Dispersal-induced social stress prolongs gestation in wild meerkats

Abstract

In the majority of mammals, gestation length is relatively consistent and seldom varies by more than 3%. In a few species, females can adjust gestation length by delaying the development of the embryo after implantation. Delays in embryonic development allow females to defer the rising energetic costs of gestation when conditions are unfavourable, reducing the risk of embryo loss. Dispersal in mammals that breed cooperatively is a period when food intake is likely to be suppressed and stress levels are likely to be high. Here, we show that pregnant dispersing meerkats (Suricata suricatta), which have been aggressively evicted from their natal group and experience weight loss and extended periods of social stress, prolong their gestation by means of delayed embryonic development. Repeated ultrasound scans of wild, unanaesthetized females throughout their pregnancies showed that pregnancies of dispersers were on average 6.3% longer and more variable in length (52–65 days) than those of residents (54–56 days). The variation in dispersers shows that, unlike most mammals, meerkats can adapt to stress by adjusting their pregnancy length by up to 25%. By doing so, they potentially rearrange the costs of gestation during adverse conditions of dispersal and enhance offspring survival.

1. Introduction

Gestation length is relatively invariable in most mammalian species [1] and although flexibility to adjust this energetically demanding period would be adaptive in unpredictable environments, only few species have been shown to possess such flexibility [2,3]. For example, in wild California leaf-nosed bats (Macrotus californicus), low body temperature slows the rate of embryonic development [4]; and in the short-tailed fruit bat (Carollia perspicillata), stressful laboratory conditions impede prenatal development [5]. Bat species have a more variable gestation length as a consequence of delayed post-implantation embryonic development and can prolong their gestation by up to 50% [4–7]. The ability to delay embryonic development could be adaptive when food is scarce or predation rates are likely to be high [5]. Stress-induced retardation of development would maximize the amount of time offspring spend in the safest stage of development (cf. safe harbour hypothesis) [8] and defer the steeply rising energetic demands of the final quartile of gestation [9], reducing the risk of embryo loss.

Mammals can adjust gestation length by either delaying implantation of the blastocyst or by delaying the post-implantation development of the embryo [3,10]. Delayed implantation occurs in nine mammalian orders (Diptrotodontia, Dasyuromorphia, Eulipotyphyta, Cingulata, Carnivora, Rodentia, Chiroptera, Lagomorpha and Cetartiodactyla) and allows mating when both the male and female are in peak physical condition [3,10]. It has been shown to be dependent on maternal [11], environmental [12] and social [13] factors. The fact that delayed implantation has been observed in a wide range of taxa shows its ecological significance and suggests that it has evolved independently several times [14]. While delayed implantation is common in mammals, delayed development has only been observed in 10 bat species and little is known about its ecological function [3].

Here, we show that cooperative meerkats exhibit variable gestation length due to delayed post-implantation development. Meerkats live in groups of 2–50 individuals with a dominant pair that monopolizes most of the reproduction and several closely related subordinate helpers [15,16]. The dominant female aggressively evicts subordinate females when subordinates are pregnant. Evictees either abort their pregnancy (stress-induced reproductive suppression) and return to their group [17], or permanently disperse in multiple-member coalitions to establish a new group and often bring their pregnancies to term [18,19]. Most dispersing females are pregnant at eviction, but those who are not typically conceive with roaming males soon after eviction [20]. One coalition member attains dominance in the new group, while the remaining coalition members remain subordinate [19]. Aggression at eviction and costs associated with dispersal increase acute stress levels (as measured by faecal glucocorticoids) as well as reduce foraging efficiency and body mass [17,21]. Despite these high dispersal costs, dispersing meerkats try to bring their pregnancies to term to rapidly augment their new group [18,19]. We hypothesized that gestation is delayed in females experiencing stressful social conditions during eviction and dispersal.

2. Methods

(a) . Study population

We followed 44 groups of habituated wild meerkats between September 2013 and January 2019 at the Kalahari Meerkat Project (KMP) located on the Kuruman River Reserve (26° 59′ S, 21° 50′ E) in South Africa. One individual per resident group carried a VHF collar as part of the long-term activities at the KMP (Biotrack Ltd, Wareham, UK: 23 g) [22]. We fitted VHF collars to subordinate females immediately after eviction from the resident group (Holohil Systems Ltd, Ontario, Canada: 25 g, approx. 3.5% of meerkat body mass). To mount the collars, individuals were sedated using a mixture of isoflurane and oxygen in compliance with the KMP protocol [22]. Dispersing and resident individuals were tracked every 4–7 days by means of VHF telemetry. We only considered pregnancies where pups were born successfully. This could be guaranteed by observing lactation marks on the mother and other group females that helped with nursing (pups stay hidden under ground for the first four weeks of life).

(b) . Visual pregnancy assessment

Meerkats were habituated to the presence of people (less than 1 m) and trained to climb onto a portable weighing scale allowing for regular weight measures to corroborate the visual pregnancy assessment. Abdominal swelling and weight gain become apparent 28 days after conception [23] (electronic supplementary material, figure S1), and we so measured the onset of 96 pregnancies in 46 dispersers and 258 pregnancies in 103 residents. We identified parturition from a sudden change in abdominal shape and weight loss [23] (electronic supplementary material, figure S1).

Some evicted females lost their pups after birth and returned to their natal groups. Hence, the same female may have been evicted and pregnant multiple times. Therefore, some dispersers—and some residents, having multiple pregnancies throughout their lives—were included in the analysis more than once. On average, we included 2.1 pregnancies per female in dispersers (s.d. = 1.5, min–max = 1–7) and 2.5 pregnancies per female in residents (s.d. = 2.5, min–max = 1–13).

(c) . Ultrasound scans

Because abdominal swelling is not an unequivocal measure of pregnancy stage, we used sequential transcutaneous ultrasound scans to precisely monitor 13 additional pregnancies, i.e. eight dispersers and five residents (electronic supplementary material, table S1). We trained animals to hold upright still for the scanning procedure by providing small amounts of water and egg (electronic supplementary material, figure S2). We performed scans using an ultrasound probe (linear array L742, SonoScape Ltd, Nanshan, China) and water-based lubricant. The procedure could usually be executed every 4–7 days without the need for anaesthesia, and we identified time of implantation and measured relevant gestational features of embryonic development until parturition (figure 1; electronic supplementary material, figure S5–S11, table S2). On the earliest ultrasound scans where we could detect pregnancies, the embryonic vesicle containing amniotic fluid was visible and embedded in the decidua (figure 1a). Shortly after, we could identify the embryonic structure in form of a hyperechogenic, grain of rice-like mass and the early formation of the endotheliochorial placenta, confirming successful implantation of the early embryo [24]. We hereafter refer to these earliest observations of implanted embryos as time of implantation.

Figure 1.

Cross-sectional ultrasound scans of embryos during developmental stages of a meerkat gestation. Embryonic development 0 days (a), 6 days (b), 12 days (c), and 36 days (d) after implantation. 1, Maternal spine; 2, amniotic fluid; 3, hyperechogenic mass; 4, thickened placenta ring; 5, outline uterus; 6, embryo; 7, embryonic vesicle; 8, yolk sac; 9, embryonal heart; 10, lungs; 11, liver; 12, intestines. (a) The early embryo has just implanted in the decidua. (b,c) The embryo and yolk sac are clearly visible inside the amniotic cavity. (d) The embryo has become large and its organs are detectable. The scales along the right picture margins are given in centimetres.

We measured the outer diameter of the embryonic vesicle as an estimate of total embryonic size (electronic supplementary material, figure S3) using the Zeiss ZEN blue software (Zeiss GmbH, Jena, Germany). To define a standardized time of implantation, we selected three pregnancies for which embryonic vesicle diameter on the day of earliest detection was smallest (less than 0.4 cm, electronic supplementary material, table S1) and signs of implantation present. From these three pregnancies, we obtained an average embryo development curve starting with the day of implantation using the loess function in R, v. 4.1.2 [25] (electronic supplementary material, figure S4). We then used the average development curve to backdate implantation of pregnancies with embryonic vesicle diameter greater than 0.4 cm on the day of earliest detection (electronic supplementary material, figure S4). On average, the earliest-detection embryo size was 0.58 cm (interquartile range = 0.54–0.61 cm, min–max: 0.30–0.86 cm) when the conceptuses were still spherical. We only used pregnancies with earliest-detection embryo size less than 1 cm to reduce noise arising from different angles of non-spherical scan cross-sections (electronic supplementary material, figure S5). Where possible, we obtained negative scans a few days prior to implantation to back up our estimate (electronic supplementary material, table S1–S2). Around parturition, we visited animals every other day and could therefore estimate birth dates with ±1 day accuracy.

(d) . Palpation

We observed two additional pregnancies by palpation during the captures of two dispersing females under anaesthesia (electronic supplementary material, table S1). By touching the animals' abdomen, we could identify the embryo palps and confirm the pregnancies previously observed by visual assessment. The identification of the embryos was done in collaboration with a veterinarian (S. Patterson, Royal Veterinary College, University of London, UK). We estimated the size of the embryo to be at least 0.9 cm at the time of palpation (i.e. max. measure of embryonic vesicle diameter on early ultrasound images), as it would be unlikely to detect palps smaller than this size. We backdated the time of implantation with the above-described method. Embryos may have been larger at the time of palpation, but we used a conservative measure to avoid overestimation of pregnancy length.

(i) . Visual assessment

We used a linear mixed effects model in the R package lme4 [26] with log-transformed gestation time (Δt first abdominal swelling-parturition) as response (right-skewed distribution) to test the difference between dispersers (n = 96) and residents (n = 258). In addition, we investigated the influence of social status (dominant versus subordinate, see electronic supplementary material text) and number of prior pregnancies on development time. It was suggested that parity [5] and competition between social ranks can affect pregnancy length [27]. To ensure that covariates were not correlated with each other, we calculated variance inflation factors for the model coefficients [28], none were correlated (max VIF = 1.28). We used individual identity as a random intercept term.

(ii) . Ultrasound and palpation

Here, we used a non-parametric Kruskal–Wallis test (KW) in the basic R environment to test the difference in gestation length (Δt implantation–parturition) between dispersers (n_ultrasound = 8 + n_palpation = 2) and residents (n_ultrasound = 5). To confirm that the two pregnancies detected by palpation did not bias the ultrasound results, we performed a second KW without the palpation pregnancies to investigate the difference between dispersers and residents. We performed a third KW to test for differences in gestation length between dominant, subordinate and lone females. We used non-parametric tests because the data were not normally distributed and did not use random terms to avoid overfitting. However, except for two individuals of which each had two pregnancies (electronic supplementary material, table S1), all pregnancies were from different individuals originating from seven different natal groups.

3. Results

Both methods—visual and ultrasound—showed that, on average, dispersing females had longer pregnancies than resident females (figure 2). Based on visual assessment, dispersers had an average post-assessment (i.e. abdominal swelling) development time of 45.1 ± s.e. 1.6 days while that of residents was 39.3 ± 0.7 days (est = 0.12, s.e. = 0.04, p = 0.005, electronic supplementary material, table S3). The ultrasound scans revealed that dispersers can indeed prolong their pregnancies by means of delayed embryonic development, with an average post-implantation duration of 57.9 ± 1.39 days in dispersers versus 54.6 ± 0.16 days in residents (χ = 4.17, p = 0.041, electronic supplementary material, table S3). Post-implantation periods ranged from 52 to 65 days in dispersers and from 54 to 56 days in residents. Gestation was on average longer when measured with ultrasound because embryos could be detected on average one week before first abdominal swelling was visible. Based on ultrasound, on average dispersers prolonged their post-implantation time by 6.3%. With the shortest observed post-implantation period being 52 days and the longest being 65 days (electronic supplementary material, table S1), we observed a maximum difference of 13 days or 25%.

Figure 2.

Variation in gestation length between resident and dispersing meerkats. (a) Days between first observed abdominal swelling in the field and parturition. (b) Days between implantation date estimate and parturition based on ultrasound. Box plots show median and interquartile range, error lines show values within 1.5 times interquartile range, transparent points show raw data. Gestation length was on average longer when measured with ultrasound because we detected pregnancies earlier.

Although the longest recorded pregnancy (65 days) was identified by palpation, the second-longest pregnancy was nearly as long (62 days) and was identified by ultrasound. When excluding the two pregnancies identified by palpation, the difference in gestation length between dispersers and residents was still marginally significant, i.e. the p-value was below 0.1 (χ = 2.91, p = 0.088, electronic supplementary material, table S3).

There was a positive correlation between gestation length and number of previous pregnancies (est = 0.02, s.e. = 0.01. p = 0.004), and there were no differences in gestation length between dominant, subordinate (est = −0.01, s.e. = 0.04, p = 0.729) and lone females (est = 0.14, s.e. = 0.10, p = 0.150, electronic supplementary material, table S3). The analysis based on ultrasound confirmed that there was no difference between dominants and subordinates (χ = 0.25, p = 0.885, electronic supplementary material, table S3).

4. Discussion

Our ultrasonographic assessment confirms that the observed delay in meerkat pregnancies was due to delayed post-implantation embryonic development. The most likely explanation of the observed flexibility in embryonic development is that it is a mechanism to allow pregnant females under adverse conditions to delay and spread the energetic costs of gestation [4,5], which increase rapidly in the last month of pregnancy [9]. There is evidence that eviction and dispersal have substantial costs in meerkats, including low foraging efficiency, body-condition loss and increased glucocorticoid levels [17,21]. Stress and malnutrition are therefore likely causes to delay the increasing energetic demands of the final stages of gestation in dispersing meerkats, which could improve the chances of giving birth to successful offspring upon settlement. An alternative, non-adaptive explanation is that embryonic development is slowed down as a by-product of reduced food intake and limited investment into embryo growth.

Adaptation of pregnancy length to reproductive conflict may be a common trait in social mongooses, as subordinates of the cooperatively breeding banded mongoose (Mungo mungo) give birth earlier and synchronize parturition with the dominant female to avoid infanticide [27]. Older and more dominant females conceive a few days before their younger group members, but all females usually give birth on the same day. By delaying mating, younger females may gain access to high-quality males after these males have finished mating with the older females in the group. Alternatively, delayed oestrus in younger females may reflect their latency in reacting to the behavioural or pheromonal cues of dominant females that enter oestrus. While it is unclear which mechanism is responsible for reduced gestation length in subordinate females of banded mongooses, the reduction seems to be caused by reproductive conflict within the group. As such, variable gestation length in both meerkats and banded mongooses is an adaptation of subordinate females to reproductive suppression.

Our findings suggest that meerkats can adjust their pregnancy length in response to reproductive suppression (i.e. eviction) and dispersal. Increased stress during eviction [17] and dispersal [21] may have led to a temporary arrest or decreased rate of embryonic development [5]. In other species where gestational delays have been documented, these appear to depend on abiotic stressors such as food availability or temperature [4,29]. In meerkats, however, gestational delays seem to be initiated by stressful evictions and are therefore socially induced. Dispersal-related external factors such as low foraging rate, unfamiliar landscapes and increased predation pressure are likely to add to the costs of social stressors.

In addition, retarded embryonic development could be responsible for stress-induced pregnancy failure in form of embryonic absorption and fetal abortions in evicted meerkats [17,30], as delayed development has been suggested to facilitate reabsorption of the embryo [31]. A stress-related mechanism to abort or prolong pregnancies in evicted meerkats may be adaptive, offering the potential to end or extend a pregnancy depending on whether a female returns to the natal group or disperses and starts a new group. If a female disperses, deferring the cost of gestation may increase the chance of giving birth successfully, and consequently of new group establishment as the success of new groups depends on their rapid augmentation.

Our study is the first confirmed observation of delayed post-implantation embryonic development in mammals outside of the Chiroptera. There is, however, some indication of delayed embryonic development in armadillos (Dasypus novemcinctus) [32] and Antarctic fur seals (Arctocephalus gazella) [29]. Considering its high adaptive value, delayed embryonic development may also occur in other mammalian species, yet it is hard to prove due to the inherent difficulty of determining implantation date in wild animals [33].

Ethics

All necessary permits to handle meerkats were granted to the KMP by the Department of Environment and Nature Conservation of South Africa and the Animal Ethics Committee of the University of Pretoria (permit ‘FAUNA 192/2014').

Data accessibility

Data and code supporting this manuscript are available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.stqjq2c6r [34]. Data include ultrasonographic images of 13 pregnancies.

Additional figures and methodological details are provided in the electronic supplementary material [35].

Authors' contributions

N.M.: conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing—original draft; G.C.: conceptualization, investigation, methodology, supervision, writing—review and editing; D.S.: conceptualization, investigation, methodology, writing—review and editing; M.M.: funding acquisition, project administration, writing—review and editing; A.S.: investigation, methodology, writing—review and editing; T.B.H.: methodology, writing—review and editing; T.C.-B.: conceptualization, funding acquisition, project administration, supervision, writing—review and editing; A.O.: conceptualization, funding acquisition, project administration, supervision, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

Data collection on resident individuals and maintenance of facilities at the KRC were funded by the ERC (grant nos 294494 and 742808) to T.C.-B., the University of Zurich, the MAVA Foundation and the Mammal Research Institute at the University of Pretoria. Analyses in this study and data collection on dispersers were funded by the Swiss National Science Foundation (grant nos CR32I3_159743 and 31003A_182286) to A.O.