Female reproductive competition explains variation in prenatal investment in wild banded mongooses

Female intrasexual competition is intense in cooperatively breeding species where offspring compete locally for resources and helpers. In mammals, females have been proposed to adjust prenatal investment according to the intensity of competition in the postnatal environment (a form of ‘predictive adaptive response’; PAR). We carried out a test of this hypothesis using ultrasound scanning of wild female banded mongooses in Uganda. In this species multiple females give birth together to a communal litter, and all females breed regularly from one year old. Total prenatal investment (size times the number of fetuses) increased with the number of potential female breeders in the group. This relationship was driven by fetus size rather than number. The response to competition was particularly strong in low weight females and when ecological conditions were poor. Increased prenatal investment did not trade off against maternal survival. In fact we found the opposite relationship: females with greater levels of prenatal investment had elevated postnatal maternal survival. Our results support the hypothesis that mammalian prenatal development is responsive to the intensity of postnatal competition. Understanding whether these responses are adaptive requires information on the long-term consequences of prenatal investment for offspring fitness.

Scientific RepoRts | 6:20013 | DOI: 10.1038/srep20013 females and helpers (called "escorts") who provision and protect pups after they emerge from the den. There is also evidence of prenatal maternal impacts on offspring competitiveness: mothers that are heavier at conception produce larger pups which have competitive advantage when competing for alloparental care; increasing pup survival 20 .
We carried out 360 ultrasound scans on 59 breeding females from 8 groups of banded mongooses to test (1) whether mothers adjust prenatal investment in response to reproductive competition, and (2) the consequences of variation in prenatal investment for mothers and offspring.

Study site. We studied a population of banded mongooses living on and around Mweya Peninsula, Queen
Elizabeth National Park (QENP), Uganda (0°12′ S, 27°54′ E) between May 2000 and November 2013. For a detailed description of the climate, habitat and the population see   18 . Rainfall data was provided by Uganda Institute of Ecology Meteorological Station and, later, using a rain gauge.
Study population. All individuals in the population are known and individually marked with either colour-coded collars (7 g) or unique shave patterns (for details of trapping protocol and anesthesia are given elsewhere; Ketamine 21 ; Isoflurane 22 ). The identity of breeding females was determined from changes in body shape, ultrasound scans and palpation 23,24 . Each group was visited daily to determine accurate parturition dates. Since parturition can be determined precisely but conception cannot, we calculated the age of fetuses retrospectively assuming an average 60 day gestation (the mean period between peak mate guarding and birth 23 ). Group size and the number of females were counted as the total number of individuals or females over 1 year old in each group for each communal litter. Individuals are habituated to step onto electronic scales to determine an accurate weight which allows regular weighing events without capture. Female weight at the time of conception was calculated using the closest weighing event prior (± 10 days from conception) to the estimated conception date; if possible weights for all females within the same group came from the same weighing event.
Measuring fetus size and number. Number of fetuses was counted under anesthesia by palpitating the abdomen, and a cross-sectional ultrasound scan of each fetus was obtained using an ultrasound scanner (SIUI CTS-900V, UK) and ultrasound gel (Anagel, UK). Trapping females within the last few weeks of pregnancy was avoided and most trapping was conducted 3-4 weeks after oestrus. Previous study has shown no adverse effects of trapping and palpitating pregnant females 24 . The age of the fetus at the time of the ultrasound scan was calculated retrospectively from the litter birth date and the scan date, assuming a gestation length of 60 days (average female gestation length 23 ).
We used the cross-sectional area (mm 2 ) of each fetus as measured from the ultrasound images as an estimate of fetus size. Fetuses were measured on average at 30 ± 7 (mean ± sd) days post conception when they are still roughly spherical in shape to minimize noise arising from different angles of the scan cross-section. The outline of a fetus was identified by the black pixilation of the fluid-filled amniotic sac and the white pixilation of the womb tissue and the amniotic sac membrane around the fetus. The mean of two perpendicular measurements of the diameter were taken using the computer software Image J (1.47c 25 ) and used to calculate the elliptical area of the fetus (see Fig. 1).

Statistics.
We analyzed fetus sizes and the number of fetuses using general linear mixed models (LMMs) and generalized linear mixed models (GLMMs) in R version 3.1.0 using lme4 package R1.1-6 26,27 . GLMMs had either a poisson error structure with log-link function or binomial error structure with logit link function. Female, litter and group identities were included as random factors in analyses to account for the repeated sampling. Fixed terms included were female weight at conception, female age (months), number of adult females present in the group, group size and the total rainfall during gestation (ml). Because groups were trapped at different stages of pregnancy, fetus age (days) was included as a covariate when analyzing fetus size. Correlations between variables fitted in the same models as fixed effects were lower than the levels indicated by Freckleton 28 to cause model fitting issues such as variance inflation in effect estimates (max r = 0.48). We obtained a minimal model via sequential removal of least significant factors, starting with 2-way interactions. Each factor was then added back into the minimum model in order to confirm removal was not contingent on the order of removal 29 .
To investigate if mothers adjust their prenatal investment in response to reproductive competition we estimated total prenatal investment by multiplying the average fetus size by the number of fetuses carried for each pregnancy. Variation in prenatal investment could be due to individual female adjustment in response to competition (a within-individual effect) or be the result of consistent differences between individuals. We tested the relative importance of within-versus between-individual effects using the method described by van de Pol & Wright 30 , which separates out the effect sizes in the fitted model attributable to variation within versus between individuals. To test the consequences of variation in prenatal investment for mothers and offspring we focused on pup survival to 3 months (y/n) using logistic regression, and pup weight (controlled age at capture < 90 days) as well as female reproductive effort and survival. Maternity assignments for pups were based on 43 microsatellite loci as described in Sanderson et al. 31 . As individual fetus scans cannot be matched to pups an average fetus size was used in these analyses. Relative fetus size was calculated as the average fetus size in each female's litter relative to average fetus size for all females within a breeding attempt. We tested whether prenatal investment predicted female participation in the next group litter (y/n) using a GLMM with binomial errors. We tested whether there was a trade-off between current investment in reproduction and female survival using Cox regression with backward selection of terms (Wald Chi-square). This analysis included total group size, number of females, and the average fetus size and number of fetuses as predictors, and to avoid repeat sampling used only the last reproductive event on record for each female. This analysis was conducted in SPSS 21.  Table  S1). Within-female variation was a better predictor of fetal size in response to reproductive competition than between-female variation (LMM, within-female variation, χ 2 1 = 4.51, N = 360, P = 0.034, between-female variation, χ 2 1 = 3.38, N = 360, P = 0.066; SI Table S2). The number of fetuses was only influenced by female age, peaking at 4 years of age before declining (GLMM poisson, χ 2 1 = 10.36, N = 361, P = 0.001). There was no significant relationship between fetus size and the number of fetuses (LMM, χ 2 1 = 1.03, N = 581, P = 0.31). Thus individual females produced larger fetuses, but no fewer of them, when faced with competition from other female breeders.
Finally, we found no evidence of a cost of prenatal investment to mothers in terms of future survival or reproduction. In fact, higher total prenatal investment was associated with higher post-scan survival of mothers (Cox regression, Wald χ 2 1 = 6.57, N = 360, P = 0.010; Fig. 3). Again this relationship was driven by fetus size rather than number (SI Table S5). Females that invested more prenatally were not less likely to reproduce in the next breeding attempt (GLMM binomial, χ 2 1 = 0.35, N = 164, P = 0.061; SI Table S6). Thus we found no evidence of a survival cost to mothers of elevated prenatal investment, nor did mothers compensate for high prenatal investment by reducing reproductive effort in the next breeding attempt.

Discussion
Female banded mongooses produced larger, but no fewer, offspring when there were more adult females in the group. Since all adult females breed in most breeding attempts, this is consistent with the hypothesis that females strategically up-regulate prenatal investment in the face of elevated postnatal reproductive competition. Such  ) show the steepest increase in fetus size with female number compared to heavier females (dark blue line). Female weight (mean ± sd = 1447 ± 201 g) and rainfall (mean ± sd = 128.3 ± 40.9 ml) are continuous variables that have been categorized for illustrative purposes using the 25% and 75% quartiles. responses may be particularly likely to evolve in breeding systems where females co-breed regularly. Females showed steeper increases in prenatal investment when ecological conditions were harsh, and when they were in relatively poor body condition, two factors which are expected to exacerbate the intensity of postnatal competition among offspring 34 . We found no evidence that increased prenatal investment incurred future costs to females in terms of reproduction or survival. On the contrary, females that invested more prenatally showed improved future survival (Fig. 3). A positive relationship between current reproductive investment and future survival is expected where females vary considerably in quality or access to resources, since high quality females may be able to divert more resources to offspring production without compromising their somatic function (the 'big house big car' effect 35,36 ). Increasing fetus size in response to increased social competition is a subtle way in which females could compete over reproduction within social groups without risking the costs of fighting or killing offspring 3,4 . However, we found no detectable benefit (in terms of short-term reproductive success) associated with increased investment in fetus size. Neither absolute fetus size nor fetus size relative to other co-breeders predicted the number of offspring that survived to emerge from the den. The lack of any detectable advantage to elevated prenatal investment is surprising, and may reflect a high level of noise associated with high pup mortality due to intra-or intergroup infanticide and predation 18,19 . It may also be that the benefits of increased prenatal investment are realised later in the life of the offspring. Studies of human famine and laboratory rodents, for example, suggest that early life environments can influence health and fitness across the lifespan, not just in the short term 13 .
Our findings offer an interesting contrast to studies of social birds and fish, in which dominant females produce smaller eggs or a larger number of eggs when there are many helpers in the group [37][38][39][40] . In banded mongooses, all group members contribute to rearing young, but prenatal investment did not vary with the potential number of helpers (measured by total group size). Our findings suggest that the intensity of reproductive competition, rather than the availability of helpers, is the main determinant of variation in prenatal investment in this species. Larger pups have better access to adult group members who provide parental care and, upon emergence, aggressively defend access to the best helpers or 'escorts' 41 . Where postnatal competition among offspring has characteristics of contest competition, the best response to competition will be to invest more resources per offspring prenatally, rather than to produce more of them 42,43 . Producing a larger number of offspring could also bring benefits, but at the unavoidable cost of intensified competition among littermates.
Our study complements previous studies which suggest that mothers use hormones to influence the development of their offspring in utero to improve their success in the postnatal environment, a form of PAR 13,44 . The PAR hypothesis has been criticized because long term forecasts of environmental conditions are inherently unreliable 14,15 . In cooperative breeders, however, the quality of the postnatal environment is largely determined by the number of breeders competing for reproduction and the number of helpers available to offspring. These features of social groups remain stable over the course of offspring development, from gestation to nutritional independence, so are highly predictable. Cooperative birds and mammals, including humans, are thus likely candidates to evolve PARs. We found evidence that female banded mongooses respond to reproductive competition by adjusting prenatal investment, consistent with the PAR hypothesis, but we did not find evidence that this response is adaptive. To test the PAR hypothesis fully will require study of the consequences of variation in prenatal investment across the lifetime of offspring in animals exposed to natural predators and pathogens.