Androgen-mediated maternal effects and trade-offs : postnatal hormone development, growth, and survivorship in wild meerkats

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Androgen-mediated maternal effects and trade-offs : postnatal hormone development, growth, and survivorship in wild meerkats © 2024 Davies, Shearer, Greene, Mitchell, Walsh, Goerlich, Clutton-Brock and Drea. Published version Davies, Charli S.; Shearer, Caroline L.; Greene, Lydia K.; Mitchell, Jessica; Walsh, Debbie; Goerlich, Vivian C.; Clutton-Brock Tim, H.; Drea, Christine M. Davies, C. S., Shearer, C. L., Greene, L. K., Mitchell, J., Walsh, D., Goerlich, V. C., Clutton-Brock Tim, H., & Drea, C. M. (2024). Androgen-mediated maternal effects and trade-offs : postnatal hormone development, growth, and survivorship in wild meerkats. Frontiers in Endocrinology, 15, Article 1418056. https://doi.org/10.3389/fendo.2024.1418056 2024 Androgen-mediated maternal effects and trade-offs: postnatal hormone development, growth, and survivorship in wild meerkats Charli S. Davies 1,2 †‡ , Caroline L. Shearer 1,3 ‡ , Lydia K. Greene 1,2,3 † , Jessica Mitchell 1,2 † , Debbie Walsh 1,2 , Vivian C. Goerlich 1,2 † , Tim H. Clutton-Brock 2,4,5 and Christine M. Drea 1,2,3,6 * 1 Department of Evolutionary Anthropology, Duke University, Durham, NC, United States, 2 Kalahari Research Trust, Kuruman River Reserve, Northern Cape, South Africa, 3 University Program in Ecology, Duke University, Durham, NC, United States, 4 Department of Zoology, University of Cambridge, Cambridge, United Kingdom, 5 Mammal Research Institute, University of Pretoria, Pretoria, South Africa, 6 Department of Biology, Duke University, Durham, NC, United States Introduction: Mammalian reproductive and somatic development is regulated by steroid hormones, growth hormone (GH), and insulin-like growth factor-1 (IGF-1). Based largely on information from humans, model organisms, and domesticated animals, testosterone (T) and the GH/IGF-1 system activate sexually differentiated development, promoting male-biased growth, often at a cost to health and survivorship. To test if augmented prenatal androgen exposure in females produces similar developmental patterns and trade-offs, we examine maternal effects in wild meerkats (Suricata suricatta), a non-model species in which adult females naturally, albeit differentially by status, express exceptionally high androgen concentrations, particularly during pregnancy. In this cooperative breeder, the early growth of daughters predicts future breeding status and reproductive success. Methods: We examine effects of normative and experimentally induced variation in maternal androgens on the ontogenetic patterns in offspring reproductive hormones (androstenedione, A 4 ;T;estradiol,E 2 ), IGF-1, growth from pup emergence at 1 month to puberty at 1 year, and survivorship. Specifically, we compare the male and female offspring of dominant control (DC or high-T), subordinate control (SC or lower-T), and dominant treated (DT or blocked-T) dams, the latter having experienced antiandrogen treatment in late gestation. Results: Meerkat offspring showed sex differences in absolute T and IGF-1 concentrations, developmental rates of A 4 and E 2 expression, and survivorship —effects that were sometimes socially or environmentally modulated. Atypical for mammals were the early male bias in T that disappeared by puberty, the absence of sex differences in A 4 and E 2 , and the female bias in IGF-1. Food availability was linked to steroid concentrations in females and to IGF-1, potentially growth, and survival in both sexes. Maternal treatment significantly affected rates of T, E 2 , and IGF-1 expression, and weight, with marginal effects on Frontiers in Endocrinology frontiersin.org01 OPEN ACCESS EDITED BY Daniel William Hart, University of Pretoria, South Africa REVIEWED BY Bo Pan, National Institutes of Health (NIH), United States Melissa Holmes, University of Toronto Mississauga, Canada Aaryn Mustoe, Texas Biomedical Research Institute, United States Michael Scantlebury, Queen’s University Belfast, United Kingdom *CORRESPONDENCE Christine M. Drea [email protected] † PRESENT ADDRESSES Charli S. Davies, University of Jyväskylä, Jyväskylä, Finland Lydia K. Greene, Duke University, Durham, NC, United States Jessica Mitchell, University of Edinburgh, Edinburgh, United Kingdom Vivian C. Goerlich, Utrecht University, Utrecht, Netherlands ‡ These authors share first authorship RECEIVED 16 April 2024 ACCEPTED 28 August 2024 PUBLISHED 30 September 2024 CITATION Davies CS, Shearer CL, Greene LK, Mitchell J, Walsh D, Goerlich VC, Clutton-Brock TH and Drea CM (2024) Androgen-mediated maternal effects and trade-offs: postnatal hormone development, growth, and survivorship in wild meerkats. Front. Endocrinol. 15:1418056. doi: 10.3389/fendo.2024.1418056 COPYRIGHT © 2024 Davies, Shearer, Greene, Mitchell, Walsh, Goerlich, Clutton-Brock and Drea. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. TYPE Original Research PUBLISHED 30 September 2024 DOI 10.3389/fendo.2024.1418056 survivorship; offspring of DT dams showed peak IGF-1 concentrations and the best survivorship. Discussion: Maternal effects thus impact offspring development in meerkats, with associated trade-offs: Whereas prenatal androgens modify postnatal reproductive and somatic physiology, benefits associated with enhanced competitiveness in DC lineages may have initial costs of reduced IGF-1, delay in weight gain, and decreased survivorship. These novel data further confirm the different evolutionary and mechanistic pathways to cooperative breeding and call for greater consideration of natural endocrine variation in both sexes. KEYWORDS female masculinization, flutamide, IGF-1, life-history trade-offs, ontogeny, prenatal programming, sex steroids, sexual differentiation Introduction Sexual differentiation is a lifelong developmental process whereby the production of male and female phenotypes arises from a complex interplay between genetic, hormonal, and environmental factors (1,2). The earliest environmental influence on an individual’s underlying biological processes derives from the maternal (or prenatal) milieu. Maternal effects on offspring are widespread and diverse in the animal kingdom, evident in the nongenetically determined, life-history markers of morphological, physiological, and behavioral development, as well as health, survival, and fitness (3). Notably, maternal androgens in egg yolks are well known to impact offspring growth, survival, and immunity in oviparous species (4,5). In mammals, maternal effects are commonly examined in the context of differences in maternal diet and stress (6,7), arising from differential access to resources (8) or social instability and conflict (9). They also owe to impacts on the fetal endocrine milieu, particularly via differential exposure of fetal females to androgens (7), be they of fraternal or maternal origin. Here, using a nonmodel system –the meerkat (Suricata suricatta)–examined during an environmental stressor (i.e., a drought), we test for effects of naturally and experimentally induced variation in maternal androgens on the endocrine development, growth, and survivorship of offspring. Our study contributes to a slow-growing literature on the effects of androgens in females, which is related both to the process of sexual differentiation and to potential epigenetic effects of early hormone exposure. Typically, in mammals, androgens of fetal testicular origin are required to produce a temporal sequence of masculinizing effects in XY individuals; when fetal XX individuals are exposed to these androgens, they incur variable, reproductively costly, masculinizing effects depending on the timing, duration, and quantity of exposure (10). In certain litter-bearing species (particularly rodents, lagomorphs, and swine), fetal females are exposed to varying concentrations of testicular hormones owing to their in utero distance from male littermates; intrauterine position produces a gradient of female morphotypes, ranging from feminized to masculinized (11–13). Beyond the well-known organizational effects of fraternal androgens on female morphology, physiology, and behavior (e.g., aggression), are the effects of androgens on various life-history traits (14). For instance, fraternally masculinized females often experience reduced growth or weight at weaning, with delayed puberty, as well as reduced fecundity (e.g. disrupted ovarian cycles), fertility (e.g. reduced litter size), and survival (15–18). Even in humans, females with male co-twins can be masculinized (19)and experience reduced fertility (20). Nonetheless, effects are inconsistent across species, may differ between captive and wild populations, and have also been shown to produce some benefits to masculinized females (21,22). More exceptionally, masculinization can arise routinely in all female members of certain species, such as the spotted hyena [Crocuta crocuta (23–25)], ring-tailed lemur [Lemur catta (26)], and meerkat (27). In these cases, natural selection operates on prenatal androgens of maternal origin, producing transgenerational effects, ultimately with female social and reproductive benefits that far outweigh any costs. Nonetheless, our understanding of postnatal ontogeny and life-history trade-offs in such species remains limited. It long has been established that mammalian reproductive, somatic, and skeletal development is regulated, in large part, by the actions of steroid hormones, particularly androgens [e.g (28–30)], with increasing appreciation for the role of estrogens [e.g (31–33)] and genetic factors (2,23,34). In early life, activation of the mammalian hypothalamic-pituitary-gonadal (HPG) axis varies depending on the sex and developmental stage of the individual. Notably, the prenatal period is critical for establishing the reproductive phenotype of embryos, as well as their future steroid production and sensitivities. Testosterone (T) is a critical organizing hormone, including in fraternally and maternally masculinized female fetuses; however, one of its androgenic precursors, androstenedione (A 4 ), is also implicated in maternally masculinized females and may be capable of producing or contributing to male-like traits without interfering with reproductive Davies et al. 10.3389/fendo.2024.1418056 Frontiers in Endocrinology frontiersin.org02 function [spotted hyenas: (35–37); ring-tailed lemurs: (26,38); meerkats (27)]. Across young mammals, there is sometimes only a brief period of HPG activity between birth and preweaning, particularly evident in males (39); otherwise, like juvenility, infancy is generally a period of reproductive quiescence, such that sex hormones may be initially at the lower limitsofdetectability,increasingastheHPGaxisis reactivated when adolescent animals approach puberty (i.e., the pubertal surge). Typically, throughout postnatal development, androgen concentrations are greater in males than in females, whereas the reverse is true of estrogens, particularly estradiol [E 2 ; reviewed in (39,40)]. Only more recently has evidence highlighted (a) the key role of insulin-like growth factor-1 (IGF-1) in these anabolic processes, (b) the integration of different neuroendocrine mechanisms, and (c) the significant trade-offs between reproductive development, growth, and longevity [reviewed in (41–43)]. Principally produced in the liver, IGF-1 is a polypeptide hormone that manages the effects of growth hormone (GH), stimulating somatic and skeletal growth; it is also a powerful regulator of reproduction, promoting steroidogenesis and growth of reproductive structures. As regards neuroendocrine integration, sex steroids are known to modulate the effects of IGF-1, and vice versa; however, androgens and estrogens can exert distinct and opposite effects (33). For instance, T typically raises, whereas E 2 typically lowers, total IGF-1 concentrations (44). Androgen and free IGF-1 concentrations thus generally correlate positively with somatic male-biased growth, suggesting potentially age-dependent sex differences in concentration [e.g (45)]. Over the course of development in both sexes, IGF-1 concentrations are initially raised and begin to plateau or decrease at or around puberty, with slower cell growth or metabolism in adulthood potentially serving as a physiological strategy for extending lifespan (46). Thus, as with T (47,48), IGF-1 promotes growth and reproduction at a cost to longevity (43), such that both hormones are key when considering life-history trade-offs. The meerkat –an aggressive (49,50), social mongoose and obligate cooperative breeder –has emerged as an important system for examining the behavioral and reproductive effects of androgens in females (27,51,52). Notably, dominant females, relative to subordinate females (and all males), express greater concentrations of A 4 and E 2 ,andgreaterorequivalentT concentrations (52). Unlike many cooperatively breeding species that experience physiological suppression of reproduction, adult meerkats of both sexes can reproduce throughout life (53). Indeed, within their natal clans, subordinate dams show normative glucocorticoid concentrations (53,54) and produce offspring that survive in relevant numbers, some even ascending to the dominant position (27,49,54,55). Nonetheless, female meerkats that express the highest androgen concentrations are, unusually, the most reproductively successful (51). These females are, however, also the most parasitized (56) and immunocompromised (57), suggesting important, androgen-mediated trade-offs. During pregnancies carried to term within the clan, the exceptionally high, albeit status-related, androgen concentrations of dams (52) provide a means for the epigenetic transmission to offspring of androgen-mediated traits. In a prior study of the reproductive endocrinology and behavior of dominant control (DC), subordinate control (SC), and dominant treated (DT) dams that received the antiandrogen flutamide throughout late gestation, we found that DC dams express their peak hormone concentrations during late gestation (27), when feeding competition, dominance interactions, and evictions are magnified. They also produce offspring that are more aggressive in early postnatal development than those of SC dams, indicating normative maternal effects on infant behavior. Moreover, blocking androgen receptors in DT dams reduces their dominance interactions and is associated with infrequent evictions and decreased social centrality within the clan, as well as increased aggression in cohabiting SC dams. Lastly, flutamide treatment also reduces offspring aggression, confirming that maternal effects on infant behavior owe specifically to late-term androgens (27). Although we lack information on fetal intrauterine position, key to examining fraternal effects (16,17), we can largely dismiss the influence of fraternal androgens on status-related differences in meerkats. Firstly, masculinization of genitalia, as would occur either from mid-gestational maternal or fraternal androgens, is not present in meerkats, diminishing the likelihood of earlier androgen exposure from either source. Secondly, although meerkats produce mixed-sex litters [a proxy of fraternal androgen exposure in field studies (18)], maternal status does not affect the relatively even offspring sex ratios observed at first capture (58). These are also estimates of fetal offspring sex ratios –given that meerkats are born underground and can differentially fall victim to early neonatal mortality, infanticide, and predation, all prior to emergence –but even with greater pup mortality in subordinate matrilines, any discordance between preand post-birth sex ratios should be randomly distributed. Accordingly, differences in pup aggressive behavior by maternal status and treatment owe to maternal, rather than fraternal, androgens. Here, therefore, we test for further maternal effects, specifically, and for potential trade-offs in the offspring of the three ‘treatment’groups of females (i.e., DC or high-T, SC or lower-T, and DT or blocked-T dams). Specifically, we examine developmental patterns in the same reproductive steroids (A 4 , T, and E 2 ), as well as IGF-1, growth, and survivorship in the first year of life. In species characterized by matrilineal dominance hierarchies, socialization effects can confound the relations between prenatal endocrine milieu and postnatal behavior or somatic development. For instance, maternally acquired dominance status across species can impact offspring aggression (59), mediate differences in food access (8), and affect IGF-1 concentrations and its consequences (41,43,45). Nevertheless, patterns observed in ‘normative’species may not hold in masculinized females that are similarly characterized by maternal rank acquisition. In spotted hyenas, for instance, maternal rank (which predicts androgen-mediated offspring aggression: 24) positively influences female juvenile size independently of IGF-1, and high IGF-1 concentrations negatively impact survival in females after reproductive maturity (60). Teasing apart the social and physiological determinants of maternal effects can be challenging. Studies in the meerkat allow us to address this challenge because intense allocare minimizes maternally determined socialization effects (61)and,relativetononcooperatively breeding species, allows better isolation of prenatal neuroendocrine effects on development (27). Davies et al. 10.3389/fendo.2024.1418056 Frontiers in Endocrinology frontiersin.org03 Wild, yet habituated, meerkats represent an excellent study system that serves as both a ‘natural experiment’of normative variation and allows ‘phenotypic engineering via hormonal manipulation’[see (47)] in the field. This combined approach uniquely allows us to address the following main predictions. Based on reproductive endocrine patterns in adult meerkats (52) and developing spotted hyenas (36), we predict (i) raised A 4 and minimal sex differences in the reproductive hormones of normative, developing offspring, including at puberty. Regarding treatment effects on reproductive hormones and other metrics, we expect (ii) the strength or rate of effects to vary by treatment group, particularly in an androgen-dependent fashion (DC > SC > DT). Regarding IGF-1 and growth, based on studies of various species experiencing raised androgen concentrations prenatally (5,60,62), we expect (iii) either a lack of sex difference in IGF-1 and growth or a reversed sex difference, with increased IGF-1 and growth in normative female meerkats. We also anticipate the typical agerelated patterns, including a decline in IGF-1, but steady gain in growth. Lastly, based on the health costs of androgens (specifically increased parasitism and reduced immunocompetence) in female meerkats (56,57) and their offspring (63), we predict (iv) an early survival cost in normative offspring –one that would be alleviated in flutamide-exposed animals. Because our study occurred during a prolonged drought, we also explore the socioecological conditions, including clan size (indicative of the number of helpers) and rainfall (indicative of the abundance of insects) [insects are the primary prey of meerkats (64)], that could have contributory effects on these various metrics. Our overall aim is thus to assess the degree to which androgen exposure is evolutionarily selected, in a manner that should contribute to our understanding of life-history theory and the trade-offs associated with routine female masculinization. Materials and methods Subjects and study site Our focal subjects were 109 wild meerkats (n=57male,n=52 female; Table 1) that were young members of 17 clans (ranging in size from 3-36 individuals) inhabiting the Kuruman River Reserve (26°58′ S, 29°49′E), South Africa, during the study period 2012-2015. These young individuals derived from healthy DC, SC, and DT dams (Table 1), whose flutamide treatment regimens had been previously validated (65) and reported (27). Briefly, while under anesthesia (see sampling procedures outlined below), dominant females in lategestation received flutamide (~15 mg/kg/day) via two, 21-day release pellets (300 mg, Innovative Research of America, Sarasota, FL), implanted subcutaneously by a licensed veterinarian. The treated animals also received a subcutaneous injection of a non-steroidal, anti-inflammatory painkiller (either 0.2–0.3 mg/kg meloxicam, Metacam, Boehringer or 0.1 ml of metacam, plus 0.1 ml of lentrax or other long-lasting penicillin, depending on availability). Dam identity was known and confirmed through genetics (66). All individuals in this study were microchipped, habituated to human observers and to voluntary weighing, and were individually recognizable via dye-marks, as previously described (67). We monitored these offspring roughly three times per week, from pup emergence (at approximately 21-30 days of age) to adulthood (at 1 year) [see (68)]. We recorded the weights and life-history metrics of the subjects up to three times per observation day. We also captured, measured, and obtained a blood sample from each focal subject at 1, 3, 6, 9, and 12 months of age, as described below. Because subadult meerkats do not disperse (69,70), individuals < 1 year of age that disappear from the clan are presumed dead; in such cases, we assigned the date of the last sighting as the date of death. Beyond the offspring’s metrics, we calculated its clan size by averaging daily values from the week preceding its capture. We obtained daily rainfall data from an onsite weather station (CR200 datalogger; Campbell Scientific) and used, for our analyses, cumulative rainfall from the month preceding capture. Sampling procedures At ~ 1 month of age (range 18-36 days), we captured and sexed the focal subjects shortly after emergence and implanted permanent Passive Integrated Transponders or PIT tags. Because the fur of pups grows quickly, we individually marked 1-month-old pups by trimming small sections of their fur rather than by using hair dye. Next, we measured right hindfoot length (mm) using digital calipers. We then obtained tissue samples for genetic analysis by cutting a small piece (2 mm) from the tail tip with sterilized scissors. Following this procedure, we collected blood samples from the tail tip using heparinized capillary tubes. At the older ages sampled, we individually captured and processed the focal subjects in the morning and at the den to minimize the time delay (mean ± S.E.M. = 8.30 ± 0.20 min) until blood draw (52). We gently picked up the subjects, by the tail base, carefully placed them into a cotton sack and anesthetized them with isoflurane (Isofor; Safe Line Pharmaceuticals, Johannesburg, South Africa) in oxygen, using a vehicle-mounted vaporizer (71). First, we drew an age-appropriate volume of 0.2–2 ml of blood from the jugular vein, using a 25-G needle and syringe. We allowed blood samplestoclotatambienttemperature in serum separator tubes (Vacutainer®,BectonDickinson,Franklin Lakes, NJ, USA) and later centrifuged them at 3700 rpm, at 24 °C, for 10 min. We stored decanted serum samples on site at −20 °C until TABLE 1 Sample sizes by maternal treatment and offspring sex. Maternal treatment # Dams # Litters #Offspring (male; female)* # Serum samples (male; female)** Dominant control (DC) 13 15 50 (27; 23) 134 (66; 68) Subordinate control (SC) 14 15 42 (23; 21) 109 (64; 45) Dominant treated (DT) 5 5 15 (7; 8) 65 (34; 31) Total 31 35 109 (57; 52) 308 (164; 144) *Numbers reported are for the start of the study. **Numbers reported are for the total number of samples available for endocrine analyses. Davies et al. 10.3389/fendo.2024.1418056 Frontiers in Endocrinology frontiersin.org04 transport, on ice, to Duke University in North Carolina, where they were stored at −80 °C until assay. Enzyme immunoassays We assayed serum samples for A 4 ,T,E 2 , and IGF-1 using commercial, competitive enzyme immunoassay (EIA) kits (ALPCO diagnostics, Salem, NH, USA). For all assays, we diluted with assay buffer, to a maximum of 1:8, any samples that had concentrations greater than the upper detection limit and multiplied the results obtained by the dilution factor. For samples with concentrations below the minimum detectable limit of the assay, we allocated the minimum assay value. We ran all samples in duplicate and, if the coefficient of variation (CV) exceeded 10%, we ran a subsequent assay. Capture-to-bleed time was recorded for nearly all (save n=6)samples and found to be non-significant for A 4 ,T,andE 2 (ANOVA: p=0.11, 0.56, and 0.31, respectively) but significant for IGF-1 (ANOVA: p= 0.01), with concentrations being weakly, but positively, related to the time delay to sample collection. Nevertheless, adding capture-to-bleed time neither improved the model fit nor altered the results, therefore we removed this variable from subsequent analyses (see below). EIA serum assays of A 4 ,T,andE 2 have been previously validated in meerkats by standard parallelism, linearity, and recovery tests (52). The serum A 4 assay has a sensitivity of 0.04 ng/ml using a 25-µl dose, with intraand inter-assay CVs of 5.23% and 8.7%, respectively. The serum T assay has a sensitivity of 0.02 ng/ml using a 50-µl dose, with intraand inter-assay CVs of 7.9% and 7.3%, respectively. The E 2 assay has a sensitivity of 10 pg/ ml using a 50-ml dose, with intraand inter-assay CVs of 7.7% and 8.7%, respectively. Owing to smaller serum volumes obtained at younger ages, we lacked sufficient representation for meerkats at 1 month and therefore excluded this age from the analyses of E 2 . The IGF-1 assay has a sensitivity of 0.09 ng/mL using a 10-ml dose, with intraand inter-assay CVs of 7.7% and 8.7%, respectively. Serial dilutions of pooled meerkat serum yielded a displacement curve parallel to the IGF-1 standard curve. Assay accuracy, measured as percent recovery of known amounts of analyte from a pooled serum sample was 102.7% (n= 6). This assay has an initial step to dissociate IGF-1 from its binding proteins by dilution in an acidic buffer; following this step, there was no cross reactivity with IGF-II, insulin, or C-peptide. Statistical analysis Subject considerations The second cohort of offspring (in 2013) experienced severe drought. There were no litters from treated dominant dams in this cohort and drought conditions resulted in extremely low survivorship (87% mortality compared to an average of 40% for cohorts one and three). Nevertheless, including or excluding this year had no impact on endocrine results, so we included individuals from this year in all hormone analyses (see below). Owing to the death of some dominant animals during the span of the study, three of our subjects acquired dominance at an unusually young age. Because of the physiological and morphological changes associated with dominance acquisition (74), we excluded these subjects’postascension blood samples (n= 5) and weight measures from our analyses. Generalized Additive Model For each hormone, we analyzed developmental trajectories of serum concentrations using a Generalized Additive Model (GAM) with gamma error distributions, a log link function, and REML penalty, using the package mgcv 1.9-0 (72) in R version 4.3.2 (73). We used additive models to account for possible nonlinear relationships across development between hormone concentrations and continuous predictors, such as age. Maternal treatment (DC/SC/ DT) and offspring sex (male/female) were included as fixed factors in the model. Because offspring weight was highly correlated with offspring age, we calculated size-corrected mass as a proxy of offspring body condition (i.e., separate from age-related size differences), using the residuals from two linear models (one per sex) wherein we regressed mass (g) with hindfoot length (mm) to correct for age-related structural size differences between individuals. We then included these residuals, as well as age in months, average clan size, and monthly rainfall as smooth terms in the model. We removed terms that were not significant and/or explained < 1% of the variation. If the empirical distribution function (EDF) was 1.000 (across all levels of factor), we removed smooths and instead included these terms as linear predictors. We included individual identity and litter as random effects to account for repeat sampling. We also initially included the interactions between sex and maternal treatment, as well as between sex and rainfall, in case the sexes responded differently to maternal treatment or environmental conditions. To enableinterpretationoffirst-order effects, we dropped interactions if they were not significant and/or explained < 1% of the deviation. Model selection was performed for the age smooth term using the lowest Akaike’s Information Criterion (AIC) score [at least 2 from the next best model; (75)] to determine which factors to include as “by”levels for this term: sex, maternal treatment, neither or both. All best models were at least 2 from the next best model except A 4 , for which another best fit model contained both “by” levels but the age by treatment (non-significant) term had an EDF approaching 0, indicating that this term contributed < 1 degree of freedom to the model. Due to REML penalization, which penalizes overfitting (76), this model was essentially equivalent to the other best model of age by sex alone. We thus report the simplified ageby-sex model. We confirmed that there were no high correlations (< 0.5) between independent terms and examined extreme residual values in quantile-quantile (QQ) plots of random effects to ensure that these values were randomly grouped and not accounted for by predictors. We used the DHARMa package (77) to confirm that there was no overor under-dispersion, or residual spatial or temporal autocorrelation in the models, and we used the function ‘k.check’in mgcv to optimize the number of knots (k value) in smooth terms. We plotted GAM model predictions on the response scale, using the base function predict and ggplot2 3.4.4 (78). Davies et al. 10.3389/fendo.2024.1418056 Frontiers in Endocrinology frontiersin.org05 We also analyzed offspring weight change using GAMs, but with gaussian error distributions on the response scale. Model selection, diagnostics, plotting, and terms were the same as those used for hormone analyses, except that we included the average hindfoot length of offspring to account for skeletal size differences. Generalized Linear Mixed Model We modeled differences between offspring hormone concentrations and weight at 12 months (maturity) using a Generalized Linear Mixed Model (GLMM) with a gamma error structure (gaussian error for weight) and log link function, using the package glmmTMB 1.1.7 (79). We used the DHARMa package for GLMM model diagnostics and used the same terms as in the GAM (excluding age and individual). We analyzed offspring survival using a GLMM with a binomial error structure and logit link function, using the package glmmTMB. We tested for collinearity between independent variables using variance inflation factors, ensuring an upper limit of three, and used the DHARMa package for GLMM model diagnostics. We included survival to the next breeding season as a binary response variable (0 for death, 1 for survival). We excluded individuals born in the second cohort (during 2013) to prevent any treatment-related survivorship bias. Additionally, we excluded individuals that had not been sexed prior to their last sighting (n = 16). We included dam treatment (DC/SC/DT), offspring sex (male/female), and both clan size and monthly rainfall at the time of pup emergence as fixed factors in the model; we included litter as a random factor. Results Our full results for the predictors of hormone concentrations (at all ages, Supplementary Table S1; at 12 months, Supplementary Table S2), offspring weight (at all ages, Supplementary Table S3;at 12 months, Supplementary Table S4), and survivorship (to 12 months, Supplementary Table S5) are presented in the Supplementary Material, with significant findings and data visualization highlighted below. For ease of visualization, Figures 1–3emphasize the model lines; respective Supplementary Figures S1-S3 present the range of individual data points and Supplementary Figure S4 presents hormone concentrations by body condition residuals. Androstenedione For offspring of both sexes, A 4 concentrations showed a quadratic age effect (GAM: females, F= 3.36, p= 0.019; males, F= 3.34, p= 0.017; for effect sizes/estimates, see Supplementary Table S1), with an atypical peak at 1 month, followed by a decrease, and then the more usual, gradual increase as juvenile meerkats approached puberty and adulthood (Figure 1A). Also unusual for most mammals, but consistent with prediction (i) based on masculinized species, was the lack of a main effect of sex in A 4 concentrations, with female values being as high as male values, even at sexual maturity (GLMM: 12 months, c 21 = 1.85, p= 0.064; Supplementary Table S2;Figure 1A). Despite a suggestive pattern, FIGURE 1 Top model-predicted serum concentrations of (A, B) androstenedione or A 4 ,(C, D) testosterone or T, (E, F) estradiol or E 2 , and (G, H) insulin-like growth factor 1 or IGF-1 for meerkats by sex (male, M; female, F), age, and/or maternal treatment (dominant control, DC; subordinate control, SC; dominant treated, DT). For significant interactions between age and either sex (A, E) or maternal treatment (D, F, H), the estimates are presented from 1-12 months with 95% CI. Otherwise, model predicted means ± s.e. are presented by sex (C, G) or maternal treatment (B). Significance of comparison between groups is as follows: N.S. p> 0.05; * p< 0.05, *** p< 0.001. Davies et al. 10.3389/fendo.2024.1418056 Frontiers in Endocrinology frontiersin.org06 there was no significant effect of maternal treatment on A 4 concentrations (Figure 1B), contrary to prediction (ii). We observed no notable patterns in A 4 concentrations in relation to either clan size or rainfall. There was, however, a significant effect of body condition (GAM: F= 6.87, p< 0.001; Supplementary Table S1), with offspring at the ‘extremes’of body condition (i.e., those at either end of the distribution) having reduced A 4 concentrations relative to offspring in average body condition (Supplementary Figure S4A). Testosterone Unlike patterns observed for A 4 , an initially typical sex effect emerged for T, with males having greater average concentrations than females throughout early development (GAM: female vs. male, t= -6.78, p< 0.001; Supplementary Table S1;Figure 1C); however, this difference no longer held at maturity, when female values matched those of males (GLMM: 12 months, c 21 = -0.61, p= 0.542; Supplementary Table S2). Prediction (i) thus held for pubertal values. T concentrations in the offspring of DT dams showed a quadratic age effect (GAM: F= 3.50, p= 0.032; Supplementary Table S1), with an atypical peak at 1 month that was absent in the offspring of DC or SC dams (Figure 1D). Instead, the latter two groups of offspring showed a more consistent increase to puberty (Figure 1D). Overall, however, T concentrations did not differ by maternal treatment. Although there were effects of treatment, they did not occur in a linear, androgen-dependent fashion, as outlined in prediction (ii). There was also a significant influence of rainfall on the T concentrations of females (GAM: t= 3.84, p = 0.031; Supplementary Table S1): whereas male values remained consistent across wet and dry conditions, female values decreased with increasing rainfall (Figure 2A). Lastly, T concentrations (as with A 4 concentrations), showed a quadratic effect of body condition (GAM: F= 28.04, p< 0.001; Supplementary Table S1), with reduced T concentrations at the extremes of body condition; Supplementary Figure S4B). Estradiol As with A 4 , overall E 2 concentrations showed no sex effect across the ages sampled, not even at maturity (GLMM: 12 months, c 21 = 1.46, p= 0.145; Supplementary Table S2;Figure 1E), consistent with prediction (i). In females, there was a quadratic age effect (GAM: F= 8.99, p< 0.001; Supplementary Table S1), evidenced by a decrease in E 2 after weaning (at 3 months) followed by a gradual increase in E 2 from 6 months to puberty (Figure 1E). Males had a non-significant linear trend with age (GAM: F= 0.88, p = 0.351; Supplementary Table S1). We also observed a significant positive and linear age effect on E 2 in the offspring of DT dams (GAM: F= 7.10, p= 0.009; Supplementary Table S1), but not in the offspring of DC or SC dams (Figure 1F). As with androgens, we detected no effect of maternal treatment on overall E 2 concentrations. Again, although there were effects of treatment, they did not occur in the order outlined in prediction (ii). There was, however, a significant linear, sex-by-rainfall interaction on E 2 concentrations (GAM: t= -3.03, p= 0.003), with male and female meerkats having opposing responses to this environmental variable: E 2 was not associated with rainfall in males but was negatively associated in females (Supplementary Table S1;Figure 2B). Lastly, whereas clan size was negatively associated with E 2 concentrations (GAM: t= -3.15, p= 0.002; Supplementary Table S1), offspring body condition positively related to E 2 concentrations (GAM: F= 3.97, p= 0.048; Supplementary Figure S4C). Insulin-like growth factor-1 We detected a significant main effect of sex on IGF-1 concentrations (GAM: t= 2.31, p= 0.023; Supplementary Table S1), with female values exceeding male values, even at maturity (GLMM: 12 months, c 21 = 1.58, p= 0.114; Supplementary Table S2; Figure 1G), consistent with prediction (iii). Age differentially predicted IGF-1 concentrations across maternal treatments (GAM: DC x age, F= 3.30, p= 0.040; SC x age, F= 3.04, p= FIGURE 2 Top model-predicted serum concentrations of (A) testosterone or T, (B) estradiol or E 2 , and (C) insulin-like growth factor 1 or IGF-1 for meerkats by sex (male, M; female, F) and total monthly rainfall (mm). The estimates are presented with 95% CI. Davies et al. 10.3389/fendo.2024.1418056 Frontiers in Endocrinology frontiersin.org07 0.016; DT x age, F= 2.47, p= 0.069; Supplementary Table S1), in that the offspring of DC and SC dams showed steadily decreasing IGF-1 concentrations with age, whereas the offspring of DT dams showed initially higher IGF-1 concentrations at emergence, that then plummeted until 6 months and stabilized thereafter (Figure 1H). Nonetheless, average IGF-1 concentrations did not differ by maternal treatment. Again, treatment effects were more variable than those outlined in prediction (ii). Social and environmental factors also impacted IGF-1 (GAM: t = -2.81, p= 0.006; Supplementary Table S1), as offspring raised in larger clans had reduced IGF-1 compared to counterparts raised in smaller clans. Moreover, there was a significant, positive association between IGF-1 and total monthly rainfall (GAM: t= 4.75, p< 0.001; Supplementary Table S1), as well as a significant sex-byrainfall interaction (GAM: t= -2.07, p = 0.041; Supplementary Table S1); in this case, whereas both sexes showed a positive relation between rainfall and IGF-1 concentrations, male meerkats were more sensitive than females to this environmental variable (Figure 2C). Lastly, as observed with E 2 , offspring in better body condition showed increased IGF-1 concentrations (GAM: t = 2.94, p= 0.004; Supplementary Table S1;Figure 3A). Body weight We detected no significant sex difference in body weight across the ages sampled (Figure 3B), consistent with prediction (iii). Also as expected, offspring body weight increased with age; however, FIGURE 4 Likelihood (%) survival of meerkat offpsring to one year of age by (A) sex and (B) maternal treatment (dominant control, DC; subordinate control, SC; dominant treated, DT). Data presented are raw means ± s.e. with sample sizes of individuals surviving to one year (top) and dying before one year (bottom) included in grey. FIGURE 3 Top model-predicted (A) IGF-1 concentrations and (B, C) weights for meerkat offspring. In (A), IGF-1 concentrations are predicted by body condition residuals. IGF-1 estimates are presented with their 95% confidence intervals (CI) and the raw data from 1-12 months are plotted by sex (male, M; female, F). In (B), mean ± s.e. weights are predicted by sex, but are not significant (data thus represent all ages, 1-12 months). In (C), weights are predicted by age and maternal treatment (dominant control, DC; subordinate control, SC; dominant treated, DT). For the significant interaction between age and maternal treatment, estimates are presented from 1-12 months with 95% CI, N.S. p> 0.05. Davies et al. 10.3389/fendo.2024.1418056 Frontiers in Endocrinology frontiersin.org08