Abstract
The genus Ardisia (Myrsinoideae, Primulaceae) has 16 subgenera and over 700 accepted names, mainly distributed in tropical Asia and America. The circumscription of Ardisia is not well-defined and sometimes confounded with the separation of some small genera. A taxonomic revision focusing on Ardisia and allies is necessary. In the Ardisia subgenus Crispardisia, symbiotic association with leaf-nodule bacteria is a unique character within the genus. The endosymbionts are vertically transmitted, highly specific and highly dependent on the hosts, suggesting strict cospeciation may have occurred in the evolutionary history. In the present study, we aimed to establish a phylogenetic framework for further taxonomic revision. We also aimed to test the cospeciation hypothesis of the leaf-nodulate Ardisia and their endosymbiotic bacteria. Nuclear ITS and two chloroplast intergenic spaces were used to reconstruct the phylogeny of Asian Ardisia and relatives in Myrsinoideae, Primulaceae. The 16S-23S rRNA were used to reconstruct the bacterial symbionts' phylogeny. To understand the evolutionary association of the Ardisia and symbionts, topology tests and cophylogenetic analyses were conducted. The molecular phylogeny suggested Ardisia is not monophyletic, unless Sardiria, Hymenandra, Badula and Oncostemum are included. The results suggest the generic limit within Myrsinoideae (Primulaceae) needs to be further revised. The subgenera Crispardisia, Pimelandra, and Stylardisia were supported as monophyly, while the subgenus Bladhia was separated into two distant clades. We proposed to divide the subgenus Bladhia into subgenus Bladhia s.str. and subgenus Odontophylla. Both of the cophylogenetic analyses and topology tests rejected strict cospeciation hypothesis between Ardisia hosts and symbiotic Burkholderia. Cophylogenetic analyses showed general phylogenetic concordance of Ardisia and Burkholderia, and cospeciation events, host-switching events and loss events were all inferred.