科研之友
微信
新浪微博
Facebook
分享链接
摘要
Despite their potential shortcomings for phylogenetic inference, sometimes only morphological data are available for systematic classification. Computational methods are used to determine morphological site patterns congruent with the molecular tree in order to improve the accuracy of the classification of taxa for which only morphological data exist. Here, we used molecular site weight calibration as implemented in RAxML to weight morphological characters based on their distribution on a maximum likelihood tree inferred from molecular data. We subsequently conducted morphology-based 'phylogenetic binning' (= assignment of morphologically defined taxa that lack molecular data to branches of the molecular reference tree) based on the morphology of the taxa included in the tree. We applied this methodology to the lichen genus Graphis s.l. (Ascomycota: Ostropales: Graphidaceae), which was recently shown to represent two separate, distantly related lineages, Graphis s.str. and Allographa. Of 313 species of Graphis and Allographa included in the study, 16 were represented by molecular sequence data and morphology, and 297 by morphology only. Using maximum likelihood and maximum parsimony site weight calibration and morphology-based phylogenetic binning, 281-290 of the 297 species represented by morphological data only could be assigned to either Graphis or Allographa with strong support (90%-100%). Two species, G. saxiseda and G. evirescens, were found to represent species of the unrelated genus Carbacanthographis. Our results showed that assignment of taxa to clades based on morphological data only substantially improved with molecular site weight calibration. Both molecular site weight calibration and branch assignment to the molecular reference tree are implemented in the RAxML v.7.2.6 Windows executable and the RAxML v.7.2.8 open-source code.
- 出版日期2011-10
