6533b837fe1ef96bd12a3306

RESEARCH PRODUCT

Time to split Salvia s.l. (Lamiaceae) - New insights from Old World Salvia phylogeny.

subject

description

Abstract Aims Salvia L. is widely known as the largest genus in the mint family. A morphological modification of the androecium (lever-like stamens) was used to support this genus. However, molecular data revealed that Salvia is polyphyletic. Since phylogenetic studies largely underrepresented Old World Salvia species, we filled this gap and combined new data with existing sequences. The aim of our study was the identification of well-supported clades that provide the basis for evolutionary and taxonomic conclusions. Methods We included ITS data (internal transcribed spacer) from 220 Salvia species, 86 of which were sequenced for the first time. Additionally, the highly variable plastid marker rpl32-trnL was sequenced, providing new data for 100 Salvia species. These sequences were combined with the accessions available from GenBank. Old World Salvia is represented herein with 57% of its species. The two datasets were analyzed separately using BI and ML approaches. Results Our data confirm that Salvia is polyphyletic with four distinct evolutionary lineages (Clade I-IV), including five additional genera. The clades strongly reflect the geographical distribution, i.e., Clade IV (East Asia), Clade III (Southwest Asia to Northern Africa), and Clade II (America). The origin of Salvia s.s. (Clade I) is most likely Southwest Asia. A high degree of parallel character evolution was identified in most of the Old World sections. Based on our results, we reconstructed the evolution and biogeography of Salvia s.l. and propose to split this large group into six genera, each supported by geographical distribution, morphology, and karyology. Conclusion Salvia s.l. is a polyphyletic group that was originally regarded as a genus because its species share a derived stamen structure. However, phylogenetic data clearly indicate that this floral trait and other morphological characters evolved in parallel. Our study illustrates that the combination of different data sets allows a comprehensive reconstruction of taxa and characteristic evolution, both of which are a precondition for future revision.