Eating one's own kind

Why a behaviour recorded in a thousand species resists a single explanation

A Eating a member of one's own species is, in most of the animal kingdom, neither rare nor pathological. It has been recorded in well over a thousand species, in every major group, and in circumstances that range from the desperate to the entirely routine. The behaviour attracts a moral vocabulary that obscures analysis: descriptions reach for words such as savage or unnatural, when what the evidence mostly shows is an animal responding to the availability of an unusually convenient package of nutrition that happens to resemble itself. The vocabulary matters because it shapes the questions asked. An investigator who begins from the assumption that a behaviour is aberrant looks for what has gone wrong, and will find something, whereas the more productive question is usually under what conditions the behaviour pays.

B The clearest cases are those in which the meal is a substitute for something scarce. Tadpoles in a drying pool, spiders in a season without insects and gulls whose fishing has failed all turn on their neighbours, and they do so when the alternative is starvation rather than as a matter of preference. Nutritionally the choice is a good one: a conspecific contains protein and minerals in precisely the proportions the eater requires, which is a property no other prey can offer. What restrains the behaviour under normal conditions is that individuals of one's own species are dangerous, are often relatives, and may carry diseases to which one is fully susceptible. The first of these is the most immediate: prey that is exactly one's own size and equipped with exactly one's own weapons is prey that may win, and the injuries sustained in the attempt are a cost that no other meal imposes.

C The last of these is the most serious and the least visible. A parasite adapted to one host can pass with unusual ease into a member of the same species, and populations that regularly eat their own accumulate infections that populations with other diets do not. The transmission of a fatal degenerative brain disease through funerary consumption in a human population is the best-documented instance, and comparable diseases spread through cannibalism have been described in other mammals. The cost is delayed, invisible at the moment of the meal, and therefore unlikely to restrain an animal that is hungry now.

D Sexual cannibalism, in which a female consumes her mate, has attracted disproportionate attention and is the case most often misdescribed. The textbook example is a mantis, and the textbook account holds that the male is routinely eaten during or after mating, sometimes to his benefit, since he supplies his partner with the nutrition to produce eggs he has fathered. Field observation complicates this considerably. The behaviour is far less frequent in the wild than in the laboratory, occurs mainly when the female is underfed, and the classic descriptions were produced under conditions — a confined container, a hungry animal, an observer at close range — that may have manufactured the phenomenon they recorded.

E In a small number of species the sacrifice is genuinely deliberate, and these cases have been documented too carefully to be dismissed. Males of certain spiders position themselves over the female's mouthparts during mating and are consumed while it proceeds; the males that are eaten fertilise more eggs than those that are not, and the females that eat them are less likely to accept another male afterwards. The behaviour is difficult to interpret as anything other than an investment, since it is initiated by the animal that dies. The relevant asymmetry is that a male of these species is unlikely to encounter a second female in his lifetime, which converts what looks like sacrifice into the only available strategy.

F Consuming one's own offspring appears at first sight to defeat the purpose of having produced them, and it is nonetheless common, particularly among fish guarding a nest. The explanation now generally accepted is coldly economic: a parent guarding eggs cannot leave to feed, a brood too small to be worth the vigil is worth more as food, and the energy recovered by eating some of it may allow the parent to raise a larger brood later. The prediction that follows — that small broods should be eaten more readily than large ones — has been repeatedly confirmed, which is the strongest kind of evidence such an argument can have: a prediction made in advance, capable of failing, and repeatedly given the opportunity to do so.

G What emerges from the whole subject is a caution against explaining behaviour by the category it falls into. Cannibalism is not one phenomenon with one cause; it is a heterogeneous collection of behaviours whose only shared feature is the identity of the meal. Some instances are adaptations refined by selection, some are errors of recognition, some are the last resort of a starving animal, and some are artefacts of the conditions under which they were observed. Grouping them produces a subject that is memorable, easily sensationalised and analytically almost useless, which may explain why it has been popular for so long and understood for so little of it.