Assessing relative variable importance across different spatial scales: a two-dimensional wavelet analysis

AimAssessing the relationship between a spatial process and environmental variables as a function of spatial scale is a challenging problem. Therefore, there is a need for a valid and reliable tool to examine and evaluate scale dependencies in biogeography, macroecology and other earth sciences.LocationCentral Europe (latitude 43.99°–54.22° N, longitude 4.79°–15.02° E).MethodsWe present a method for applying two-dimensional wavelet analysis to a generalized linear model. This scale-specific regression is combined with a multimodel inference approach evaluating the relative importance of several environmental variables across different spatial scales. We apply this method to data of climate, topographic and land cover variables to explain variation in annual greening of vegetation (i.e. phenology) in Central Europe.ResultsLand use is more important to explain the variation in greening than climate at smaller resolution while climate is more important at larger resolution with a shift at c. 1000 km2.Main conclusionsTo the best of our knowledge, this is the first study analysing the scale dependency of an ecosystem process, clearly distinguishing between the different components of scale, namely grain, focus and extent. The obtained results demonstrate that our newly proposed method is particularly suitable for studying scale dependencies of various spatial processes on environmental drivers keeping grain and extent constant and changing focus (i.e. resolution).