Message - 7th IAL Symposium

The 7 th International Association for Lichenology Symposium 2012
(3A-O7) Submission ID: IAL0103-00004
NEW APPROACHES TO INCORPORATE AMBIGUOUSLY ALIGNED SEQUENCE PORTIONS
AND MORPHOLOGICAL DATA INTO PHYLOGENETIC ANALYSIS
Lücking R. 1
1 Botany, The Field Museum, Chicago, Illinois, United States
Two novel methods are presented and briefly discussed to incorporate non-DNA data into phylogenetic
analysis: PICS-Ord ambiguous region coding and morphology-based phylogenetic binning. PICS-Ord is
a new approach to recode ambiguously aligned sequence portions in multiple sequence alignments that have
low alignment confidence, and incorporate the codes into phylogenetic analysis using a maximum likelihood
approach. It works by computing pairwise distance between the ambiguously aligned sequence portions, using
the distance options provided by the software NGILA, then ordinating the distance matrix by means of principal
coordinates analysis (using the R cmdscale function), and then transforming the axis scores into integer codes.
The method can handle an unlimited number of OTUs and is comparatively fast: computing an ML tree in RAxML
including 100 bootstrap replicates with 700 OTUs of ITS sequences and three coded portions takes about 36
hours on a single-core PC computer. Morphology-based phylogenetic binning is a novel method to incorporate
OTUs known by their morphological data only into phylogenies based on molecular data. Instead of just combining
the molecular and morphological data into a supermatrix with gaps for the molecular data, the method
first calculates a reference tree for all taxa for which both molecular and morphological data are known. It then
computes weights for each morphological character based on its distribution in the molecular tree and the level
of homoplasy displayed. These weights can be calculated using either maximum likelihood or maximum parsimony.
In a third step, each OTU known by morphological data only is added to the dataset individually and its
topological position in the tree computed by invoking the morphological character weights. Bootstrapping is performed
to estimate the level of confidence for the topology. This is repeated for each OTU separately and gives
an objective prediction for taxonomic placement of taxa even if no molecular data are available. The method is
implemented in RAxML, available at http://www.exelixis-lab.org/software.html and https://github.com/stamatak/.
A simple method to assess the level of homoplasy in molecular and morphological data prior to phylogenetic
analysis is also discussed.
53
3A-O