Cephalotes

Moreau, Corrie S. & Wray, Brian D., 2017, An Empirical Test of Reduced-Representation Genomics to Infer Species-Level Phylogenies for Two Ant Groups, Insect Systematics and Diversity 1 (2), pp. 1-8 : 4-5

Cephalotes

Our analyses recovered as monophyletic all Cephalotes turtle ant species for which we had multiple samples represented, and the two Procryptocerus cluster together outside of the Cephalotes clade. Cephalotes lineages had been previously grouped into 18 different ‘species groups’ based on morphological data ( de Andrade and Baroni Urbani 1999). In all instances in this study where there were multiple representative species from the same species group, our analysis recovered the group as being monophyletic. This topology matches that of the phylogeny from Price et al. (2014), which was based on Sanger gene-based sequencing combined with morphological data.

Taxonomic group is the target group plus the outgroup. Five samples were excluded from the original 24 Caphalotes / Procryptocerus samples due to low depth of coverage (see Table 1). Mean number of loci passing quality filter is the average number of clusters formed at 85% similarity, after sequences were trimmed and filtered.The minimum taxa dataset of a sample is the set of loci for which there were at least four samples with representative data. Average number of parsimonyinformative variable sites is the total number for each respective taxonomic group.

Results of our analyses also lend strong support (bootstrap support> 0.95) to some nodes, where support was lacking (posterior probability <0.95) in the Sanger/morphology tree for both the Bayesian and ML topologies ( Figs. 1 and 2 ). All nodes in our GBS tree were well supported, while there were three nodes in the trimmed Sanger/ morphology tree of Price et al. (2014) for which support was lacking ( Fig. 2 ). For example, the node representing the split of the basalis group from the clade including the laminatus / pusillus , pallens , pinelli, coffee, and angustus groups lacks strong support in the Sanger/morphology topology but has high support in our GBS phylogeny ( Fig. 1 ).

However, there were several differences in well-supported relationships between the species groups in the Sanger/morphology tree of Price et al. (2014) and our GBS topology. For example, in

Prices based from quotes from June 2016. the phylogeny of Price et al. (2014), Cephalotes targionii ( angustus group) is sister to the clade containing the laminatus / pusillus , pallens , pinelli, and coffae groups, with the latter two groups sister to each other. However, in the GBS phylogeny from this study, C. targionii is sister to Cephalotes crenaticeps (coffae group, bootstrap support> 0.95), and the two together are sister to a clade containing the laminatus / pusillus , pallens , and pinelli groups in a strongly supported split ( Figs. 1 and 2 ).

Figures

Fig. 1. ML molecular phylogenies generated for this study using the RRGS method GBS. On the left is a phylogeny of the ant genus Cephalotes and the outgroup Procryptocerus.This Cephalotes topology including all samples that had sufficient data generated for our analyses. On the right is the full phylogeny of all samples included for our analysis of the ant genus Polyrhachis, along with the outgroup Camponotus. Both phylogenies were inferred from Illumina data generated from GBS libraries and were inferred using RAxML. Nodes with ML bootstrap support>0.95 are represented by black dots on nodes.

Fig. 2. Comparison between phylogenetic topologies generated by different molecular methods for the turtle ant genus Cephalotes. On the left is the tree inferred by Price et al. (2014), trimmed with the Mesquite prune clade tool to represent only the taxa represented in the present study.The Price et al. (2014) phylogeny was inferred with Sanger DNA sequence data from three protein-coding nuclear genes (1,457 bp) and three mitochondrial genes (2,022 bp). On the right is the phylogeny inferred for this study, also pruned with the Mesquite prune clade tool to remove duplicate taxa.This GBS topology was inferred from a dataset that had on average 25,000 loci per sample, each 100 bp in length. In the middle are the morphological species groups assigned by de Andrade and Baroni Urbani (1999). Nodes with Bayesian posterior probabilities>0.95 and/or ML bootstrap support>0.9 are represented by black dots on nodes; Nodes with only Bayesian posterior probabilities>0.95 and ML bootstrap support>0.7 are represented with black diamonds on nodes.