Phylotranscriptomics resolved phylogenetic relationships and divergence time between 20 golden camellia species

Golden camellia species are endangered species with great ecological significance and economic value in the section Chrysantha of the genus Camellia of the family Theaceae. Literature shows that more than 50 species of golden camellia have been found all over the world, but the exact number remains undetermined due to the complex phylogenetic background, the non-uniform classification criteria, and the presence of various synonyms and homonyms; and phylogenetic relationships among golden camellia species at the gene level are yet to be disclosed. Therefore, it is necessary to investigate the divergence time and phylogenetic relationships between all golden camellia species at the gene level to improve their classification system and achieve accurate identification of them. Phenotypic data and transcriptomic sequences of 20 golden camellia species commonly found in Guangxi, China were obtained. PCA and OPLS-DA analyses were conducted based on phenotypic data, and agglomerative clustering was performed to generate the clustering tree of the 20 golden camellia species. Single-copy homologous genes were used to generate phylogenetic trees using Neighbor-Joining, Maximum Likelihood, and Bayesian Inference methods, and the results obtained with these three methods were compared. Then the molecular dating analysis was performed to reveal the divergence time and evolutionary relationships. Rhododendron griersonianum, Diospyros lotus, and Impatiens glandulifera were used as outgroups. The phylogenetic tree based on single-copy homologous genes showed that golden camellia species with shorter geographical distances were closer phylogenetically. Phylogenetic relationships based on phenotypic traits and those based on single-copy homologous genes were inconsistent, suggesting that species with a close genetic evolutionary relationship may show high variation in phenotypic traits and thus the analysis of evolutionary relationships based on phenotypic traits may result in inaccurate outcomes. Among three phylogenetic trees constructed by the three methods, the evolutionary sequences were different, but evolutionary relationships between most species were consistent. For 6 species, the divergence time estimated by Maximum Likelihood and Bayesian Inference varied much, that estimated by Bayesian Inference later than that estimated by Maximum Likelihood. Using these two methods, the resulting divergence time of 14 species was 3.452 Mya. The divergence time predicted in our study is later than that in the literature. In the present study phylogenetic relationships among 20 golden camellia species were analyzed at the transcriptome level to provide a supplement to the phylogenetic classification and evolutionary relationships explored using morphological traits and some molecular markers. Our findings show that the 20 golden camellia species diverged at a later time than other known species in the genus Camellia. Since our analyses were based on the failed molecular clock hypothesis, our conclusions are tentative. Further research using more systematic analyses and more methods should be conducted to confirm the phylogenetic relationships among golden camellia species.