So what do systematists do for spring break?

Thanks to a post on Darren Naish’s blog Tetrapod Zoology about plastic pelicans the taxon of interest was pelicans. For people who actually care how I got the tree hear are the methods. Otherwise just skip it and go check out the tree. I only did a parsimony analysis, but if I get bored I’ll do a Bayesian analysis also. As I’m not a pelican taxonomist I’ll leave the result interpretation to someone who actually knows something about these birds.

Methodology

Sequence data

Due to my inability to get morphological data and the fact I was on spring break so didn’t want to spend the entire time tracking down original descriptions etc to code a morph matrix my analysis is only of molecular data mined off of Genbank. I narrowed the sequence data to genes that had sequences for at least 3 taxa included in the analysis (5 pelicans and 1 outgroup a shoebill, Balaeniceps rex). Genes included in my analysis were cytochrome oxidase subunit 1, ATP synthase subunit 8 and 6, 12s, Fibrinogen beta chain gene, cytochrome b, and ornithine decarboxylase gene. A total of 3893 bases were included in the analysis of which 3409 were constant. Of the remaining characters 313 were parsimony uninformative leaving 171 characters as parsimony informative. Because sequences were not from the same specimen they were combined at random in order to have one sample of each taxa with the most possible sequences for analysis.

Taxon sampling

Data were only available for 5 of the 8 species of Pelecanus. Included taxa were the brown pelican (P. occidentalis), American white pelican (P. erythrorhnchous), great white pelican (P. onocrotalus), Dalmatian pelican (P. crispus), and the Australian pelican (P. conspicillatus). Only a single gene (which was not available for any other species) was available for the spot-billed pelican (P. phillippensis). The remaining two species, the Perunvian pelican (P. thagus) and the pink-backed pelican (P. rufescens) lacked any sequence data.

Data analysis

Data were extracted from genbank, aligned in clustal w (using bioedit) then manually checked by eye. Data were then analyzed using parsimony in PAUP*. A heuristic search was done using 1000 random adition sequences. Next duplicates of taxa were removed and an exhaustive search performed (tree shown below)

Results

A single most parsimonious tree was recovered with a length of 547. CI and RI were 0.956 and 0.870 respectively. The ingroup was monophyletic with a bootstrap value of 95. The Australian pelican was the most basal of the ingroup sampled. I haven't done bremer supports or looked at the Ti/Tv ratios etc if people are curious though it could be arranged.