R code for "Nepotism masks evidence for reciprocity in cooperation networks"

Nepotism and reciprocity are not mutually exclusive, because cooperation among kin can depend on both kinship cues and past reciprocal help. The presence and importance of these strategies can therefore be difficult to disentangle using observational data. The reliability of genetic marker-based genetic relationships are often assessed through resampling procedures, but the reliability of symmetrical cooperative relationships are less often assessed, even though they are harder to estimate. We developed a resampling procedure for inferring statistical power to detect observational evidence of nepotism and reciprocity and applied this to both simulated and real data. We simulated datasets of perfectly symmetrical helping rates and added varying degrees of nepotism. We then assessed how nepotism and the number of observations influence the probability of detecting reciprocity. Due to collinearity, nepotism consistently masked evidence for reciprocity especially with fewer observations of helping. Even under perfect reciprocity, kinship bias was often detected faster and appeared larger, because fewer observations decreases the accuracy of given and received helping rates, but not kinship. Our conclusions were supported by the same analysis applied to empirical data on food sharing in vampire bats and allogrooming in mandrills and Japanese macaques. Results highlight the need to estimate the relative reliability of estimates of kinship and helping rate before comparing their relative importance. We provide R scripts to assess the reliability of kinship and reciprocal help estimates in other cooperation networks.