Morphological Comparison of the Epimedium franchetii Stearn Species Complex Based on Population Observation and Implications for Taxonomy

Many doubts and difficulties still exist in regards to the taxonomic treatment of the species complex in the genus Epimedium. For example, the E. franchetii species complex consists of four taxa difficult to distinguish, namely, E. franchetii Stearn, E. baojingense Q.L. Chen et B.M. Yang, E. zhushanense K.F. Wu et S.X. Qian, and E. lishihchenii Stearn. To understand the morphological characters and variation patterns of the E. franchetii species complex, a comprehensive herbarium specimen inspection and field investigation (during flowering periods) were carried out based on the "population" concept. Results showed that E. zhushanense was the most unique species due to its purple spur, making it the easiest to distinguish. However, the remaining taxa were very similar in flower and inflorescence characters, and key features among taxa were complex and unclear. For example, the contrast between the compact rhizome of E. franchetii and the long-creeping rhizome of E. lishihchenii appeared to be neither stable nor reliable. In the present study, 20% of individuals from the population collected from Lushan (type locality) had compact rhizomes. Although closely resembling E. franchetii in flower, inflorescence and rhizome, E. lishihchenii diverged by having different hairs on the underside of the leaflets. Specifically, E. franchetii had leaflets with short bristle-like hairs, while E. lishihchenii had multicellular long hairs. An interesting population (JXJA) distributed near Lushan had the hair of E. lishihchenii, but rhizome of E. franchetii. This population was defined as a transition type. In addition, E. baojingense was closely allied to E. franchetii, but differed in the petiolule, petiole and flowering stem, with pubescent hairs densest at the nodes and with long multicellular hairs on the leaflets. According to our field investigations, however, the density of hairs depended on the individuals and/or populations, and was often densest when young and gradually thinned in old age. The color of the pubescent hairs was not confined to dark yellow, but ranged from white on young twigs, to yellow when mature, and dark yellow when old. In addition, two populations (HBMP and HBFX), which were transition types between E. baojingense and E. franchetii, were observed with sparse hairs on the petioles, petiolules and flowering stems. A high-morphology-population structure, estimated by the unweighted pair group method with averaging (UPGMA) cluster analysis, was observed. This method allowed for the separation of the populations into three groups. The results suggested that E. zhushanense was a good and independent species, but the rest were complicated and close. Therefore, E. zhushanense should be removed from the E. franchetii species complex. Both E. franchetii and E. lishihchenii were originally described from cultivated plants by Stearn, which appeared to have fairly narrow species concepts. As far as E. franchetii and E. baojingense are concerned, they are geographically sympatrically distributed, with the fundamental difference being their leaf hairs. Therefore, E. baojingense should be treated as E. franchetii var. baojingense (ecological race). Both E. lishihchenii and E. franchetii have isolated distribution, with definite differences in leaf hair and rhizome styles. Thus, E. lishihchenii should be treated as E. franchetii ssp. lishihchenii (geographical race).