Interpreting selection when individuals interact

Summary A useful interpretation of quantitative genetic models of evolutionary change is that they (i) define a set of phenotypes that have a causal effect on fitness and on which selection acts, and (ii) define a set of breeding values that change as a correlated response to that selection because they covary with the phenotypes. When the expression of one trait causes variation in other traits then there are multiple paths by which a trait can cause fitness variation. Because of this, there are multiple ways in which selection can be defined, and still be consistent with a causal effect of traits on fitness. We use this result to show that genetical theories of natural/kin selection ignore causation and because of this we suggest they shed little light on the nature of selection. When traits expressed by an individual are affected by traits of their social partners (indirect genetic effects), we suggest a causal partitioning that allows selection to be cast in terms of Hamilton's costs and benefits. We show that previous attempts to understand Hamilton's rule in the context of indirect genetic effects either lack generality, or do not adequately describe all the ways in which an individual's actions constitute a cost to the individual or a benefit to its social partner(s). Our results allow us to explore Hamilton's rule in a multitrait setting. We show that evolution always increases inclusive fitness, and when the traits are measured in units of generalised genetic distance evolutionary change in the traits is in the direction in which inclusive fitness increases the fastest. However, we show that Hamilton's rule only holds in a multitrait context when the suite of traits are at equilibrium. When they are out of equilibrium, the conditions for altruism to evolve may be more or less stringent depending on genetic architecture and how costs and benefits are defined.