Predicting tropical insect herbivore abundance from host plant traits and phylogeny

Abstract

Phylogenetic ecology has identified patterns of diversity in communities that may find explanation in trophic interactions, and yet there have been few attempts to directly relate such patterns among trophic levels. Density‐dependent processes involving pests and pathogens, for example, have been invoked to account for plant community phylogenetic patterns, but relatively little is known about how plant relatedness might affect community structure at other trophic levels. We examined the degree to which the abundance of herbivores in a rain forest community is explained by the phylogeny and functional traits of host plants. We destructively sampled all stems ≥5 cm diameter in two 1‐ha plots of New Guinea primary and secondary lowland forest to test predicted relationships between herbivore abundance and plant resources. We analyzed per‐tree caterpillar and leaf miner abundance, total leaf biomass (kg), percentage of immature foliage, specific leaf area (cm2/g), leaf nitrogen content (percentage of dry mass), and presence of exudates in the context of a plant community phylogeny estimated from DNA barcodes.Apart from nitrogen content and exudates, neither plant resources nor herbivore abundance showed evidence of phylogenetic signal in our community sample. The plant traits we measured could account for only 30% and 16% of variation in caterpillar and leaf miner abundance, respectively, among individual trees. Leaf biomass was a stronger predictor of herbivore abundance than either resource quality (leaf nitrogen content) or palatability (percentage of immature foliage, specific leaf area). The primary importance of resource quantity was also observed at the plant species level in analyses of species means and phylogenetic generalized least‐squares regression. Plant relatedness did not account for much variation in herbivore abundance, but significant effects of exudates and leaf nitrogen content on caterpillar abundance illustrate how conserved traits at one trophic level may influence community‐wide patterns at another.