks. Historic aerial maps provide strong support for model accuracy, as verification estimates correlated highly with model predictions (rs = 0.701). One particular, novel distribution modelling technique we think has high potential when dealing with large, spatially-biased species data such as the ATI is zero-inflated regression models. Results from preliminary simulation studies show that when using these models, the accuracy of distribution maps produced from biased species data is comparable to that of maps produced from data collected via a random sampling protocol. Our next steps will be focused on developing and understanding this method and applying it to the ATI and other citizen science databases. We hope that our research will facilitate improved, targeted surveys of ancient trees based on unbiased distribution maps, and consequently the addition of undiscovered ancient trees to the ATI for their conservation and protection.