While the main goal of this work is not to elucidate a novel phylogeny for Siphonophora, we did expand on the most recent transcriptome based phylogeny [@munro2018improved] to accommodate a larger taxon sampling. In order to do this, we ran a constrained analysis on an extensive 18S+16S dataset. The phylogenetic analysis included 55 siphonophore species and 6 outgroup cnidarian species (*Clytia hemisphaerica*, *Hydra circumcincta*, *Ectopleura dumortieri*, *Porpita porpita*, *Velella velella*, *Staurocladia wellingtoni*). The gene sequences we used in this study are available online (accession numbers in Dryad repository). Some of the sequences we used were accessioned in [@dunn2005molecular], and others we extracted from the transcriptomes in [@munro2018improved]. Two new 16S sequences for *Frillagalma vityazi* (MK958598) and *Thermopalia* sp. (MK958599) sequenced by Lynne Christianson using the primers from [@cunningham1993molecular] (read 3' to 5' F: TCGACTGTTTACCAAAAACATAGC , R: ACGGAATGAACTCAAATCATGTAAG) were included and accessioned to NCBI. Additional details on the phylogenetic inference methods can be found in the Supplementary Methods.
Unconstrained ML and Bayesian phylogenies were congruent (S2,S5). Given the broader sequence sampling of the transcriptome phylogeny, we ran constrained inferences (using both ML and Bayesian approaches, which produced fully congruent topologies (S4,S6)) after clamping the 5 nodes (S3, blue circles in Fig. \@ref(figure4)) that were incongruent with the topology of the consensus tree in [@munro2018improved]. This topology was then used to inform a Bayesian relaxed molecular clock time-tree in RevBayes, using a birth-death process (sampling probability calculated from the known number of described siphonophore species) to generate ultrametric branch lengths (S7-8). Scripts and tree files are available in the Dryad repository.
Published May 07, 2021 on Dryad. https://doi.org/10.5061/dryad.p2ngf1vp2