Anim. Behav., 1997, 53, 487–498

Colony member recognition and xenophobia in the naked mole-rat

M. J. O’RIAIN & J. U. M. JARVISDepartment of Zoology, University of Cape Town

(Received 14 February 1995; initial acceptance 1 September 1995;final acceptance 26 April 1996; MS. number: 4850ʀ)

Abstract. The ability of naked mole-rats, Heterocephalus glaber, to discriminate between familiar andunfamiliar conspecifics and their response to intruders was investigated. Odour cues used by mole-ratsin recognition contexts were identified in a three-way choice apparatus and ‘decision rules’ for acceptingor rejecting conspecifics were explored in a series of odour manipulation experiments. Naked mole-ratswere highly xenophobic, even to closely related foreign conspecifics, and a division of labour existedamongst the non-breeders in colony defence. The principal mechanism of recognition appeared to bedistinct colony odour labels, contributed by each colony member and distributed among, and learnedby, all colony members. Differences in the mixture of these odours may provide even genetically similarcolonies with a unique odour label. These odours persisted despite controlling for exogenous cues.Fitness consequences of these phenomena are interpreted with respect to the need for closely relatedneighbouring colonies to maintain autonomy and the importance of excluding foreign competitors fromwithin-colony rivalry for reproductive succession. � 1997 The Association for the Study of Animal Behaviour

Kin bias is a feature of social organization inmany species (Gadagkar 1985; Sherman &Holmes 1985; Fletcher & Michener 1986;Blaustein & Porter 1990; Hepper 1991; Pfennig &Sherman 1995) and can occur for a number ofreasons, some involving kin discrimination, othersreflecting incidental consequences of discrimi-nation at other levels (Barnard 1991). In socialorganisms bias towards close kin in the form ofcolony mates is particularly important where itserves to facilitate cooperation amongst colonymembers and to decrease the chances of socialparasitism, theft of brood and food stores orkilling of the colony (Michener 1974; Crosland1990; Fishwild & Gamboa 1992). Social insectstypically have well-developed nestmate recog-nition abilities, by which they admit nestmates to,and exclude non-nestmates from, their colony(Buckle & Greenberg 1981; Fletcher & Michener1986; Gamboa et al. 1986; Venkataraman et al.1992). In these instances discrimination may beeffected at the level of group member recognitionrather than the degree of kinship (Barnard 1991).If groups tend to be composed of relatives (e.g.

parent/offspring associations) such discrimina-tion will necessarily be correlated with geneticsimilarity. According to Grafen (1990) kin dis-crimination at this level does not represent kinrecognition (see, however, Blaustein et al. 1991;Byers & Bekoff 1991; Stuart 1991).

Evolutionary explanations for kin bias are oftenbased on the assumption that the animals dis-criminate between conspecifics according to theirgenetic relatedness (Hepper 1991). It is arguedthat it is only by responding differentially on thebasis of genetic similarity that individuals canobtain the fitness benefits as espoused byHamilton’s kinship theory (Hamilton 1964a, b).Experimental tests of this assumption and themechanisms that enable such genetic discrimina-tion are thus essential to theories on kin selection.

Colonies of the naked mole-rat, Heterocepha-lus glaber, a cooperatively breeding subterraneanrodent (Jarvis 1981), are highly xenophobic toforeign conspecifics in the laboratory (Lacey &Sherman 1991). Intriguingly, neighbouring col-onies, the most likely source of foreign con-specifics within naturally occurring populations,are also close genetic relatives (Faulkes et al.1990; Reeve et al. 1990). Such high inter-colonyrelatedness is due in part to the high viscosity(low vagility) of naked mole-rat populations,

Correspondence: M. J. O’Riain, Department ofZoology, University of Cape Town, Rondebosch 7700,South Africa (email: joriain@botzoo.uct.ac.za).

0003–3472/97/030487+12 $25.00/0/ar960299 � 1997 The Association for the Study of Animal Behaviour

487

with limited dispersal and the apparent forma-tion of new colonies by fissioning (Brett 1991).Their close physical proximity makes them themost likely competitors in any interactions whiletheir extremely high relatedness would suggest alimited potential for the use of genetic cues ininter-colony recognition. This raises interestingquestions as to the mechanisms used by nakedmole-rats to recognize colony members andreject genetically similar (i.e. siblings), butforeign, conspecifics.

Our aims in this study were to investigate (1)the ability of the naked mole-rat to discriminatebetween foreign conspecifics and resident mem-bers of a colony in a homogeneous environment,(2) how kinship and familiarity affect recognitionability, and (3) the possible sources of cues used indiscrimination and the maintenance of colonyintegrity.

METHODS

All animals used in these experiments were captiveborn and all experiments were performed in thelaboratory. Except where noted, housing andmaintenance procedures are the same as thosegiven in Jarvis (1991a).

Transfer Experiments

We investigated the behavioural responses ofresident mole-rats to a foreign mole-rat by trans-ferring individuals from one experimental colonyto another. The transferred mole-rat is referred toas the foreigner and the colony into which it istransferred as the resident colony. We chose for-eigners at random (amongst non-breeders) fromtheir colony and then introduced them into thetoilet chamber of the resident colony within 1 minof their removal. We chose the toilet chamber asthe site of introduction as it was the least fre-quented location within the resident burrow sys-tem thus enabling us to introduce the foreignerwith minimal disturbance to resident colony mem-bers. In addition, the latency time for foreignersleaving the toilet chamber was much less than forforeigners leaving either food chambers or diggingarenas. This effectively decreased the time fromintroduction to contact between foreigner andresident mole-rats, minimizing the duration ofeach trial.

We used two ‘relatedness classes’ of foreignersin all experiments to explore the effects of famili-arity versus relatedness: (1) foreign kin, whichwere mole-rats that had been removed from theirnatal colony (the resident colony) at least 6months prior to the onset of these experimentsand housed together in small groups of same-sexed individuals; and (2) foreign non-kin, whichwere mole-rats that were born and reared in acompletely separate colony to resident mole-rats.Whilst the exact degree of relatedness betweenforeign non-kin and resident mole-rats was notmeasured (i.e. by DNA fingerprinting) previousstudies (Faulkes et al. 1990; Reeve et al. 1990)have shown that intra-colony relatedness is con-sistently higher than inter-colony relatedness.

We performed 80 trials on four resident col-onies. These included 16 foreign kin trials (fourforeign kin to each of four resident colonies), 44foreign non-kin trials (11 foreign non-kin to eachof four resident colonies) and 20 control trials(five resident mole-rats to their respective residentcolonies). The latter experiment was designed tocontrol for the experimental effects of handlingand transferring mole-rats. The overall exper-imental design ensured that all resident colonieswere exposed to a similar cross-section of foreign-ers, to minimize the possible effects of variation ininter-colony recognition/aggression. In addition,the random selection of individuals within sourcecolonies ensured variation in the body mass offoreigners thus allowing us to analyse the effectsof body size on the acceptance or rejection levels.

Diet and nest-bedding were the same in allcolonies, minimizing the possible influence ofexogenous odour sources and thus allowing us toassess recognition responses in a relatively homo-geneous odour environment.

We recorded the behaviour of the transferredindividual, as well as its interactions with thecolony residents it encountered, for a maximum of10 min. Only three of the 60 trials involvingforeigners lasted longer than 30 s, and in all thesetrials the foreign mole-rat was accepted. Accept-ance or rejection of the transferred animal in eachtrial was noted on the basis of either aggressive(i.e. biting and toothfencing, sensu Lacey &Sherman 1991) or non-aggressive interactions (i.e.sniffing and ignoring) by resident colony mem-bers. Each introduced mole-rat was usually testedagainst several members of the resident colonybecause more than one animal was usually active

Animal Behaviour, 53, 3488

in the burrow system at the time of introduction.We performed control experiments by removingan individual (for less than 1 min) and thenreintroducing it to its own colony.

Although it was not possible to control forbehavioural idiosyncrasies of the introducedmole-rats, their behaviour was not observed tovary much. Foreign mole-rats typically advancedcautiously along the section of burrow that leddirectly from the toilet chamber and it was herethat all aggressive interactions between them andresident colony members were observed to takeplace.

Odour Choice Experiments

Nestmate recognition cues were investigated ina three-way choice test employing a T-maze(Fig. 1). All the non-breeders from two colonies(colonies 100 and 7400, N=45 and N=37, respect-ively) were used in this study. Each trial involvedremoving an individual from its colony and plac-ing it in a clean glass terrarium. After a 5-minadjustment period we introduced the mole-rat intothe T-maze. Each mole-rat was presented with thechoice of an odour source from its own colony, aforeign colony and a blank (a chamber containingnesting or toilet material that had not previouslybeen in contact with any mole-rats). Two colonyodour sources were used separately: the soiledlitter of the communal toilet and the soiled bed-ding material of the nest. Each test lasted 10 minduring which preference was inferred from thefrequency and duration of attempts to gain access

(by gnawing at the perforated partition) to anodour source. We tested each animal once only, tominimize the effects of familiarity and learning onthe results. We randomized the position of thedifferent odour sources relative to one another tocontrol for any directional bias amongst membersof the two colonies. The entire apparatus wascleaned, wiped down with 70% alcohol and thenrinsed with water, between each trial to removeresidual odours from previous animals.

Odour Manipulation Experiments

To investigate the relative importance of anindividual’s odour versus the colony odour innestmate recognition experiments we manipulatedthe odour of individual non-breeding mole-rats.We altered the odours of individuals by exposingthem to different odour sources. We conducted 45trials (15 per colony) using 15 individuals fromeach of three colonies. Individuals from eachcolony were subjected to three different odourenvironments, two experimental and one control.Experimental odours involved exposing individ-uals (five individuals per odour) to either a foreigncolony odour source or a blank colony odoursource for 12 h and then reintroducing them intotheir natal burrow system. We exposed a mole-ratto a foreign colony’s odour source by introducingit to a section of the burrow system of a foreigncolony that was temporarily sealed-off from itsinhabitants (to prevent physical contact). Weintroduced a mixture of soiled material from boththe nest and toilet of the foreign colony to thissealed section to ensure the animal was exposed toa diverse compliment of foreign colony odourcues. We obtained blank colony odour sources byplacing the mole-rat into a section of previouslyunused burrow provided with clean toilet (wood-shavings) and nesting material (wood-wool). Weperformed control experiments by confining ananimal to a section of its own burrow system andthus exposing it to its own colony odour sources.When we returned these animals, who had beenexposed to these three different odour environ-ments in separate trials, to their colony, werecorded the behavioural response of the residentmole-rats to these individuals. The aggression rate(and acceptance/rejection) was determined as thenumber of aggressive acts (i.e. shoving) permanipulated individual per trial.

A

B

CD

Figure 1. Odour choice apparatus. Individual mole-ratswere introduced into the apparatus at D. A, B and Crepresent chambers into which the different odoursources were introduced. Perforations at the end of eacharm enabled olfactory detection of the odour sources ineach chamber. The arms of the apparatus were allinterchangeable.

O’Riain & Jarvis: Recognition in naked mole-rats 489

Data Analyses

We recorded acceptance or rejection of foreign-ers in transfer experiments as a categorical vari-able with two states, accept or reject. Here a singleattack on the foreigner was sufficient for a trial tobe categorized as reject. We could not accuratelyrecord the aggression rate because of ethical con-siderations: to avoid any mole-rat being physicallyharmed we intervened whenever the potential forserious injury existed (see below). Differences inthe proportion of rejected versus accepted foreignkin and foreign non-kin, and males and females,were collated for all colonies and tested usingFisher’s exact test for comparing proportions (Zar1984). We explored differences in the body massof accepted versus rejected mole-rats using theMann–Whitney U-test for unpaired samples.

The reduced severity of aggressive encountersbetween interactants in odour manipulation trialsenabled us to assess the rate of aggression. Weanalysed these latter data using a multifactoralANOVA with aggression rate as the dependentvariable, odour treatment, colony and sex asfactors and the manipulated individuals’ bodymass as a covariate. To determine the effects ofan intruder on the general activity levels withinthe colony we used a paired t-test to compare thelevels of activity of all non-breeders within thecolony before the introduction and immediatelyafter the removal of a foreign individual. Weanalysed the relationship between the behaviourof resident mole-rats and their body mass usingSpearman’s rank correlation tests. We analysedall odour trials using the Friedman’s analysis ofvariance model for related samples. In all theabove analyses we used parametric tests when thedistribution of data, transformed where necessary,permitted. Otherwise, we used non-parametrictests.

Ethical Note

We performed preliminary experiments todetermine the most effective and yet least stressfulmeans of testing the proposed hypotheses. Giventhe potentially harmful design of the transferexperiments it is important to point out that weconsidered alternatives before proceeding with thefinal protocol and, once adopted, precautionswere taken to prevent the physical injury ofanimals in all trials.

We first attempted placing foreign individualsin separate containers that could be placed inolfactory but not physical contact with residentcolonies. However, the results from these trialsshowed that there was no measurable response toeither controls or foreigners. Resident mole-ratsgnawed at the perforated dividing wall for bothcontrols and foreign conspecifics. There was noevidence of threats (i.e. open mouthed gaping),and obviously no chance of mole-rats performingany other aggressive interactions in their naturalrepertoire. Furthermore, while mole-rats chosetheir own colony odours over foreign odours inthe odour choice trials they never displayed anyagonistic response to the foreign odour. Giventhese limitations we decided to adopt a moredirect method of measuring inter-colony aggres-sion, but not without ensuring that no individualswere subjected to physical harm.

Of paramount importance was our ability toterminate any experiment at the point when theforeigner and the residents had initiated theritualized act associated with conspecific aggres-sion but prior to the occurrence of any poten-tially injurious aggression. The former took theform of shoving, open mouthed gapes, incisorfencing, toothlocking and biting, typically fol-lowed by a rapid retreat by the foreign individ-ual. Here biting as defined by Lacey et al. (1991,page 237; ‘the jaws of one animal close over thebody of another animal’) does not imply that theskin is broken. By ensuring that we had immedi-ate access to interactants (by removing the roofof the burrow) at all stages during the exper-iment we were able to intervene prior to theescalation of ritualized aggression. When theroof of the burrow is removed all mole-ratsfreeze, enabling us to terminate immediately anyinteractions. We intervened in every experimentin which we witnessed an escalation in the ritual-ized form of aggression (i.e. all rejections). Inthis way no individual in any of the experimentssustained any physical injuries (i.e. the skin wasnever broken).

The apparently excessive duration of transfertrials (maximum of 10 min) was a function of thefact that foreign mole-rats, in preliminary trials,occasionally remained undetected in the residentcolony for up to 7 min. The decision to makethe total time 10 min was thus an attempt to caterfor these instances. Once olfactory/physical con-tact had been made between a foreigner and a

Animal Behaviour, 53, 3490

resident then the trial was invariably terminatedwithin 30 s.

Transferred individuals in odour manipulationtrials were typically shoved following their returnto their natal colony. Other behaviour includedbiting (see above) and toothfencing. No physicalinjuries were sustained by any individuals. Wealways monitored transferred individuals as thiswas an integral part of the experiment. We termi-nated observations once colony activity and thebehaviour of the transferee had returned to nor-mal (usually within 10 min of reintroduction). Afurther point was that mole-rats exposed toforeign odours were not observed to show anyovert signs of distress and all their exhibitedbehaviour patterns were part of their behaviouralrepertoire when in their natal burrow system.

RESULTS

Transfer Experiments

Naked mole-rats were highly xenophobic toforeign conspecifics. Resident mole-rats aggres-sively rejected foreign mole-rats (non-kin) in93.2% of all trials (Fig. 2). There was no signifi-cant effect of sex (Fisher’s exact test: Z=0.514,P>0.05) on acceptance or rejection. Similarlythere was no significant difference in the bodymass of rejected versus accepted mole-rats(Mann–Whitney U-test: U=0.102, N1=41, N2=3,P=0.978).

Resident mole-rats showed similar levels ofrejection (Fisher’s exact test: Z=0.916, P>0.2) ofboth foreign non-kin and foreign kin (Fig. 2).These results suggest that acceptance or rejectionis dependent on recent association and familiarityof odours and not genetic relatedness.

Given that naked mole-rats live in the dark,have poor visual acuity, and that introduced ani-mals were always sniffed following physical con-tact, it seems likely that discrimination wasachieved through olfaction. In all colonies dis-crimination and aggressive rejection of foreignerswas positively correlated with the body mass ofthe aggressors (Spearman rank correlation test:P<0.05, for data presented in Fig. 3), most for-eigners being rejected by a small group of highlyaggressive, physically robust individuals. Interest-ingly, despite being the principal colony defend-ers, these larger individuals were seldom the firstto encounter introduced animals (see also Lacey &

Sherman 1991). Smaller more active colony mem-bers were responsible for the majority of firstencounters, a probabilistic outcome of theirheightened activity in the burrow (Jarvis et al.1991). Upon encountering a foreigner these indi-viduals typically (92% of trials) retreated andalarm-called (described by Pepper et al. 1991).Alarm callers invariably returned directly tothe communal nest to alert colony members to thepresence of an intruder. This behaviour resulted ina significant increase (paired t-test: P<0.05 for allcolonies) in the number of mole-rats activelypatrolling the burrow system (Fig. 4), in additionto mobilizing the colony defenders.

Reproductive animals, despite being amongstthe largest individuals in the colony, were notobserved to engage in colony defence againstforeigners in three of the four study colonies (Fig.3a, c and d). In contrast, the reproductive femalein colony 7400 (Fig. 3b) was responsible forthe majority of aggressive acts associated with theousting of intruders.

Controls for all transfer experiments indicatethat introduction procedures did not significantly

Experimentalcolony(N = 4)

Foreignnon-kin(N = 44)

Resident(N = 20)

Foreignkin

(N = 16)

93.2%

100%

0%

Introducedmole-rat

Residentmole-rats

% Rejection

Figure 2. The percentage rejection of introduced mole-rats by resident colonies. The number of foreign andcontrol mole-rats for each ‘relatedness/familiarity’ classis provided in parentheses.

O’Riain & Jarvis: Recognition in naked mole-rats 491

influence the behaviour of the resident colonymembers (paired t-test: P>0.05 for all colonies).

Odour Choice Experiments

Mole-rats spent significantly more time(Friedman ANOVA: P<0.001 for all trials)attempting to access their own colony odours thanforeign colony odours. Results were consistent forboth colonies and for both experimental odoursources (Fig. 5). Mole-rats in both colonies spentmore time on average attempting to access theodour source from the nest than that of the toilet.Close olfactory contact with soiled toilet materialappeared to cause mild irritation of the mole-rats’nasal region with many individuals groomingtheir nose between efforts to access the odoursource. This behaviour reduced the total amountof time that mole-rats spent attempting to accessthis odour source, as reflected in Fig. 5c, d.

Odour Manipulation Experiments

There were no significant interaction effectsbetween odour treatment, colony or sex and con-sequently the main effects were calculated withoutthem. Furthermore body mass did not covarysignificantly (F1,44=0.005, P=0.947) and was thusexcluded from the analysis. There was a signifi-cant difference in the aggression rate betweenodour treatments (F2,44=20.46, P<0.0001) withmole-rats subjected to foreign and blank odourenvironments being shoved significantly more fre-quently than mole-rats exposed to their own col-ony odour (Fig. 6; Tukey multiple range test).There was no significant difference in the aggres-sion rate of individuals exposed to either foreignor blank odour sources (Fig. 6; Tukey test). Therewas also a significant difference in the rate ofaggression (F2,44=4.308, P=0.02) between col-onies 100 and 1000 (Tukey test), with the latter

8

0

Body mass (g)

Tota

l n

um

ber of

ago

nis

tic

acts

4

2

10

(c)

6

20 30 40 50 60 70

FM

8

0

4

2

20

(d)

6

30 40 6050 80

F

70

6

0

4

2

20

(a)5

30 40 50 70 80

FM

10

0

4

2

(b)

6

40 60 80

M

3

1

60

8

F

20

Figure 3. The relationship between aggression towards foreigners and the body mass of resident mole-rats. (a)Colony 100, rS=0.46, N=45, P=0.0023; (b) colony 7400, rS=0.48, N=37, P=0.0067; (c) colony 1000, rS=0.55,N=44, P=0.0008; (d) colony 2000, rS=0.644, N=12, P=0.0416. The mean body mass for each colony is indicated bythe arrow. F: breeding female; M: breeding male. There was no breeding male in colony 2000 (d).

Animal Behaviour, 53, 3492

exhibiting the highest levels of aggression. The sexof the manipulated individuals did not have asignificant effect on aggression rates (F1,44=2.97,P=0.09). All experimental animals were acceptedback into their resident colony within 10 min (i.e.no more shoving or tugging at the skin).

DISCUSSION

These results demonstrate that naked mole-ratsrecognize colony members and aggressively dis-criminate against foreign conspecifics. This abilitypersisted despite controlling for exogenous cues inall colonies. Naked mole-rats, however, failed todiscriminate between foreign kin (siblings nolonger resident in the colony) and foreign non-kin,and rejected both. This indicates that genetic

similarity per se is not the criterion used fordiscrimination. It is possible that because inter-colony relatedness in populations of naked mole-rats is high, foreign colony recognition systemsbased on a direct assessment of relatedness wouldlead to ambiguity. Furthermore, because nakedmole-rats are predictably more likely to associatespatially and/or temporally with kin, and/or morelikely to be unfamiliar with non-kin, colony rec-ognition mechanisms need not involve an abilityto recognize genetic relatives (Sherman 1980;Davis 1982; Barnard 1991). Our finding that col-ony members interact aggressively even with kinthat have been removed from their natal burrowsystem for a short period (12 h), suggests thatodour familiarity in naked mole-rats needs to becontinually reinforced if it is to serve as an effec-tive mechanism of colony member recognition.

50

0

Experimental

Act

ivit

y le

vel

40

30

20

10

Pre Post Pre

Control

(c)50

0

Experimental

40

30

20

10

Pre Post Pre

Control

(d)

50

0

40

30

20

10

Pre Post Pre

(a)50

0

40

30

20

10

Pre Post Pre

(b)

Post Post

Post Post

Experimental Control Experimental Control

Figure 4. The mean number (+��) of mole-rats active within the burrow system before and after the introduction ofa foreign (experimental) and resident (control) mole-rat. Experimental data: (a) colony 7400, t18=�5.17, P<0.001;(b) colony 100, t20=�14.68, P<0.001; (c) colony 1000, t16=�6.69, P<0.001; (d) colony 2000, t18=�13.56,P<0.001. Control data: t-test, P>0.05 for a, b, c and d.

O’Riain & Jarvis: Recognition in naked mole-rats 493

This may be mediated by some of the communalbehaviour patterns of naked mole-rats. Thus allcolony members depend on the communal nest forefficient thermoregulation and routinely use thecommunal toilet area. In both these places theymay contribute to and acquire distinctive colonyodours.

These results suggest, therefore, that colonymember recognition in the naked mole-rat isachieved through the mixing of individual odoursto form a unique and dynamic recognition cuebetween colonies of closely related individuals.Thus, physically separated groups of siblings willcontribute to and acquire a unique colony odouror ‘badge’ (sensu Linsenmair 1987). Similarly,newly acquired foreign odours are capable ofinducing aggression in otherwise non-aggressive

colony members. Similar findings have beenreported for the spiny mouse, Acomys cahirinus(Porter & Wyrick 1979) and house mouse, Musmusculus (Hurst et al. 1993). In contrast, in theabsence of familiarity, both Richardson’s groundsquirrels, Spermophilus richardsonii (Davis 1982)and white-footed mice, Peromyscus leucopus(Grau 1982) can still recognize their siblings.

The odour choice experiments suggested thatboth the nest and toilet areas are distinctivelymarked with colony odours. When individualswere given a choice between their own and aforeign colony odour, they were able to distin-guish between them and preferentially chose theirown colony odours. Because these odour sourcesare derived from all members of the colony, it islikely that colony odours themselves are a

Figure 5. The mean amount of time (+��) mole-rats spent actively attempting to gain access to the respective odoursources. (a) and (b) show data for individuals from colony 100 (Friedman ANOVA: �2

r 2,132=65.899, P<0.001) and7400 (Friedman ANOVA: �2

r 2,96=50.0, P<0.001), respectively, when presented with soiled nesting material. (c) and(d) show data from colony 100 (Friedman ANOVA: �2

r 2,132=41.0556, P<0.001) and 7400 (Friedman ANOVA,�2r 2,96=32.758, P<0.001), respectively, when presented with soiled toilet material.

Animal Behaviour, 53, 3494

combination of individual odour cues. This ‘cock-tail’ of odours is shared among all members of acolony but would appear to be different from thatof other, even closely related, colonies (such asthose formed by siblings). The colony-specificgroup labels are thus reliable cues for the recog-nition of familiar colony members but convey nodirect information on genetic relatedness.

The most likely recognition mechanism innaked mole-rats would thus appear to be throughthe sharing of familiar odours that are spread

through close physical association. Holmes &Sherman (1983) stated that sometimes animalstreat as kin those conspecifics with whom theyhave close association during their life. Nakedmole-rats thus exhibit kin bias because, undernatural conditions, there is normally a reliablecorrelation between genetic relatedness and thespatial/temporal component of association.

This recognition mechanism in naked mole-ratsclosely parallels nestmate recognition cues in somesocial insects. Various authors have suggested thatnestmate recognition in many social insects mayhave both genetic and environmental components(Wilson 1971; Jutsum et al. 1979; Carlin &Hölldobler 1983; Page & Breed 1987; Hölldobler& Wilson 1990; Gastreich et al. 1991;Venkataraman et al. 1992), and that colonyodours, derived from the mixing of individualodours or a common environmental odour, mayhave evolved as the simplest and most effectivemeans of facilitating nestmate recognition(Hölldobler & Michener 1980).

If the penalities for failing to exclude foreignconspecifics were trivial, colony-specific recog-nition and colony defence would have a limitedselective advantage. However, both the existenceof a highly efficient defence system in nakedmole-rat colonies and the existence of a reliablecolony member recognition mechanism suggestthat protection of the burrow system and itsresources from potential usurpers (irrespective oftheir relatedness) is of paramount importance totheir continued survival. Indeed the existence of adefence-based division of labour with small indi-viduals recruiting large colony defenders withalarm calls suggests an evolutionary history ofdefence against conspecifics. Interestingly, for-eigners never alarm-called whilst in the residentburrow system or following physical contact witha resident individual. These foreigners did, how-ever (94% of all trials), alarm-call following theirreturn to their natal burrow system, suggestingthat the alarm call of the naked mole-rat is contextspecific, serving to alert only colony members tothe presence of foreign conspecifics.

A further important selective advantage ofxenophobia in naked mole-rats may be that of theselfish ‘desire’ to reproduce. Despite reproductionbeing monopolized by a minority of dominantanimals in the colony, every mole-rat born to thatcolony has the potential to be a reproductive(Jarvis et al. 1991) and colony members are

2.5

0

Odour source

Shov

es/tri

al

2

1.5

1

0.5

(c)

Own BlankForeign

2.5

0

2

1.5

1

0.5

(b)

2.5

0

2

1.5

1

0.5

(a)

Figure 6. The mean number (+��) of agonistic acts(shoves) performed by resident mole-rats on individualsthat were returned to their colonies after separate expo-sure to three different odour sources. (a) Colony 100;(b) colony 7400; (c) colony 1000.

O’Riain & Jarvis: Recognition in naked mole-rats 495

therefore principal competitors for fulfilling thisrole. The acceptance of a foreign conspecificwould only serve to diminish an individual’schances of attaining reproductive status, and gain-ing the associated direct fitness benefits. Dispersalto avoid this competition is a costly alternative fornaked mole-rats and one that appears to havebeen selected against, judging from its rarity ofoccurrence (Brett 1991; O’Riain et al. 1996) andthe fact that inbreeding is common both in thefield (Faulkes et al. 1990; Reeve et al. 1990) andthe laboratory (Jarvis 1991b).

This suggested reproductive threat is supportedby our finding that colony defence against foreignconspecifics is predominantly performed by thelargest/oldest members of the colony (Fig. 3).These individuals are the colony’s most likelyreproductive replacements in the event of thedeath of either of the reproductives. In anotherseries of experiments designed to investigatereproductive succession in naked mole-rats, thenew reproductives, both male (N=4) and female(N=4), were consistently amongst the largest (top5%) individuals in their colony (M. J. O’Riain,unpublished data). Their large size and experiencemakes them worthy contestants for what is often abloody battle to the death for reproductive rightswithin the colony (Jarvis 1991b). It is these indi-viduals, therefore, that stand to lose the most, inpotential direct fitness terms, by accepting a for-eign conspecific into their social hierarchy. Fur-thermore, whilst colony defence is positivelycorrelated with body mass in naked mole-ratcolonies, the breeding female, typically one of thelargest animals in the colony, rarely participates incolony defence (N=3 this study, N=4 personalobservation). The exception to this observationwas in colony 7400 (Fig. 3b) in which the repro-ductive female had recently acquired her sexualstatus through intense physical competition withcolony members. This breeding female was con-sistently the most aggressive defender of thecolony against foreigners, emphasizing the closelink between colony defence and reproductiveopportunity.

The ability to maintain colony integrity, in themidst of closely related neighbouring colonies, isobviously important and reflects an evolutionaryhistory of repeated contacts between colonies. Wesuggest that the use of a recognition system inwhich the individual odours are mixed to form aunique colony odour may have evolved to func-

tion primarily as a mechanism to ensure theacceptance of colony members and the rejection ofnon-colony members, and as a mechanism topermit the segregation of genetically similar kin indifferent neighbouring colonies. This is essentialfor the efficient functioning of individual coloniesas separate entities (Stuart 1992). Naked mole-ratcolonies are essentially independent functionalunits with a characteristic composition of a repro-ductive pair, their non-reproductive offspring andother close kin. Their efficient functioning maydepend on this composition, and their ability tomaintain colony autonomy, in the midst of closelyrelated neighbouring colonies, may be essentialfor survival.

ACKNOWLEDGMENTS

We thank Joseph Booysen and Mary Vosloo forassisting with the maintenance of the animals inthe laboratory. We thank Nigel Bennett, JonathenBloomer, David Jacobs, Daniel Polakow, PaulSherman and Andrew Spinks and two anonymousreferees for comments on an early version of themanuscript. This research was supported by anFRD bursary and grant and a grant from theUniversity of Cape Town.

REFERENCES

Barnard, C. 1991. Kinship and social behaviour: thetrouble with relatives. Trends Ecol. Evol., 6, 310–312.

Blaustein, A. R. & Porter, R. H. 1990. The ubiquitousconcept of recognition with special reference to kin.In: Methods, Inference, Interpretation, and Expla-nation in the Study of Behavior (Ed. by M. Bekoff &D. Jamieson), pp. 123–148. Boulder, Colorado:Westview Press.

Blaustein, A. R., Bekoff, M., Byers, J. A. & Daniels,T. J. 1991. Kin recognition in vertebrates: what do wereally know about adaptive value? Anim. Behav., 41,1079–1083.

Brett, R. A. 1991. The population structure of nakedmole-rat colonies. In: The Biology of the Naked Mole-Rat (Ed. by P. W. Sherman, J. U. M. Jarvis & R. D.Alexander), pp. 97–136. Princeton, New Jersey:Princeton University Press.

Buckle, G. R. & Greenberg, L. 1981. Nestmate recog-nition in sweat bees (Lasioglossum zephyrum): does anindividual recognize its own odour or only odours ofits nestmates? Anim. Behav., 29, 802–809.

Byers, J. A. & Bekoff, M. 1991. Development, theconveniently forgotten variable in ‘true kin recog-nition’. Anim. Behav., 41, 1088–1090.

Animal Behaviour, 53, 3496

Carlin, N. F. & Hölldobler, B. 1983. Nestmate and kinrecognition in interspecific mixed colonies of ants.Science, 222, 1027–1029.

Crosland, M. W. J. 1990. Variation in ant aggressionand kin discrimination ability within and betweencolonies. J. Insect Behav., 3, 359–379.

Davis, L. S. 1982. Sibling recognition in Richardson’sground squirrels (Spermophilus richardsonii). Behav.Ecol. Sociobiol., 11, 65–70.

Faulkes, C. G., Abbott, D. H. & Mellor, A. L. 1990.Investigation of genetic diversity in wild colonies ofnaked mole-rats (Heterocephalus glaber) by DNAfingerprinting. J. Zool., Lond., 221, 87–97.

Fishwild, T. G. & Gamboa, G. J. 1992. Colony defenceagainst conspecifics: caste-specific differences in kinrecognition by paper wasps, Polistes fuscatus. Anim.Behav., 43, 95–102.

Fletcher, D. J. C. & Michener, C. D. 1986. Kin Recog-nition in Animals. New York: John Wiley.

Gadagkar, R. 1985. Kin recognition in social insects andother animals. A review of recent findings and aconsideration of their relevance for the theory of kinselection. Proc. Indian Acad. Sci., 94, 587–621.

Gamboa, J. G., Reeve, H. K., Ferguson, I. D. &Wacker, T. L. 1986. Nestmate recognition in socialwasps: the origin and acquisition of recognitionodours. Anim. Behav., 34, 685–695.

Gastreich, K. R., Queller, D. C., Hughes, C. R. &Strassmann, J. 1991. Kin discrimination in the tropi-cal swarm-founding wasp, Parachartergus colobop-terus. Anim. Behav., 40, 598–601.

Grafen, A. 1990. Do animals really recognize kin? Anim.Behav., 39, 42–54.

Grau, H. G. 1982. Kin recognition in white-footeddeermice (Peromyscus leucopus). Anim. Behav., 30,497–505.

Hamilton, W. D. 1964a. The genetical evolution ofsocial behaviour I. J. theor. Biol., 7, 1–16.

Hamilton, W. D. 1964b. The genetical evolution ofsocial behaviour II. J. theor. Biol., 7, 17–52.

Hepper, P. G. 1991. Recognizing kin: ontogeny andclassification. In: Kin Recognition in Animals (Ed. byP. G. Hepper), pp. 259–278. Cambridge: CambridgeUniversity Press.

Hölldobler, B. & Michener, C. D. 1980. Mechanisms ofidentification and discrimination in social Hymenop-tera. In: Evolution of Social Behaviour: Hypotheses andEmpirical Tests (Ed. by H. Markl), pp. 35–58.Weinheim: Verlag Chemie.

Hölldobler, B. & Wilson, E. O. 1990. The Ants.Cambridge, Massachusetts: Belknap Press.

Holmes, W. G. & Sherman, P. W. 1983. Kin recognitionin animals. Am. Scient., 71, 46–55.

Hurst, J. L., Fang, J. & Barnard, C. 1993. The role ofsubstrate odours in maintaining social tolerancebetween male house mice, Mus musculus domesticus:relatedness, incidental kinship effects and the estab-lishment of social status. Anim. Behav., 48, 157–167.

Jarvis, J. U. M. 1981. Eusociality in a mammal: coop-erative breeding in naked mole-rat colonies. Science,212, 571–573.

Jarvis, J. U. M. 1991a. Methods for capturing, trans-porting, and maintaining naked mole-rats in captivity.

In: The Biology of the Naked Mole-Rat (Ed. by P. W.Sherman, J. U. M. Jarvis & R. D. Alexander),pp. 467–483. Princeton, New Jersey: PrincetonUniversity Press.

Jarvis, J. U. M. 1991b. Reproduction of naked mole-rats. In: The Biology of the Naked Mole-Rat (Ed. byP. W. Sherman, J. U. M. Jarvis & R. D. Alexander),pp. 384–425. Princeton, New Jersey: PrincetonUniversity Press.

Jarvis, J. U. M., O’Riain, M. J. & McDaid, E. 1991.Growth and factors affecting body size in nakedmole-rats. In: The Biology of the Naked Mole-Rat(Ed. by P. W. Sherman, J. U. M. Jarvis & R. D.Alexander), pp. 358–383. Princeton, New Jersey:Princeton University Press.

Jutsum, A. R., Saunders, T. S. & Cherrett, J. M. 1979.Intraspecific aggression in the leaf-cutting ant Acro-myrmex octospinosus. Anim. Behav., 27, 839–844.

Lacey, E. A. & Sherman, P. W. 1991. Social organis-ation of naked mole-rat colonies: evidence for divi-sions of labour. In: The Biology of the NakedMole-Rat (Ed. by P. W. Sherman, J. U. M. Jarvis &R. D. Alexander), pp. 275–336. Princeton, NewJersey: Princeton University Press.

Lacey, E. A., Alexander, R. D., Braude, S. H., Sherman,P. W. & Jarvis, J. U. M. 1991. An ethogram for thenaked mole-rat nonvocal behaviors. In: The Biologyof the Naked Mole-Rat (Ed. by P. W. Sherman, J. U.M. Jarvis & R. D. Alexander), pp. 209–242.Princeton, New Jersey: Princeton University Press.

Linsenmair, K. E. 1987. Kin recognition in subsocialarthropods, in particular in the desert isopod Hemi-lepistus reaumuri. In: Kin Recognition in Animals (Ed.by D. J. C. Fletcher & C. D. Michener), pp. 121–208.New York: John Wiley.

Michener, C. D. 1974. The Social Behaviour of Bees.Cambridge, Massachusetts: Belknap Press.

O’Riain, M. J., Jarvis, J. U. M. & Faulkes, C. G. 1996.A dispersive morph in the naked mole-rat. Nature,380, 619–621.

Page, R. E. & Breed, M. D. 1987. Kin recognition insocial bees. Trends Ecol. Evol., 2, 272–275.

Pepper, J. W., Braude, S. H., Lacey, E. A. & Sherman,P. W. 1991. Vocalizations of the naked mole-rat. In:The Biology of the Naked Mole-Rat (Ed. by P. W.Sherman, J. U. M. Jarvis & R. D. Alexander),pp. 243–274. Princeton, New Jersey: PrincetonUniversity Press.

Pfennig, D. W. & Sherman, P. W. 1995. Kin recognition.Scient. Am., 272, 68–73.

Porter, R. H. & Wyrick, M. 1979. Sibling recognition inspiny mice (Acomys cahirinus). Behav. Ecol. Sociobiol.,3, 61–68.

Reeve, H. K., Westneat, Noon, W. A., Sherman, P. W.& Aquadro, C. F. 1990. DNA ‘fingerprinting’ revealshigh levels of inbreeding in colonies of the eusocialnaked mole-rat. Proc. natn Acad. Sci. U.S.A., 87,2496–2500.

Sherman, P. W. 1980. The limits of ground squirrelnepotism. In: Sociobiology: Beyond Nature/Nurture?(Ed. by G. W. Barlow & J. Silverberg), pp. 505–544.Boulder, Colorado: Westview Press.

O’Riain & Jarvis: Recognition in naked mole-rats 497

Sherman, P. W. & Holmes, W. G. 1985. Kin recog-nition: issues and evidence. In: Experimental Behav-ioural Ecology (Ed. by B. Hölldobler & M. Lindauer),pp. 437–460. Stuttgart: Fischer Verlag.

Stuart, R. J. 1991. Kin recognition as a functionalconcept. Anim. Behav., 41, 1093–1094.

Stuart, R. J. 1992. Nestmate recognition and theontogeny of acceptability in the ant, Leptothoraxcurvispinosus. Anim. Behav., 30, 403–408.

Venkataraman, A. B., Swarnalatha, V. B., Nair, P. &Gadagkar, R. 1992. The mechanism of nestmatediscrimination in the tropical social wasp Ropalidiamarginata and its implications for the evolution ofsociality. Anim. Behav., 43, 95–102.

Wilson, E. O. 1971. The Insect Societies. Cambridge,Massachusetts: Belknap Press.

Zar, J. H. 1984. Biostatistical Analysis. EnglewoodCliffs, New Jersey: Prentice-Hall.

Animal Behaviour, 53, 3498