Taxonomy, phylogeny, and secondary sexual character evolution of diving beetles, focusing on the genus Acilius

Sexual conflict can lead to antagonistic coevolution between the sexes, but empirical examples are few. In this thesis secondary sexual characters in diving beetles are interpreted in the light of sexual conflict theory. Whether the male tarsal suction cups and female dorsal modifications are involved in a coevolutionary arms race is tested in two ways. First eight populations of a species with dimorphic females that varied in frequency of the morphs were investigated and male tarsal characteristics quantified. The frequency of female morphs is shown to be significantly correlated to the average number and size of male tarsal suction cups in the population, a prediction of the arms race hypothesis. Second, the hypothesis is tested in a phylogenetic perspective by optimizing the secondary sexual characters on a phylogeny. A full taxonomic revision of the genus Acilius is presented, including new synonyms, lectotype designations, geographic distributions based on more than five thousand examined museum specimens and the description of a new species from northeastern USA. Specimens of all species (except one possibly extinct that failed to be found in Yunnan, China 2000), were field collected between 2000 and 2003 in Sardinia, Sweden, Russia, Honshu and Hokkaido in Japan, New York, Maryland, California and Alberta. Three genes (CO1, H3 and Wingless) were sequenced from the fresh material as well as scoring a morphological character matrix all of which was used to derive a robust and complete hypothesis of the phylogenetic relationship in the group. The phylogeny was derived using Bayesian phylogenetics with Markov Chain Monte Carlo techniques and received a posterior probability of 0.85. Changes in male and female characters turned out to be perfectly correlated across the phylogeny, providing one of the best empirical examples to date of an antagonistic arms race between the sexes in a group of organisms. Finally, a review of a pitfall to phylogenetic analysis known under the name long-branch attraction (LBA), is provided. The problem is well known theoretically but has been questioned to occur in real data, and LBA has been in the core center of the hard debate between parsimony and likelihood advocates since different inference methods vary in sensitivity to the phenomenon. Most important conclusions from the review are; LBA is very common in real data, and is most often introduced with the inclusion of outgroups that almost always provide long branches, pulling down long terminal ingroup branches towards the root. Therefore it is recommended to always run analyses with and without outgroups. Taxon sampling is very important to avoid the pitfall as well as including different kind of data, especially morphological data, i.e. many LBA-affected conclusions have recently been reached by analyses of few taxa with complete genomes. Long-branch extraction (incl. outgroup exclusion), methodological disconcordance (parsimony vs modelbased), separate partition analyses (morphology vs molecules, codon positions, genes, etc), parametric simulation (incl. random outgroups), and split graphs are available relevant methods for the detection of LBA that should be used in combinations, because none alone is enough to stipulate LBA