Urbanization significantly changes the composition of many animal communities. One possible reason for this is the dramatic alteration of habitat structure that accompanies urban development. Frequently, urbanization leads to the replacement of forests, characterized by complex habitat structure occurring at a variety of spatial scales, with simpler and larger-scaled habitats in the form of lawns and buildings. In other words, habitat structure becomes increasingly coarsely-textured with increasing urbanization. According to the Textural-Discontinuity Hypothesis (TDH), an increase in coarsely-textured habitat structure should be associated with an increase in animal body sizes. We tested this prediction using bird and carabid beetle data and high-resolution aerial photographs collected for sites representing a gradient of increasing housing density in Ottawa and Gatineau, Canada. We identified discontinuities in the body size distributions of birds and beetles using hierarchical cluster analysis, classification and regression tree analysis, the body mass difference index, and the gap rarity index. We determined the characteristic spatial scales of structure within sites, a measure of their habitat texture, using wavelet analysis and the aerial photographs. We used redundancy analyses separately for each taxon to test for an effect of scale on the distribution of species among body size classes.
Sites with higher housing density had significantly larger characteristic scales and significantly less variability in scales. Bird and beetle body sizes grouped into five and eleven classes, respectively. For birds, scale of habitat structure had a significant effect on the distribution of bird species among body size classes. As predicted, more large-bodied species and fewer small-bodied species were found in sites with larger characteristic spatial scales. Thus, for birds, the TDH is supported, indicating that the spatial scaling of urban habitat elements influences community structure. For beetles, scale also had a significant effect on the distribution of species among body size classes but, contrary to our prediction, more small-bodied species were associated with increasing scale. To explain this, we speculate that remotely-sensed characteristic scale did not represent the scale of habitat texture for ground beetles but rather urbanization intensity. As such, our results for beetles support the hypothesis that the increasing stress and instability of urban habitats favor smaller-sized species. This hypothesis is also supported by a significant effect of ground cover composition measured in the field at our sites on the distribution of beetle body sizes, with the covers of pavement and bare ground positively associated with smaller-bodied species.