Bivariate model of maternal provisioning rates (feeds/hour) and number of female helpers.

This model was fitted in R using the MCMCglmm package (v2.34; [71]), using an inverse Gamma as prior distribution for random and residual (co)variances (V = 1, nu = 1.002), 50,000 MCMC iterations, with 1,000 as initial burn-in and sampling every 10 iterations. Effective sample sizes for all model terms were always higher than 2,500, and MCMC traces were visually inspected to confirm convergence. We calculated the slope between maternal provisioning rates and female helper number within and among mothers by dividing the corresponding covariance between maternal provisioning rates and female helper number by the estimated variance in female helper number. We found that both the within- and among-mother slopes of female helper number on maternal provisioning rates were negative (within mothers [95% CrI] = −0.451 [−1.077, 0.151]; among mothers [95% CrI] = −0.311 [−1.381, 0.667]). Mean posterior estimates (“Mean”) and their 95% credible intervals (“95% CI”) are provided. (DOCX)