TREE vol. 15, no. 9 September 2000 0169-5347/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0169-5347(00)01923-6 3 4 9
K
in recognition is differentialtreat-ment of conspecifics as a function of their genetic relatedness. In nature, the widespread occurrence of nepotism (assisting kin) and the rarity of close inbreeding testify to the ubiquity of kin recognition, thus raising intriguing ques-tions about the ontogeny and mecha-nisms of discrimination. The sources of recognition cues and the fitness conse-quences of kin recognition have been foci of intense research interest and debate for three decades (reviewed in Ref. 1). A new report by Mateo and Johnston2 pro-vides evidence for the most controver-sial and difficult-to-demonstrate mecha-nism: self-referent phenotype matching. The investigators cross-fostered newborn female golden hamsters (Mesocricetus auratus) into unrelated litters and found that at maturity they responded differ-ently to odors of unfamiliar conspecifics depending on whether they were geneti-cally related. Apparently, these rodents use their own smell as a standard against which to compare novel odors.
There are two general categories of kin recognition mechanisms3. On the one hand, individuals might behave nepotis-tically toward, or avoid mating with, any conspecific they encounter in a location that predictably contains only relatives (e.g. a nest or burrow) – such site-spe-cific behavior results in ‘indirect’ recog-nition. On the other hand, individuals might recognize relatives ‘directly’ by comparing their phenotypes with mental images (‘templates’) of kin formed during previous social encounters. For example, most birds and mammals learn character-istics of kin during early development, based on physical features of parents and siblings. Later on, they recognize individ-uals who resemble the remembered tem-plate closely enough as relatives. This works because parental care is obligatory in all mammals and most birds, because young are typically reared in groups, and because there is a correlation between phenotypic similarity and genotypic simi-larity1. Therefore, phenotype matching mediated by social learning reliably results in kin discrimination.
However, in some circumstances, social learning predictably yields inade-quate or misleading recognition tem-plates (Box 1). For example, obligate brood parasites are reared by het-erospecifics, thus learning phenotypes of nestmates or foster parents would result
in species misidentification. Within species, multiple mating by females with different males potentially creates broods of mixed paternity (i.e. full- and half-siblings reared together) and multi-ple mating by males with different females might result in half-siblings that are reared apart (in different nests or dens). In these cases, individuals could conceivably discriminate conspecifics from heterospecifics or close kin from distant kin if they inspect and learn salient aspects of their own phenotype, and then match features of littermates or unfamiliar conspecifics to themselves.
Circumstantial evidence of self-referent phenotype matching, which Dawkins5 called the ‘armpit effect’, exists for species satisfying several of these condi-tions; for example, multiple paternity (Belding’s ground squirrels4 and honey bees6) and dispersal of juveniles (pea-cocks7). However, in none of these cases was self-inspection proven experimen-tally nor social learning eliminated. More-over, the existence of self-inspection as a mechanism facilitating nepotism has been doubted on the grounds that genes promoting such abilities are ‘outlaws’ if their selfish interests conflict with those of the rest of the genome8. Presumably, the resulting intragenomic conflict would favor genes that suppress self-inspection. Indeed, under this hypothesis, genetic conflict should characterize and destroy not only self-referencing but also pheno-type matching based on social learning of cues from any relatives. Alternatively, however, if genes encoding production and perception of self-recognition cues were well spread among chromosomes, the entire genome would benefit from
their nepotism-promoting effects9, thus maintaining phenotype matching, in-cluding the armpit effect. Mateo and Johnston’s new paper provides much needed empirical evidence regarding this controversy.
Golden hamsters are native to arid regions of Syria. Although little is known about their social behavior in nature, the home ranges of several males overlap with a female’s range. In captivity, females will mate with multiple males, thus resulting in mixed paternity of lit-ters2. Therefore, M. auratus apparently satisfy several conditions favoring self-inspection (Box 1). In addition, despite decades of inbreeding, females are known to discriminate between flank-gland odors of close and distant rela-tives, and between unfamiliar kin and familiar non-kin10. However, these behav-iors are consistent with both self-referent phenotype matching and social learning.
To disentangle the two mechanisms, Mateo and Johnston manipulated litter sizes and composition of simultaneously parturient mothers. They cross-fostered 18 female pups within 12 hours of birth, allowing each lactating female to rear three young: a biological son and daugh-ter, and one foster daughter. When cross-fostered females were sexually mature (41–61 days), their responses to glass plates smeared with flank-gland odors of various conspecifics were quantified. The use of flank-gland odors minimized the possibility of early social learning (in uteroor ,12 hr after birth) of siblings’ odors, because hamsters’ flank glands do not begin secreting until pups are .one month old.
Cross-fostered females approached plates containing odors of unfamiliar nonsiblings of both sexes significantly more quickly than odors of unfamiliar sib-lings (separated since birth) and of famil-iar nonsiblings (e.g. foster littermates that were reared together) (Fig. 1). Discrimi-nation between odors of unfamiliar
NEWS & COMMENT
The armpit effect in hamster kin
recognition
Box 1. Circumstances favoring self-referent phenotype matching
In parental species there are several ecological and social circumstances in which social associa-tions are predictably inadequate or misleading indicators of relatedness, thus favoring self-referent phenotype matching to enhance nepotism and preclude inbreeding (after Ref. 4).
Multiple mating
• Full-siblings and maternal half-siblings are reared together • Paternal half-siblings are reared apart
Interbrood aggregation
• Communal and cooperative breeding • Posthatch brood amalgamation • Adoption and kidnapping
Parasitism
• Intraspecific brood parasitism • Interspecific brood parasitism
Dispersal
3 5 0 TREE vol. 15, no. 9 September 2000 siblings and nonsiblings was confirmed
by observations of other behaviors of cross-fostered females toward the plates, such as duration of investigation and frequency of flank-marking. Because the only social referents available to these females during development were nonrelatives, they apparently compared smells of unfamiliar conspecifics with their own flank-gland odors.
Interestingly, experimental females must also have incorporated odors from their foster family into their recognition template, because they investigated and marked plates containing flank-gland secretions of familiar nonsiblings about as often as plates with secretions of unfamil-iar siblings. To assess the relative impor-tance of these cue sources, Mateo and Johnston compared females’ behavior toward secretions of their own unfamiliar sisters and unfamiliar sisters of their foster littermates. The hamsters investigated the odors of unfamiliar biological sisters sig-nificantly longer, thus suggesting that a female’s own odors were weighted differ-entially in her kin recognition template than those of her foster family.
Mateo and Johnston’s results consti-tute the best experimental evidence of self-referent phenotype matching. How-ever, the data leave open the question of whether a hamster’s recognition tem-plate is learned (for example, by memo-rizing its own flank-gland odors) or is genetically determined (for example, owing to restrictive architectural design of perceptual filters11 in its olfactory
processing system). Distinguishing these possibilities requires attempting to ‘fool’ individuals by experimentally manipulat-ing their phenotype and causmanipulat-ing pre-dictable recognition errors. Under the self-learning hypothesis, if a hamster’s own odor were altered experimentally (e.g. by changing its diet12), it should treat similar smelling but unrelated, unfamiliar conspecifics like unfamiliar relatives.
Mateo and Johnston’s results do not directly address the function of kin recog-nition in nature. Captive hamsters flank mark in both territorial and mate-choice contexts, thus recognition might function to either minimize incest, maximize nepotism or both. Theoretically, self-referencing would be the more effective mechanism in forming templates used in nepotistic contexts where individuals are attempting to discriminate close from dis-tant kin, whereas social learning would be more effective in forming templates asso-ciated with mate choice, when additional referents (e.g. parents and siblings) are necessary to identify everyone to whom an individual is related2,13. Thus, evidence of self-inspection in hamsters suggests that kin recognition probably facilitates nepotism (e.g. female cooperation in ter-ritory defense, communal nesting or food sharing).
In nature, self-referent phenotype matching might enable female hamsters to discriminate full- from half-sisters among nestmates and paternal half-sisters versus non-kin upon emergence from the natal burrow, as in Belding’s ground squirrels4,14. Alternatively, the discriminations might be lab artifacts – errors in individuals’ recognition of their own odor, an ability which normally func-tions in some purely selfish context (e.g. maintenance of an exclusive territory). Under this interpretation, what appear to be fine-scale discriminations actually result from increasing errors with increasing relatedness (owing to greater odor similarity among closer kin; see Ref. 8). This raises the question of whether M. auratus frequently interact with maternal and paternal half-sisters in nature. If so, persistence of recognition mistakes is unlikely because, over time, selection would eliminate fitness-reduc-ing errors; for example, by sharpenfitness-reduc-ing the recognition template (through inclu-sion of more chemical cues) or adjusting the rejection threshold more conserva-tively1. The importance of evaluating the context and function of hamster kin dis-crimination in nature is re-emphasized2.
In summary, Mateo and Johnston’s results clearly demonstrate the feasibil-ity of self-referent phenotype matching. They also yield specific hypotheses about the social behavior of golden
hamsters for testing in the field. This work, along with other studies implicat-ing self-referencimplicat-ing4,6,7,15, should galva-nize the search for evidence of this intriguing but controversial kin-recogni-tion mechanism among species in which social learning predictably is inadequate or misleading as a guide to relatedness.
Acknowledgements
We are grateful to M.C.B. Andrade, R.E. Johnston, J.M. Mateo, H.K. Reeve and P.T. Starks for discussions. Financial support was provided by a Howard Hughes Medical Institute Predoctoral Fellowship, the National Science Foundation and Cornell University.
Mark E. Hauber Paul W. Sherman
Dept of Neurobiology and Behavior, Cornell University, Ithaca, NY 14853-2702, USA ([email protected]; [email protected])
References
1 Sherman, P.W. et al. (1997) Recognition systems.
In Behavioural Ecology (Krebs, J.R. and Davies, N.B., eds), pp. 69–96, Blackwell Science 2 Mateo, J.M. and Johnston, R.E. (2000) Kin
recognition and the ‘armpit effect:’ evidence of self-referent phenotype matching. Proc. R. Soc. London Ser. B 267, 695–700
3 Waldman, B. et al. (1988) Problems of kin recognition. Trends Ecol. Evol. 3, 8–13 4 Holmes, W.G. and Sherman, P.W. (1982) The
ontogeny of kin recognition in two species of ground squirrels. Am. Zool. 22, 491–517 5 Dawkins, R. (1982) The Extended Phenotype,
Freeman
6 Getz, W.M. and Smith, K.B. (1986) Honey bee
kin recognition: learning self and nestmate phenotypes. Anim. Behav. 34, 1617–1626 7 Petrie, M. et al. (1999) Peacocks lek with
relatives even in the absence of social and environmental cues. Nature 401, 155–157 8 Alexander, R.D. (1990) Epigenetic rules and
Darwinian algorithms: the adaptive study of learning and development. Ethol. Sociobiol. 11, 241–303
9 Hamilton, W.D. (1987) Discriminating nepotism: expectable, common and overlooked. In Kin Recognition in Animals (Fletcher, D.J.C. and Michener, C.D., eds), pp. 417–437, Wiley 10 Heth, G. et al. (1998) Kin recognition in golden
hamsters: evidence for phenotype matching. Anim. Behav. 56, 409–417
11 Marler, P. (1997) Three models of song learning:
evidence from behavior. J. Neurobiol. 33, 501–516 12 Schellinck, H.M. et al. (1997) Odors of individuality
originating from the major histocompatibility complex are masked by diet cues in the urine of rats. Anim. Learn. Behav. 25, 193–199
13 Penn, D. and Potts, W. (1998) MHC-disassortative
mating preferences reversed by cross-fostering. Proc. R. Soc. London Ser. B 265, 1299–1306 14 Holmes, W.G. (1986) Kin recognition by
phenotype matching in female Belding’s ground squirrels. Anim. Behav. 34, 38–47
15 Grosberg, R.K. and Quinn, J.F. (1986) The
genetic control and consequences of kin recognition by the larvae of a colonial marine invertebrate. Nature 322, 456–459
NEWS & COMMENT
Fig. 1. Mean time in seconds (6 SE)
between the introduction of a glass plate with conspecific flank-gland odors smeared on it, and the first approach and investigation by female golden hamsters (Mesocricetus auratus) that had been in-dividually cross-fostered and reared among non-kin. Stimulus odors were from familiar nonsiblings (i.e. foster littermates; black bars), unfamiliar siblings (siblings separated
,12 hrs after birth; open bars) and com-pletely unfamiliar nonsiblings (gray bars). Horizontal lines denote significant differ-ences in responses between two odor classes, based on repeated-measure ANOVAs (*p,0.05, **p,0.01). Repro-duced, with permission, from Ref. 2.
Trends in Ecology & Evolution
Odor donors 0
5 10 15 20 25
Male odors
Female odors
Mean latency to
in
vestigate (s)(
SE
)
* *
