Spatial dependence in Northern Flicker habitat-reproduction relationships: An application of Dutilleul's modified t-test

Abstract

While there is growing awareness in ecology of spatial dependency, the application in avian ecology of spatially explicit statistical methods is rare in areas such as habitat-reproduction relationships. We compared nonspatial vs. spatially explicit tests of correlation between a measure of reproduction and a habitat attribute associated with Northern Flicker (Colaptes auratus) nests. While nonspatial tests showed a significant positive correlation between nest productivity and canopy cover of conifers, Dutilleul's modified t-test, which reduces the effective sample size by accounting for dependence among sampled nest trees, showed no significant correlation. These results may be due to spatial autocorrelation of both canopy cover of conifers and nest productivity at the scale of <1 km. Low localized nest productivity appears to have been driven by nest predation rather than by conifer cover. Our results illustrate the utility of this simple modification for examining ecological correlations in the presence of spatial structure. We encourage further use of this and other spatially explicit statistical tests in avian ecology. Failing to do so risks potentially spurious and overstated conclusions, as we demonstrate here. © The Cooper Ornithological Society 2010.