Phylogeny of carabid beetles as inferred from 18S ribosomal DNA (Coleoptera : Carabidae)

The phylogeny of carabid tribes is examined with sequences of 18S ribosomal DNA from eighty-four carabids representing forty-seven tribes, and fifteen outgroup taxa. Parsimony, distance and maximum likelihood methods are used to infer the phylogeny. Although many clades established with morphologica l evidence are present in all analyses, many of the basal relationships in carabids vary from analysis to analysis. These deeper relationships are als o sensitive to variation in the sequence alignment under different alignmen t conditions. There is moderate evidence against the monophyly of Migadopin i + Amarotypini, Scaritini + Clivinini, Bembidiini and Brachinini. Psydrini are not monophyletic, and consist of three distinct lineages (Psydrus, Lac cocenus and a group of austral psydrines, from the Southern Hemisphere cons isting of all the subtribes excluding Psydrina). The austral psydrines are related to Harpalinae plus Brachinini. The placements of many lineages, inc luding Gehringia, Apotomus, Omophron, Psydrus and Cymbionotum, are unclear from these data. One unexpected placement, suggested with moderate support, is Loricera as the sister group to Amarotypus. Trechitae plus Patrobini fo rm a monophyletic group. Brachinini probably form the sister group to Harpa linae, with the latter containing Pseudomorpha, Morion and Cnemalobus. The most surprising, well supported result is the placement of four lineages (C icindelinae, Rhysodinae, Paussinae and Scaritini) as near relatives of Harp alinae + Brachinini. Because these four lineages all have divergent 18S rDN A, and thus have long basal branches, parametric bootstrapping was conducte d to determine if their association and placement could be the result of lo ng branch attraction. Simulations on model trees indicate that, although th eir observed association might be due to long branch attraction, there was no evidence that their placement near Harpalinae could be so explained, The se simulations also suggest that 18S rDNA might not be sufficient to infer basal carabid relationships.