Sampling error, effective paternity, and estimating the genetic structure of honey bee colonies (Hymenoptera: Apidae)

Abstract

Multiple mating by social Hymenopteran queens significantly lowers the average genetic relatedness among female nestmates, which subsequently affects a wide range of social behaviors. Honey bees (Apis spp.) have among the highest levels of multiple mating in social insects, and have received the most empirical effort to quantify the effective paternities within colonies. We reviewed 24 studies that estimated paternity frequencies of individual, naturally-mated honey bee queens using molecular techniques. We summarize the methods used to estimate effective paternity (me) and intracolonial genetic relatedness (G). We then concentrate on the effect of sample size on estimates of me using Monte Carlo simulations. The results demonstrate that me estimates may vary, significantly as a result of sampling error, particularly at low worker sample sizes and high paternity numbers. From these simulations, we arbitrarily define a "threshold" worker sample size to effective paternity ratio (n/me) that, at best, reduces the error of estimating me to less than one subfamily. The literature review illustrates that no study with an n/me ratio above this threshold estimates an me above 15 subfamilies. Finally, we briefly discuss other factors that may serve to over-estimate me including numerous sampling biases. We conclude that although 152 colonies in the various species of Apis have been tested, the extremity of their paternity frequencies may be somewhat exaggerated, although not drastically.