Sibling rivalry: Related plants in particular clades are much less likely to share similar pollinators or herbivores than predicted by their phylogeny
Plants’ mutualistic and antagonistic interactions (e.g., pollination and herbivory) are thought to be explained well by their phylogenetic relationships. This hypothesis relies on the phylogenetic conservation of many plant traits, including those such as flower structure that constrain plant-animal interactions. Phylogenetic conservation of interactions has been tested in plant-pollinator networks, but generally from the perspective of related pollinators visiting similar sets of plants. Similar tests in plant-herbivore networks or from the perspective of plants are, however, rare. Moreover, at least some traits affecting plant-animal interactions are poorly conserved. Hence, plants’ phylogenies may not accurately predict their interaction partners. Here we test whether related plants have statistically similar sets of pollinators or herbivores using a set of 59 plant-pollinator and 9 plant-herbivore networks. We first determined whether the interactions in each network showed a phylogenetic signal at the community level. We then examined the relationship between interaction-partner overlap and phylogenetic distance between pairs of plants within each network. Lastly, to account for the possibility that lack of overlap at the network level might be due to differences in interaction partners within specific clades, we tested the relationship between overlap and phylogenetic distance within well-represented plant families.
At the network level, we found that neither pollination nor herbivory interactions were conserved across the plant phylogenies. When considering pairs of plants, however, we found that interaction-partner overlap decreased significantly with increasing phylogenetic distance. This suggests that more closely-related plants do tend to share similar sets of interaction partners, but that this tendency may be restricted to species that have diverged relatively recently such that the phylogenetic signal breaks down across more distantly-related clades and hence there is little network-level phylogenetic signal. The relationship between interaction-partner overlap and phylogenetic distance was stronger in plant-herbivore networks than plant-pollinator networks, possibly due to greater generality among the herbivores in our dataset. Besides these overall trends, we observed substantial variation across networks of each type. This variation carried over to the family-level analysis, with some families showing greater similarity of interaction partners between close relatives and other families showing the reverse pattern. This suggests that while some plant families have strong conservation of interactions, driving the overall relationship between interaction-partner overlap and phylogenetic distance, other families have developed divergent interactions. As such, the usefulness of plant phylogenies in predicting plant-animal interactions depends strongly on the particular community and taxa in question.