Why and how do animals recognise their relatives?

Greg Detre

Monday, May 28, 2001

Prof. Marian Dawkins

Animal Behaviour V

Explanations of why animals recognise their relatives can be given at the level of the organism or the level of the gene. �Selfish gene theory� ties the two together. In Richard Dawkins� helpful terminology, animals, humans, plants and any other life are termed �survival machines�. This serves as a continual reminder that all life can be considered an expression of its genes. If the phenotype is not adaptive, i.e. gives rise to a body which survives long enough to bring forth lots of children who are themselves adaptive, then the genes which are being expressed will not proliferate into the population. If the genes do give rise to adaptive traits, then that organism will be more likely to survive and produce more offspring, producing more organisms with similar genes. Dawkins is advocating what Daniel Dennett might term the �intentional stance�: ascribing intentions, purpose and forward-looking mentality to decidedly non-conscious objects as a shortcut to describing purposive-seeming activity. Rather as we say that our alarm clock will �remember� to wake us up in the morning, or that the thermostat �realised� that it was getting too hot, our selfish genes �want� us to be as adaptive as possible, so that they get passed on through generations and around the population. We have evolved to recognise our relatives either because it may help us survive as organisms, or because they are sufficiently genetically similar to us that their survival (even at our expense) benefits both sets of our shared genes. This idea that behaviour towards other organisms will vary according to their relatedness was put forward most persuasively by Bill Hamilton (1964) in quite mathematical terms.

Before considering kin recognition specifically, it is worth noting that co-operation in kin groups makes sense for animals for all the same reasons that any sort of altruism or co-operation in groups makes sense. Axelrod & Hamilton�s (1984) theoretical work on the Prisoners� Dilemma or Wilkinson�s (1984) observations of regurgitation of blood by vampire bats lend credence to our strong intuition that it may well be in our own individual best interests to co-operate, or even act in altruistic-seeming ways. Kin selection is just one of four explanations for the evolution of co-operation, the others being mutualism, manipulation and reciprocity. Mutualism is where all co-operating individuals receive an immediate benefit from co-operation, e.g. lionesses are far more effective hunting zebras together than alone (Caraco & Wolf 1975). Manipulation is where one animal deceives the other into benefitting it unwillingly and/or unknowingly. Reciprocation is like mutualism, but spread out over time or across numerous encounters, as in the Prisoners� Dilemma model.

The index of relatedness is a simple measure of how genetically similar two animals are likely to be, based on how the genes are passed from one generation to another, and the number of generations between them. Identical twins have a relatedness of 1. In the case of humans, the index of relatedness between a son and a father is 0.5, since 50% of the son�s genome is taken from the father�s genome (mutations can be ignored for our purposes), and there is only one generation separating them. Similarly, the index of relatedness between two siblings is also 0.5, since two siblings� genomes are derived from the same parental genetic pool, but may contain anywhere between 0 and 100% identical genes. On average though, 50% of their genes will be shared. It is worth noting that parent-child relations are asymmetrical because the child is likely to produce more offspring over its remaining lifetime than its parent, as well as being less able to fend for itself.

As the number of generations (the total �generation distance� separating the two individuals from their most recent common ancestor(s)) increases, the probable genetic similarity between individuals halves. I am only likely to share �, i.e. (�)², of my grandfather�s genes, or 1/8 of my first cousin�s. From the selfish gene�s point of view, the survival of one genetically-similar organism is as good as another.

The situation is slightly different for non-diploid species. Human beings and most mammals are diploid, that is, their chromosomes are ordered in pairs, and offspring derive their genome equally from both parents. Some members of the order Hymenoptera, for instance, derive their genetic code solely from the female side. Bees provide the best example of this, and it is one of the reasons that their eusocial hive system is so well-developed. �Eusociality� describes an advanced level of social organization, in which a single female or caste produces the offspring and non-reproductive individuals cooperate in caring for the young. They illustrate the �caring� vs �bearing� distinction, in that the bearer of a child need not necessarily be the person that rears, feeds and teaches it. In hives, although the queen bee bears all of the bees in a given generation, all the sister bees in the hive care for them, since the genetic similarity between the bearer and the carers is so high.

The mole-rats provide the first known example of a mammalian eusocial society. Jarvis (1994) considers two species of mole-rat, the naked and the Damaraland mole-rats. Being able to compare the two species is helpful, because we are better able to exclude �red herring� factors like the naked mole-rats� glabrousness, poikilothermy and extreme inbreeding. Although diploid, both species live in small colonies composed of close kin, with one breeding female and one to three mates. Living in groups allows them to survive the long droughts, finding enough food for the entire colony when it rains enough for them to extend their burrows. Like other cases of mutual reciprocity, the probability of the co-operating colony surviving over time is far higher than the chances of any individual mole-rat trying to find its own food in the limited time-window, defending the burrow against predators and being able to dig a large enough burrow.

However, there is a further reason why we might not expect siblings to treat each other with the same care that parents treat children in most real world situations: certainty. A mother knows for a fact that an infant is her child because she gave birth to it. However, a sibling does not know for certain that another infant is genetically related. The other infant could have fallen into the nest at hatching time, the eggs could have been swapped, or it might be a half-brother or half-sister. In the same way, a father cannot ever be as certain as his mate that an infant is his, despite his best efforts to fend off other males.

Animals are able to recognise each other in a number of ways. Recognising relatives is even more difficult, since it not only requires differentiating between and remembering different individuals, but also assessing their relatedness (sometimes in cases where they�ve never even seen the individual before).

The simplest and most obvious method is simply to label inside the nest or home as kin. In most cases, this method works well. However, other animals have devised various means of exploiting such na� criteria. The cuckoo is the most well-known example, depositing its eggs inside other birds� nests to be cared for unwittingly as if part of the family. Over evolutionary time, victims of such deception like the reed warbler have learnt to be extremely aggressive towards adult cuckoos near their nest, and to become better and better at distinguishing cuckoo eggs from their own species. In opposition, cuckoos have evolved to lay eggs that are harder and harder to tell apart from reed warblers�.

An only slightly more sophisticated method focuses on those you grow up with. Lorenz�s famous �imprinting� experiments with young geese illustrate that they will extend the notion of kin outside the species, following the first conspicuous moving object they see after hatching, whether that object is their mother, Konrad Lorenz himself, or an oil tanker. Similarly, Holmes and Sherman (1982) showed that sibling recognition in Belding�s ground squirrels is heavily skewed by those they have been reared with. The pups were separated into four groups at birth: siblings reared together, siblings reared apart, non-siblings reared together and non-siblings reared apart. The frequency of agonistic interactions in a closed environment between two older squirrels was almost entirely dependent on whether they had been reared together. However, there was also a small relatedness effect, and in a separate experiment, year-old full sisters were less aggressive towards each other, and more co-operative, than half-sisters.

The previous two methods for guessing relatedness use contingent, non-genetic cues based on social context. Hamilton�s initial paper anticipated that phenotypic clues would play an important role. Indeed, humans rely to some extent on visible similarity between relatives, but this only really works at all for very close relatives. Some other explanation is required to explain how animals are able to recognise kin to some degree even without experience with parents or siblings.

Hamilton considered what is now jokingly known as the �green beard� effect, where a gene codes for a highly noticeable phenotypic signal, such as a green beard, as well as coding in altruism for anyone else with a green beard. Such green bearded altruists would thrive as a group, and the genes would quickly proliferate amongst kin members. Assuming that each family could lay claim to one or more such distinguishing features, it would be very easy to tell who is related to whom, and maybe even the degree of relatedness according to the shade of green. Of course, such an effect would eventually be subject to deception. For instance, if a gene were to code for a green beard without the altruism, then another non-kin member of the species could free-ride on his fellow greenbeards, and the usual animal deception cycle would result. However, the green beard effect has been largely dismissed as a serious explanation for kin recognition, since it requires a single gene to perform both of two very dissimilar functions. Though green bearded mutants might not be considered particularly unlikely, without the altruism towards other green beards, the system would not operate as a marker for kin.

One emerging candidate for identifying kin is based around the Major Histocompatibility Complex. This is the genetic region found in vertebrates best known for mediating cell-cell recognition, enabling the body to distinguish and fight off foreign proteins, usually germs and viruses, and more troublesomely for modern medicine, transplants. Containing 74 genes in humans, many with extreme variability (up to 60 alleles in some loci), the MHC� can give rise to an enormous range of genotypes. Behaviour ecologists first started to take note of the MHC in relation to kin recognition after results from the Sloan Kettering Institute started to show that mice may employ a unique phenotypic signal based on the MHC in mate preference. In mice, this discrimination is based (at least in part) on a urinary odor cue, and in humans sweat also appears to play a role in the chemical source of the signal.

Such a cue could be used in a number of ways (Lenington, 1994). It could be used to identify kin � indeed, female mice are more likely to nest communally with females that are the same at the MHC than with females that differ in MHC genes. It may be used in individual recognition, for example in �pregnancy blocking� or aborting litters when female mice are exposed to strange males who differ in their MHC from the stud male. It could also be used to avoid in-breeding, as their early results suggested � mice preferred to mate with mice that differed at the MHC rather than with mice that had the same MHC, even when genetic background was controlled (Yamazaki et al., 1976).

The role of the MHC as a means of assessing the genetic similarity between two individuals appears to outweigh the effect of the entire remaining genetic background (Brown & Eklund, 1994). As the potential ways in which it might be employed listed above show, it might be employed both in recognising individuals and quantifying overall genetic similarity (with reference to one�s own MHC phenotypic signals). However, most of the work so far has centred around mice, with some on rats and humans, but further studies are required on other species.

Hamilton�s main thesis appears correct: genes express a predisposition towards other animals on the basis of genetic similarity in a number of ways. There is no single means by which animals recognise kin, although they appear to employ a mixture of social context and phenotypic clues. Because such clues may be open to deception, certainty plays as much of a role as actual relatedness in mediating behaviour towards kin.