Phylogeny of the gall midges (Diptera, Cecidomyiidae, Cecidomyiinae): Systematics, evolution of feeding modes and diversification rates

Abstract

Gall midges (Cecidomyiidae) constitute one of the largest and most diverse families of Diptera, with close to 6600 described species and thousands of undescribed species worldwide. The family is divided into six subfamilies, the five basal ones comprising only fungivorous taxa, whereas the largest, youngest and most diverse subfamily Cecidomyiinae includes fungivorous as well as herbivorous and predatory species. The currently accepted classification of the Cecidomyiinae is morphology-based, and the few phylogenetic inferences that have previously been suggested for it were based on fragmentary or limited datasets. In a first comprehensive phylogenetic analysis of the Cecidomyiinae we sampled 142 species representing 88 genera of 13 tribes from all feeding guilds and zoogeographic regions in order to test the validity of the systematic division of the subfamily and gain insight into patterns of diversification and the evolution of feeding modes. We used sequences from five mitochondrial and nuclear genes to reconstruct maximum likelihood and Bayesian, time-calibrated phylogenies and conducted ancestral state reconstruction of feeding modes. Our results corroborate to a great extent the morphology-based classification of the Cecidomyiinae, with strong support for all supertribes and tribes, all were apparently established in the Upper Cretaceous concordant with the major radiation of angiosperms. We infer that transitions from fungus-feeding to plant-feeding occurred only once or twice in the evolution of the subfamily and that predation evolved only once, contrary to previous hypotheses. All herbivorous clades in the subfamily are very species rich and have diversified at a significantly greater rate than expected, but we found no support for the assertion that herbivorous clades associated with symbiotic fungi in their galls diversify faster than clades that do not have such associations. Currently available data also do not support the hypothesis that symbiotic clades have broader host ranges than non-symbiotic clades.