Competition, breeding success and ageing rates in female meerkats

Authors


Stuart Sharp, Large Animal Research Group, Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK.
Tel.: +44 0 1223 336673; fax: +44 0 1223 336676; e-mail: sps44@cam.ac.uk

Abstract

Competition between females is particularly intense in cooperatively breeding mammals, where one female monopolises reproduction in each group. Chronic competition often affects stress and may therefore have long-term consequences for fitness, but no studies have yet investigated whether intrasexual competition has effects of this kind and, in particular, whether it affects rates of reproductive senescence. Here, we use long-term data from a wild population of meerkats to test whether reproductive success and senescence in dominant females are affected by the degree of intrasexual competition experienced prior to dominance acquisition. Females that experienced greater competition had lower breeding success and higher rates of reproductive senescence. Furthermore, females that were evicted from the group more frequently as subordinates had lower breeding success when dominant. We conclude that the intense intrasexual competition between females in cooperatively breeding groups may carry fitness costs over a longer period than is usually recognised.

Introduction

Intrasexual competition for breeding opportunities is a prominent feature of many animal societies. Most studies of reproductive competition in vertebrates have focused on males, but recent research has shown that females also commonly compete for access to breeding territories or other resources required for producing and raising offspring (Clutton-Brock, 2007, 2009). In group-living species, reproduction is often monopolised by a small proportion of group members (Hager & Jones, 2009) and females must therefore compete for social rank and breeding status (Woodroffe & Macdonald, 1995; Holekamp et al., 1996; Sterck et al., 1997; Koenig et al., 1998). Higher-ranked females usually have greater reproductive success (e.g. Holekamp et al., 1996) and, in some cases, survival rates (e.g. Buffenstein, 2005), but the competition for and acquisition of status may carry significant fitness costs, which are often overlooked.

In many cooperatively breeding mammals, a single dominant female is responsible for most or all of the breeding attempts in the group (Solomon & French, 1997) and prevents subordinate females from breeding by physiological suppression, infanticide or eviction from the group (Digby, 1995; Faulkes & Abbott, 1997; Creel & Creel, 2002). Competition for dominance status is therefore particularly intense in these species, often more so in females than males (Clutton-Brock, 2007, 2009). Although competition for dominance frequently entails short-term costs associated with physical conflict (e.g. Faulkes & Abbott, 1997), prolonged periods of intense competition are likely to be stressful and may therefore carry longer-term fitness costs, but this has yet to be studied. The long-term fitness consequences of reproductive suppression for those subordinates that later become dominant have also received little attention, despite evidence showing that suppression may induce chronic stress (Creel, 2001). In vertebrates, reproductive competition and prolonged exposure to glucocorticoid stress hormones can reduce breeding success (Woodroffe & Macdonald, 1995; Cyr & Romero, 2007) and increase the rate of reproductive senescence (Bonduriansky et al., 2008; Monaghan et al., 2008), but these processes have never been investigated in a cooperative breeder.

We investigated the effect of intrasexual competition on breeding success and reproductive senescence in dominant female meerkats (Suricata suricatta, Desmarest), a cooperatively breeding mongoose endemic to arid regions of southern Africa. Meerkats live in social groups consisting of a single dominant female, who largely monopolises reproduction in the group, a dominant male, who is the father of most of her offspring, and between 2 and 50 subordinate males and females that guard and feed the young born to the dominant pair (Clutton-Brock et al., 1999, 2006, 2008; Griffin et al., 2003; Spong et al., 2008). Breeding success in dominant females initially increases during their tenure before declining as they age (Sharp & Clutton-Brock, 2010). Subordinate females are able to breed from 1 year of age, but reproductive attempts are infrequent and usually unsuccessful because of suppression by dominants, which includes eviction from the group and infanticide (Clutton-Brock et al., 1998, 2008; Young et al., 2006). This behavioural suppression is not known to lead to the cessation of reproductive cycles in subordinates, but evicted females do suffer a chronic elevation of their stress hormone levels, which is associated with reduced conception rates and increased abortion rates (Young et al., 2006). Following the death of the dominant female or during the formation of a new group, subordinate females compete for the vacant breeding position. If one subordinate is older than all other females in the group, then she becomes the new dominant soon after; where competitors are members of the same litter and therefore the same age, however, there is usually a longer period of conflict that is ultimately won by the heaviest female (Clutton-Brock et al., 2006; Hodge et al., 2008).

Here, we used over 15 years of data to test whether reproductive success and senescence in dominant female meerkats are affected by the degree of intrasexual competition experienced prior to the acquisition of dominance status. As competition for dominance and reproductive suppression often involve prolonged periods of physical aggression and are highly stressful, we predicted that breeding success would be lower and the rate of reproductive senescence higher in females who (i) experienced a higher level of competition for dominance and (ii) were evicted more frequently before becoming dominant.

Methods

Data collection

Data were collected from a wild population of meerkats in the South African Kalahari (26°58′S, 21°49′E) between January 1994 and November 2009. Around 2000 individuals living in more than 40 social groups were closely monitored, all of whom were habituated to observation from < 2 m. All individuals were marked with subcutaneous transponder chips and were recognisable in the field by unique dye marks applied to their fur. Each group was visited approximately once every 3 days to record all key life-history events and changes in group composition, including births, deaths, evictions and changes in pregnancy or dominance status. In addition, more than 95% of individuals were trained to climb onto an electronic balance and could be weighed most mornings before they went foraging. Rainfall was measured daily with a rain gauge. Further details of the study population, including the habitat and climate, are given by Russell et al. (2002).

We analysed the effect of intrasexual competition on age-related variation in the reproductive success of 38 dominant females in 18 groups. Analysis was restricted to females for several reasons: (i) reproductive competition is more intense in females than males (Clutton-Brock et al., 2006); (ii) the exact age is known for a greater proportion of females (> 90%), because males are the dispersive sex and many in the population are immigrants (Clutton-Brock et al., 1999; Spong et al., 2008); (iii) parentage is more easily determined for females, and hence, their breeding success is better known (Griffin et al., 2003; Hodge et al., 2008; Spong et al., 2008); and (iv) reproductive success is already known to senesce in females (Sharp & Clutton-Brock, 2010). Females were excluded if they were still alive in November 2009. A small number were also excluded because they emigrated (= 2) or were dominant for less than a year (= 4). We focused on a single measure of reproductive success that combines breeding frequency, fecundity and offspring survival and thus provides a useful measure of fitness: the number of pups produced per year that survived to independence. Following previous studies, 90 days was taken as the age of independence as pups are fed by all group members until they can forage without assistance, when about 3 months old (Brotherton et al., 2001).

Statistical analysis

For each female, age was measured in years after birth date (i.e. a female was considered age zero until she was 365 days old), and the number of independent pups produced was measured for each full year of life as a dominant up to and including the age of 8; only two females lived beyond this age and were therefore excluded to minimise bias and heteroscedasticity. The relationship between age and reproductive success was analysed using generalised linear mixed effect models (GLMMs) in the ‘lme4′ package (Bates & Maechler, 2009) for the R environment, version 2.9.1 (R Development Core Team, 2009). Poisson models with a log-link function were used, and female identity and group identity were fitted as random terms to control for the nonindependence of repeated measures of the same individual or group. A small number of dominant females moved from one group to another during their lives, so these random terms were not nested.

Previous analyses have shown a quadratic relationship between age and reproductive success (Sharp & Clutton-Brock, 2010), so both age and its square were fitted as fixed effects in the initial model. Other terms known to affect annual breeding success in dominant females were also fitted as fixed effects, which are as follows: female weight, rainfall and group size (Hodge et al., 2008; Sharp & Clutton-Brock, 2010). Female weight was calculated as the mean weight during the year excluding measurements made during pregnancy and, for females of age zero, those made prior to nutritional independence. Rainfall was the total rainfall (in mm) for that year, and group size was calculated as the mean group size (excluding pups) during the year. To control for the selective disappearance of individuals from the population according to their quality, age at last reproduction was fitted as a fixed effect in the initial model together with its square (van de Pol & Verhulst, 2006). This is a widely used procedure for minimising bias when estimating within-individual ageing patterns from mixed effect models with individual fitted as a random effect (e.g. Nussey et al., 2007; Bouwhuis et al., 2009). Previous analyses of age-specific breeding success have shown that selective appearance does not occur in this population (Sharp & Clutton-Brock, 2010) and, in the sample of females included here, the variance in age at first reproduction was extremely low, so this variable was not fitted. Two aspects of intrasexual competition were measured and fitted as fixed effects: the level of competition for dominance (high or low), and the number of times a female was evicted prior to the acquisition of dominance (range = 0–9). The level of competition for dominance reflected whether or not females shared the group with female littermates of the same age (i.e. sisters) in the year before dominance acquisition, a year being the minimum age at which a female became dominant. Females were considered to have experienced high competition if they had at least one same-age sister present in the group for more than half of the year (> 182 days) prior to the start of their dominance tenure. The remaining females were classified as having experienced low competition, either because they had no female littermates or because all such littermates had died more than 6 months before dominance acquisition. Finally, the size of each female’s natal litter and her weight when she first emerged from the natal burrow were fitted as fixed effects. Natal litter size was measured at emergence, because litter size at birth is rarely known (although the two measures are correlated, Russell et al., 2003), and this term was included to separate the effects of competition for dominance between sisters and the effects of sibling competition more generally. Weight at emergence was measured as the mean daily weight during the 7 days following emergence and was fitted to control for variation in the condition of the individuals and the environment during early life. All biologically meaningful first-order interactions were fitted. A significant interaction between the age terms and the level of competition for dominance or eviction frequency would indicate an effect of that factor on the rate of senescence; a significant interaction between the age terms and natal litter size or weight at emergence would suggest that senescence rates are influenced by other aspects of early-life conditions.

Prior to model selection, colinearity between all explanatory variables was assessed visually and by calculating variance inflation factors (VIFs), following Zuur et al. (2009). All variables were included in the initial model because pairwise correlations between covariates were weak (< 0.5 in all cases except age and group size, for which = 0.5) and no variable had a VIF of > 3. The initial model was then refined using backwards stepwise deletion. Model terms were removed in order of increasing test statistic value if likelihood ratio tests indicated that they did not explain any significant variation. Terms were assessed sequentially, with the impact of a lower-order term only considered if no corresponding higher-order term was present in the model. This process was repeated until the minimal model was obtained; each removed term was then put back into the minimal model to obtain the level of nonsignificance and to ensure that significant terms had not been inappropriately dropped. Interactions are only presented if they were found to explain significant variation. The final model was validated by plotting the distribution of the residuals, residuals vs. fitted values and residuals vs. each of the covariates (Zuur et al., 2009). To further verify the results of the final model and, in particular, to check that quadratic effects and interactions were not primarily driven by increases in reproductive success during the first few years of life (for which the sample size is greater), all of the same model terms except for age2 were fitted in a new GLMM using a subset of data covering only the period of senescence. Reproductive success in dominant females typically peaks at 4 years of age (Sharp & Clutton-Brock, 2010), so all data for females aged three or younger were excluded from this second analysis, giving a sample size of 16 dominant females in 12 groups.

Results

Reproductive success throughout the lifetime

The number of independent pups produced per year by dominant females increased with age up to a peak at around 4–5 years and declined thereafter, the final model containing a significant quadratic effect of age (Table 1, Fig. 1). This relationship varied significantly with the level of competition for dominance; females that experienced a high level of competition produced fewer independent pups per year and suffered a higher rate of decline in this trait (Table 1, Fig. 1). This could not be attributed to high competition amongst larger groups of siblings as there was no significant effect of natal litter size (Table 1). Furthermore, there were no significant interactions between the age terms and natal litter size or weight at emergence. There was a significantly negative relationship between the number of evictions that females experienced before becoming dominant and the number of pups they subsequently produced per year that survived to independence (Table 1, Fig. 2). Finally, there was a significant effect of female weight, with heavier females producing more independent pups per year; no other terms were retained in the final model (Table 1).

Table 1.   A GLMM of age-specific variation in the reproductive success of dominant female meerkats throughout their lifetimes.
 No. of pups produced per year that survived to independence
Random effectsVariance
Female identity0.113
Group identity0.154
Fixed effectsEstimate ± SEχ2P*
  1. GLMM, generalised linear mixed effect model; ALR, age at last reproduction.

  2. *Significant values (< 0.05) are highlighted in bold.

Intercept−2.301 ± 0.841
Age (years)0.243 ± 0.18228.458< 0.001
Age2−0.024 ± 0.01932.943< 0.001
ALR (years)3.0370.081
Weight (g)0.005 ± 0.0015.8370.016
Rainfall (mm)0.0420.839
Group size0.6300.427
Natal litter size1.8800.170
Weight at emergence (g)3.3340.068
Level of competition−1.702 ± 0.54015.4180.001
No. of evictions experienced−0.157 ± 0.04611.2130.001
Age: level of competition0.795 ± 0.2559.5360.002
Age2: level of competition−0.096 ± 0.02811.723< 0.001
Figure 1.

 The effect of age and the level of competition for dominance on the number of independent pups produced per year by dominant female meerkats. The plots show generalised linear mixed effect model (GLMM) predictions with partial residuals for low competition (solid line, solid circles) and high competition (dashed line, hollow circles), plotted for the median weight (750 g) and median number of evictions experienced prior to dominance acquisition (1).

Figure 2.

 The effect of the number of evictions experienced prior to dominance acquisition on the number of independent pups produced per year by dominant female meerkats. The plot shows generalised linear mixed effect model (GLMM) predictions and partial residuals for the median age (3 years), median weight (750 g) and low competition for dominance.

Reproductive success after the onset of senescence

An analysis restricted to data for females aged four and above produced similar results to the model that included data from throughout their lifetimes. Reproductive success declined with age, and the rate of decline was significantly higher in females that experienced a high level of competition for dominance (Table 2). Heavier females were again found to produce more independent pups per year, but the negative relationship between the number of evictions experienced before becoming dominant and subsequent reproductive success was marginally nonsignificant in this smaller sample (Table 2).

Table 2.   A GLMM of senescent declines in the reproductive success of dominant female meerkats aged four and above.
 No. of pups produced per year that survived to independence
Random effectsVariance
Female identity0.018
Group identity∼0
Fixed effectsEstimate ± SEχ2P*
  1. GLMM, generalised linear mixed effect model.

  2. *Significant values (< 0.05) are highlighted in bold.

Intercept−0.346 ± 1.091
Age (years)0.017 ± 0.06525.855< 0.001
Weight (g)0.003 ± 0.0016.0930.014
Level of competition1.423 ± 0.49211.7720.003
No. of evictions experienced3.3850.066
Age: level of competition−0.326 ± 0.09414.022< 0.001

Discussion

Dominant females that experienced a higher level of competition for the dominant position had lower breeding success and higher rates of reproductive senescence. Furthermore, females that were more frequently evicted as subordinates had lower breeding success when dominant. The effects of competition for dominance between females did not simply reflect sibling competition, because the size of a female’s natal litter did not affect her reproductive success or senescence when dominant. In addition, there was no effect of an individual’s weight at emergence, indicating that dominant females with higher reproductive success and slower ageing rates were not merely in better condition during the early part of their lives. Together, these results support our prediction that the intense intrasexual competition between females in cooperatively breeding groups carries long-term fitness costs.

When dominant positions in meerkat groups become available, they are normally taken by the oldest subordinate female, but where multiple females from the same litter are present, there is usually a more prolonged period of conflict before the heaviest female eventually acquires the role (Clutton-Brock et al., 2006; Hodge et al., 2008). As subordinate females cannot predict when such a vacancy will arise, selection should favour those that can increase and maintain high body weight relative to that of their female littermates from an early age (Hodge et al., 2008). Furthermore, selection pressure for high body weight is intensified by the positive effect on breeding success shown here and in previous studies (Russell et al., 2003; Hodge et al., 2008). High growth rates in early life are, however, known to be costly in many animals and may lower breeding success and increase ageing rates (Metcalfe & Monaghan, 2001, 2003). In addition, extended periods of competition are likely to increase the levels of stress hormones, exposure to which can affect parental behaviour, reduce breeding success and accelerate senescence (Cyr & Romero, 2007; Monaghan et al., 2008; Spencer et al., 2010). Finally, there may be significant costs in maintaining the dominant position (Creel, 2001). Female meerkats that experience high levels of competition often continue to share the group with same-age sisters after gaining the dominant position, and it may be harder to suppress reproduction in these older females as suggested by a previous study showing that the breeding frequency of subordinates increases with their age (Clutton-Brock et al., 2008).

Although intrasexual competition has been shown to affect ageing rates in polygynous mammals (e.g. Mysterud et al., 2005), our study is the first to investigate this in a cooperatively breeding species. It has become widely recognised that sociality affects senescence (Bourke, 2007), but most research has focused on the relationship between social structure and longevity using captive model systems (e.g. Keller & Genoud, 1997; Buffenstein, 2005). Our results provide the first evidence from a wild animal population that intrasexual competition arising from the nature of the social system is associated with higher rates of senescent decline in reproductive success. This supports the previous findings that adverse conditions experienced in the early years of life can increase the rate of reproductive senescence (Nussey et al., 2007; Bouwhuis et al., 2009). Caution is needed when determining the causality and mechanisms underlying correlations such as this, and it may be that competition for dominance is associated with some unmeasured trait, which drives the observed patterns of senescence. However, it is clear that social factors should be given greater consideration in future attempts to explain intraspecific variation in ageing rates.

The number of times that a female was evicted prior to becoming dominant did not affect the rate of reproductive senescence, but was negatively correlated with breeding success. The relationship was weaker in the model that only included females aged four or above, but this is presumably because of the reduction in sample size. Eviction in meerkats is thought to be one example of stress-mediated reproductive suppression, and evicted females suffer elevated levels of stress hormones (Young et al., 2006). Our results indicate that the costs of being evicted and the associated increase in stress are more pervasive than has previously been appreciated, and support suggestions that exposure to stress in early life can carry long-term fitness costs (Monaghan et al., 2008). Stress levels are also known to influence longevity (Parsons, 1995), and the number of evictions experienced by females may affect the rate of age-specific declines in survival rather than breeding success.

To conclude, our study of female meerkats shows that intrasexual competition experienced in the early years of life may carry significant fitness costs over a longer period than is often appreciated. The results also indicate that social factors should be an important consideration when investigating individual variation in reproductive success and senescence. In cooperatively breeding species, the intense competition for dominance status may also impact on survival, and future studies of the relationship between sociality, intrasexual competition and longevity should prove enlightening.

Acknowledgments

We are grateful to the Kotze family and Northern Cape Conservation for allowing us to conduct research in the Kalahari; Martin Haupt, Johan du Toit, Elissa Cameron and Penny Roth for logistical support; Sinead English and two anonymous reviewers for providing helpful comments on an earlier draft; and the many volunteers, field staff, Ph.D. students and postdocs who assisted us in data collection. NERC, The Newton Trust, The Leverhulme Trust, the University of Cambridge and the Earthwatch Institute provided financial support.

Ancillary