PubMed2000
A key transition in the evolution of the insect suborder Caelifera (Orthoptera; Insecta) was from predominantly non-angiosperm-feeding basal lineages to the modern acridomorph fauna (grasshoppers and related insects). However, because of conflicts in the distribution of several complex morphological characters, the relationships of the presumed intermediates, and in particular of the superfamily Pneumoroidea, are presently unclear. We undertook a phylogenetic study of representatives of all of the transitional acridomorph families using mitochondrial and nuclear DNA sequences. No support for pneumoroid monophyly was obtained from nonparametric bootstrap analysis. Furthermore, adopting a maximum-likelihood approach, specific hypotheses of relationships within the Pneumoroidea were firmly rejected using parametric bootstrapping and Kishino-Hasegawa tests. The results indicate that the Pneumoroidea are at best a grade. This distinction implies that the evolution of the proposed pneumoroid synapomorphies, femoro-abdominal stridulation and simple male genital structure, might previously have been misinterpreted as cases of single character gains or losses within lineages. Reconstructions of character states for the femoro-abdominal stridulation indicate that, in fact, multiple losses or gains are equally likely. An important implication of our findings is that, in grasshoppers, auditory tympana may have evolved before stridulation, supporting the argument that the original function of tympana may have been related not to conspecific communication but to predator detection. Overall, the results of this study emphasize the high information content of these minor groups (in this case, the four intermediate families under consideration contain only 0.2% of extant orthopteran species diversity). Our analyses also demonstrate the advantages of model-based methods in analyzing systematic problems and, in particular, of the importance of testing specific phylogenetic hypotheses when a priori support for groupings (e.g., from nonparametric bootstrapping) is marginal.