How Vampire Bats Share Blood: The Strange Reciprocal Altruism of Desmodus rotundus

The vampire bat lives on a diet of blood, which is a remarkably bad caloric strategy. A successful nightly foraging trip yields about 20 milliliters of blood; the bat needs roughly that much per night to survive, and it cannot store much. A vampire bat that goes 60 hours without a meal starves to death, and on any given night, perhaps 30 percent of foraging trips fail. The species should have gone extinct from straightforward energetic arithmetic, except that vampire bats found a workaround: they feed each other.

The behavior was first documented in detail by Gerald Wilkinson at the University of Maryland in a 1984 Nature paper, "Reciprocal food sharing in the vampire bat." Wilkinson's observations of free-ranging Desmodus rotundus in Costa Rica revealed that bats that failed to feed on a given night were often regurgitated blood by roost-mates, and that the regurgitation pattern was not random. Bats that had received food on a previous night were more likely to give food when the donor was the one who needed it. The behavior looked, even on first inspection, like classical reciprocal altruism: a cost-of-now for benefit-of-later, contingent on the other party returning the favor.

Why reciprocal altruism is hard

The textbook problem with reciprocal altruism is that it should not evolve. A strategy of helping others is invadable by a strategy of accepting help without giving it back; the accepters get the benefit of receiving without the cost of giving, and out-reproduce the givers. Robert Trivers's 1971 theoretical paper laid out the conditions under which reciprocal altruism could be stable: the same individuals must interact repeatedly, the cost of helping must be small relative to the benefit of being helped, and there must be a mechanism for the helpers to identify cheaters and stop helping them. The conditions are restrictive, and through the 1970s and 1980s, decisive examples of reciprocal altruism in wild mammals were rare.

Vampire bats turned out to meet all three conditions unusually well. They live in stable roosts of 20-50 individuals that persist for years, so the same bats interact night after night. The cost of regurgitating 20 milliliters of blood is small for a well-fed bat (energetically equivalent to a few hours of foraging) compared to the benefit for a hungry bat (the difference between life and death). And the bats appear to track who fed them, both through direct memory and through ongoing social grooming that probably encodes reciprocity history in tangible behavioral signals.

The mechanism of recognition

The recognition mechanism is the hardest part of the reciprocal altruism story, and the one that took the longest to pin down. How does a bat know which roost-mate fed it last time? The candidate mechanisms include individual recognition (the bat remembers individuals), kin selection (related bats help each other regardless of reciprocity), and indirect reciprocity (helping bats that have helped third parties).

Wilkinson's original analysis suggested that recognition was based partly on kinship and partly on reciprocity-of-individuals. Subsequent work by Gerald Carter and others (especially at Smithsonian Tropical Research Institute and Ohio State) has refined the picture. Carter's 2013 Proceedings of the Royal Society B paper used controlled food deprivation experiments to show that vampire bats preferentially fed individuals who had fed them previously, even controlling for kinship. Reciprocity is a real signal independent of relatedness.

The recognition appears to use multiple modalities: vocal calls (each bat has an individual signature), olfactory signatures (urine-marking and skin chemistry), and the social-grooming network that bats maintain through long sessions of mutual grooming. Bats that share a strong grooming relationship are more likely to share food, and grooming relationships are built over weeks to months of repeated interaction.

The friendship hypothesis

The 2015 and onward work by Carter and Wilkinson takes the analysis further: the relationships between specific pairs of bats in a roost are stable, asymmetric, and individualized in ways that look more like the structures of human friendship than the simple tit-for-tat that early theoretical models proposed. Pairs of bats develop strong preferential bonds that persist for years; food sharing within these bonded pairs is more reliable than between roost-mates without strong bonds; and the bonds appear to be built up through ongoing grooming investment that creates the social capital that gets called on during food sharing.

The 2020 Carter, Schino, Farine paper in Current Biology used social-network analysis of grooming and food-sharing across multiple captive roosts to show that the network has clustered structure: subsets of bats are highly interconnected with each other and have weaker connections with the rest of the roost. The clusters are roughly stable across years. Food sharing happens preferentially within clusters, suggesting that the bats have invested in specific long-term relationships rather than maintaining a uniform reciprocity policy with all roost-mates.

The captive-versus-wild distinction matters. Most of the detailed work has been in captive roosts where individual identification and food manipulation are tractable; the 2017 Carter et al PNAS paper extended the relationship-tracking work to wild roosts using PIT tags and automated tracking, and found that the captive patterns largely held in the wild, with some additional complexity from the larger and more fluid roost structures.

The genetic and neural substrate

The molecular biology of vampire bat social behavior is now a small but active field. The 2018 Zepeda Mendoza et al Nature Ecology and Evolution paper sequenced the Desmodus rotundus genome and identified gene expression changes in the brain associated with the social-bond formation period (weeks of repeated interaction between bats that become bonded pairs). Oxytocin and vasopressin pathway genes show patterns consistent with mammalian pair-bonding research in voles and primates, suggesting that the neuroendocrine substrate for vampire bat friendship is recognizably the same machinery that produces pair bonds and parent-offspring attachment in other mammals, repurposed for non-reproductive social bonding.

The brain regions implicated include the medial preoptic area and the nucleus accumbens, both of which are heavily involved in mammalian social behavior generally. The cognitive demands of the vampire bat social system (tracking dozens of individuals across years, maintaining differentiated relationships, computing reciprocity histories) are substantial for a bat brain, and may explain why vampire bats have relatively large brains for their body size compared to other bat species.

The wider significance

The vampire bat case matters for the theory of cooperation because it is one of the cleanest documented examples of reciprocal altruism in any wild mammal, and because the detailed mechanisms (relationship-tracking, partner-choice, asymmetric bond strength) anticipated by abstract theoretical models turn out to be present in observable form. The original Trivers theory required relationship-tracking and cheater-detection as preconditions; vampire bats show that the preconditions are realized through specific behavioral and neural mechanisms that are tractable to study.

The pattern probably extends to other species. Reciprocal altruism has been observed in chimpanzees, capuchin monkeys, and some other primates; the mechanisms are similar. Recent work suggests reciprocal behavior in some corvids, in some elephants, and even in some fish (the cleaner-wrasse pattern includes reciprocity-of-clients dynamics). The vampire bat case is the most-thoroughly characterized in a mammal at present, and it has influenced how researchers approach cooperation in other species.

The conservation angle is straightforward: vampire bats are widely persecuted because their feeding on cattle transmits rabies and produces livestock losses, and culling programs aimed at vampire bats often kill other bat species that share their roosts. The social structure of vampire bat roosts means that culling has secondary effects on the surviving bats, who lose social partners and may have reduced ability to weather food shortages. The species is not currently endangered, but the social-ecology dynamics are relevant to conservation policy in ways that simple population counts do not capture.

The deeper observation

The deeper observation is that cognitive and social capabilities long associated with primates and great apes turn out to be more widely distributed across mammals than the canonical hierarchy suggests. Friendship-like long-term differentiated social bonds, reciprocity tracking, and individualized relationships across years are present in a small flying mammal with a brain the size of a peanut. The pattern is consistent with what other comparative work in elephant cognition and corvid cognition and cetacean cognition keeps finding: the capabilities are old, widespread, and built on neural machinery that mammals share. The interesting questions are now less about whether other species have these capabilities and more about exactly how the implementations differ across lineages with very different brain architectures.