Multi-model selection procedure.

We compared generalised linear mixed effects models (GLMM) with triad_ID nested in male type as random effect and the following combination of fixed effects: male dominance status (Dominance), kinship (Kin), male body size (Body Size), interaction between male dominance status and kinship, interaction between male body size and kinship, interaction between male body size and male dominant status. Number of parameters retained (K), Akaike information criterion weighted for small sample size (AICc), differences between AICc (Δ AICc) and normalized Akaike weights (wi) for each model are shown. Data are showed for sexual display and mate choice analyses.