We thank Fletcher et al. [1] for their comments on our recent TREE article [2]. These can be condensed down to two key points. First, a change in the definition of altruism: to claim that altruism can evolve without relatedness, Fletcher et al have simply changed and broadened the definition used by Wilson and Hölldobler [3] and ourselves [2] to include reciprocal altruism. Second, the authors argue that group selection preceded kin selection (i.e. inclusive fitness) theory as a theoretical tool to explain altruism in the social insects. We discuss both but note that neither point weakens our article [2], or rescues the problems in that by Wilson and Hölldobler [3].
Shifting semanticsRecent work by Fletcher and Zwick [4] showed that the altruism of insect workers, and reciprocal altruism, where one individual helps another and gains a delayed reproductive benefit, can both be modeled with a form of Hamilton's rule. We reached an identical conclusion in a recent model of mutualisms that is also based upon a directfitness version of Hamilton's rule [5]. The central point is that the delayed feedback benefit in reciprocal altruism can be captured with a phenotypic correlation term, which emerges in the same way as genetic relatedness. On this basis, Fletcher et al. claim that altruism can occur through a phenotypic correlation alone, without genetic relatedness. However, this is only true if one changes the definition of altruism used by Wilson and Hölldobler [3], and ourselves [2]: 'Altruism is defined as behavior that benefits others at the cost of the lifetime production of offspring by the altruist. ' [3] Crucially, reciprocal altruism and the related phenomena that occur in mutualisms [5] cause a phenotypic feedback that benefits personal reproductive fitness. We appreciate that it is unfortunate that the nomenclature of both phenomena contain the word 'altruism', but we were careful to define our terms and, as with Wilson and Hölldobler [3], reciprocal altruism was not being discussed.Where does this leave us? We are happy to accept that reciprocal altruism and mutualism can evolve without genetic relatedness between actor and recipient [5]. This much is obvious as partners in a mutualism need not even be of the same species. However, an altruistic act that decreases the lifetime reproduction of the actor will only be selected when it increases propagation of the causal genes through individuals that are related at one or more loci.Which is the best: kin selection or group selection? We view this as an empty question. There are three different ways of partitioning social selection: (i) the inclusive fitness extension of individual selection; (ii) the direct fitness model of individual selection; (iii) and the withinand-between group selection model [6,7]. Fletcher et al. spend most of their time advocating the second (a form of kin selection theory) but then conclude that group selection is best [1]. In reality, all three models are important and useful tools for investigating and modeling social evolution ...