Abortions
There was evidence of variation between individual mothers in the probability of abortion for the nutritional stress group (model with random intercept ω = 0.985; model without ω = 0.015), but not for the control or mating delay groups, based on comparison of models with and without random effects (Table S4.5). The predicted probability of abortion increased with age across all treatments, up to c. 0.2 for the mating delay group and c. 0.5 for the control group by 100 days (Fig. 3, Table S4.5). Females in the nutritional stress group had higher expected probability of abortion than those in the control or mating delay groups for any given age, increasing to c. 0.7 by 100 days (Fig. 3, Table S4.6).