Queller 1985 ("Kinship, reciprocity and synergism in the evolution of social behavior") provides a generalization of Hamilton's rule that allows for non-additivity. To accomplish that, Queller writes the fitness of an individual as \begin{align} W = W_{0} - CP + BP' + DPP' \end{align} where $W_0$ is the baseline fitness; $P$ is the phenotypic value (1 for altruist or 0 for nonaltruist); $P'$ is the phenotypic value of the interacting individual; $C$ is fitness cost if $P=1$; $B$ is the fitness benefit if $P'=1$; $D$ is the deviation from additivity of $C$ and $B$ if both were altruistic (i.e., $PP' = 1$).

Queller then substitutes this equation into Price's equation $\Delta G = Cov (G, W)$, where $G$ is the frequency of altruism allele in the focal individual($0$, $1/2$ or $1$). The resulting equation is: \begin{equation} - \bar c + \bar b \frac{Cov(G,P')}{Cov(G,P)} + \bar d \frac{ Cov(G,PP') }{Cov(G,P)} > 0 \end{equation} where $\bar c$, $\bar b$, and $\bar d$ are the sample means of $C$, $B$, and $D$.

My question is: how did Queller arrive at the sample means $\bar c$, $\bar b$, and $\bar d$? I understand that $Cov(aX,Y) = a\, Cov(X,Y)$, provided that $a$ is a constant. But what Queller is doing here is different; he seems to be saying that $Cov(AX,Y) = E(A)\, Cov(X,Y)$.


1 Answer 1


It is just worded a little wierdly in my opinion. The key line in the paper is: 'Fitness components are also defined for all individuals, for example, $C$ is defined, even for a non-altruist, as the cost it would incur if were altruistic.'

Essentially, if it doesn't matter what individual you are you always pay the same cost, then $C$ is a constant. This is pretty much the definition of a constant! The expectation of a constant is just the constant, $E(C) = C$.

The reason it works is that to find whether or not the cost is realised, Queller simply multiplies by the focal individual's phenotype value $P$, which is either 0 or 1. So either the individual pays the cost or it doesn't.

  • 1
    $\begingroup$ I think Queller's discussion of his 1985 model in his 1992 paper "Quantitative Genetics, Inclusive Fitness, and Group Selection" (p. 552) is better worded. There he writes the C, B, and D as regression coefficients. $\endgroup$
    – falsum
    May 30, 2015 at 16:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.