Although the subfamily Nepetoideae (Lamiaceae) is considered to be monophyletic, relationships between tribes, subtribes and genera within the subfamily are poorly understood as complex and possi- bly homoplasious morphological characters make taxa difficult to delimit. DNA sequence data from three regions (chloroplast: trnK intron; trnL-F; nuclear: ITS) in total including 278 accessions, representing 38 out of 40 genera of subtribe Menthinae and 11 outgroup genera, were used to reconstruct the evolution- ary history, test previous hypotheses of classification, explain biogeographic patterns and elucidate char- acter evolution. Using maximum parsimony (MP) and Bayesian analysis phylogenetic reconstructions based on nuclear and chloroplast sequence data were incongruent, consequently the data were analyzed separately.
Both nuclear and chloroplast datasets provide strong support for three major lineages: the ``Satureja'', ``Micromeria'' and ``Clinopodium'' group. The first contains members of Satureja and Gontscharovia. In the second lineage Micromeria s.str. and Origanum were resolved as monophyletic, Pentapleura and Zataria indicated as sister groups. Thymbra includes two species of Satureja turning the latter genus polyphyletic. Thymus is revealed as paraphyletic with respect to Argantoniella and Saccocalyx in both and Origanum in the plastid dataset only. In the third lineage, the Clinopodium-group, branching pattern is highly incon- gruent among datasets and possibly influenced by recent and ancient hybridization, chloroplast capture and incomplete lineage sorting. However, identical terminal groups are inferred in both analyses. A Madagascan lineage of ``Micromeria'', sister to the recently described South African Killickia, is suggested to represent a new genus. The Himalayan Clinopodium nepalense group and the tropical African C. abyssinicum alliance are monophyletic but indicated in different positions. Both groups appear in the ITS phylogeny in a clade with Cyclotrichium and Mentha, relationships not suggested previously. The enig- matic Micromeria cymuligera is close to Mentha and possibly is a representative of this genus. Species of Acinos, now regarded as part of Clinopodium, are mixed up with species of Ziziphora, questioning either the inclusion of Acinos in Clinopodium or generic distinctness of Ziziphora. All data sets suggest a mono- phyly of the New World taxa and argue for long distance dispersal from the Old World, rather than a vicariance explanation. Bystropogon marks the split up between the two lineages. Inclusion of 22 genera intermixed with Clinopodium spp. in the New World clade provides evidence that the current circum- scription of the genus is highly unnatural. Low sequence divergence resulting in low phylogenetic reso- lution especially at the base of the clade indicate a rapid radiation accompanied by considerable ecological diversification and speciation.