2.1 | Phylogeny selection and supertree construction
We constructed three genus-level phylogenies using publicly-available data. For analyses requiring informative branch lengths, we used the RAxML (Stamatakis, 2014) phylogram and treePL (Smith & O’Meara, 2012) chronogram of Opatova et al. (2020) both generated using an anchored hybrid enrichment (AHE) dataset consisting of 472 loci. We used the R-package ape (Paradis et al., 2004) to prune these trees down to a single representative per genus and a single outgroup (Liphistius : Liphistiidae), resulting in an 89-taxon genus-level chronogram and phylogram.
For analyses not requiring informative branch lengths, we constructed a more inclusive supertree using several recent mygalomorph-focused genomic phylogenies. We downloaded the maximum-likelihood phylogenies of Opatova et al. (2020) – Mygalomorphae-focused (AHE data); Hedin et al. (2018) – Atracidae-, Actinopodidae-, and Hexathelidae-focused (ultra-conserved elements [UCE]); Hedin et al. (2019) – Atypoidea-focused (UCE); and Montes de Oca et al. (2022) – Nemesoidina-focused (AHE). For the latter, the raw tree file was not available, so we generated a new maximum-likelihood phylogeny using IQtree (Nguyen et al., 2015) using the alignment and partition files from the study (Appendix A). We pruned these phylogenies down to a single representative per genus, rooted them, and used them as input trees for supertree construction using matrix representation with parsimony (MRP) in the R-package phangorn (Schliep, 2011), resulting in a 110-taxon final supertree (Fig. 1). The supertree topology was uncontroversial except in the position of the Venom Clade + Stasimopidae (from here on referred to as the Venom Clade+), which was recovered as either sister to the Domiothelina or of the clade including the Domiothelina and Crassitarsae. We chose to use the first of these topologies as it agrees with Opatova et al. (2020), which represents the most robust mygalomorph phylogeny currently available.