Sweet osmanthus (Osmanthus fragrans) is one of the top ten famous native horticultural plants in China. According to different flowering seasons, flower colors and inflorescence types, the cultivars are divided into four groups: O. fragrans Asiaticus Group, O. fragrans Albus Group, O. fragrans Luteus Group  and O. fragrans Aurantiacus Group. Despite long-term cultivation of Osmanthus, little information was recorded on the formation of so many cultivars. O. fragrans Asiaticus Group was considered as the most primitive cultivars, and the ones with light color flowers formed earlier, then followed by deep color  flowers. The cluster results based on various types of molecular markers were quite different from traditional classification system, indicating diversity of phenotypic traits might account for a tiny part of the whole genetic diversity of sweet osmanthus. However, at present, few studies can give evidence to  further understand the evolution process of sweet osmanthus cultivars.
In this study, to further understand the evolution theory, a rooted phylogenetic tree for the cultivars of sweet osmanthus was constructed based on population structure analysis through sequence-related amplified polymorphism (SRAP) technology. Forty-five cultivars were used as plant materials; O. heterophyllus, O. fordii, O. cooperi “Yujie” and O. cooperi “Xuegui” were used as controls, and O. matsumuranus was used as an outgroup. Ten pairs of SRAP primers with high polymorphism were applied to amplify DNA of all samples, and the fragments were examined by capillary electrophoresis. POPGENE 1.32 software was applied to analyze genetic diversity and genetic differentiation. Structure 2.34 software was used to analyze population structure and divide cultivars into subgroups. Nei’s genetic distance among subgroups was calculated by NTSYSpc, then applied to construct a rooted phylogenetic tree by MEGA 6. The rate of hermaphrodite flower cultivars on each level of the phylogenetic tree was calculated to understand the sexual system evolution. Moreover, the genetic mechanism of flower color variation for sweet osmanthus was further speculated based on the result.
Results showed that the 10 pairs of SRAP primers produced 137 polymorphic bands among all the samples with an average of 13.7 bands per primer. Polymorphism information content ranged from 0.202 8 to 0.302 7, with an average of 0.250 7. Nei’s genetic diversity index ranged from 0.220 3 to 0.350 2, with an average of 0.283 5. The Shannon’s genetic diversity index ranged from 0.348 3 to 0.519 3, with an average of 0.436 4. There was significant population structure among sweet osmanthus cultivars, and 36 cultivars could be divided into seven subgroups with simple genetic background. Nine cultivars had complicated genetic background, which were identified as a mixed group. Gene differentiation coefficient (Gst) was 51.32% among subgroups, much higher than that of four cultivar groups. Moreover, less gene flow was observed among subgroups than that of four cultivar groups. These results indicated that the cultivars in the same subgroup had much closer genetic relationship than those in the same cultivar group. Using subgroups as the unit of evolution, a rooted phylogenetic tree was constructed. The sweet osmanthus cultivation had experienced about 10 stages (A-J level): subgroup 3 composed of major cultivars in O. fragrans Asiaticus Group and O. fragrans Albus Group formed first, and subgroup 5 composed of the male cultivars in O. fragrans Aurantiacus Group formed the latest, and the cultivars in O. fragrans Luteus Group formed in each stage after D level. With the evolution process, the rate of hermaphrodite flower cultivars dramatically reduced, proving that androdioecy sexual system of sweet osmanthus originated from monoecism. Moreover, the flower color of sweet osmanthus may be controlled by multiple homologous genes in the CCDs family, and the mutations resulted in flower color changing continuously from white to orange red. Loss of new alleles due to genetic drift led to rare individuals with orange red flower in wild population.
In conclusion, a new efficient method is offered to further understand the process for formation of sweet osmanthus cultivars and provide a significant reference for genetic mechanism study on evolution of flower colors.