Thus, given the taxon sample size, we think that the approximate overall concordance of our trees with the "backbone" hypothesis in Fig. 2A is unlikely to be accidental, despite the lack of bootstrap support.
This implements a coalescent-based method that uses the concordance of gene trees across multiple loci as evidence for a particular species delimitation model, but does not rely on reciprocal monophyly for each individual locus.
The concordance of the SNPs tree from raw reads was more adequate than the one from assembled genomes, which is not surprising.
The resulting 111,760 phylogenies and 22,352 alignments have been deposited in PhylomeDB [12] (http://www.phylomedb.org; phylome codes SceP60, SceP21, SceP12a and SceP12b) Some methodologies have been previously proposed to explicitly address the issue of concordance between species trees and individual gene trees [13], [14].
We obtained >90% of concordance between trees, and a Gelman-Rubin test statistic below 1.1 for all parameters, indicating convergence.
As a fine-tune evaluation of convergence, we also used the phylogenetic tree comparison tool, which assesses the degree of concordance between trees from replicate runs.
The concordance of the orthologous gene trees with the best estimate of the species tree is affected by stochastic error related to gene size, although weakly and the effect is not distributed evenly among functional categories.
We chose to use distance matrices rather than gene trees for the evaluation of gene concordance, because trees are brittle: a small change in sequence can dramatically change tree probabilities or the topology of the likeliest tree.
The evaluation of accuracy revealed lack of convergence for all parameters (values highly above the acceptable cut-off), and the PSG data set was the bottom-ranked group in analysis of the concordance between trees from independent runs.
The remarkably high concordance of this 31-protein tree and my groupings by ultrastructure and molecular cladistic characters, entirely independent evidence, strongly suggests that both are close to the truth; the multigene tree would probably be further improved by having more representatives of weakly sampled taxa, but 31 proteins are probably not enough for complete resolution.
