Statistical analyses
All statistical analyses and graphic illustrations were performed in R
v. 3.6.3 (R Core Team, 2020). We tested whether the slopes of the
reaction norms of the measured life-history traits in response to food
abundance differed among dopamine treatments (control versus dopamine
exposure). To assess this for dry mass at maturation and somatic growth
rate, we used linear mixed effects models (LME, implemented using thelme functions in the package nlme , Pinheiro et al., 2020),
with fixed effects of treatment and food ration and the interaction
between these, and with beaker id as a random effect. For offspring dry
mass, LME models were fitted with treatment, food ration and their
interaction as fixed predictor variables, and maternal id nested in
beaker as a random predictor variable. Finally, we tested the effects of
treatment, food ration and their interaction on age at maturation and
offspring longevity, using Poisson generalized linear mixed-effects
(GLMM, implemented using the glmer function in the packagelme4 , Bates et al., 2015). For these latter analyses, beaker id
and maternal id nested in beaker id were included as random effects,
respectively. For LME models, the VarIdent command from the nlmepackage was used to allow residual variance to differ among treatments,
food ration and the two-way interaction between these (Pinheiro &
Bates, 2000). The appropriate random structure with respect to this was
chosen based on AICc comparisons. GLMM models were tested for
overdispersion and their Pearson and deviance residuals were checked for
patterns and lack of fit. For all analyses, fits of alternative models
with different fixed effects structures were compared using AICc.