When controlling for body weight, one individual still significantly passes over the other more than expected by chance. This suggests that passing dominance is not purely because it is mechanically easier for larger individuals to pass over smaller ones in the tunnel system, as previously hypothesised (Lacey et al., 1991), and that other factors are involved. In Norway rats (Rattus norvegicus) passing and crawl-over behaviours are good indicators of social dominance with passing dominance being associated with agonistic dominance and perhaps also implicated in the maintenance or establishment of dominance relationships (Ziporyn & McClintock, 1991). Dominance rank calculated using passing behaviour also appears to be a good indicator of dominance relationships among naked mole-rats. Passing dominance was associated with agonistic dominance, and correlated with body weight, age, and urinary testosterone levels, all implicated as important factors in the attainment of high dominance rank in other mammals (Clutton-Brock, 1988).
Schieffelin & Sherman (1995) also found a dominance hierarchy in naked mole-rat colonies with a dominance rank based on body weight but not age or sex, using tugging contests over food to determine dominance relationships. Body weight in naked mole-rats is labile (perhaps responding to colony needs), and does not necessarily co-vary with age (Jarvis, et al., 1991; O’Riain, 1997), perhaps explaining why Tank was correlated with weight but not age in their study. However, O'Riain (1997) found that in naked mole-rat colonies formed from pairs, individuals in the first bom litter receive little aggression from littermates, whereas individuals in later litters receive aggression from animals in all older litters, providing some support for the correlation between dominance and age in this study.
Dominance hierarchies in all three colonies were highly linear. It has been suggested that, for large group sizes strict linearity is unlikely unless differences in the resource holding power (size, strength or fighting ability) of contestants are extreme (Chase, 1974; Pusey & Packer, 1997). The large variation in adult body size and urinary testosterone levels of colony members, even within litters, may explain the high linearity of naked mole-rat hierarchies. It remains unclear how large phenotypic variation can arise between littermates in highly inbred species such as naked mole-rats. Although multiple paternity within a litter has been recorded and colony members are not genetically identical (Faulkes et al., 1997), it is not yet clear if those traits that contribute to the attainment of dominance relationships in naked mole-rats are heritable. Phenotypic variation could arise as a result of a stochastic developmental process: in other rodents it has been shown that the intra-uterine position of a fetus can have an effect on its future reproductive success and sexual behaviour (vom Saal, 1989; Clarke et al., 1992). Additionally, the probability of strict linearity in large group sizes is expected to increase where individuals assess their
relative resource holding power and fight if this exceeds an evolutionary stable threshold and/or where there is psychological reinforcement of ‘losing status’ through winner-loser effects (Mesterton-Gibbons & Dugatkin, 1995).
Dominance rank appears to be the most reliable predictor of reproductive status: queens are the highest ranking females and are succeeded by the next highest ranking females. In field studies, body weight may be the most useful predictor, since behavioural observation of colony members is not usually possible. Colony O was formed by pairing a non-breeding male and a female at random, in contrast to colony N l and NN, where resident queens had attained their breeding status naturally through competition in long established colonies, perhaps explaining why the queen in colony O was not the highest ranking colony member.
In this study dominance rank was correlated with the urinary testosterone titres of males and females. In males, testosterone regulates spermatogenesis and mating behaviour but also mediates aggression (Barfield, 1984; Wingfield et al., 1994). Among females the relationship between aggression, dominance and testosterone is poorly understood. Nevertheless, at least in female mice and rats, testosterone appears to form the foundation of hormone-dependent aggression (mice: Zielinski & Vandenbergh, 1991; Rats: Albert et al., 1991; 1992; 1993). Although socially dominant animals are often the most aggressive animals, a simple relationship between dominance and testosterone levels has been difficult to establish. Such correlations are to an extent dependent on factors such as taxonomic class, age, social context and experience (Wingfield et al., 1994). It is important to realise that while testosterone promotes aggression, aggression does not necessarily facilitate dominance.
There is extensive evidence that social environments can be stressful (Sapolsky, 1990; Wingfield et al., 1994; Davies, 1996). Social stress often leads to an increase in the release of glucocorticoids, such as cortisol, from the adrenal cortex (Kaplan, 1986; Broom & Johnson, 1993; Wingfield et al., 1994). In this study urinary cortisol levels of colony members increased during the period of social instability following removal of the queen. Such changes in adrenocortical activity in response to social stress are often related to dominance status (Davies, 1996). Whereas several studies have concluded that low ranking animals are more stressed than high ranking individuals, evidenced by increased adrenocortical activity (sugar gliders: Mallick et al., 1994; female cynomolgous monkeys: Shively et a l., 1997; male olive baboons: Sapolsky, 1982; 1990), others have found the opposite to be true (dwarf mongoose: Davies, 1996; African wild dogs: Creel e ta l., 1997; female yellow-bellied marmots: Armitage, 1990). There is also evidence that an increase in glucocorticoid secretion, resulting from social stress, can suppress reproductive function (Dunbar, 1985; Wingfield et al., 1994). In this study dominance rank was correlated with urinary
cortisol levels: high ranking colony members had higher cortisol levels than low ranking individuals. Although urinary cortisol titres of individuals in only one study colony were determined, this finding suggests that high ranking naked mole-rats are more ‘stressed* than low ranking individuals. That the dominant breeding female in colony NN exhibited one of the highest urinary cortisol titres of colony members, suggests that there does not appear to be a causal link between cortisol levels and reproductive suppression in naked mole-rats (Faulkes, 1990; Faulkes & Abbott, 1997; Davies, 1996). Finally, in this study correlations between dominance and both androgens and glucocorticoids disappeared under periods of social instability. This is in contrast to several studies which have found that adrenocortical differences between dominants and subordinates are only seen during periods of social stress (Kaplan, 1986; Sachser & Lick, 1989) and studies which found a relationship between testosterone and dominance that was absent during social equilibrium (Sapolsky,
1983; Sachser & Prove, 1986; Wingfield etal., 1994).
The reproductive activation of one or several females and increase in intra- colony aggression, social instability and body weight, following queen removal, confirms the results of previous studies that the presence of the queen in colonies has a suppressive influence on all colony members of both sexes (Faulkes, 1990; Faulkes & Abbott, 1993, Jarvis, 1991; Jarvis et al., 1991). An increase in body weight by both males and females following queen removal has also been reported by other researchers (Jarvis, 1991; Jarvis et al., 1991; Lacey & Sherman. 1991). Field studies also suggest a similar pattern of weight gain (Braude, 1991). These authors also reported that removal of both the queen and the breeding male(s) or several other large non-breeders, results in severe colony disruption and an increase in intra-colony aggression and the death of colony members (Jarvis, 1991, Lacey & Sherman, 1991). We report here that removal of only the queen is sufficient to cause colony social instability and severe aggression leading to the death of some of colony members. Social order is restored when a new female has taken over, and agonistic interactions become relatively rare and subtle events. In contrast, other social bathyergids such as the Damaraland mole-rat Cryptomys damarensis and the common mole-rat Cryptomys h. hottentotus, appear to show strong incest avoidance. In these species the death of a breeder results in either a cessation of reproduction until a foreign animal joins the colony or colony fissioning and outbreeding (Bennett et al., 1996; Burda, 1995; Jarvis & Bennett, 1993; Jarvis etal., 1994;).
Following Hamilton’s rule (Hamilton, 1964), the high level of inbreeding in naked mole-rat colonies due to strong ecological constraints on dispersal (Jarvis et al.,
1994), should promote altruistic behaviour among colony members. However, despite high intra-colony relatedness (Reeve etal., 1990; Honeycutt, etal., 1991; Faulkes et al., 1997b), this and other studies have shown that colonies are not perfectly
harmonious societies (Reeve & Sherman, 1991). There is underlying competition over resources such as food (Schieffelin & Sherman, 1995; O'Riain, 1997) or breeding opportunities (Jarvis, 1991; Lacey & Sherman, 1991). The increase in linearity of dominance hierarchies following queen removal may be due to increased competition between colony members which reinforces dominance relationships through winner- loser effects. Optimal skew theory predicts intense dominance-related aggression in high skew societies (Keller & Reeve, 1994) such as naked mole-rats where, despite high intra-colony relatedness, the payoffs from becoming a breeder are large relative to a single helper’s effect on colony reproduction (Reeve & Sherman, 1991). The death of colony members following queen removal may be an extreme attempt by new queens to establish their dominance and re-impose reproductive suppression on subordinates. High ranking males, who are often reproductively active, may be killed because they are a threat to new queens if they mate with other females who have become reproductively active. That dominance rank appears to be the most important determinant of reproductive status in female naked mole-rats, may explain why intense aggression is limited to the highest ranking colony members.