Debate over neutral models and other macroecological approaches has recently reinvigorated research into explaining species abundance patterns.  Focusing on a well-characterized moth caterpillar species assemblage found on a riparian tree (Acer negundo L., boxelder) in the eastern US, I investigated the spatial dynamics of an herbivore species abundance distribution by sampling the community in a nested spatial design in summer 2006.  Additive beta diversity partitioning, a technique to evaluate the contribution of different scales of analysis to the overall regional species diversity, revealed strong significance of lower than expected alpha diversity and higher than expected contributions to beta diversity at higher scales, but only when relative abundance was considered.  This result is consistent with high intraspecific clustering, for which multiple plausible explanations exist.  Assemblages might be shaped by dispersal limitation, or by environmental effects such as alternative host plant density for these mostly generalist caterpillars.  I used partial Canonical Correspondence Analysis (pCCA) to independently assess the explanatory power of spatial arrangement and tree community on the variance in the caterpillar fauna, at two different scales.  At the forest scale, both geographic distance and tree community explained significant variation in the assemblages once the effect of the other variable set had been incorporated.  No significant effects were seen at the scale of tree stands.  Although all the moths in the study were sampled from the single focal host plant species, many are generalists on broadleaf trees to varying degrees; these results show the vegetative context of a particular site can influence which members of the potential assemblage appear on the host plant, and in what abundances.