Social dominance hierarchies

Given the dynamic nature of dominance hierarchies among animals, it would be very unlikely to get a well formed control hierarchy in nature.
For example, here is the dominance hierarchy of a single population of elephant seal males during the mating season:

From Marianne Riedman, The Pinnipeds, page 206.

You’ll notice that very few relations are unequivocally transitive. Even among what might be considered the “same” level there are inconsistencies. Why is this? Well consider what happens in a single act of engagement, or as ethologists sometimes call it, a transaction. An animal can approach another animal along a continuum of possible responses from outright aggressive to a positive reinforcement (such as grooming in primates). Likewise the respondee can do the same. In simple terms, each transaction can be ++ (both positive), — (both negative) or mixed (+- or -+). Of course, there are no discrete states here – the positive and negative poles are merely extremes defined in terms of costs to the agents, and a mixed strategy can be more positive than negative, or vice versa, as well as exactly half way. So while for convenience, ethologists and game theorists represent a matrix of outcomes in any given transaction as a table like this:

in fact the more realistic representation is a field of states that has a contouring based on the properties of the individuals engaged in the interaction:

And since these payoff contours will vary for each transaction between different individuals, the overall social organisation of the population will be more like a probabilistic and dynamic field than the traditional game theoretic matrix. Still, that is simply accommodated by modern simulations. I mention it here merely to prevent people thinking in terms of overall simplicity.

Given that we have such a probabilistic and populational rather than class-based notion of hierarchy developing out of individual interactions, a note is required on what counts as an individual or agent here. Nothing should be take to mean that only the archetypical individual counts. As many people have noted over the decades since Darwin, groups of archetypical individuals can count as agents, as in the case of bee hives, colonial organisms, and even social groups like a troop or village. But that is for another time – all that we need here is that players or agents transact according to their relevant properties or traits, and that we do not need to assign each attitude or behaviour as being positive or negative, in a simple manner.

So, consider some kind of idealised primate. In that species, the typical social arrangement is done by an alpha male who controls some majority of mating opportunities (but not all – sneaky mating still occurs in nearly every studied case), who dominates the troop through a series of alliances with subordinate males who get access to food and some mating opportunities as a result. Not all subordinate males will be part of the status quo alliance, and at any time, if the alpha male fails to defend his position, the others, even if they are the deputies, may make a move and take the alpha position.

High status males pay a cost, in terms of defence activity and also in terms of payola to the subordinates, but there appears to be a widespread benefit in terms of endocrinal activity, particularly testosterone in vertebrates (I have no idea what the biochemistry of arthropod dominance is), that means that higher status animals are on average healthier than their subordinates when other factors like diet and care are taken into account. This has also been the case in numerous studies on humans, beginning with the famous Whitehall Study.

Most primate hierarchies are based on male status, but this is not set in stone. Hierarchies are fundamentally “about” mating access, so male hierarchies apply solely in those species in which mating opportunities are controlled by males. This is not universal, and some species, bonobos being the most obvious, have either a female primary hierarchy, or a mixed hierarchy. At some later time, I will discuss the phylogenetic distributions of these, when I have managed to gather all the data, but for now, let us consider the primacy of a sex-related hierarchy to correlate roughly with the degree of sexual dimorphism; that is, the bigger the sex, the more it controls the social structural hierarchy.

Gorillas have a major case of sexual dimorphism in the males favour, for example, and the social structure is a male silverback controlling all sexual access, while “bachelor males” live on the fringe, seeking “sneaky mating” opportunities. Incidentally, the average cuckoldry rate for gorillas is, it seems, around 15%. This might be called the “harem” model, but I think it is a stable outcome where males of any clade, not merely primates, are much bigger than their females, so I will call it the “bull herd model”. It is open whether the males started out bigger and thus impose that structure on their populations, or whether the populations started out as mildly bull herds, and competitive selection drove greater dimorphism. I think probably both – a novel species will “inherit” the social and sexual dimorphism of its parental species, and selection will subsequently modify that structure depending on the overall necessities of the new species’ ecology. It will not always drive greater dimorphism, for example, and two species in particular demonstrate this: bonobos, and humans.

As Frans de Waal has explored in various books, especially his Bonobo: The Forgotten Ape (with Frans Lanting), bonobos have a mostly female hierarchy, but there is also a male hierarchy. It’s just that the male hierarchy is about being attractive to the females. Since mating is not controlled by either sex, fitness here is determined by how well the individual is accepted by the entire troop, unlike with gorillas or the common chimp, where male control is the norm.*

Humans are a complicated case, compared to their primate cousins. First of all, humans have moved out of the local troop (usually around 50 individuals) and regional band (usually around 500 individuals) model of other apes. We travel over vast distances in virtue of our bipedality, and hence we trade across distances even back when we were in what is called the Paleolithic period, around 300 thousand to 30 thousand years ago. Second, like bonobos, we are only marginally sexually dimorphic. Third, we are language users. This means that human social organisation, at least before the eventuation of agriculture, was quite a lot more complex than even a bonobo band. That’s for the next post.

• Not all chimp troops are male dominated. There is a case of a troop where the alpha male died but the subordinate males were too junior to take the alpha role, and a female became the alpha individual. This suggests that sex-based hierarchies may be a bit more plastic than we might otherwise think.

©2009 All rights reserved.