Gall-induction in insects: evolutionary dead-end or speciation driver?

BACKGROUND: The tree of life is significantly asymmetrical - a result of differential speciation and extinction - but general causes of such asymmetry are unclear. Differences in niche partitioning are thought to be one possible general explanation. Ecological specialization might lead to increases in diversification rate or, alternatively, specialization might limit the evolutionary potential of specialist lineages and increase their extinction risk. Here we compare the diversification rates of gall-inducing and non-galling insect lineages. Compared with other insect herbivores feeding on the same host plant, gall-inducing insects feed on plant tissue that is more nutritious and less defended, and they do so in a favorable microhabitat that may also provide some protection from natural enemies. We use sister-taxon comparisons to test whether gall-inducing lineages are more host-specific than non-galling lineages, and more or less diverse than non-gallers. We evaluate the significance of diversity bipartitions under Equal Rates Markov models, and use maximum likelihood model-fitting to test for shifts in diversification rates. RESULTS: We find that, although gall-inducing insect groups are more host-specific than their non-galling relatives, there is no general significant increase in diversification rate in gallers. However, gallers are found at both extremes - two gall-inducing lineages are exceptionally diverse (Euurina sawflies on Salicaceae and Apiomorpha scale insects on Eucalytpus), and one gall-inducing lineage is exceptionally species-poor (Maskellia armored scales on Eucalyptus). CONCLUSIONS: The effect of ecological specialization on diversification rates is complex in the case of gall-inducing insects, but host range may be an important factor. When a gall-inducing lineage has a host range approximate to that of its non-galling sister, the gallers are more diverse. When the non-galler clade has a much wider host range than the galler, the non-galler is also much more diverse. There are also lineage-specific effects, with gallers on the same host group exhibiting very different diversities. No single general model explains the observed pattern.