Phylogenomics of tubeworms (Siboglinidae,Annelida) and comparative performance of different reconstruction methods |
| |
| Authors: | Yuanning Li Kevin M. Kocot Nathan V. Whelan Scott R. Santos Damien S. Waits Daniel J. Thornhill Kenneth M. Halanych |
| |
| Affiliation: | 1. Department of Biological Sciences & Molette Biology Laboratory for Environmental and Climate Change Studies, Auburn University, Auburn, AL, USA;2. Department of Biological Sciences & Alabama Museum of Natural History, The University of Alabama, Tuscaloosa, AL, USA |
| |
| Abstract: | Deep‐sea tubeworms (Annelida, Siboglinidae) represent dominant species in deep‐sea chemosynthetic communities (e.g. hydrothermal vents and cold methane seeps) and occur in muddy sediments and organic falls. Siboglinids lack a functional digestive tract as adults, and they rely on endosymbiotic bacteria for energy, making them of evolutionary and physiological interest. Despite their importance, inferred evolutionary history of this group has been inconsistent among studies based on different molecular markers. In particular, placement of bone‐eating Osedax worms has been unclear in part because of their distinctive biology, including harbouring heterotrophic bacteria as endosymbionts, displaying extreme sexual dimorphism and exhibiting a distinct body plan. Here, we reconstructed siboglinid evolutionary history using 12 newly sequenced transcriptomes. We parsed data into three data sets that accommodated varying levels of missing data, and we evaluate effects of missing data on phylogenomic inference. Additionally, several multispecies‐coalescent approaches and Bayesian concordance analysis (BCA) were employed to allow for a comparison of results to a supermatrix approach. Every analysis conducted herein strongly supported Osedax being most closely related to the Vestimentifera and Sclerolinum clade, rather than Frenulata, as previously reported. Importantly, unlike previous studies, the alternative hypothesis that frenulates and Osedax are sister groups to one another was explicitly rejected by an approximately unbiased (AU) test. Furthermore, although different methods showed largely congruent results, we found that a supermatrix method using data partitioning with site‐homogenous models potentially outperformed a supermatrix method using the CAT‐GTR model and multispecies‐coalescent approaches when the amount of missing data varies in a data set and when taxa susceptible to LBA are included in the analyses. |
| |
| Keywords: | |
|
|
