Effects of local versus global competition on reproductive skew and sex differences in social dominance behaviour

Abstract

Social hierarchies are often found in group-living animals. The hierarchy position can influence reproductive success (RS), with a skew towards high-ranking individuals. The amount of aggression in social dominance varies greatly, both between species and between males and females within species. Using game theory we study this variation by taking into account the degree to which reproductive competition in a social group is mainly local to the group, emphasizing within-group relative RS, or global to a larger population, emphasizing an individual's absolute RS. Our model is similar to recent approaches in that reinforcement learning is used as a behavioural mechanism allowing social-hierarchy formation. We test two hypotheses. The first is that local competition should favour the evolution of mating or foraging interference, and thus of reproductive skew. Second, decreases in reproductive output caused by an individual's accumulated fighting damage, such as reduced parenting ability, will favour less intense aggression but should have little influence on reproductive skew. From individual-based simulations of the evolution of social dominance and interference, we find support for both hypotheses. We discuss to what extent our results can explain observed sex differences in reproductive skew and social dominance behaviour.

Keywords: aggression; foraging interference; game theory in biology; hard and soft selection; mating interference; social hierarchy.

Figure 1.

Illustration of model elements, with a one-season life cycle. The focus is on one of the sexes, either males or females. (a) Dominance-hierarchy formation comes first, followed by risk of mortality, which depends on contest damage, and reproduction. An individual’s reproductive success (RS) is the net result of different effects. (b) After hierarchy formation, dominant individuals can interfere with subordinates, reducing their acquired resources (AR), for instance matings for males or foraging opportunities for females. Interference strength is a genetically determined trait κ. The red curve shows the decrease in AR for a subordinate and the blue curve shows the decrease for a dominant from performing interference. The grey vertical lines indicate three values of κ. (c) Expected RS as a function of dominance position (k = 1 is top ranked) for local (L₁, L₂, L₃) and global (G₁, G₂, G₃) competition, corresponding to trait values κ₁, κ₂, κ₃ from (b). The curves do not include effects of contest damage. Other local groups are assumed to have a total RS of 16 offspring. (d) The orange curve shows the influence of accumulated contest damage on parenting ability and the sloping dashed curve shows the risk of mortality from contest damage. (Online version in colour.)

Figure 2.

The outcome of individual-based simulations, showing effects of local versus global competition and contest damage. (a) Expected RS as a function of dominance position for the six cases in table 1 (colour coding given in legend). Red, green and blue show global, intermediate and local competition, respectively, with risk of mortality from figure 1d. The light-coloured curves in addition have reduced parenting ability from contest damage, as given in figure 1d. (b) Accumulated contest damage as a function of dominance position, for the six colour-coded cases in (a). Points and bars give median and first and third quartiles for the distribution over simulated local groups. (c) Median and first and third quartiles for the distribution over groups of the skew index M [18] (see electronic supplementary material, table S1), as a function of the proportion of local competition (the six colour-coded cases are shown). (d) Effects of interference and contest damage on the total group contested AR (summed over competing individuals), expressed as proportions of the maximal possible value. Unfilled parts indicate costs of interference, light grey parts costs of damage (reduced parenting ability), and colour-coded parts the remaining AR, respectively. (a,b) Include only surviving individuals; these have a dominance position. There were 500 groups in the simulated populations. (Online version in colour.)

Figure 3.

Simulated data (light grey points), density violins and colour-coded fitted curves of log-transformed total number of AA rounds (fighting rounds, loess fits) for cases with either no or full local competition. (a–d) Cases 1, 3, 4 and 6 in table 1, with colour coding as in figure 2. The x-values of the points are jittered for visibility. (Online version in colour.)