Abstract: The Huperzia serrata (Thunb.) Trevis. complex in China includes taxonomically confusing taxa with unclear boundaries. The number of recognized taxa in this group has long been debated and with this, the group has undergone several taxonomic changes. The species circumscription of H. serrata complex in China is inadequate and unresolved and is in need of comprehensive systematic studies. To address these issues, we have attempted to infer diagnostic differences between the putative taxa and circumscribe species boundary of H. serrata complex in China using an integrative multivariate approach. Morphological characters were subjected to multivariate tests (principal component analysis, discriminant analysis, χ²-test, cluster analysis) followed by univariate analysis of variance and Tukey's post-hoc test. MaxEnt modeling was used to create distribution models of the species followed by niche equivalency and similarity tests. The morphometric analysis clearly separated the specimens into four distinct clusters representing H. crispata (Ching) Ching, H. nanlingensis Y. H. Yan & N. Shrestha, and two clusters of H. serrata comprising specimens from geographically isolated regions. The recognition of these four distinct taxa was supported by climatic data as well. Based on our analysis of morphological with climatic data and ecological niche modeling we recognized H. crispata, H. nanlingensis, the northern Chinese H. serrata, and the southern Chinese H. serrata as four different species. We identified the northern Chinese H. serrata as true H. serrata and the southern Chinese taxa as H. javanica. Our results, however, did not support the recognition of other infraspecific taxa previously recognized under H. serrata.

Key words: Huperzia crispata, H. javanica, Lycopodiaceae, MaxEnt, multivariate analysis, species distribution modeling, UPGMA