Mental glands and their associated delivery behaviors during courtship are unique to the plethodontid salamanders. Because previous interpretations of the evolution of these features were conducted using older phylogenetic hypotheses, we reanalyzed these traits with newer courtship descriptions and contemporary phylogenetic methods. Using Bayesian ancestral state reconstruction methods that have been developed since the first phylogenetic analyses were conducted in the mid-1990s, we reconstructed mental gland and courtship behavior evolution on a novel molecular (Rag1) topology of plethodontids. The most probable ancestral condition for plethodontids was resolved as presence of a mental gland. Loss of a mental gland occurred in each subfamily and was recovered as the most probable ancestral condition for the Spelerpinae. In contrast, parsimony reconstruction recovered the presence of a mental gland in the ancestor to Spelerpinae with multiple secondary losses. We hypothesize that that absence of a mental gland is possibly ancestral in some clades (i.e., Spelerpinae), and secondary in others (e.g., paedomorphic Eurycea). The most probable ancestral form of the mental gland is likely to be the large pad-type distributed extensively in Plethodontinae and Bolitoglossinae. Desmognathans have the most unique mental glands, occurring in an anterior protrusion or bifurcated form (in Desmognathus wright). Fan-shaped mental glands evolved independently in Eurycea and Oedipina. Small pads arose independently in Bolitoglossinae, Plethodontinae, and Spelerpinae. Head-rubbing behavior for mental gland delivery mode was recovered as the most probable and parsimonious ancestral state for the Plethodontidae, with independent losses of this behavior in Plethodontinae and Spelerpinae. Because head-rubbing was observed in outgroups, we hypothesize that head-rubbing behavior predated mental gland evolution. Pulling, snapping, slapping, and biting behaviors evolved independently in the Plethodontinae and Spelerpinae and are not homologous with head rubbing. All hypotheses of mental gland and courtship evolution invoke homoplasy.
Supplementary table 1
Species and GenBank accession numbers for rag1 sequences used in this study.
Supplementary figure 1
Unconstrained topology for selected taxa in the salamander family Plethodontidae using Rag1 sequence obtained from GenBank.
Supplementary figure 2
Topology for selected taxa in the salamander family Plethodontidae using Rag1 sequence obtained from GenBank. This topology is constrained to form a monophyletic unit for Pseudotrition, Gyrinophilus, and Stereochilus.
Supp. Fig. 3 - Gland presence parsimony
Parsimony reconstruction of mental glands within Plethodontidae. Two states were considered: absent (white) and present (black). Pie charts at each node provide the most parsimonious reconstruction for each node. The phylogeny was reconstructed using a Bayesian analysis of Rag1 sequences (see Materials and Methods).
Supp. Fig. 4 - Gland morphology parsimony
Parsimony reconstruction of mental gland morphology within Plethodontidae. Six unordered states were considered: absent (white), small pad (black), large pad (red), anterior protrusion (yellow), bifurcated (green), and fan-shaped (blue). Pie charts at each node provide the most parsimonious reconstruction for each node. The phylogeny was reconstructed using a Bayesian analysis of Rag1 sequences (see Materials and Methods).
Supp. Fig. 5 - Head rubbing parsimony
Parsimony reconstruction of heading rubbing within Plethodontidae. Two states were considered: absent (white) and present (black). Pie charts at each node provide the most parsimonious reconstruction for each node. The phylogeny was reconstructed using a Bayesian analysis of Rag1 sequences (see Materials and Methods).
Supp. Fig. 6 - Pulling parsimony
Parsimony reconstruction of pulling behavior within Plethodontidae. Two states were considered: absent (white) and present (black). Pie charts at each node provide the most parsimonious reconstruction for each node. The phylogeny was reconstructed using a Bayesian analysis of Rag1 sequences (see Materials and Methods).
Supp. Fig. 7 - Snapping parsimony
Parsimony reconstruction of snapping behavior within Plethodontidae. Two states were considered: absent (white) and present (black). Pie charts at each node provide the most parsimonious reconstruction for each node. The phylogeny was reconstructed using a Bayesian analysis of Rag1 sequences (see Materials and Methods).
Supp. Fig. 8 - Slapping parsimony
Parsimony reconstruction of slapping behavior within Plethodontidae. Two states were considered: absent (white) and present (black). Pie charts at each node provide the most parsimonious reconstruction for each node. The phylogeny was reconstructed using a Bayesian analysis of Rag1 sequences (see Materials and Methods
Supp. Fig. 9 - Biting restraint parsimony
Parsimony reconstruction of biting/restraint behavior within Plethodontidae. Two states were considered: absent (white) and present (black). Pie charts at each node provide the most parsimonious reconstruction for each node. The phylogeny was reconstructed using a Bayesian analysis of Rag1 sequences (see Materials and Methods).