The effect of spatial resolution on projected responses to climate warming

Aim To determine how changing the resolution of modelled climate surfaces can affect estimates of the amount of thermally suitable habitat available to species under different levels of warming. Location Lake Vyrnwy RSPB Reserve, which covers around 9700hectares of a topographically diverse landscape in Wales. Methods A recently published microclimate model was used to predict maximum, minimum and mean temperatures at 5 x 5m resolution for the study site, under current and possible future conditions. These temperature surfaces were then averaged to produce coarser resolution surfaces, up to a maximum of 1X1km resolution. Ground beetles were collected using pitfall traps between May and August 2008. Generalized linear models (GLMs) were fitted to the temperature surfaces to predict the amount of landscape suitable for a northerly-distributed ground beetle, Carabus glabratus, and the most southerly-distributed ground beetle found at the site, Poecilus versicolor, under current and possible future conditions. Results A wider range of temperatures are expected within our site when temperature is modelled at finer resolutions. Fitting GLMs at different resolutions resulted in the inclusion of different temperature variables in the best models. Coarser resolution models tended to have higher prediction error, and different resolution models predicted that different amounts of the landscape would remain or become suitable in future. There was less agreement between models for C.similar to glabratus than for P.similar to versicolor. Main conclusions In our example system, different resolution analyses result in different predictions about the ability of populations to survive climatic warming. Higher resolution analyses are not only likely to provide more accurate estimates of expected patterns of change, but also to highlight potential microclimatic refugia for the conservation of species that otherwise might appear to be threatened with regional or global extinction.