ii) Transition rate analyses of categorical and continuous social behaviour data
When analysed as a categorical variable, we recorded 10 G1, five G2 and eight G3 species out of 75 included in Kozak et al. (2015). The phylogenetic signal is strong (λ = 0.829) and not significantly different from 1.0 (X2(1) = 3.044, p = 0.081). Transitions between behavioural states are estimated to occur at different rates (X2(11) = 27.548, p = 0.004), with the highest transition rate estimated to be from G2 to G3 (transition rate = 0.320 vs next highest G1 to solitary = 0.177). We did not constrain the root node in our analysis of categorical social behaviour, but solitariness was estimated as the most likely ancestral state (likelihood of 0.950 for solitariness vs 0.018 next likeliest).
When analysed as a continuous variable, larval social behaviour is estimated to follow an Ornstein-Uhlenbeck model of evolution (comparison with BM model: X2(1) = 8.800, p = 0.003), with a relatively weak mean reverting force (α = 0.104) towards the optimum clutch size (z0 = 11.290). Again, the phylogenetic signal is strong (λ = 0.862) and not significantly different from 1.0 (X2(1) = 1.587, p = 0.208).