Children's fingernail cortisol among BaYaka foragers of the Congo Basin: associations with fathers' roles

Abstract

Children and mothers’ cortisol production in response to family psychosocial conditions, including parenting demands, family resource availability and parental conflict, has been extensively studied in the United States and Europe. Less is known about how such family dynamics relate to family members' cortisol in societies with a strong cultural emphasis on cooperative caregiving. We studied a cumulative indicator of cortisol production, measured from fingernails, among BaYaka forager children (77 samples, n = 48 individuals) and their parents (78 samples, n = 49) in the Congo Basin. Men ranked one another according to locally valued roles for fathers, including providing resources for the family, sharing resources in the community and engaging in less marital conflict. Children had higher cortisol if their parents were ranked as having greater parental conflict, and their fathers were seen as less effective providers and less generous sharers of resources in the community. Children with lower triceps skinfold thickness (an indicator of energetic condition) also had higher cortisol. Parental cortisol was not significantly correlated to men's fathering rankings, including parental conflict. Our results indicate that even in a society in which caregiving is highly cooperative, children's cortisol production was nonetheless correlated to parental conflict as well as variation in locally defined fathering quality.

This article is part of the theme issue ‘Multidisciplinary perspectives on social support and maternal–child health’.

1. Introduction

Compared to young in other mammalian and primate species, human children are dependent on caregivers during a relatively prolonged developmental period that stretches into the second decade of life [1]. Humans have evolved to engage in cooperative caregiving to successfully raise our costly offspring across this period, meaning that evolutionarily both children and mothers probably drew on support from other group members to help ensure their survival and health [1,2]. Across their long development period, especially early in life, children are highly sensitive to their relationships with their array of caregivers, which helps them acquire understanding about their broader social worlds [3]. Through this attunement, early life caregiving experiences can help shape the long-term function of children's physiological axes that respond to psychosocial stress, with implications for later life behaviour, cognitive function and health [4]. In particular, children's production of cortisol from the hypothalamic–pituitary–adrenal (HPA) axis in response to variation in family function, including parenting styles and conflict within their families, has been extensively researched in the United States (US) and Europe in which the nuclear family is often the culturally valued and institutionally emphasized primary caregiving unit [4–7]. Results from such studies are generally consistent with psychobiological frameworks regarding how the HPA axis increases cortisol output in response to social experiences that involve perceptions of social and physical threat, unpredictability, and lack of control [4,8]. In US families, women have also been found to have higher cortisol (measured through various biological media) or less healthy daily salivary cortisol profiles (i.e. flatter patterns of diurnal decline across the day) when their childcare demands are higher, they receive less partner assistance in domestic tasks, and they are in poorer-functioning marital relationships (including extreme forms, such as involving partner violence) [9–11]. By contrast, much less is known about the psychobiology of family systems, including child and maternal HPA axis function, in societies in which cooperative caregiving and other forms of extensive sharing of social and material support within the community are more widely valued and practised [2,6].

In many societies, there is a strong cultural emphasis on cooperative caregiving by kin and the broader community. Such cooperative caregiving often takes multiple forms, including instrumental or practical support (e.g. physical resources; direct child-rearing assistance), especially within egalitarian smaller-scale societies [1,12,13]. Both children and mothers probably benefit from the extensive cooperative caregiving support in these societies, in terms of their physical health [13–16] and potentially also their mental well-being, consistent with the focus of the current theme issue in which this article appears. In what follows, we outline some of the potential pathways through which cooperative caregiving and correlated forms of social support could relate to HPA axis function and cortisol production in children and parents.

Broad systems of community cooperative caregiving stand in contrast to institutionalized forms of paid non-parental care (e.g. daycare, preschool) that parents in large, market-based economies frequently rely on. This reliance on paid care is often necessary for parents to participate in the labour force of industrial and post-industrial economies [17]. In cultures where children are raised with more substantial care from invested, routinely available, non-parental cooperative caregivers, they may be more buffered from psychosocial stressors that activate the HPA axis through the support and availability of those individuals [2]. For example, in an extensive long-term study of child health and physiology in Dominica, children had elevated salivary cortisol during challenging family events, including the absence of a parent, change in residence or caregiving instability [2,18]. However, children with grandmothers available to them during these periods had attenuated salivary cortisol responses and also fared relatively better for other health outcomes [2]. In the US, Europe and similar contexts, children's experiences in their immediate nuclear family settings, such as exposure to conflict between parents and chaotic or unstable home environments, have been linked to greater salivary cortisol production across the day, including through flatter diurnal curves, especially for younger children [19–22]. As noted, in societies practising cooperative caregiving, those HPA axis-activating nuclear family effects may be more mitigated, although to our knowledge this is relatively under-explored [2].

Mothers may also experience less activation of stress-related physiology through direct and indirect pathways related to cooperative caregiving. As briefly mentioned above, there is some evidence from US-based research that mothers showed healthier salivary cortisol recovery at the end of the day (i.e. cortisol declining more, rather than remaining relatively elevated) if their partners were more involved with domestic tasks and if their professional and domestic demands were relatively lower [9,10]. These patterns indicate that paternal cooperative care in US family settings can aid in maternal psychosocial and physiological recovery [9,10] and are generally consistent with a larger body of the literature exploring the importance of social support and child-rearing assistance to mothers' mental well-being in the US and Europe [23]. Along those lines, there is evidence that greater social support helps couples avoid some forms of family related conflict [24], and, in many societies, concerns about the sufficiency of resources to meet family needs are also a major source of psychological strain and conflict for parents, which can trickle down to children's psychosocial development and well-being [25]. Thus, cultural norms around cooperative caregiving and variation in levels of received cooperative caregiving and resources could also shape family conflict, with implications for HPA axis activation in parents and children.

In addition to those potential buffering effects of cooperative caregiving, adults' generosity and cooperative behaviour may have strong ties to their social capital within the community, with implications for family well-being in some societies [7]. For example, in cultures that still rely on foraging (i.e. procuring non-domesticated foods via gathering, fishing and hunting) as a core component of their subsistence strategies there is often an emphasis on egalitarianism and cooperation in daily life, including sharing of physical resources and direct rearing of young [1,12–14,26]. Past research in two Amerindian forager-horticulturalist societies found that men's community generosity was linked to benefits to their families. Specifically, Aché men who were considered generous sharers of resources received more aid for their families during difficult periods [27]. Meanwhile, Tsimane’ men with higher prestige, which is based partially on generosity, had children who survived at higher rates [16]. Somewhat similarly, Agta (Philippines) and BaYaka (Republic of Congo) forager women who were more centrally located within social networks had more surviving offspring [15]; it is plausible that women accrue such network positions in part through generosity and cooperation [15]. Collectively, these findings indicate there are child health and survival implications of parents' cooperation and generosity in egalitarian communities and hint at pathways that could link those behaviours to psychosocial stress and HPA axis function in children and adults.

In the present study, we drew on data from BaYaka forager children (n = 48) and their parents (n = 49) to examine the links between a cumulative measure of individual family members’ cortisol (via fingernails) and peer rankings of men according to indicators of locally defined fathering quality, which relate to domains regarding both nuclear family function and cooperative caring. BaYaka foragers are highly egalitarian and engage in cooperative caregiving, which we define here as resource sharing within the community as well as communal child rearing [13]. According to local cultural definitions, core roles of a ‘good’ father include being a generous sharer within the community, providing resources for one's family and avoiding marital conflict [13]. Despite cultural values around the avoidance of marital distress, BaYaka spouses do have periodic conflicts and their frequency and severity vary by family, as in other cultures. Moreover, despite community resource sharing, fathers' provisioning of resources can be a source of strain between spouses. Here, we tested whether greater exposure to psychosocial stressor domains, such as more family conflict and resource concerns, that are germane to family members’ mental well-being and cortisol production in contexts like the US and Europe also correlated with higher cortisol in parents and children in an egalitarian society with cooperative caregiving. We specifically tested the prediction that BaYaka children and parents, respectively, would have higher cortisol in families in which fathers were ranked as being poorer providers or (separately) engaging in more conflict with their wives. Moreover, building from prior relevant work [15,16,28], we tested the prediction that parents and children, respectively, would have higher cortisol in families where fathers were ranked as less generous resource sharers within the community.

2. Material and methods

(a). Study population and site

Our research team collected data from BaYaka families in 2017 in Likouala Province in northern Republic of the Congo. These communities are 3–6 days (via truck and motorboat) from Brazzaville, the capital of the Republic of the Congo. A main focus of our overall project was on fathering, and consequently men and their families were eligible to participate in the study if they had at least one dependent child who was less than 18 years old; all participating men had at least one biological child that met this age criterion. Data collections included demographic, anthropometric (i.e. triceps skinfold thickness; height), fingernail clippings and family role ranking measures for fathers. Here, we focused primarily on cortisol data analysed from fingernails as our core outcome variable.

We collected data from 31 BaYaka men in this community, representing all eligible fathers present at the time of data collection, and 27 of those men had fingernail cortisol data, along with 22 of their spouses. We do not have a full 2017 census of all children in this BaYaka community; however, based on 2018 demographic data, we estimate there were approximately 100–125 children (in total) at the site in 2017. There were 81 children whose fathers were enrolled in the study, which was a required criterion for inclusion in the present analyses, along with anthropometric data. Of the eligible children who did not have full anthropometric data (n = 22), the majority (16 of 22) were 2 years of age or younger. Among the 59 BaYaka children who had the requisite data (e.g. fathers' rankings; anthropometrics) for inclusion here, approximately 81% (n = 49) of them had fingernail cortisol data for these analyses.

Given this focus on cortisol, boys were over-represented in the present analyses for two main reasons (table 1). First, BaYaka girls generally trim their own fingernails, and their household roles wear their nails down. Thus, a smaller proportion of girls had sufficient nail growth to provide samples. Second, BaYaka girls are generally more introverted than boys, who were more willing to provide samples for the study. Ethically, we could not strongly compel participation, especially in light of BaYaka cultural values around respect for individual autonomy, including for children.

Table 1.

Linear mixed models predicting BaYaka family members' fingernail cortisol from fathers' rankings.

	children's cortisola,b			parents' cortisolb,c
	coef	95% CI	p	coef	95% CI	p
Share
Share	−0.39	(−0.63, −0.14)	0.002	0.17	(−0.05, 0.38)	0.128
Age	−0.18	(−0.39, 0.02)	0.082
triceps skinfolds	−0.28	(−0.48, −0.08)	0.006	0.06	(−0.35, 0.46)	0.777
step-childd	0.87	(−0.14, 1.87)	0.091
sex				0.51	(−0.27, 1.30)	0.202
model AIC		201.1			220.2
Dispute
Dispute	0.31	(0.05, 0.56)	0.018	−0.10	(−0.32, 0.12)	0.362
age	−0.16	(−0.38, 0.05)	0.142
triceps skinfolds	−0.19	(−0.40, 0.01)	0.068	0.01	(−0.40, 0.42)	0.954
step-childd	1.29	(0.20, 2.39)	0.021
sex				0.60	(−0.19, 1.39)	0.136
model AIC		204.0			221.7
Provider
Provider	−0.33	(−0.61, −0.06)	0.016	0.14	(−0.07, 0.36)	0.192
age	−0.19	(−0.40, 0.02)	0.083
triceps skinfolds	−0.29	(−0.50, -0.08)	0.006	0.01	(−0.39, 0.41)	0.956
step-childd	0.90	(−0.15, 1.94)	0.092
sex				0.58	(−0.20, 1.35)	0.146
model AIC		203.9			220.8

^aResults reflect analyses of n = 77 cortisol data points from 48 children.

^bAll independent and dependent variables have been converted to standard deviation units (z-scores). Statistically significant results are listed in italics.

^cResults reflect analyses of n = 78 cortisol data points from 49 adults.

^dResults shown for comparison of step-children with a deceased biological father to biological children with a present father.

(b). Bayaka ethnographic data

The study community is home to approximately 400 people, with around half the population being BaYaka and the other half being Bondongo fisher-farmers [13]. The focus of the present study is solely on BaYaka, as we do not have comparable physiological measures from Bondongo residents. BaYaka society is status-averse, with cultural practices that help mitigate hierarchy. The community strongly values cooperation, sharing and respect for the autonomy of all individuals [13]. As noted, they engage in cooperative caregiving, including widely sharing resources within the community, such as hunted animal protein, honey and collected and cultivated plants [29], and engaging in communal rearing of children, with non-parents routinely assisting families with direct care and supervision of children.

Within families, avoidance of routine marital conflict is also culturally valued, and marital relationships are equitable, with men and women often working collaboratively on domestic and foraging activities. To forage, BaYaka individuals often leave the village to acquire forest resources and obtain a major part of their daily subsistence from forest foods. In the forest, they engage in hunting, fishing, and trapping and collect honey and plant resources. Men primarily do the hunting in the community, but they also meaningfully contribute to subsistence and provisioning through other forest activities. Families also maintain low-intensity gardens of cassava, maize and plantains. Finally, in the village, Bondongo families often employ BaYaka adults for domestic and subsistence activities, including to collect forest resources for both consumption and other needs (e.g. building materials). Given their overall patterns of subsistence, we refer to the BaYaka as ‘foragers’ while recognizing that this term glosses over complex histories and identities [26,30,31].

(c). Bayaka fathering and ranking data

One of our major goals was to use mixed methods to understand and model culturally defined roles for BaYaka fathers. In 2017, AHB and SLL conducted qualitative ethnographic interviews with BaYaka men and women to define the relevant local cultural domains for the qualities that made a ‘good’ father. Consistent themes emerged in these interviews, and we had fathers rank one another according to each characteristic (for further information, see [13]).

BaYaka participants characterized good fathers as those who: hunted, gathered honey and collected other forest resources for their families (Provider); engaged in less conflict with their wives (Dispute); welcomed others to the community and shared resources well within the community (Share) and taught their children to forage (Teach), although we did not include this latter domain in the present analyses [13]. After establishing these domains, we then had fathers rank one another according to each of those characteristics. We recruited an opportunistic subsample of BaYaka fathers from the larger study (n = 21). During the ranking task, we showed the participant a set of photographs of their peers and then asked them to place these in order, in piles, from ‘first’ to ‘last’ for each domain. We asked that fathers attempt to uniquely rank each of their peers, but allowed for ties. We recorded the last position as a score of ‘1’ and the highest score as the total number of piles made. Using the ranking scores from the 21 BaYaka participants, we then calculated an average father ‘quality’ peer-ranking score from their fellow fathers in each domain. The Cronbach's alphas for these peer rankings ranged from 0.76 to 0.86, indicating relatively strong internal reliability [13]. The sub-set of BaYaka fathers who provided the rankings (n = 21) did not significantly differ from the fathers who did not rank their peers but who had ranking data (n = 10) for any key study variables [32].

(d). Socio-demographic data

Parents reported family demographic data during interviews, such as number of children and age of youngest child. Because the BaYaka do not record age, we estimated an approximate age following methods and with assistance from Diekmann et al. [33].

(e). Anthropometric data

We collected triceps skinfold thickness data using Lange skinfold calipers by standard techniques and measured height with a Seca stadiometer. As in our past work on BaYaka children [13], we created standardized values of the anthropometrics adjusted for age and sex, to allow for comparison across the age range of the sample of children. We regressed each anthropometric measure on age and sex. We then saved the residuals of each model, added them back to the mean of the relevant measure, and transformed the resulting variables to z-scores to create population-specific reference values. We treated triceps skinfold thickness as a shorter-term indicator of energetic condition, and height-for-age as a longer-term marker of energetic exposures during childhood growth [34].

(f). Fingernail cortisol

At two time points approximately two weeks apart (15 days ± 3.0 s.d.), we collected fingernail clippings from participants from all digits on both hands. Approximately 61% of participants provided repeated nail samples. Cortisol in fingernails is a cumulative indicator of the body's exposure to the hormone in the four to five months prior to collection, which conceptually aligns with the goals of our study to test whether aspects of family function correlate with chronic, cumulative production of the hormone [35]. We chose to measure cortisol via fingernails because BaYaka individuals generally keep their hair cut very close to the scalp, particularly for men and boys, which made nail sampling a better cultural fit in this setting. Moreover, though we have drawn on saliva sampling at this site [31,32], we did not have the research infrastructure to collect multiple saliva samples across the day, repeated across multiple days, for each participant, as is recommended for assessment of trait-like individual cortisol profiles, and to then freeze that large volume of samples on site during fieldwork [36]. After shipment to Dr Jerrold Meyer's laboratory at the University of Massachusetts-Amherst, the nail samples were analysed for cortisol using a commercially available kit (Arbor Assays; Ann Arbor, MI; no. K003-H1). Fingernail samples were weighed and then washed twice for 1 min with isopropanol. Afterward, nail grinding was performed for 2 min using a Mini-BeadBeater-16 (BioSpec, Bartlesville, OK) with a single 6.35 mm chrome steel bead. Cortisol in each ground nail sample was extracted overnight (18–24 h) with 1.5 ml HPLC-grade methanol [37]. Of the 1.5 ml methanol extract, 1.0 ml was evaporated and then reconstituted in 0.25 ml of Arbor Assays assay buffer for subsequent procedures. The reconstituted extract was filtered using a Corning Costar Spin-X 0.45 µm cellulose acetate filter (VWR, Philadelphia, PA, USA). Output of the cortisol assay was adjusted by the stated pipetting volumes to calculate total sample content measured as pg of cortisol. This value was then divided by the weight of the sample in mg, resulting in final units of pg mg⁻¹ [35]. For children, the nail powder weights in the analyses were a mean of 15.33 mg ± 11.39 s.d. and for adults the mean was 30.78 mg ± 29.93 s.d. The intra- and inter-assay coefficients of variation (CVs) were 11.0% and 13.6%, respectively, for cortisol; these CVs provide a measure of assay precision. We excluded two cortisol values that were 3+ s.d. from the mean for the sample, and a small number of samples had insufficient powder weight for the assay. We natural log-transformed cortisol values in all analyses.

(g). Statistical analyses

We conducted all statistical analyses using Stata v. 14.0 (Stata Corporation). In our core models, we used linear mixed models with maximum-likelihood estimation (Stata's ‘mixed’ command) to predict parent and children's cortisol, respectively, from fathers' ranking scores for Share, Dispute and Provider. In each model, we included a random intercept effect for each adult or each child nested within a family ID (linked through the father, for children). By including these nested random effects, we accounted for the multi-level structure of the data, given that some individuals had two data points for cortisol and that children and parents, respectively, were nested within families and thus have ranking values in common (since those rankings are from the father in each family). The parents’ cortisol models also had heteroskedastic error variance by sex, thus we included a random effect for sex, allowing for the difference in error variance between mothers and fathers. To allow for comparison of standardized coefficients across models, we converted the dependent variables and our core predictors to z-scores. In the full models for each outcome (table 1), we then adjusted for theoretically grounded and empirically informed covariates (see the next paragraph and the electronic supplementary material, tables S2 and S3). We evaluated statistical significance at p < 0.05.

(h). Covariates and model selection

To help reduce the likelihood of overfitting our core analyses, we identified theoretically and empirically supported covariates that could potentially confound or help to explain links between fathers' rankings and family members’ cortisol, including: anthropometrics (triceps skinfolds; height-for-age for children), number of children and age of youngest child in the family, age and step-child status. Given space constraints, we have provided information regarding the background and rationale for these covariates in the electronic supplementary material, table S2. Using Akaike information criterion (AIC), we then assessed whether the covariates improved the model fit for our core analyses predicting cortisol from fathers' rankings. Although conventions vary, we evaluated whether the model including the covariate had a lower AIC value, compared to the base model [38]. We report the AIC values for these models in the electronic supplementary material, table S3. In models for children's cortisol, age, triceps skinfold thickness and step-child status each improved the model fit, compared to the base model (electronic supplementary material, table S3). In the parents' cortisol models, triceps skinfold thickness and sex improved the model fit (electronic supplementary material, table S3). Along with the full model results, we report the final models’ AIC values in table 1.

3. Results

(a). Children's cortisol and fathers' rankings

In linear mixed models, we found that children's cortisol was lower if their fathers were viewed as more generous sharers of resources within the community (p = 0.002) and if their fathers were ranked as better providers (p = 0.018), respectively, compared to children whose fathers were ranked as less effective in those domains (figure 1 and table 1). Children also had higher cortisol if their parents were viewed as having greater marital conflict (p = 0.016; figure 1), compared to families seen as having less conflict. In the models for Share and Provider, children with greater triceps skinfold thickness (age- and sex-adjusted) had lower cortisol than those with lower triceps skinfold thickness (both p < 0.01). See table 1 for the full model results.

Figure 1.

Coefficient plot following linear mixed models for fathers' Share, Dispute and Provider rankings and other continuous covariates predicting children's fingernail cortisol. All of the variables are in s.d. units. The plotted coefficients reflect the linear mixed models in table 1.

(b). Parents’ cortisol and fathers' rankings

In complementary linear mixed models for parents, fathers’ rankings for Share, Dispute or Provider were not significantly correlated with parents' cortisol (all p > 0.1; table 1). We adjusted these models for sex and triceps skinfold thickness, but neither covariate was a significant predictor (all p > 0.1). See table 1 for the full model results.

4. Discussion

In the present study, we tested whether a cumulative measure of cortisol in children and parents was related to fathers’ peer rankings for locally valued domains of fathering. Our research provides an important socio-ecological contrast to the majority of research on children's stress-related physiology, which has been conducted predominately in US–European contexts in which community social (and physical/material) support of families is often less robust than in the cooperative caregiving BaYaka society we focused on here [4,5]. We found that children had lower cortisol if their fathers were ranked higher on valued parenting roles, including sharing in the community, providing resources to the family and engaging in less marital conflict. We contextualize these results below based on psychobiological and cooperative caregiving frameworks and in relation to past research on family influences on children's cortisol production.

(a). Fathers’ rankings and children's cortisol

We found that children had lower cortisol if their fathers were seen as better community sharers and providers, compared to their peers whose fathers were ranked as less effective in these domains. As we have described previously, these locally valued dimensions of fathering (i.e. Share and Provider) are strongly correlated, such that men who are viewed as better parents in one domain are also ranked higher in the others [13]. Thus, the individual findings linking children's cortisol to fathers' Share and Provider rankings in each domain potentially reflect an aggregate pattern that higher quality BaYaka fathers have children with lower cortisol production. These patterns align with the notion that children are not necessarily ubiquitously buffered from variation in parental quality in a cooperative caregiving society like the BaYaka where they can typically rely on multiple non-parental caregivers for emotional and instrumental/practical support. It is well characterized that experiences of social or physical threat, lack of control and unpredictability activate cortisol production from the HPA axis [4,8], which can lead to chronic upregulation in cortisol production, as is measured in cumulative indicators like hair and fingernails [39]. We suggest that BaYaka fathers who are higher quality based on local standards may help attenuate these routine, day-to-day experiences for children. Reciprocity within the community in response to fathers’ sharing and generosity could be one potential pathway through which this occurs.

Among BaYaka communities, the sharing of resources and in rearing of children occurs routinely, as is common in many other forager societies [1,12–14]. It is plausible that BaYaka fathers who are seen as more generous may receive greater reciprocity of energetic resources, especially during difficult times [16,27]. Because cortisol rises in response to energetic stress and food insecurity could also be experienced by children as a social and physical threat, such reciprocity could help attenuate cortisol levels in BaYaka children with generous fathers. However, we have previously shown that BaYaka men's fathering rankings, including for sharing, were not meaningfully linked to children's growth and energetic well-being [13]. We argued that it is likely that community resource sharing buffers BaYaka children from routine nutritional short-falls (e.g. being underfed) that might otherwise be linked to fathering quality, including provisioning, in other settings [13]. Along those lines, BaYaka children are typically not food insecure or routinely undernourished but do commonly face pathogen stress, especially from helminth infection in this rainforest ecology. Here, we adjusted our models for children's triceps skinfold thickness (a marker of shorter-term energetic condition) and considered height-for-age (a longer-term indicator of energetic experiences during childhood growth) as a covariate. The latter did not improve the model fit predicting children's cortisol and was not significantly or meaningfully correlated with it. While BaYaka children with lower triceps skinfold thickness (age- and sex-adjusted) had higher cortisol, this did not account for relationships between fathers' sharing rankings and child cortisol.

This is consistent with the possibility that the relationship between fathers’ sharing and children's cortisol could be mediated by psychosocial mechanisms (e.g. how children are treated socially within the community) as opposed to an energetic pathway. In past research among Hadza foragers in Tanzania, Crittenden & Marlowe [12] noted that women who were better at acquiring resources and had higher social standing might receive ‘not-in-kind reciprocity’ (e.g. resources exchanged for direct care or child monitoring). Among BaYaka, it is possible that children of more generous fathers benefit from social support derived from this type of not-in-kind reciprocity. As we mentioned in the Introduction, prior work among Tsimane’ Amerindian forager-horticulturalists found that more prestigious men (i.e. status accrued partly through generosity) had children with higher survival rates [16]. More recent research has extended these patterns to Tsimane’ women's social status, as women with more political influence had children with better indicators of growth and health [40]. Complementing these studies, our results highlight a potential physiological pathway through which men's reputations as generous community sharers may relate to their children's health and well-being in an egalitarian, cooperative caregiving society.

Finally, we also found that BaYaka children in families who were perceived as having greater parental conflict had higher cortisol relative to those in families ranked as having less parental disputing. This pattern is generally consistent with findings from other cultural settings in which children experiencing more conflict within their families exhibit higher basal salivary cortisol [2,18,19] as well as other physiological indicators of psychosocial stress [41,42]. While verbal and occasional physical aggression do occur among BaYaka couples, the community generally disapproves of conflict between wives and husbands, and there are social mechanisms through which individuals attempt to mediate and attenuate conflict, reflecting a form of social support that could have benefits for children, mothers and fathers [7,13]. In our results, the observed positive correlation between parental conflict and children's cortisol is consistent with the idea that such family contexts are stressful to BaYaka children (e.g. as experiences of social threat or unpredictability), despite the cultural emphasis on cooperative caregiving and social support children may receive thereof. Indeed, BaYaka children form close relationships with their parents and the prospect of losing one of them (e.g. through divorce resulting from conflict) is socially threatening [7,43]. Along these lines, according to our respondents, one reason that ‘good’ BaYaka fathers avoid marital conflict is because it negatively affects children's emotional well-being.

(b). Fathers’ rankings and parents' cortisol

BaYaka couples have periodic verbal disputes and occasional physical aggression does occur, which are theoretically plausible pathways that could connect to chronic upregulation of the HPA axis. However, we did not find a significant association between fathers’ marital conflict rankings and parents' cortisol. This contrasts with the observed patterns among BaYaka children. It also differs from findings for couples in the US and European contexts, where marital conflict is typically associated with greater cortisol production, measured via saliva and blood [44], and women exposed to intimate partner violence had higher hair cortisol levels than unexposed women [11]. We must be cautious in over-interpreting non-significant results, particularly in a relatively small study. Nonetheless, the bivariate effect size linking marital conflict to parental cortisol was small and close to zero (r = −0.03), which can be broadly compared to relevant effect sizes elsewhere [44], with attention paid to differences in research design and cortisol measurement. This raises important questions regarding whether parental stress physiology within families may be expressed differently in contexts with greater cooperative caregiving and other aligned cultural norms, compared to the US, Europe and similar societal settings. For example, BaYaka communities have strong cultural values of autonomy, rather than spousal inter-dependence, which may contribute to diminished links with marital conflict. This is speculative at this time and would need to be tested in future studies.

We also did not find significant links between parents’ cortisol and men's reputations as providers or sharers within the community, respectively. Recent complementary research among Hadza women in Tanzania also found that two markers of women's social status (foraging reputation and popularity) were not significantly linked to their hair cortisol levels [45]. However, past work among Tsimane’ Amerindian forager-horticulturalists found that men who were more politically influential had lower urinary cortisol than less influential men, which was partially related to differences in social support [28]. Our measure of fathers' reputations as sharers within the community is potentially (if not likely) distinct from the Tsimane’ study's measure of political influence. Although we reiterate our cautiousness about over-interpreting null results, these sets of findings set up future questions regarding physiological correlates of social capital, social networks and cooperative caregiving in relatively egalitarian societal settings.

(c). Limitations

A primary limitation of this study is the under-sampling of BaYaka girls (see Material and methods). While boys' and girls’ cortisol did not significantly differ in the present study, we cannot rule out that aspects of their social experiences within the community could potentially vary in ways that interrelate with cortisol production. In addition, the fingernail measure of cortisol production that we used in the current study is an aggregate measure of the body's exposure in prior months, sharing similarities with hair cortisol [39,46]. Non-invasive sampling of nail/hair hormones cannot capture dynamic aspects of cortisol's diurnal rhythm that are known to be responsive to chronic stress and predictive of poor health outcomes in the US and Europe [47] and increasingly in other global contexts [48]. Nonetheless, important insights on inter-relationships between psychosocial stress, health and cortisol can be gained from cumulative measures of cortisol production [39]. Moreover, while fingernail cortisol is less commonly studied than hair cortisol, the incorporation of the hormone into both sources is biologically similar and studies indicate similar validity for nail cortisol relative to hair cortisol [35,46].

However, an additional methodological limitation is that there could be sources of fingernail cortisol variation that our study does not capture, such as frequency of hand washing, which could potentially affect detected cortisol levels, similar to wash-out effects in hair [39,46]. Individual differences in nail growth rates and engagement in manual, day-to-day tasks that affect the exterior of the nails (e.g. breaking or wearing down) could also add between-individual variation to the time frames reflected in nail cortisol values. These concerns would generally decrease data reliability for our analyses, increasing the likelihood of Type II error. Thus, we do not think they are likely to explain our consistent results linking fathers' rankings and child cortisol. However, they may merit consideration in terms of our null findings for parental cortisol and for future research in this area, in general, particularly since measuring nail cortisol is a relatively new methodological approach that has been studied and validated less widely than hair cortisol [39,46].

5. Conclusion

In a study of BaYaka forager families residing in the Congo Basin, we found that children whose fathers were considered more effective in locally valued parenting domains, including engaging in less marital conflict, providing more resources to the family and sharing generously within the community, had lower cortisol. Our finding linking children's cortisol and fathers' rankings as community sharers is particularly salient for this context. In highly egalitarian settings, such as this BaYaka community, cooperative caring through shared commitments to the demands of daily life are central to group functioning and well-being, and this type of hyper-cooperation may have been critical to the evolution of human life history [1,13]. Given cortisol's immune system modulation and energetic roles in the body, our results provide evidence of a physiological pathway through which children's health and well-being could be shaped by variation in their fathers' community cooperation and generosity in addition to paternal roles within the family.

Ethics

Initial permission to conduct research in the village was given by the village council during a community meeting with A.H.B. in 2015. The Centre de Recherche et D'Etudes en Sciences Sociales et Humaines granted permission to conduct the research in Republic of the Congo, and the Institutional Review Boards of Duke University (Protocol no. 2017-0038) and the University of Notre Dame (no. 18-02-4397) also approved the study. The verbal consent process as well as all data collections and methods were conducted based on Duke University and University of Notre Dame ethics guidelines. Individual informed verbal consent was obtained from all adult participants and children old enough to provide consent. Parental consent and child assent were obtained for all other children.

Data accessibility

We have included a codebook and data necessary to replicate the patterns in the current analyses in files that we have appended to the current submission. Because the BaYaka community is small, we took steps to help protect their identities and privacy. First, we used a published R script for a package known as ‘synthpop’ to generate synthetic demographic data that approximate the real data. We converted the synthetic demographic variables to z-scores. For other study variables, we have included the non-synthetic data in the attached .csv files. See https://elifesciences.org/articles/53275.

Authors' contributions

L.T.G. and A.H.B. conceived of the study and secured funding for the research. L.T.G. analysed the data and wrote the article with routine input from A.H.B. and S.L.L. M.S.S. assisted with editing and framing. A.H.B., S.L.L., M.S.S. and V.M. collected the data. M.D. conducted all of the laboratory procedures for cortisol in consultation with J.S.M. All authors reviewed the content of the article.

Competing interests

We declare we have no competing interests.

Funding

The Jacobs Foundation and the University of Notre Dame supported this work.