There is no doubt that selection acts at the individual level but, there is heated debate over the relative importance of higher and lower levels of selection. Kin selection is the dominant paradigm explaining the evolution of cooperation. Whereby, individually detrimental traits can be selectively favored if they increase the fitness of genetic relatives. Kin selection operates at the individual level, biasing cooperation towards those sharing the most genes. However, kin selection may also act at the group level when efficiently functioning groups are more productive than dysfunctional ones, provided group success correlates to kin structure. Finally, kin selection

can occur at the genome level within individuals, where paternally and maternally-inherited genes may favor different behaviors and actions. The imprinting of genes to parent-of-origin could also have important ramifications for social evolution. Here, I exploit the conflict over male production in honey bees to examine how these three levels of selection operate. Honey bee workers could ‘police’ eggs laid by other workers either to maintain colony-level productivity, favor more closely related individuals or as a result of intragenomic conflict (i.e. the paternal genome favors laying, the maternal one favors policing). Firstly, I found that although African workers lay eggs more rapidly than European workers, there is no difference in their times to ovary activation. Significant effects of both the juvenile and adult social environment on ovary activation, suggest that environment has a larger effect on the propensity to activate ovaries rather than subspecies. Secondly, I mathematically simulated a typical eusocial colony where I varied the number of mates per queen, viability of worker-laid males, colony efficiency costs of reduced worker helping, and whether or not intragenomic conflict could be expressed. Genome level selection dominated over both individual and group levels, and group level selection was more significant than the individual level in determining when queens dominate male production. Thirdly, individual level selection predicts policing late into larval development whilst benefits accrued through colony efficiency predicts workers should stop policing viable larvae soon after hatching. To this end I reared queen and worker laid male larvae in a queenless colony and transferred male larvae, from both sources, of differing ages, into a queenright colony. Post transferal (4 and 24 h), I found that workers equally removed larvae regardless of age or maternal source. With the observed high efficiency of policing eggs, these results suggest no mechanism has evolved to police larvae. Alternatively, drones may have a higher level acceptance threshold than female larvae, due to the possibility that they are laid by workers. Finally, I examined genome level selection by crossing African and European honey bees and then placing the emerging worker offspring into a queenless colony. I observed behavior from day 8 to 28 and collected marked workers on day 16 and 28. I predicted that parent-of-origin effects would occur, but instead found workers of both crosses have higher levels of ovarial development than their purebred counterparts. This suggests an imprinting mismatch such that only the paternal imprint is expressed. Together these results indicate that selection is acting at levels besides that of the individual. Continued research is needed to understand how selection, interacting over multiple levels, impacts behavior, across the animal kingdom.