Hamilton (1964a) has pointed out that social behavior can evolve through kinship selection, wherein the probability of an altruistic act is proportional to the probable genetic relationship between altruist and recipient. Therefore, to calculate likelihoods of altruistic behavior being evolved in certain situations, the probable proportions of genes identical by descent in the individuals involved must be known. Haplodiploidy, the basic genetic system in Hymenoptera, is particularly favorable for the evolution of altruistic behavior, since full sisters have three-fourths of their genes identical by descent (Hamilton 1964b). This is because they each receive the same haploid set of chromosomes from their father, making up half their genotypes, while genes from their mother, being the products of a normal meiosis, are identical by descent half of the time. Full sisters under normal autosomal inheritance, where both sexes are diploid, have only half their genes identical by descent. When I compared a number of other coefficients of relationship calculated by Hamilton (1964b) with those given by Li (1955), however, I observed several discrepancies. In some cases it was also clear that Hamilton's values must be in error. For example, since brothers, being haploid and impaternate, are the equivalents of gametes produced by the meiosis of one individual, they clearly have half their genes in common; yet Hamilton (1964b) gives a coefficient of relationship of r 1/4 for this relationship. This erroneous result is due to Hamilton 's treatment of males in calculating coefficients of relationship: each male is artificially "made diploid" by the addition of a "cipher " gene that is unique to it. Although Hamilton 's calculations are in error for some relationships involving males, the coefficients of relationship given by Wright (1922; see Li 1955) are also inadequate for determining the optimum strategy of an individual based on the probable degree of relatedness of another to it. This is because the coefficient of relationship as usually employed is actually a measure of phenotypic correlation assuming simple additive gene effects, rather than an estimate of the proportion of genes held in common through identical descent, although in normal autosomal inheritance the values obtained are correct for either purpose. Another failing of the coefficient of relationship is that it makes no distinction between the relationship of a given male to a given female and the converse female-male relationship. Thus the value r = 1/2 is obtained for the father-daughter relationship (Li 1955), whereas in fact all the father's genes are identical by descent with genes present in the daughter, while only half the daughter's genes are identical by descent with genes present in the father. The situation is, however, clearly indicated by Lji's diagrams (1955, p. 182). Letting GA(B) be the proportion of individual A's
Read full abstract