Statistical analysis

From 2010 to 2013, we conducted a set of 3-month feeding experiments on adults aged 310–870 days and on juveniles aged 111–215 days to investigate whether unfed littermates (challenged individuals) would increase their growth rate in response to experimentally elevated growth rates of their fed siblings (challengers). We identified pairs containing at least two same-sex littermates and fed the individual that was lightest (or as heavy as its sibling) when the experiment started (mean weight difference (±SD) in juveniles: 9.8±30.6g, in adults: 29.9±28.2g). The fed individuals received half an egg twice daily four times a week during three months. Competitive growth has never been described previously, so no prior information was available for power analyses to establish adequate sample sizes. For 17 fed adults including 8 females, the shortest feeding bout lasted 55 days and the mean±SD feeding duration was 84±11 days. For 31 fed juveniles including 12 females, the shortest feeding bout lasted 21 days and the mean±SD feeding duration was 76±21 days. For one adult female litter and one juvenile male litter, there were three same-sex siblings and the two lightest individuals were very close in weight (i.e., their average weight difference was lower than 10g in the 15 days preceding the experiment); one of them was fed, and the two unfed siblings were included in the cohort of challenged individuals. Experiments were interrupted when a pregnancy was detected in an experimental female (fed or unfed), and corresponding data were excluded from analysis. In other cases where the experiment was aborted (e.g., if an individual disappeared), data collected during the shortened period were included in analyses; note that for three juvenile dyads, food supplementation lasted respectively 21, 23 and 26 days, so these individuals were excluded from all calculations related to measures describing the second half of the experiment. Observations and weighing sessions were not subjected to blinding, because weight gained by fed individuals during the experiment was often detectable by observers.