Intracolony aggression in the eusocial naked mole-rat, Heterocephalus glaber

ANIMAL BEHAVIOUR, 2001, 61, 311–324 doi:10.1006/anbe.2000.1573, available online at http://www.idealibrary.com on

Intracolony aggression in the eusocial naked mole-rat, Heterocephalus glaber F. M. CLARKE* & C. G. FAULKES†

*Institute of Zoology, London, U.K. †School of Biological Sciences, Queen Mary & Westfield College (Received 18 June 1999; initial acceptance 17 September 1999; final acceptance 1 September 2000; MS. number: 6260R)

In colonies of the eusocial naked mole-rat, breeding is monopolized by one dominant female (the ‘queen’) and one to three males. Aggression in the form of shoving (prolonged pushes involving nose to nose contact) is frequently observed in captive colonies and is principally initiated by breeders. It has been suggested that shoving by queens functions primarily to suppress reproduction in subordinates (threat reduction hypothesis) or to incite work activity in nonbreeding ‘helpers’ (work conflict hypothesis). We tested predictions of both hypotheses by examining shoving in five captive colonies before and after removing breeders. Shoving was strongly associated with reproductive status. The vast majority of shoves were carried out by the queen, and to a lesser extent breeding males, and the onset of reproductive activity coincided with the onset, or greatly increased rates, of shoving. No evidence for the work conflict hypothesis was found. Queens shoved high-ranking and large colony members most, a finding that does not allow discrimination between hypotheses. Those males that posed the greatest threat to the queen’s reproductive dominance, namely future breeding males, were shoved most, providing some support for the threat reduction hypothesis. However, evidence that queens targeted those females that posed the greatest threat, females that were to succeed them, was equivocal. Queen shoving may have several functions, depending on social context, in inhibiting reproduction in subordinates of both sexes, maintaining social order, and in inciting work-related behaviours in colony members, all of which ultimately increase the reproductive success of queens. 

experimental manipulation of colonies (during periods of presumed social stability), agonistic behaviours are not uncommon (Reeve & Sherman 1991). The most frequently observed agonistic behaviour in captive colonies is shoving. Shoves are vigorous, prolonged head to head pushes between two individuals that result in one animal being pushed backward along the tunnel system (Lacey et al. 1991). Queens initiate shoves far more often than other colony members and most of the remaining shoving in colonies is carried out by breeding male(s), and to a lesser extent by large nonbreeders (Reeve & Sherman 1991). Several hypotheses have been proposed for the function of shoving by the queen, although they are not mutually exclusive. The ‘work conflict’ hypothesis proposes that there is a conflict of interest between the queen and other colony members over the extent of aid they provide in support of her reproduction, and that shoving serves primarily to incite activity in ‘lazy’ workers (Reeve 1992). The ‘threat reduction’ hypothesis suggests that the queen shoves colony mates to reduce the threat that they will challenge her for breeding rights (Reeve & Sherman

Naked mole-rats are fossorial and eusocial mammals (Jarvis 1981; Sherman et al. 1992) endemic to eastern Africa. Colonies occupy discrete burrow systems and may number up to 300 animals (Brett 1991). Breeding is restricted to one dominant female and one to three males (Jarvis 1991; Lacey & Sherman 1991), while other subordinate colony members are infertile but not sterile (for review see Faulkes & Abbott 1997). Genetic studies indicate high levels of inbreeding (Reeve et al. 1990; Honeycutt et al. 1991; Faulkes et al. 1997) and dispersal is rare (O’Riain et al. 1996). Nonbreeders are typically the offspring of breeders or their close relatives, and remain in the natal colony acting as ‘alloparents’ or ‘helpers’ by performing ‘work’ behaviours such as foraging, maintaining and defending the colony, and assisting both directly and indirectly in caring for pups (Sherman et al. 1992). Despite high levels of relatedness within colonies, they are not perfectly harmonious societies. Even without Correspondence: C. G. Faulkes, School of Biological Sciences, Queen Mary & Westfield College, Mile End Road, London E1 4NS, U.K. (email: [email protected]). F. M. Clarke is now at the Department of Zoology, University of Aberdeen, Aberdeen AB24 2TZ, U.K. 0003–3472/01/020311+14 $35.00/0

2001 The Association for the Study of Animal Behaviour

311



2001 The Association for the Study of Animal Behaviour

312

ANIMAL BEHAVIOUR, 62, 2

1991). Behavioural contact with the queen, rather than a primer pheromone in her urine, is necessary for the reproductive suppression of nonbreeding naked mole-rats (Faulkes & Abbott 1993; Smith et al. 1997). It has been suggested that both subtle and overt agonistic interactions between the queen and nonbreeders of both sexes inhibit gonadal function in the latter, and that shoving may be a key behaviour facilitating suppression (Faulkes & Abbott 1997). However, empirical tests designed to discriminate between the competing hypotheses have so far proved equivocal. Whereas Reeve (1992) found evidence supporting the work conflict hypothesis, Jacobs & Jarvis (1996) did not. To test some of the predictions of the hypotheses for the function of shoving, we carried out focal sampling of colony members from five captive colonies and recorded which individuals initiated and received shoves, as well as the frequency of shoving. We also used scan sampling to determine the ‘work’ levels of colony members from three colonies. Whereas Reeve & Sherman (1991) relied upon theoretical predictions about which colony members posed the greatest threat to the reproductive dominance of breeders, we removed breeders to determine empirically which colony members subsequently became reproductively active and also to examine changes in shoving when individuals became reproductively active. We addressed a number of questions. How do the shove rates of reproductively active and inactive colony members differ? How is shoving by the queen influenced by the rank, weight, age, sex and reproductive status of recipients? What is the relationship between the rate of shoving by queens and the urinary testosterone and cortisol levels of recipients? Finally, we examined the ontogeny of shoving in individuals that became reproductively active after a queen’s removal and during the subsequent period of social instability.

Behavioural Sampling During periods of behavioural observation, we caught all the animals in the study colony weekly, weighed them and, to facilitate recognition of individuals during observations, drew their unique toe clip number on their back with an indelible marker pen. We used focal animal observation (Martin & Bateson 1986) to quantify the following behavioural interactions between individuals within the study colonies. (1) Individual mole-rats were observed for a 10-min period and the frequency of shoving recorded on checksheets, together with the identities of the animals involved. We made the observations between 0700 and 1900 hours with approximately one focal sample of each individual each week. We collected a minimum of six focal samples (range 6–11) for each individual both before and after removal of the breeders. (2) The proportion of times that one animal passed over another individual in dyadic encounters was obtained, and summed for each individual encountered to provide a rank score which we used to assign a dominance rank. Passing is a good metric of dominance relationships among both male and female naked mole-rats (Clarke & Faulkes 1997). From behavioural observations, we constructed a ‘sociometric matrix’ (Altmann 1974) for males and females in each colony, using the frequency of passing over between interacting mole-rat dyads as cell entries. We used scan sampling (Martin & Bateson 1986) to record work behaviours at 30-min intervals in colonies NN, 2200 and N2. Work behaviours were digging (which included foreleg digging, gnawing and backshovelling), mouth carrying, dragging and sweeping. A detailed description of these behaviours is given by Lacey et al. (1991). Work level is the proportion of scans in which an individual was engaged in work behaviours. Finally, rare but important events, such as sexual behaviours, were recorded whenever observed.

METHODS

Composition of Study Colonies

Urine Sampling and Hormone Determination

Captive colonies of naked mole-rats were maintained at the Institute of Zoology, in artificial Perspex burrow systems comprising nest, food and toilet chambers as described by Faulkes et al. (1990). All captive colonies were derived from wild/genetic stock from Mtito Andei and Lerata, Kenya. Animals were numbered and identified by a system of toe clipping. Three colonies, N1, O and NN, observed as part of a study on reproductive succession among females, contained 16 males and 12 females, eight males and seven females, and nine males and 10 females, respectively (Clarke & Faulkes 1997). Colony members of both sexes were subject to behavioural observation and hormone determination. Another two colonies, 2200 and N2, observed as part of a study on reproductive succession among males, contained 19 males and 14 females, and 21 males and 24 females, respectively (Clarke & Faulkes 1998). No hormonal or behavioural data were available for nonbreeding females in colonies 2200 and N2. All five colonies contained only one breeding female (queen).

We collected urine samples from both sexes in colonies N1, O and NN, but not from nonbreeding females in colonies 2200 and N2. We used urine rather than blood for hormone determination, as we could sample it repeatedly over the same period as the behavioural observations with minimum disturbance to the colonies, and for ethical considerations. Prior to urine sampling, we removed wood shavings from the toilet chamber of the colony, and wiped the chamber clean with damp tissue paper. Immediately after each urination, we collected the deposited urine in a glass Pasteur pipette, and wiped the toilet chamber clean as before. Urine samples were put on ice immediately, then stored at
20C until hormone determination. We first determined creatinine levels in the urine samples as described by Bonney et al. (1982), to correct for differing dilutions. Urinary hormone concentrations are expressed as mass per mg creatinine (mg/Cr). Progesterone in female urine was measured in petroleumether extracted samples (50–100
l), by radioimmunoassay; testosterone and cortisol in the urine of both sexes

CLARKE & FAULKES: AGGRESSION IN NAKED MOLE-RATS

Table 1. Schedule of behavioural observations and sampling in each of the five colonies before and after removal of breeders Before breeder removal

Colony N1 O NN N2 2200

After breeder removal

No. days observed

No. hours of focal observation

No. scan samples

No. days observed

No. hours of focal observation

No. scan samples

112 56 145 87 87

45 27 120 27 25

— — 108 79 83

55 55 238 92 58

45 27 122 26.5 17

— — 68 — —

were measured in diethyl ether-extracted samples (50– 100
l), by radioimmunoassay. For a complete description of urine sampling and hormone determination, see Clarke & Faulkes (1997, 1998).

Breeder Removal Experiments Focal observation of shoving behaviour and urine collection from all colony members were undertaken for a control period in colonies N1, O and NN (Table 1). We then removed the single queen in each colony (simulating natural death or predation), and continued focal observation and urine collection to determine which females attained breeding status (Table 1). Additionally, scan sampling was used to record work behaviours in colony NN before and after queen removal. In colonies N2 and 2200, we recorded the shoving behaviour of queens and males by focal sampling, and work behaviours by scan samples, and collected their urine, for a control period (Table 1). We then removed the single breeding male in both colonies, and continued focal observation and urine collection to determine which males attained breeding status (Table 1). Reproductively active colony members in both queen removal and breeding male removal experiments were identified by a number of criteria: elevated urinary hormone levels (progesterone in females and testosterone in males), and observation of sexual behaviours such as copulation and mutual ano-genital nuzzling. The latter behaviour occurs most frequently between the queen and breeding males (Jarvis 1991).

Data Analysis For all statistical correlations we used Spearman rank correlations. To control for the possible confounding effect of male reproductive status on shoving, we carried out all correlation tests twice, once including, and then excluding, breeding males from the analysis. However, correlations held whether breeding males were included or excluded and correlation coefficient values and probability levels given in the Results are exclusive of breeding male data. To test for an effect of sex on shoving, we compared shoving data for nonbreeding males and nonbreeding females in colonies N1, O and NN. To test for an

effect of reproductive status on shoving, within all five study colonies shoving data were partitioned out into the following groups: breeding females; breeding males; nonbreeding males; nonbreeding females; and those colony members that became new breeders. For shoving rate and hormone level comparisons between two groups we used Mann–Whitney U tests and for comparisons between three or more groups we used Kruskal–Wallis tests. In the Results, ‘all P<0.05’ indicates where multiple comparison tests were significant between all groups. All statistical analysis was two tailed and the cutoff point for statistical significance was P=0.05.

Ethical Note To minimize stress or injury during breeder removal experiments, we removed harassed or injured individuals and housed them separately. Two animals with bite injuries were removed and put to sleep later the same day when they failed to recover from their wounds. The unpredictability of severe aggression meant that early intervention was not always possible. Even without experimental manipulation of colonies, a small number of mole-rats die during fighting each year. Mole-rats were individually marked by toe clipping. The minimum number of toe tips was removed from each animal to allow unambiguous identification (assigning hind feet toes as 1, 2, 4, 7 and 10, 20, 40 and 70, respectively). Although a controversial procedure for marking animals, tattooing, the other common method of marking this hairless species, was considered a more painful and distressing alternative. Similarly, subcutaneous implantation of transponders has been shown to be less effective than toe clipping in studies of naked mole-rats in the wild (Braude 1998). This study was covered by a Home Office project licence. RESULTS

Before Breeder Removal Queen shove initiation rates In colonies N1, O and NN, where quantitative data on shoving were available for all colony members, the queen initiated significantly more shoves per unit time than any

313

ANIMAL BEHAVIOUR, 62, 2

5

(a)

0.5

4

0.4

3

0.3

2

0.2

1

0.1

0

0.5

Q

BM

NonBM NonBF

(c)

0.4

0.3

0.2

0.1

0

0.0

New Q

Mean shoves received from queen/10 min

Mean shoves initiated by colony members/10 min

314

Q

BM

NonBM NonBF

New Q

New Q

BM

NonBM NonBF

BM

NonBM NonBF

(d) 0.25

0.2

0.15

0.1

0.05

0

(e)

New Q (f)

4

0.4

3

0.3

2

0.2

1

0.1

0

(b)

Q

NonBM NonBF

New Q

0.0

New Q

NonBM NonBF

Figure 1. Mean±SE number of shoves initiated by colony members and received from the queen in (a, b) colony N1, (c, d) colony O and (e, f) colony NN before experimental removal of queens. Q: queen; BM: breeding male(s); NonBM: nonbreeding males; NonBF: nonbreeding females; New Q: females that became reproductively active after removal of resident queens.

other colony member (colony N1: Mann–Whitney U test: U=17.5, N1 =10, N2 =23, P<0.001; colony O: Kruskal– Wallis test: H4 =34.8, P<0.001; multiple comparison test: all P<0.01; colony NN: Kruskal–Wallis test: H3 =321.0, P<0.001; multiple comparison test: all P<0.01; Fig. 1). In colony 2200 and colony N2, focal observation of nonbreeding females was not undertaken and, therefore, quantitative data on shoving were available only for

queens, breeding males and nonbreeding males. Colony 2200’s queen was the only colony member observed shoving (Fig. 2a), whereas in colony N2, the queen initiated significantly more shoves per unit time than both breeding and nonbreeding males (Kruskal–Wallis test: H3 =22.4, P<0.001; multiple comparison test: all P<0.01; Fig. 2c). In all colonies queens were never observed receiving shoves.

CLARKE & FAULKES: AGGRESSION IN NAKED MOLE-RATS

0.2

(a)

1.5

0.15

1

0.1

0.5

0

3

Q

BM

NonBM

New BM

(c)

2.5 2

Mean shoves received from queen/10 min

Mean shoves initiated by colony members/10 min

2

0.05

0

0.3

1

0.1

0.5

0.05

BM

NonBM

New BM

NonBM

New BM

BM

NonBM

New BM

(d)

0.2 0.15

Q

BM

0.25

1.5

0

(b)

0

Figure 2. Mean±SE number of shoves initiated by colony members and received from the queen in (a, b) colony 2200 and (c, d) colony N2 before experimental removal of breeding males. Q: queen; BM: breeding male(s); NonBM: nonbreeding males; NewBM: those males that became breeders after removal of resident males.

Breeding male shove-initiation rates Other than queens, breeding males in colony N1 and colony O initiated the majority of shoves (Fig. 1a, c). No breeding males were identified in colony NN (Fig. 1e). In colony N1 the breeding males were the only colony members other than the queen to show shoving behaviour (Fig. 1a), whereas in colony O, the breeding male initiated significantly more shoves per unit time than other colony members (Kruskal–Wallis test: H4 =34.8, P<0.001; multiple comparison test: all P<0.05), shoving at over five times the rate for an average nonbreeding male or nonbreeding female (Fig. 1c). Similarly, colony N2’s breeding male shoved at a significantly higher rate than nonbreeding colony males (Kruskal–Wallis test: H3 =22.4, P<0.001; multiple comparison test: P<0.05), shoving at over 20 times the rate for an average nonbreeding male (Fig. 2c). In contrast to the other colonies, colony 2200’s breeding male never shoved (Fig. 2a).

Nonbreeder shove initiation rates Nonbreeding colony members of both sexes initiated shoves at a low rate relative to breeders (colony O and colony NN), or did not shove at all (colony N1). Nonbreeding females initiated shoves at a significantly higher

rate than nonbreeding males in colony NN (Kruskal– Wallis test: H3 =321.0, P<0.001; multiple comparison test: P<0.05; Fig. 1e), but not in colony O where both sexes shoved at a similar rate (Fig. 1c). In both colony N1 and colony O, the female that became reproductively active after removal of the resident queen did not shove (Fig. 1a,c). In contrast, the three females in colony NN that became reproductively active after removal of the queen shoved at a significantly higher rate than other nonbreeding females and nonbreeding males (Kruskal–Wallis test: H3 =321.0, P<0.001; multiple comparison test: all P<0.05; Fig. 1e). Nonbreeding males in colony 2200, including the male that was to become the new breeding male, were not observed shoving (Fig. 2a). In contrast, colony N2’s nonbreeding males shoved at a low rate relative to breeders, with the two males that were to become breeders shoving at a significantly higher rate than other nonbreeding males (Kruskal–Wallis test: H3 =22.4, P<0.001; multiple comparison test: P<0.05; Fig. 2c). Qualitative data suggest that nonbreeding females shoved at similar rates to nonbreeding males in colony 2200 and colony N2. In colonies NN and N2, an appreciable number of shoves were initiated by several nonbreeders (colony NN: 42% of the total; colony N2: 39% of the total), allowing

315

316

ANIMAL BEHAVIOUR, 62, 2

Table 2. Spearman rank correlations between the rate of shoving by the breeding female and the recipients’ dominance rank, body weight and urinary testosterone titres, in each study colony before removal of the breeding female Colony Variable Dominance rank Body weight Testosterone

Sex

N1

0

NN

Males Females Males Females Males Females

0.04 −0.42 −0.05 −0.38 −0.13 −0.43

0.79* 0.66 0.71* 0.66 0.80* 0.13

−0.69* −0.38 −0.88** −0.54 −0.82* −0.21

*P<0.05; **P<0.005.

correlation analysis. The rate of shoving by nonbreeders in colony NN was positively correlated with their dominance rank (Spearman rank correlation: rs =0.61, N=18, P<0.05): high-ranking individuals shoved at a higher rate than low-ranking animals. The rate of shoving by nonbreeding males in colony N2 was positively correlated with their rank (Spearman rank correlation: rs =0.53, N=20, P<0.05), weight (rs =0.52, N=20, P<0.05) and urinary testosterone titres (rs =0.65, N=15, P<0.05).

colony N2, the two nonbreeding males that were to become breeders were shoved by the queen at a significantly higher rate than other nonbreeding males (Mann– Whitney U test: U=4268, N1 =18, N2 =145, P<0.001; Fig. 2d). In contrast, the queen was not seen to shove the current breeding male. Although focal observation of nonbreeding females was not undertaken, qualitative data suggest that the queens in colonies 2200 and N2 shoved nonbreeding females at a similar rate to nonbreeding males.

Effect of recipient’s sex on queen shoving No significant difference was found between nonbreeding males and nonbreeding females in the rate that they received shoves from the queen in colony N1 (Mann– Whitney U test: U=63.5, N1 =13, N2 =10, NS; Fig. 1b), colony O (U=20.5, N1 =7, N2 =6, NS; Fig. 1d) and colony NN (U=34.5, N1 =N2 =9, NS; Fig. 1f).

Effect of recipient’s reproductive status on queen shoving The recipients of shoving from colony N1’s queen were breeding males, nonbreeding males and females, and F102, the female that was to become the new queen (Fig. 1b). No significant difference was found between these groups in the rate they received shoves from the queen. In colony O the breeding male and F73, which was to become the new queen, were subject to similar rates of shoving from the queen. Both were shoved at a significantly higher rate than nonbreeding males (Kruskal– Wallis test: H3 =6.9, P<0.001; multiple comparison test: all P<0.05). Nonbreeding females were not shoved by the queen (Fig. 1d). Nonbreeding females were the prime recipients of queen shoving in colony NN, followed by breeding males, and lastly, the three females that were to become reproductively active after removal of the queen. Differences in the rate groups were shoved by the queen were significant (Kruskal–Wallis test: H2 =9.6, P<0.001; multiple comparison test: all P<0.05; Fig. 1f). In colony 2200 the nonbreeding male that was to become the new breeding male was shoved by the queen at a significantly higher rate than other nonbreeding males and the resident breeding male (Kruskal–Wallis test: H2 =37.8, P<0.001; multiple comparison tests: all P<0.05; Fig. 2b). The breeding male and nonbreeding males were shoved at similar rates by the queen. In

Effect of recipient’s rank and body weight on queen shoving In three of the five colonies (colonies O, 2200 and N2), the rate of shoving by queens was positively correlated with the dominance rank and body weight of nonbreeding male recipients (Tables 2, 3). In contrast, in colony NN the rate of shoving by the queen was negatively correlated with the rank and weight of nonbreeding male recipients (Table 2). No significant correlations between these variables were found for nonbreeding males in colony N1. In each colony where focal observation of nonbreeding females was undertaken (colonies N1, O and NN), no correlation was found between the rate of shoving by queens and the rank or body weight of nonbreeding female recipients (Table 2).

Effect of recipient’s work level on queen shoving Work levels were determined for males in colonies NN, 2200 and N2, and also for colony NN females. In all three colonies, no correlation was found between the rate of shoving by queens and the work levels of recipients.

Urinary testosterone and cortisol titres The rate of shoving by colony O’s queen showed a strong positive correlation with the urinary testosterone titres of nonbreeding male recipients (Table 2). In contrast, the shove rate of colony NN’s queen showed a strong negative correlation with the urinary testosterone levels of nonbreeding male recipients (Table 2). For nonbreeding males in the other three study colonies, no significant correlations between these variables were found (Tables 2, 3). In three colonies the urinary cortisol levels of nonbreeding males were determined and no

CLARKE & FAULKES: AGGRESSION IN NAKED MOLE-RATS

Table 3. Spearman rank correlations between the rate of shoving by breeding female(s) and the male recipients’ dominance rank, body weight, urinary testosterone and urinary cortisol titres, in each study colony before and after removal of the colony breeding males Colony Before removal Variable Dominance rank Body weight Testosterone Cortisol

2200 0.54* 0.49* −0.20 0.23

After removal N2

0.64*** 0.65*** 0.43 −0.24

2200 0.58* 0.60** 0.71*** 0.08

N2 0.75**** 0.73**** 0.39 −0.13

*P<0.05; **P<0.01; ***P<0.005; ****P<0.001.

correlation was found between their cortisol levels and the rate at which they were shoved by queens (colony NN: Spearman rank correlation: rs =
0.66, N=6, NS; colonies 2200 and N2: Table 3). Only in colony NN were the urinary cortisol titres of nonbreeding females determined and no correlation was found between their cortisol levels and the rate at which they were shoved by the queen (Spearman rank correlation: rs =
0.38, N=9, NS).

Effect of body weight and rank on work level In colony NN the work levels of both males and females were determined. For colony 2200 and colony N2, the work levels of males only were calculated. In colony NN the work level of colony members was negatively correlated with their body weight (Spearman rank correlation: rs =
0.46, N=18, P=0.05). Work level was negatively correlated with the body weight of males in colony 2200 (Spearman rank correlation: rs =
0.43, N=19, P=0.05), but no such correlation was evident in colony N2. In all three colonies the work level of colony members was not correlated with their dominance rank.

After Queen Removal Shoves initiated After removal of the queen from colonies N1 and O, one female in each colony became reproductively active (Clarke & Faulkes 1997). The onset of reproductive activity coincided with the onset of shoving behaviour by each of these females (Fig. 3a, c). In contrast, in colony NN three females became reproductively active after removal of the queen. These three females initiated shoves before (Fig. 1e), and increased their rate of shoving (Fig. 3e) after, queen removal. In each colony, reproductively active females shoved at a significantly higher rate than any other colony member (colony N1: Kruskal– Wallis test: H3 =130.1, P<0.001; multiple comparison test: all P<0.05; colony O: Kruskal–Wallis test: H3 =48.4, P<0.001; multiple comparison test: all P<0.05; colony NN; Kruskal–Wallis test: H2 =113.5, P<0.001; multiple comparison test: all P<0.05). Other individuals in colonies N1 and NN shoved at similar low rates. In contrast, in colony O the former breeding male shoved at a significantly higher rate than nonbreeding males, which in turn shoved at a higher rate than nonbreeding

females (Kruskal–Wallis test: H2 =48.4, P<0.001; multiple comparison test: all P<0.05).

Shoves received The former breeding males in colonies N1 and O were the prime recipients of shoves from reproductively active females (Fig. 3b, d). In colony N1 the new queen shoved the former breeding male at a significantly higher rate than nonbreeding males or reproductively inactive females (Kruskal–Wallis test: H2 =36.1, P<0.001; multiple comparison test: all P<0.05; Fig. 3b). Correspondingly, nonbreeding males were shoved by female 102 at a significantly higher rate than nonbreeding females (Kruskal–Wallis test: H2 =36.1, P<0.001; multiple comparison test: P<0.05). Two of the former breeding males and two other high-ranking males were killed by the new queen during this period. Similarly, the new reproductively active female in colony O shoved the former breeding male at a significantly higher rate than nonbreeding males, and nonbreeding males were shoved at a significantly higher rate than reproductively active females (Kruskal–Wallis test: H2 = 5.9, P<0.01; multiple comparison test: all P<0.5; Fig. 3d). The former breeding male and one other high-ranking male were killed by this female during this period. In both colonies, the rate of shoving by reproductively active females showed a strong positive correlation with the male recipient’s dominance rank (Table 4). In colony N1, but not colony O, the rate of shoving by reproductively active females was also correlated with the recipient’s body weight and urinary testosterone levels (Table 4). In colony NN the rate of shoving by females showed a strong positive correlation with the rate at which they received shoves from other females (Spearman rank correlation: rs =1.0, N=8, P<0.01) and also their dominance rank (Table 4), that is shoving was primarily initiated by, and directed towards, those high-ranking reproductively active females competing for reproductive dominance. One of these females was killed during this period. The rate of shoving by reproductively active females was not correlated with the urinary testosterone titres (Table 4) or urinary cortisol titres of male and female recipients (Spearman rank correlation: males: rs =
0.27, N=7, NS; females: rs =0.18, N=7, NS).

317

ANIMAL BEHAVIOUR, 62, 2

(b)

(a)

2

0.6

1.5

0.45

1

0.3

0.5

0.15

0

2

New Q

FBM

NonBM

(c)

1.5

1

0.5

0

0

NonBF Mean shoves received/10 min

Mean shoves initiated by colony members/10 min

318

NonBM

NonBF

FBM

NonBM

NonBF

FBM

NonBM

NonBF

(d) 0.6

0.45

0.3

0.15

New Q

FBM

NonBM

NonBF

(e)

0

1

1.2

0.75

0.9

0.5

0.6

0.25

0.3

0

FBM

New Q

NonBM

NonBF

0

(f)

Figure 3. Mean±SE number of shoves initiated by colony members and received from reproductively active females in (a, b) colony N1, (c, d) colony O and (e, f) colony NN, after removal of queens. New Q: new breeding females; FBM: former breeding males; NonBM: nonbreeding males; NonBF: nonbreeding females.

Work levels After removal of colony NN’s queen, there was a small but significant decrease in the mean work levels of colony members, determined from the proportion of scan

samples in which mole-rats exhibited work behaviours (before: 0.39; after: 0.30; Wilcoxon signed-ranks test: Z=2.22, N=18, P<0.05). A prediction of the work conflict hypothesis is that those individuals that decreased their

CLARKE & FAULKES: AGGRESSION IN NAKED MOLE-RATS

Table 4. Spearman rank correlations between the rate of shoving by new breeding female(s) and the recipients’ dominance rank, body weight and urinary testosterone titres, in each study colony after removal of the original breeding female Colony Variable Dominance rank Body weight Testosterone

Sex

N1

O

NN

Males Females Males Females Males Females

0.69** 0.37 0.67** 0.71* 0.85* −0.27

0.75* 0.22 0.50 0.44 −0.11 0.11

0.35 0.79* 0.47 0.67 0.48 0.28

*P<0.05; **P<0.005.

work level when the queen was removed (‘lazy’ individuals) would be the prime recipients of queen shoving before queen removal. There was no significant difference in the rate of shoving by the queen for mole-rats that increased (0.28 shoves/10 min) and those that decreased their work level (0.19 shoves/10 min) after queen removal (Mann–Whitney U test: U=26.5, N1 =N2 =5, P>0.5). Molerats that increased their work levels (mean 36.1) and those that decreased their work levels (mean 30.1) after queen removal did not differ in weight (Mann–Whitney U test: U=17.0, N1 =5, N2 =10, P>0.10). Females that became reproductively active after queen removal were not included in the above analysis.

N2’s queen shoved the two new breeding males at a significantly higher rate than nonbreeding colony males (Mann–Whitney U test: U=6.3, N1 =15, N2 =129, P<0.001; Fig. 4d). The rate of shoving by colony N2’s queen was positively correlated with the recipients’ rank and weight, but not urinary testosterone levels (Table 3). Colony N2’s queen also directed a large number of shoves at a highranking female. The queen was the prime recipient of shoving from this female and after several weeks was killed by her. In both colonies no correlation was found between the rate of shoving by queens and the recipients’ urinary cortisol titres (Table 3).

Urinary Hormone Levels After Breeding Male Removal Shoves initiated After removal of the breeding male from colonies 2200 and N2, the queen in each colony continued to shove at a similar rate, initiating shoves at a significantly higher rate than new breeding males and nonbreeding males (colony 2200: Kruskal–Wallis test: H2 =29.0, P<0.001; multiple comparison test: all P<0.05; Fig. 4a; colony N2: Kruskal–Wallis test: H2 =79.1, P<0.001; multiple comparison test: all P<0.05, Fig. 4c). In colony 2200 the new breeding male began to shove for the first time, shoving at a significantly higher rate than nonbreeding males (Kruskal–Wallis test: H2 =29.0, P<0.001; multiple comparison test: P<0.05; Fig. 4a). The two new breeding males in colony N2 initiated shoves at similar rates both as nonbreeders (Fig. 2c), and on attaining breeding status (Fig. 4c). As breeders they initiated shoves at a significantly higher rate than nonbreeding males (Kruskal– Wallis test: H3 =79.1, P<0.001; multiple comparison test: P<0.05), shoving at ca. 25 times the rate for an average nonbreeding colony male (Fig. 4c).

Shoves received Colony 2200’s queen shoved nonbreeding males but was not observed shoving the new breeding male (Fig. 4b). One high-ranking male was killed by the queen during this period. The rate of shoving by colony 2200’s queen was positively correlated with the rank, weight and urinary testosterone titres of recipients (Table 3). Colony

Urinary testosterone titres of reproductively active females were significantly higher than those of reproductively inactive females (Mann–Whitney U test: U=25.0, N1 =10, N2 =15, P<0.005; Fig. 5a). Breeding male testosterone levels were significantly higher than those of nonbreeding males (Mann–Whitney U test: U=64.0, N1 =10, N2 =46, P<0.001; Fig. 5b). No significant difference was found between the urinary cortisol titres of reproductively active females, reproductively inactive females, breeding males and nonbreeding males (Kruskal–Wallis test: H3 =1.92, NS; Fig. 5c). DISCUSSION

Incidence and Frequency of Shoving Behaviour Although shoving behaviour varied between colonies, general trends do emerge from this study. It appears that for male and female naked mole-rats their rate of shoving is strongly associated with their dominance and reproductive status. Queens are generally the highest-ranking colony members (Clarke & Faulkes 1997), and initiated far more shoves than any other colony members. Breeding males are among the highest ranking colony males (Clarke & Faulkes 1998), and accounted for the majority of remaining shoves observed within the study colonies. Furthermore, for both males and females in our colonies, the onset of reproductive activity, evidenced by increased urinary testosterone levels in males and progesterone

319

ANIMAL BEHAVIOUR, 62, 2

1.5

(a)

1.0

(b)

1.25 0.75 1.0 0.5

0.75 0.5 0.25 0

1.5

Q

New BM

NonBM

(c)

1.25

Mean shoves received/10 min

Mean shoves initiated by colony members/10 min

320

0.25

0

1.0

New BM

NonBM

New BM

NonBM

(d)

0.75 1.0 0.75

0.5

0.5 0.25 0.25 0

Q

New BM

NonBM

0

Figure 4. Mean±SE number of shoves initiated by colony members and received from the queen in (a, b) colony 2200, (c, d) colony N2, after experimental removal of breeding males. Q: queen; NonBM: nonbreeding males; NewBM: those males that became breeders after removal of resident breeding males.

levels in females (indicative of ovarian cyclicity), coincided with the onset, or greatly increased rates, of shoving. Other researchers have reported a similar relationship between male and female reproductive status and shoving in their colonies, but had no information on the dominance status of their animals (Reeve & Sherman 1991; Reeve 1992; Jacobs & Jarvis 1996). In agreement with these other studies, nonbreeders of both sexes in this study shoved at low rates relative to breeders, or did not shove at all (Reeve & Sherman 1991; Reeve 1992; Jacobs & Jarvis 1996). Where a significant number of nonbreeding colony members initiated shoves, we found a tendency for high-ranking and large individuals to shove more than low-ranking and small colony members. Reeve & Sherman (1991) also found a correlation between body size and shoving for nonbreeders. It is well established that testosterone enhances aggression in mammals, particularly in rodents (Barfield 1984; Wingfield et al. 1994). We found some evidence for an association between shoving and urinary testosterone levels. In one of two colonies where a significant number of nonbreeders shoved, their rates of shoving were corre-

lated with their urinary testosterone titres. Additionally, breeders had significantly higher urinary testosterone levels than same-sexed nonbreeders, and shoved far more often than nonbreeders. Whilst there is extensive evidence that social stress can increase glucocorticoid secretion and suppress reproductive function (Wingfield et al. 1994), there is no evidence that this occurs in naked mole-rats. The rate of queen shoving was not correlated with the recipient’s urinary cortisol levels and breeders and nonbreeders of both sexes had similar urinary cortisol titres. In this study both quantitative and qualitative data suggest that overall there is no sex bias in shoving. The exception was colony NN, where among nonbreeders, females shoved more than males. Specifically, three nonbreeding females in colony NN shoved at a higher rate than other nonbreeders of both sexes. However, reproductive suppression of nonbreeding females in this colony was not complete. We have previously shown that these females had elevated progesterone titres relative to other nonbreeding females during this period, although no signs of ovarian cyclicity were found (Clarke &

CLARKE & FAULKES: AGGRESSION IN NAKED MOLE-RATS

(a)

*

300 Urinary testosterone ng/mg Cr

Urinary testosterone ng/mg Cr

30

10

20

10 16

0

400

Urinary cortisol ng/mg Cr

350

** 10

200

100 46

0

NonBF

Queen

(b)

NonBM

BM

(c)

6

300 250 200 8

150

27 5

100 50 0

Queen

NonBF

BM

NonBM

Figure 5. Urinary testosterone concentrations (X±95% CI; ng/mg creatinine) of (a) females and (b) males, and (c) the urinary cortisol titres (X±95% CI; ng/mg creatinine) of colony members of both sexes. Numbers of animals are given above bars. Queen: reproductively active females; NonBF: reproductively inactive females; BM: breeding males; NonBM: nonbreeding males. *P<0.005, **P<0.001.

Faulkes 1997). In contrast, in the other two colonies, nonbreeding females that were to become queens did not shove at all. During this period they had progesterone levels similar to other nonbreeding colony females, that is low or undetectable (Clarke & Faulkes 1997).

Work Conflict versus Threat Reduction Reeve & Sherman (1991) and Reeve (1992) observed six captive colonies, and Jacobs & Jarvis (1996) two captive colonies, to examine the function of queen shoving. They found that larger colony members were shoved most by queens. Similarly we found that high-ranking and large colony members were shoved more than low-ranking and small colony members. These findings, however, do not allow discrimination between the hypotheses. This is because the work conflict and threat reduction hypotheses both predict that high-ranking and/or large individ-

uals should be shoved most (Reeve & Sherman 1991; Reeve 1992). High-ranking and/or large individuals are more likely to attain breeding status (Clarke & Faulkes 1997, 1998), and therefore pose the greatest threat to a queen’s reproductive dominance, and should be shoved most according to the threat reduction hypothesis (Reeve & Sherman 1991). These same individuals that are likely to become breeders should be lazy, thereby reducing their energy expenditure and risk of predation, and hence increasing the probability that they, rather than other colony members, will replace the breeders and hence should be shoved most according to the work conflict hypothesis (Reeve 1992). This is because a queen benefits most if all her workers are active, and hence should preferentially direct shoves at ‘lazy’ individuals. In our study, work levels of colony members were not correlated with their rank, but were with body weight in two colonies, with smaller individuals working more. Other

321

322

ANIMAL BEHAVIOUR, 62, 2

studies have also reported a correlation between work levels and weight, suggesting that large individuals are indeed ‘lazier’ in terms of colony maintenance activities, although they show higher frequencies of ‘defence’ behaviours than smaller animals (Lacey & Sherman 1991; Reeve 1992; Jacobs & Jarvis 1996). The work conflict hypothesis proposes that queen shoving serves to incite nonbreeders to work. Demonstration that shoving increases the work rate of recipients would provide strong evidence for the work conflict hypothesis, but again the evidence is equivocal. Whereas Reeve (1992) found that queen shoving increased the work rate of recipients, Jacobs & Jarvis (1996) did not. In fact in the latter study, colony members were more likely to be shoved when active in the tunnel system than resting in the nest, suggesting that shove rate is, in part, a function of how often the recipient is encountered when the queen is patrolling the tunnel system (Jacobs & Jarvis 1996). Another prediction of the work conflict hypothesis is that frequent recipients of shoving should work less, at least in the absence of the queen (Reeve 1992). Reeve (1992) found that when queens were experimentally removed from their colonies for a short period (4 h) frequent recipients of queen shoving did indeed work at a lower rate. Although in one of our colonies work levels decreased after queen removal, those individuals that decreased their work rate most were not the most frequently shoved colony members, as predicted by the work conflict hypothesis. Additionally, our results are in agreement with those of Jacobs & Jarvis (1996), who found that those mole-rats that decreased their work levels after queen removal were not the largest colony members, as predicted by the work conflict hypothesis. What evidence is there for the threat reduction hypothesis? The lack of sex bias in queen shoving found in this study, and by other researchers (Reeve & Sherman 1991; Reeve 1992; Jacobs & Jarvis 1996), could be interpreted as evidence against the threat reduction hypothesis. This is because females are thought to pose the greatest threat to the queen’s reproductive dominance (Reeve & Sherman 1991). However, the threat reduction hypothesis does not necessarily predict a sex bias in queen shoving. Both nonbreeding males and females are reproductively suppressed (Faulkes & Abbott 1997), and there is considerable evidence that behavioural contact with the queen is a key factor in the imposition of reproductive suppression in both nonbreeding males and females (Faulkes & Abbott 1993; Margulis et al. 1995; Smith et al. 1997). Queens may suppress male reproduction because males are a threat to a queen’s reproductive dominance if they mate with other reproductively active colony females. Naked mole-rats appear to lack an incest avoidance mechanism (Jarvis et al. 1994), and there is evidence that within colonies the degree of suppression among nonbreeding females is not uniform (Westlin et al. 1994; van der Westhuizen 1997). Older nonbreeding females tend to have higher bioactive luteinizing hormone concentrations, suggesting they may be less reproductively suppressed; additionally nonbreeding females with perforate vaginas have been found within colonies (Jarvis 1991). In our study three high-ranking, nonbreeding females from

one colony had elevated urinary progesterone titres indicative of reproductive activity. The lack of an incest avoidance mechanism and the occurrence of nonbreeding females with signs of reproductive activity may mean that males, if not suppressed, would pose a significant threat to a queen’s reproductive hegemony. In two colonies where the identities of those nonbreeding males that were to become breeders were known, these males were shoved by queens at a far higher rate than other nonbreeding males and breeding males. This finding appears to support the threat reduction hypothesis, as it is reasonable to assume that among males, those individuals that were to become breeders would pose the greatest threat to a queen’s reproductive dominance if they were not suppressed. If the threat reduction hypothesis is correct, and queen shoving functions primarily to suppress reproduction in colony mates, why are breeding males often shoved by queens? We found that breeding males were either shoved more than, or at similar levels to, nonbreeding males, although in one colony the breeding male was not shoved. That breeding males are frequently shoved by queens does not necessarily negate the threat reduction hypothesis. The queen appears to control the reproductive physiology of breeding males as well as of nonbreeders. Testosterone concentrations in breeding males peak during the follicular phase of the queen’s cycle, just prior to oestrus and ovulation, and are low at other times (Faulkes et al. 1991). Reproductive control by the queen over breeding males may minimize the chance that they will mate with other colony females, or may help reduce aggression associated with elevated testosterone titres, which might otherwise cause social instability (Faulkes & Abbott 1997). Alternatively, suppression of male testosterone levels with the queen’s oestrous period may benefit males by reducing the potentially harmful effects of sustained, elevated testosterone titres, such as immune suppression. However, a causal link between shoving by queens and the changes in testosterone titres of breeding males observed over the queen’s ovarian cycle has not yet been shown empirically. Disperser males have higher basal levels of plasma luteinizing hormone than nondispersers and are not the focus of shoving by the queen, a finding that lends support to the threat reduction hypothesis (O’Riain et al. 1996). Finally, in two of our colonies where the queen was removed, new queens primarily directed shoves and bites at the mates of the former queen and other high-ranking males, killing several of them. This behaviour appears to support the threat reduction hypothesis, as it may represent an extreme attempt by queens to suppress reproduction in these dominant and reproductively active males. Evidence from this study that queens target those females that pose the greatest threat to their reproductive dominance is equivocal. In only one of three colonies was the nonbreeding female that was to become the new colony queen shoved more than other nonbreeding females. In contrast, the three nonbreeding females in colony NN showing signs of reproductive activity that were to succeed the queen were shoved far less than other nonbreeding females, despite being the prime initiators of shoving

CLARKE & FAULKES: AGGRESSION IN NAKED MOLE-RATS

among nonbreeders. This was an unusual colony in that one of these three females initially replaced the queen, but was later killed and superseded by the other two, to form a dual-queened colony (Clarke & Faulkes 1997).

Conclusion Rates of shoving in colonies of naked mole-rats were strongly associated with dominance and reproductive status, and with urinary testosterone levels of males and females. The vast majority of shoves were carried out by the queen, and to a lesser extent breeding males. We found no evidence for the work conflict hypothesis, other than that high-ranking and large colony members were shoved most, a finding that supports both hypotheses. We found some evidence for the threat reduction hypothesis. Those males that posed the greatest threat to the queen’s reproductive dominance were shoved most, although evidence that queens targeted those females that posed the greatest threat was equivocal. Nonbreeding females on becoming reproductively active directed most shoves and bites at other high-ranking individuals showing signs of reproductive activity, until their reproductive dominance was established. It is possible that shoving is used to establish reproductive dominance, and, once achieved, other cues may maintain suppression in subordinates. Naked mole-rats show several other nonvocal (Lacey et al. 1991) as well as vocal agonistic behaviours (Pepper et al. 1991) which may be involved in suppression, perhaps in conjunction with more subtle behavioural cues. Cortisol did not appear to be causally implicated in the reproductive suppression of subordinates, nor did it appear to be associated with social stress. The disparity between different studies examining queen shoving may reflect differences in the composition of the colonies studied, for example, sex ratios, age, size and dominance structure. Social context at the time of observation may be important, for example, periods of reproductive quiescence or activity, social stability or competition over breeding opportunities and/or food. We suggest that the most parsimonious explanation for queen shoving is that it serves several functions, in inhibiting reproduction in subordinates of both sexes, maintaining social order, and also in inciting workrelated behaviours in colony members, all of which ultimately increase the reproductive success of queens.

Acknowledgments This work was supported by a BBSRC Ph.D. studentship, ref. no. 94316743, to F.M.C. We thank T. Burland, M. Beaumont, the anonymous referees for comments on the manuscript, and the animal technicians at the Institute of Zoology for care of the mole-rats.

References Altmann, J. 1974. Observational study of behavior: sampling methods. Behaviour, 49, 227–265.

Barfield, R. 1984. Reproductive hormones and aggressive behaviour. In: Biological Perspectives on Aggression (Ed. by K. G. Flannely, R. J. Blanchard & D. C. Blanchard), pp. 105–134. New York: A. R. Liss. Bonney, R. C., Wood, D. J. & Kleiman, D. G. 1982. Endocrine correlates of behavioural oestrous in the female giant panda (Ailuropoda melanoleuca) and associated hormonal changes in the male. Journal of Reproduction and Fertility, 64, 209–215. Braude, S. 1998. Survival of naked mole-rats marked by implantable transponders and toe-clipping. Journal of Mammalogy, 79, 360– 363. Brett, R. A. 1991. The population structure of naked mole-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. Clarke, F. M. & Faulkes, C. G. 1997. Dominance and queen succession in captive colonies of the eusocial naked mole-rat, Heterocephalus glaber. Proceedings of the Royal Society of London, Series B, 264, 993–1000. Clarke, F. M. & Faulkes, C. G. 1998. Hormonal and behavioural correlates of male dominance and reproductive status in captive colonies of the naked mole-rat, Heterocephalus glaber. Proceedings of the Royal Society of London, Series B, 265, 1–9. Faulkes, C. G. & Abbott, D. H. 1993. Evidence that primer pheromones do not cause social suppression of reproduction in male and female naked mole-rats, Heterocephalus glaber. Journal of Reproduction and Fertility, 99, 225–230. Faulkes, C. G. & Abbott, D. H. 1997. The physiology of a reproductive dictatorship: regulation of male and female reproduction by a single breeding female in colonies of naked mole-rats. In: Cooperative Breeding in Mammals (Ed. by N. G. Solomon & J. A. French), pp. 302–334. New York: Cambridge University Press. Faulkes, C. G., Abbott, D. H. & Jarvis, J. U. M. 1990. Social suppression of ovarian cyclicity in captive and wild colonies of naked mole-rats, Heterocephalus glaber. Journal of Reproduction and Fertility, 88, 559–568. Faulkes, C. G., Abbott, D. H. & Jarvis, J. U. M. 1991. Social suppression of reproduction in male naked mole-rats, Heterocephalus glaber. Journal of Reproduction and Fertility, 91, 593–604. Faulkes, C. G., Abbott, D. H., O’Brien, H. P., Lau, L., Roy, M., Wayne, R. K. & Bruford, M. W. 1997. Micro- and macrogeographical genetic structure of colonies of naked mole-rats, Heterocephalus glaber. Molecular Ecology, 6, 615–628. Honeycutt, R. L., Nelson, K., Schlitter, D. A. & Sherman, P. W. 1991. Genetic variation within and among populations of the naked mole-rat: evidence from nuclear and mitochondrial genomes. In: The Biology of the Naked Mole-Rat (Ed. by P. W. Sherman, J. U. M. Jarvis & R. D. Alexander), pp. 195–208. Princeton, New Jersey: Princeton University Press. Jacobs, D. S. & Jarvis, J. U. M. 1996. No evidence for the work conflict hypothesis in the eusocial naked mole-rat (Heterocephalus glaber). Behavioral Ecology and Sociobiology, 39, 401–409. Jarvis, J. U. M. 1981. Eu-sociality in a mammal: cooperative breeding in naked mole-rat, Heterocephalus glaber colonies. Science, 212, 571–573. Jarvis, J. U. M. 1991. Reproduction of naked mole-rats. In: The Biology of the Naked Mole-Rat (Ed. by P. W. Sherman, J. U. M. Jarvis & R. D. Alexander), pp. 384–425. Princeton, New Jersey: Princeton University Press. Jarvis, J. U. M., O’Riain, M. J., Bennett, N. C. & Sherman, P. W. 1994. Eusociality: a family affair. Trends in Ecology and Evolution, 9, 47–51. Lacey, E. A. & Sherman, P. W. 1991. Social organization of the naked mole-rat: evidence for divisions of labour. In: The Biology of the Naked Mole-Rat (Ed. by P. W. Sherman, J. U. M. Jarvis & R. D. Alexander), pp. 274–337. Princeton, New Jersey: Princeton University Press.

323

324

ANIMAL BEHAVIOUR, 62, 2

Lacey, E. A., Alexander, R. D., Braude, S. H., Sherman, P. W. & Jarvis, J. U. M. 1991. An ethogram for the naked mole-rat: nonvocal behaviors. In: The Biology of the Mole-Rat (Ed. by P. W. Sherman, J. U. M. Jarvis & R. D. Alexander), pp. 209–242. Princeton, New Jersey: Princeton University Press. Margulis, S. W., Saltzman, W. & Abbott, D. H. 1995. Behavioral and hormonal changes in female naked mole-rats (Heterocephalus glaber) following removal of the breeding female from a colony. Hormones and Behavior, 29, 237–247. Martin, P. & Bateson, P. 1986. Measuring Behaviour: an Introductory Guide. Cambridge: Cambridge University Press. O’Riain, M. J., Jarvis, J. U. M. & Faulkes, C. G. 1996. A disperser morph in the naked mole-rat. Nature, 380, 619–621. 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. 243–276. Princeton, New Jersey: Princeton University Press. Reeve, H. K. 1992. Queen activation of lazy workers in colonies of the eusocial naked mole-rat. Nature, 358, 147–149. Reeve, H. K. & Sherman, P. W. 1991. Intracolonial aggression and nepotism by the breeding female naked mole-rat. In: The Biology of the Naked Mole-Rat (Ed. by P. W. Sherman, J. U. M. Jarvis & R. D. Alexander), pp. 337–357. Princeton, New Jersey: Princeton University Press.

Reeve, H. K., Westneat, D. F., Noon, W. A., Sherman, P. W. & Aquadro, C. F. 1990. DNA ‘fingerprinting’ reveals high levels of inbreeding in colonies of the eusocial naked mole-rat. Proceedings of the National Academy of Sciences, U.S.A., 87, 2496–2500. Smith, T. E., Faulkes, C. G. & Abbott, D. H. 1997. Combined olfactory contact with the parent colony and direct contact with non-breeding animals does not maintain suppression of ovulation in female naked mole-rats (Heterocephalus glaber). Hormones and Behavior, 31, 277–288. Sherman, P. W., Jarvis, J. U. M. & Braude, S. H. 1992. Naked mole-rats. Scientific American, 267, 72–78. van der Westhuizen, L. A. 1997. Social suppression of reproduction in the naked mole-rat, Heterocephalus glaber: plasma LH concentrations and differential pituitary responsiveness to exogenous GnRH. M.Phil. thesis. University of Cape Town. Westlin, L. M., Bennett, N. C. & Jarvis, J. U. M. 1994. Relaxation of reproductive suppression in non-breeding female naked molerats, Heterocephalus glaber. Journal of the Zoological Society of London, 234, 177–188. Wingfield, J. C., Whaling, C. S. & Marler, P. 1994. Communication in vertebrate aggression and reproduction: the role of hormones. In: The Physiology of Reproduction (Ed. by E. Knobil & J. D. Neill), pp. 303–342. New York: Raven.