Background/Question/Methods Although a number of studies show differences in arthropod structure between canopy and understory, the factors that contribute to these differences are much less known. 
Leaftying caterpillars, acting as ecosystem engineers on oak (Quercus) trees in Missouri USA, influence arthropod composition on those trees. In a factorial experiment we explored the impacts of leaf quality and plant architecture on the probability of leaftier attack in canopy versus understory individuals of white oak, Quercus alba. Previous results suggested that both the number of touching leaves and phenolic content influence attack by leaftying caterpillars in this system. We increased the number of touching leaves by tying twigs together and did so on plants we previously had identified as being either high in total phenolics or low in total phenolics. We replicated these treatments on understory saplings and on branches of canopy level trees. We then censused the abundance of ties, and arthropods occurring within those ties, at three different times during the season. Results/Conclusions Both manipulation of plant architecture and phenolics influenced leaftie abundance in the understory. Increasing the number of touching leaves increased the number of ties by approximately 25% while tie abundance was 25% less on high phenolic versus low phenolic saplings. In the canopy, there were only half the number of ties as in the understory and modification of architecture had no effect. In contrast, branches of high phenolic canopy trees had 41% fewer ties than branches of low phenolic trees. We suggest that the factor influencing differences in treatment effects between understory and canopy is the much higher level of phenolics in canopy leaves in general, obviating any potential influence of plant architecture. Because the presence of leafties influenced the abundance and composition of arthropods in both the canopy and understory, our results demonstrate that the ecosystem engineering effect of leaftying caterpillars on arthropod communities of white oak is driven by both variation in plant architecture and phenolic content in understory saplings, but only by variation in phenolics in the canopy. These results contribute to our understanding of the factors that structure arthropod communities in tree canopies and to our understanding of the contingencies of the impacts of ecosystem engineers.