Cannibalism

Abstract

Cannibalism is a form of animal behavior—specifically a feeding behavior—the killing and eating of a member of one's own species, and far from a pathological aberration it is a widespread, adaptive feeding and reproductive strategy. This article treats cannibalism as a decision: an animal weighs the nutritional, competitive, or reproductive gain of consuming a conspecific against the risks of injury, disease, and the loss of kin. It covers the ecological conditions that favor cannibalism, the paradox of sexual cannibalism in which a female eats her mate, filial cannibalism in which a parent eats its own offspring, the cannibalistic polymorphisms that develop in some amphibians, and the role cannibalism plays in regulating population density. A worked example and three interactive demonstrations accompany the text.

Keywords: cannibalism, sexual cannibalism, filial cannibalism, intraspecific predation, population regulation

A female praying mantis bites the head from her mate while he is still copulating. A male stickleback guarding a nest full of eggs eats a portion of the clutch he is tending. A spadefoot toad tadpole grows a wide jaw and a muscular gut and begins to eat its poolmates. Each act looks, at first, like a breakdown—of courtship, of parental care, of the ordinary rule that an animal does not eat its own kind. None of them is. Each is a solution to a problem the animal faces, and the problem is what makes the behavior legible.

This article treats cannibalism as a behavioral and evolutionary strategy rather than a disorder. It begins with what the category covers and why eating a conspecific should be regarded as an adaptive option at all, then turns to the three forms that have drawn the most study—sexual cannibalism, filial cannibalism, and cannibalistic polymorphism—before examining how cannibalism feeds back on the density of a whole population. Throughout, the act is read as a decision weighing gain against cost, not as an accident or a pathology.

What Cannibalism Is

Cannibalism is the killing and consumption of an individual of the same species. In the Medical Subject Headings vocabulary it is classified as a feeding behavior, itself a form of animal behavior—a placement that frames it, correctly, as a normal dietary option rather than a pathology. Laurel Fox's 1975 review was the first systematic treatment of the phenomenon across natural populations, assembling scattered reports into the recognition that cannibalism occurs routinely in a great many taxa and under predictable conditions (Fox, 1975).

The conceptual shift that made cannibalism a subject of serious study was Gary Polis's reframing of it as intraspecific predation: a predator–prey interaction in which predator and prey happen to belong to the same species. Polis's 1981 synthesis showed that cannibalism is neither rare nor incidental but a major interaction with consequences for individual fitness, life history, and population dynamics, governed by the same economics as any act of predation (Polis, 1981). The edited volume by Mark Elgar and Bernard Crespi later documented the behavior across animals from protozoa to mammals, confirming that it is taxonomically widespread and takes many functional forms (Elgar & Crespi, 1992).

Read this way, the question is never whether an animal eats its own kind—many do—but when, whom, and why. A conspecific is a package of nutrients that is, by definition, exactly the right diet; it is also a potential rival removed, and sometimes a current or future offspring destroyed. The behavior is the resolution of those competing considerations.

Why Cannibalism Evolves

If eating a conspecific were always costly, it would not persist. It persists because, under the right conditions, the eater gains more than it loses. Polis identified the conditions that reliably favor cannibalism: a shortage of alternative food, a high density of conspecifics, and a size structure in the population that lets larger individuals safely subdue smaller ones (Polis, 1981). Each condition shifts the balance toward the act. Scarce food raises the nutritional value of a conspecific meal; crowding raises the rate of encounter; and a size difference lowers the risk that the prey injures the predator or turns the tables.

The costs are real and shape who is eaten. Cannibalizing a relative destroys shared genes, so kin recognition often suppresses the behavior toward close kin even where it is readily directed at strangers—a constraint documented in the amphibian systems discussed below (Pfennig, 1992). Eating a conspecific also carries an elevated risk of acquiring species-specific parasites and pathogens, because a member of one's own species is the ideal host for them. And the prey, being of the same species, is often nearly the same size and fully capable of fighting back. These costs explain why cannibalism is conditional rather than constant: it appears when the balance tips and recedes when it does not (Fox, 1975; Elgar & Crespi, 1992).

Sexual Cannibalism

The most studied and most counterintuitive form is sexual cannibalism: the killing and eating of a potential or actual mate, usually the male by the female, in or around the act of mating. It is widespread in spiders and mantids and has become the textbook case of sexual conflict, because the two sexes have opposed interests in the outcome. A male gains by surviving to mate again; a female gains a meal that can be converted directly into eggs. Jutta Schneider's synthesis frames the behavior as exactly this clash of interests, in which the female's foraging decision and the male's reproductive decision cannot both be satisfied (Schneider, 2014).

The timing of the attack is decisive, which is why sexual cannibalism is a problem in game theory rather than a simple loss for the male. Sandra Lawrence's field study of the European mantis showed that cannibalism in the wild is common and that a male's fate depends on his approach: a male eaten before copulation loses everything, while a male eaten during or after sperm transfer may still father offspring, and the consumed body may even raise the number of eggs the female lays (Lawrence, 1992). Where consumption follows insemination, the male's loss is partly recovered as paternity, and the behavior can be stable for both sexes.

In the most extreme cases the male is not a victim but an active participant. Maydianne Andrade's work on the Australian redback spider showed that the male actively somersaults his abdomen into the female's mouthparts during copulation, and that cannibalized males father more offspring and sire longer because the female is less likely to remate—a case of sexual selection for male self-sacrifice rather than against it (Andrade, 1996). Male complicity makes plain that cannibalism is a strategic outcome negotiated between the sexes, not simply a predatory act by the female.

Cannibalism is not confined to the mating moment. Henrik Roggenbuck and colleagues, studying the European garden spider, found that females attack males both before and after copulation and that a male's defense is to time and position his approach to maximize sperm transfer before any attack (Roggenbuck et al., 2011). Across species, the puzzle of why a female should destroy a mate at all has several non-exclusive answers—a meal that funds egg production, the choosy rejection of a poor-quality male, or an aggressive disposition that spills over from foraging into mating—and distinguishing them remains an active question (Zuk, 2016).

Filial Cannibalism

Stranger still is filial cannibalism: a parent eating its own offspring. It is common in fish that guard eggs, where the very male tending a nest will consume some or all of the clutch. Andrea Manica's review of the phenomenon in teleost fish is the standard treatment, cataloguing the conditions under which guarding parents eat eggs and distinguishing total filial cannibalism, in which the whole brood is consumed, from partial filial cannibalism, in which only a fraction is (Manica, 2002).

The behavior is adaptive because a guarding parent is itself an organism with an energy budget. Tending a nest is costly, and a starving parent that abandons or eats the brood to survive may do better, in lifetime offspring, than one that guards a doomed clutch to exhaustion. Hope Klug and Michael Bonsall built this into a formal model placing filial cannibalism on a continuum with parental care and offspring abandonment: eating some offspring can raise the survival or growth prospects of those that remain, so partial consumption of a brood is frequently the optimal strategy rather than a failure of care (Klug & Bonsall, 2007). The worked example below makes that optimum explicit.

Why a parent should ever eat the offspring it has invested in is answered by the same currency throughout: lifetime reproductive success, not the survival of any single brood. Klug and Bonsall later showed that the behavior coevolves with the traits of the offspring and the parent together, so the level of filial cannibalism a species shows is a joint evolutionary outcome rather than a fixed parental vice (Klug & Bonsall, 2019). Related work on unconventional forms of care reinforces the point that eating offspring and caring for them are two ends of one strategic continuum, not opposites (Davenport et al., 2019).

Cannibalistic Polymorphism

In some amphibians, cannibalism is not merely a behavior but a developmental path that reshapes the body. Larvae of certain spadefoot toads and salamanders develop into a distinct cannibal morph—wider-mouthed, larger, with a modified gut—specialized for eating other larvae, alongside the ordinary omnivore morph that eats detritus. David Pfennig's work on spadefoot toad tadpoles showed that this polyphenism is conditional, induced by diet and crowding, and that it is also kin-sensitive: cannibal-morph tadpoles preferentially avoid eating close relatives and preferentially consume unrelated individuals (Pfennig, 1992).

This kin discrimination is the signature of an adaptive strategy under the constraint of inclusive fitness. Eating a sibling destroys shared genes, so a cannibal that can tell kin from non-kin and bias its diet accordingly captures the nutritional benefit while minimizing the genetic cost. Eric Hoffman and Pfennig demonstrated the proximate side of the same system in larval tiger salamanders, identifying the ecological cues—density and the size distribution of poolmates—that trigger the switch to the cannibal morph (Hoffman & Pfennig, 1999). The morph is produced when conditions make cannibalism pay and suppressed when they do not, a developmental decision matching the behavioral one seen elsewhere.

Density Dependence and Population Regulation

Because the rate of cannibalism rises with the density of conspecifics, the behavior feeds back on the size of the whole population. When a population is crowded, encounters between potential cannibals and victims are frequent and food is scarce—exactly Polis's conditions—so cannibalism intensifies, cutting numbers; when the population thins, cannibalism relaxes, letting it recover (Polis, 1981). This makes cannibalism a density-dependent regulator, a negative feedback that can stabilize a population and damp the boom-and-bust cycles it would otherwise show.

The strength of that feedback depends on precisely how the cannibalism rate scales with density, which is not fixed. Jay Rosenheim and Sebastian Schreiber showed that the density-dependent expression of cannibalism can follow several distinct pathways—through changes in encounter rate, in the fraction of encounters that end in an attack, or in the vulnerability of the victims—and that these pathways imply very different consequences for how strongly cannibalism regulates the population (Rosenheim & Schreiber, 2022). The behavior is thus not only a driver of population dynamics but one whose regulatory effect has to be measured, not assumed.

Worked Example

Partial filial cannibalism looks self-defeating—why eat the very offspring the parent is guarding?—until the parent's budget is written down. Consider a guarding parent with a clutch of N = 100 eggs. Left alone, each egg survives to independence with a baseline probability s₀ = 0.2. But the parent can eat k of its own eggs, and doing so improves conditions for the rest—more oxygen, more room, more energy for the parent to keep guarding—so that each surviving egg's probability rises linearly, s(k) = s₀ + a·k, with a = 0.004.

The number of offspring that survive is the number left times their improved survival:

W(k) = (N − k)(s₀ + a·k).

Maximizing over k gives k\ = N / 2 − s₀ / (2a*) = 50 − 25 = 25. Eating 25 of the 100 eggs is optimal. Put in numbers:

The parent that eats none raises 20 offspring; the parent that eats the optimal 25 raises 22.5—more surviving young from a smaller clutch, because each remaining egg is worth more. Eat too many and the gain reverses: at k = 50 the improved survival no longer compensates for the eggs destroyed, and the yield falls back to 20. The optimum is partial, which is exactly the pattern the fish data show: guarding parents typically eat a fraction of the brood, not all of it and not none (Klug & Bonsall, 2007; Manica, 2002). The FilialCannibalismDemo above lets the reader move the survival-benefit a and the clutch size N and watch the optimal fraction and the surviving-offspring curve shift.

Discussion

The forms of cannibalism share one logic: an animal eats a conspecific when the gain exceeds the cost, and the gain and cost are always measured in lifetime fitness. The sexually cannibalized male and the filially cannibalized egg look like victims of a malfunction, but each is accounted for by the same budget that explains ordinary predation—the eater converts a conspecific into energy or eggs, and the behavior is favored whenever that conversion pays more than the alternative. Polis's reframing of cannibalism as intraspecific predation is what unifies the cases, because it puts them all under one cost–benefit account rather than treating each as a separate oddity (Polis, 1981; Elgar & Crespi, 1992).

The costs are what give cannibalism its characteristic structure. Because eating kin destroys shared genes, kin recognition shapes who is eaten; because a conspecific is the ideal host for one's own parasites, disease risk limits when it is worthwhile; and because the prey can fight back, a size advantage is usually required. These constraints are why cannibalism is conditional—switched on by scarcity, crowding, and size structure, and switched off otherwise—and why it rarely runs to the extreme of a population eating itself to extinction (Fox, 1975; Pfennig, 1992).

Cognitive Implications

For cognitive science, cannibalism is a compact case study in decision-making under conflicting incentives and in the recognition problems that decision requires. The clearest is kin recognition: a cannibal that discriminates relatives from strangers and biases its diet accordingly is solving a classification problem whose payoff is inclusive fitness, and the spadefoot and salamander systems show that developing larvae can and do make exactly this discrimination (Pfennig, 1992; Hoffman & Pfennig, 1999). The behavior therefore implies a sensory and cognitive machinery for telling kin apart—by odor, by familiarity, or by phenotype matching—that is interesting well beyond the act of eating.

Sexual cannibalism poses a decision problem of a different kind: the male must assess a female's state and time his approach to transfer sperm before, not after, an attack, while the female weighs a meal against a mate. This is a genuine strategic interaction, with each party's best move depending on the other's, and it is why the behavior is modeled with the tools of sexual conflict and game theory rather than as a reflex (Schneider, 2014; Roggenbuck et al., 2011). In each form, reading cannibalism as a decision rather than a malfunction is what reveals the information-processing problem the animal is solving.

Current Directions

A central current question is how, mechanistically, the rate of cannibalism scales with population density, because the answer determines how strongly the behavior regulates a population. Rosenheim and Schreiber's recent analysis separates the distinct pathways—encounter rate, attack probability, victim vulnerability—through which density-dependence can arise, and argues that lumping them together has obscured real differences in their population consequences; measuring which pathway operates in a given system is now the empirical task (Rosenheim & Schreiber, 2022).

On the evolutionary side, the active work treats cannibalism not as a fixed trait but as one that coevolves with the traits around it. Klug and Bonsall's coevolutionary models show the level of filial cannibalism emerging jointly with offspring and parental traits, so predicting a species' behavior means modeling the whole system rather than the parent alone (Klug & Bonsall, 2019). In sexual cannibalism, the open question remains adjudicating among the competing adaptive and non-adaptive explanations—nutritional benefit, mate choice, and aggressive spillover—which make different, testable predictions about when females should attack (Zuk, 2016).

Common Misconceptions

Cannibalism is a rare pathology or a sign of stress.

Across the animal kingdom cannibalism is common and, under the right conditions, adaptive. It is a normal feeding option that appears predictably when food is scarce, conspecifics are dense, and a size advantage makes it safe—not a breakdown of normal behavior (Fox, 1975; Polis, 1981).

A female that eats her mate is simply being aggressive.

Sexual cannibalism is better understood as sexual conflict with several possible payoffs: a meal converted into eggs, the rejection of a low-quality male, or aggression carried over from foraging. The male is not merely a victim—if eaten after sperm transfer he may still father the resulting young (Schneider, 2014; Lawrence, 1992).

A parent eating its own offspring is a failure of parental care.

Filial cannibalism is often adaptive care, not its opposite. A parent that eats part of a brood can raise the survival of the rest and preserve its own capacity to reproduce again, so partial consumption of a clutch is frequently the strategy that maximizes lifetime offspring (Manica, 2002; Klug & Bonsall, 2007).

Cannibals will eat any conspecific indiscriminately.

Many cannibals discriminate kin from non-kin and bias their attacks toward the unrelated, because eating a relative destroys shared genes. Cannibal-morph spadefoot tadpoles preferentially spare their siblings, evidence that the behavior is constrained by inclusive fitness rather than blind (Pfennig, 1992).

Glossary

Cannibal morph.

A distinct developmental form—wider-mouthed and larger, with a modified gut—that some amphibian larvae develop, specialized for eating other larvae.

Cannibalism.

The killing and eating of an individual of the same species; a special case of intraspecific predation and, in MeSH, a feeding behavior.

Cannibalistic polymorphism.

The existence within one species of a cannibal morph alongside an ordinary morph, with development of the cannibal form induced by cues such as diet and crowding.

Density dependence.

A relationship in which a process changes in intensity with population density; cannibalism is density-dependent because its rate rises as conspecifics become crowded.

Filial cannibalism.

The consumption by a parent of its own offspring, commonly eggs in nest-guarding fish; may be partial or total.

Inclusive fitness.

An individual's total genetic contribution, counting the reproduction of relatives; the reason eating kin is costly and is often suppressed by kin recognition.

Intraspecific predation.

Predation in which predator and prey belong to the same species; the framework, due to Polis, that treats cannibalism as a normal predator–prey interaction.

Kin recognition.

The ability to distinguish relatives from non-relatives, allowing a cannibal to bias its attacks toward unrelated individuals and spare its kin.

Partial filial cannibalism.

The consumption by a guarding parent of only a fraction of its brood, frequently the fitness-maximizing option because it raises the survival of the remainder.

Polyphenism.

The production of two or more discrete phenotypes from one genotype according to environmental cues; the cannibal and omnivore morphs of spadefoot tadpoles are an example.

Sexual cannibalism.

The killing and eating of a mate, usually the male by the female, before, during, or after mating; a textbook case of sexual conflict.

Sexual conflict.

A divergence between the evolutionary interests of the two sexes; in sexual cannibalism the male's interest in surviving collides with the female's nutritional gain.

Size structure.

The distribution of body sizes in a population; a wide size structure favors cannibalism by letting larger individuals safely subdue smaller ones.

Total filial cannibalism.

The consumption by a parent of its entire brood, typically when the clutch is too small or the parent too energetically stressed for guarding to pay.

Key Researchers

Laurel R. Fox

(University of California, Santa Cruz). Ecologist whose 1975 Annual Review was the first systematic treatment of cannibalism in natural populations. ORCID

Hope Klug

(University of Tennessee at Chattanooga). Evolutionary biologist whose models place filial cannibalism on a continuum with parental care and offspring abandonment. ORCID

Andrea Manica

(University of Cambridge). Evolutionary ecologist whose 2002 review is the standard account of filial cannibalism in teleost fish. ORCID

David W. Pfennig

(University of North Carolina at Chapel Hill). Evolutionary biologist whose spadefoot-toad and tiger-salamander work established cannibalistic polyphenism as a conditional, kin-sensitive developmental strategy. Faculty profile

Gary A. Polis

(University of California, Davis; 1946–2000). Desert ecologist whose 1981 synthesis reframed cannibalism as intraspecific predation and established it as a major, dynamically important ecological interaction.

Jutta M. Schneider

(University of Hamburg). Behavioral ecologist whose work on spiders frames sexual cannibalism as a manifestation of sexual conflict. ORCID

Frequently Asked Questions

What is cannibalism in animals?

It is the killing and eating of a member of one's own species. The Medical Subject Headings vocabulary classifies it as a feeding behavior, a form of animal behavior, because across a great many taxa it is a normal, adaptive dietary option that appears under predictable ecological conditions rather than a pathology (Fox, 1975; Polis, 1981).

Why would an animal eat its own kind?

Because under the right conditions the gain exceeds the cost. A conspecific is a high-quality meal and a rival removed, and cannibalism is favored when alternative food is scarce, conspecifics are dense, and a size advantage makes the act safe. It recedes when those conditions do not hold (Polis, 1981; Elgar & Crespi, 1992).

Why do female praying mantises and spiders eat their mates?

Sexual cannibalism is a case of sexual conflict with several possible payoffs for the female—a meal converted into eggs, the rejection of a poor male, or aggression spilling over from foraging. The male's fate turns on timing: eaten before copulation he loses everything, but eaten after transferring sperm he may still father offspring (Schneider, 2014; Lawrence, 1992).

Is it normal for a parent to eat its own offspring?

In many nest-guarding fish it is, and it is often adaptive. Eating part of a brood can raise the survival of the rest and preserve the parent's capacity to breed again, so partial filial cannibalism frequently maximizes lifetime offspring rather than signaling a failure of care (Manica, 2002; Klug & Bonsall, 2007).

Why do parents eat only some of their offspring, not all?

Because the fitness payoff is usually maximized at a partial fraction: eating some eggs improves conditions for the rest, but eating too many destroys more than the improvement is worth. The optimum is an intermediate number, which is why guarding fish typically consume a portion of the clutch (Klug & Bonsall, 2007; Manica, 2002).

What is a cannibal morph?

In some spadefoot toads and salamanders, larvae can develop into a distinct cannibal form—wider-mouthed, larger, with a modified gut—specialized for eating other larvae. The form is induced by cues such as crowding and diet, and cannibal-morph larvae preferentially spare their close kin (Pfennig, 1992; Hoffman & Pfennig, 1999).

Do cannibals avoid eating their relatives?

Often, yes. Because eating a relative destroys shared genes, many cannibals use kin recognition to bias their attacks toward unrelated individuals. Cannibal-morph spadefoot tadpoles, for example, preferentially consume non-kin and spare their siblings (Pfennig, 1992).

How does cannibalism affect population size?

Because its rate rises with crowding, cannibalism is a density-dependent brake: it intensifies when a population is dense and relaxes when it thins, which can stabilize numbers and damp fluctuations. How strongly it regulates depends on the precise pathway linking density to the cannibalism rate (Polis, 1981; Rosenheim & Schreiber, 2022).

References