Effects of early-life competition and maternal nutrition on telomere lengths in wild meerkats

Early-life adversity can affect health, survival and fitness later in life, and recent evidence suggests that telomere attrition may link early conditions with their delayed consequences. Here, we investigate the link between early-life competition and telomere length in wild meerkats. Our results show that, when multiple females breed concurrently, increases in the number of pups in the group are associated with shorter telomeres in pups. Given that pups from different litters compete for access to milk, we tested whether this effect is due to nutritional constraints on maternal milk production, by experimentally supplementing females’ diets during gestation and lactation. While control pups facing high competition had shorter telomeres, the negative effects of pup number on telomere lengths were absent when maternal nutrition was experimentally improved. Shortened pup telomeres were associated with reduced survival to adulthood, suggesting that early-life competition for nutrition has detrimental fitness consequences that are reflected in telomere lengths. Dominant females commonly kill pups born to subordinates, thereby reducing competition and increasing growth rates of their own pups. Our work suggests that an additional benefit of infanticide may be that it also reduces telomere shortening caused by competition for resources, with associated benefits for offspring ageing profiles and longevity. Data collection was conducted in the context of a long-term study, monitoring a naturally regulated population of wild meerkats at the Kuruman River Reserve, South Africa (26°58′ S, 21°49′ E), between 1994 and 2015. All meerkats were habituated to close observation (less than 1 m) and individually recognizable using small dye-marks (approx. 2 cm2, for adults and older pups) or trimming small patches of fur (approx. 0.5 cm2, for newly emerged pups) [36]. Virtually all (greater than 95%) meerkats could be voluntarily weighed on electronic scales (±0.1 g, Durascale, UK) before they commenced foraging in the morning, at midday and after sunset. Groups were visited two to three times per week to collect behavioural, life-history and bodyweight data. Observations of pregnancy, birth, infanticide, dominance, group size and rainfall were made using protocols detailed elsewhere [36,37]. Mother and father identity were assigned genetically [38,39]. Meerkats are born in an underground burrow, emerging for the first time at age 3–4 weeks. Shortly after the litter’s first emergence, a small biopsy of skin from the tail tip was collected from each pup (age 28.3 ± 3.4 days) for the determination of telomere length and parentage [39]. Skin samples were immediately transferred to 96% ethanol and stored at −20°C until DNA extraction. To investigate the effects of early nutritional environment on telomere lengths, we fed pregnant females during gestation and lactation. To minimize inter-individual differences in body condition, our experimental procedure was limited to dominant females. The supplementary feeding protocol consisted of one hard-boiled egg per day (divided equally between the morning and afternoon observation sessions) commencing 6 weeks after the end of a dominant female’s pregnancy, and continuing until the next parturition [40]. Thereafter, fed dominant females received four eggs per week until the pups were weaned. This feeding protocol occurred between August and November in 2011 and 2012. Control females were pregnant during the same period and did not receive supplemental food. We investigated how infanticide by dominant females affects the number of competing pups and the likely consequences for telomere lengths in her own litter. While previous analyses of the distribution of infanticide have focused on consequences for the victim mother (i.e. whether her litter survives or is killed [35,37]), we quantified the benefits of infanticide for the perpetrator (i.e. how many competitor pups she removes). We identified periods when the dominant female is most likely to kill pups born to other females (the 30 days prior to her own parturition, hereafter termed ‘high infanticide period’) and least likely (the 30 days immediately after giving birth, hereafter termed ‘low infanticide period’) [27]. We then assessed subordinate litter survival probabilities and the total number of subordinate pups surviving to emergence during these two periods. Parturition for all females could be identified by sudden weight loss and change in body shape [36], and pup production for each period was measured as the number of pups born that survived to emergence from the birth burrow. We used quantitative PCR (qPCR) analysis to measure telomere length in skin samples, based on published protocols with some modifications [41,42]. This measure represents the average telomere length across cells in a sample and is reported as the level of telomeric sequence abundance relative to a reference non-variable copy number gene (T/S ratio). Further details of DNA extraction and qPCR analysis can be found in the electronic supplementary methods. Statistical analyses were carried out in R v. 3.2.3, using a stepwise model simplification approach [43,44]. Initially all fixed terms of interest were fitted, followed by the stepwise removal of terms whose removal from the model resulted in a non-significant change in deviance (using maximum log-likelihood estimation), until the minimal adequate model (MAM) was obtained, in which only significant terms remained. Dropped terms were then added back in to the MAM to confirm their non-significance. The homoscedasticity and normality of residuals were confirmed by visual inspection, and all continuous predictors were scaled to a mean of 0 and standard deviation of 1. The significance of all terms was tested either by removing the terms from the MAM (if the term was in the MAM) or by adding the terms to the MAM (if the term was not included in the MAM). Analysis using Akaike’s information criterion correcting for small sample size (AICc) and inspection of the top model set (for which AICc differed by less than 2) yielded qualitatively identical results [45]. We ran four sets of statistical models: first to investigate the determinants of pup telomere lengths in the large correlative dataset, second to investigate how experimental supplementary feeding of mothers impacted pup telomere lengths, third to investigate whether early-life telomere lengths predict survival into adulthood and fourth to investigate the consequences of infanticide for pup competition. Our primary interest was the effect of the number of competing pups on telomere lengths at emergence from the natal burrow. For each sampled pup, we assessed the number of rival pups (aged under 90 days) present in the group, every day between the focal pup’s birth and day of sampling for telomere length. The average of these daily rival counts represents our measure of overall competition experienced by the focal pup prior to sampling, hereafter termed ‘pup number’. This estimate of pup competition includes littermates and pups from older and younger litters born to the dominant female and subordinate females. We controlled for maternal factors that may influence offspring quality, including weight at conception, age (mean 4.9 years, range 1.2–8.0) and dominance status (dominant or subordinate) [46]. Social group size (average number of adult group members calculated as above for pup number) and rainfall (mm) in the month before birth can also both influence offspring quality [47]. Pup sex (male, female or unknown) and age at capture were also controlled for. We included these individual, maternal, environmental and social predictors, with our estimate of pup number, in a general linear mixed-effects model (GLMM), with pup telomere length as the response. Cohort year, group identity (ID), mother ID and litter ID were included as random terms, to account for the non-independence of pups within years, groups, mothers and litters. Telomere lengths were available for 230 pups from 63 litters in 13 groups, born between 2009 and 2012. We also tested the effect of paternal age (mean 4.1 years, range 1.4–6.1) on pup telomere lengths in a reduced dataset for which the father’s date of birth could be accurately determined (78 pups from 23 litters in seven groups). To test the effect of supplementary feeding of the pregnant and lactating mother on pup telomere lengths, we included experimental treatment (fed/control) as a two-level factor in a GLMM, with pup telomere length as the response and litter ID as the random term. Given our smaller sample size for the experimental dataset, only terms found to be significant in the larger correlative model were included, and two-way interactions between these and treatment. Telomere lengths were available for 25 pups from eight litters in each treatment. We investigated whether pup telomere lengths predicted survival to adulthood (1 year old). Sub-adult meerkats do not disperse [31,48], and any disappearance from the group before reaching adulthood is therefore likely to reflect mortality. We removed any individuals dying before reaching nutritional independence (90 days) as death at this early stage typically occurs due to starvation, predation or becoming separated from the group, these sources of mortality are unlikely to reflect variation in telomere lengths. We used a binary term for survival to adulthood as the response in a binomial mixed-effects model. We included pup telomere length as a predictor. We also controlled for other predictors known to influence telomere lengths and survival in young meerkats: sex, group size, rainfall, maternal dominance status and maternal age [47]. We controlled for the effects of pup body-weight on survival, by including their bodyweight at age 40 days in the model. Group ID, mother ID and litter ID were included as random terms. This model was fitted to a dataset of 178 individuals: 161 pups from 51 litters born to dominant females and 17 pups from seven litters born to subordinates. The maximum confirmed lifespan for meerkats in our population is 12.2 and 12.4 years, for males and females, respectively. We contrasted the fates of subordinate litters born in periods of high and low dominant female infanticides. First, for each dominant female parturition (n = 158), we counted subordinate parturitions during the two periods (30 days before and after dominant parturition). Infanticide typically takes places shortly after birth, so we classed each subordinate parturition as a ‘success’ or ‘infanticide’ according to whether the litter survived its first 2 days (litter loss after this point is more likely to be due to starvation or predation [35,37]). Although newborn litters remained in the burrow for up to 4 weeks, their survival could be recorded daily by observing whether the group continued to leave babysitters during foraging trips [35]. The number of successes and infanticides were then used as the response term in a binomial mixed-effects model, with the high/low infanticide period fitted as a two-level predictor. The random terms were dominant female pregnancy ID, dominant female ID and group ID. Second, for each dominant female parturition, we calculated the total number of emerging subordinate pups born during the two infanticide periods, and fitted this as the response term in a GLMM with a Poisson distribution. The main predictor of interest was the two-level high/low dominant female infanticide period, and we controlled for the number of subordinate females giving birth.