Extended Data Fig. 4: Female song and cooperative breeding transition likelihoods based on the median difference from expected transition counts, calculated using alternate territory classification, alternative phylogenies to account for phylogenetic uncertainty, or alternative cooperative breeding classification methods.

Using the default cooperative breeding classification method: Panels A and B show the numbers of transitions between female song states and cooperative breeding states that occurred in lineages that were A) non-territorial or weakly/seasonally territorial and B) year-round territorial. C) Pooled results over 200 trees, 20 simulations per tree. D) Results from 500 simulations using the real trait distribution on the consensus phylogeny obtained from using the “mean.edge” method with absent edges ignored in the mean edge length calculations. Using the default consensus tree: E) MeanCoopTie2Noncoop, F) MeanCoopTie2Coop, G) MeanCoopOmitTies, H) AnyCoopEqualsCoop, and I) AnyNoncoopEqualsNoncoop. Arrow weights indicate the log-transformed median relative number of state switches that occurred on that topography. Deviations of observed from expected transition counts were assessed across stochastic character map simulations using the transition-count framework described in Methods: log-transformed observed vs. expected counts were analyzed with a two-way linear model (observed vs expected × transition), with significance assessed by ANOVA and Tukey-adjusted post-hoc comparisons within each transition. Grey arrows indicate transitions not significantly different from expectation; red/blue arrows indicate significantly more/fewer transitions than expected. Otherwise, color denotes the median difference from the expected number of transitions across simulations, with red indicating that the observed number of state switches was greater than expected, and blue indicating fewer state switches than expected. Asterisks indicate that the percent of simulations where the number of transitions is greater (red arrows) or less (blue arrows) than the calculated expected number of transitions for the simulation in question in >95% of simulations (**) or 90-95% of simulations (*). Sample sizes: A) n = 637 species, B) n = 404 species, C) n = 1041 species, D) n = 1041 species, E) n = 1075 species, F) n = 1075 species, G) n = 1066 species, H) n = 1075 species, I) n = 1075 species.