Cooperation beyond consanguinity: post-marital residence, delineations of kin and social support among South Indian Tamils

Abstract

Evolutionary ecologists have shown that relatives are important providers of support across many species. Among humans, cultural reckonings of kinship are more than just relatedness, as they interact with systems of descent, inheritance, marriage and residence. These cultural aspects of kinship may be particularly important when a person is determining which kin, if any, to call upon for help. Here, we explore the relationship between kinship and cooperation by drawing upon social support network data from two villages in South India. While these Tamil villages have a nominally male-biased kinship system (being patrilocal and patrilineal), matrilateral kin play essential social roles and many women reside in their natal villages, letting us tease apart the relative importance of genetic relatedness, kinship and residence in accessing social support. We find that people often name both their consanguineal and affinal kin as providing them with support, and we see some weakening of support with lesser relatedness. Matrilateral and patrilateral relatives are roughly equally likely to be named, and the greatest distinction instead is in their availability, which is highly contingent on post-marital residence patterns. People residing in their natal village have many more consanguineal relatives present than those who have relocated. Still, relocation has only a small effect on an individual’s network size, as non-natal residents are more reliant on the few kin that they have present, most of whom are affines. In sum, marriage patterns have an important impact on kin availability, but the flexibility offered by the broadening of the concept of kin helps people develop the cooperative relationships that they rely upon, even in the absence of genetic relatives.

This article is part of the theme issue ‘The evolution of female-biased kinship in humans and other mammals’.

1. Introduction

Evolutionary ecology has a longstanding interest in the relationship between relatedness and cooperation [1]. Inclusive fitness theory suggests that the shared genetic material among kin may facilitate altruism, because by helping a relative, an organism can increase its own fitness (i.e. the replication of its own genes) [2,3]. Considerable research has found that consanguineal relatives, and especially close consanguineal relatives, are more likely to have cooperative ties with each other than with non-relatives [4–6]. The presence of female relatives in particular has been shown to have a positive impact on female reproductive performance in numerous mammalian species, including non-human primates, providing diverse benefits such as coalitionary support, reduced stress and improved offspring nutrition and survival [7].

While the benefits of support from relatives is straightforward, the actual availability of relatives is far more complicated. Sex-biased dispersal (where one sex leaves their natal territory) mitigates inbreeding and resource competition among relatives, but also shapes relatedness within local groupings. Female philopatry is the norm in mammals, possibly because foraging efficiency has a stronger impact on female reproductive success [8]. In primates, cooperative relationships are sometimes stronger in the non-dispersing sex, for example, among male chimpanzees [9,10] and among females in matrilineal cercopithecines [11,12]. Social bonds in the dispersing sex have been considered to be less developed, although they are not absent [13,14]. The consequences of dispersal in primates makes clear the importance of social systems for shaping individuals’ prospects, strategies and behaviour, especially in terms of cooperative alliances with kin.

In our own species, many societies have exogamous marriage practices that lead to one gender more consistently dispersing than the other. Such practices have the consequence that those who leave have few (or at least fewer) relatives present, while those who stay in place will be primarily surrounded by relatives from either the maternal or paternal side. Our unique life-history pattern—especially our extended period of childhood dependency—means that women can particularly benefit from help during their reproductive years [15,16], and, across numerous cultural contexts, maternal relatives have been found to be particularly important for assisting women and their children [17]. Yet, cross-cultural data suggest that, in contrast to other mammals, in human societies it is most often women who leave their natal communities [18].

As dispersing individuals often join groups with fewer relatives, they may be at a disadvantage in terms of their ability to build cooperative relationships. However, cultural kinship systems, as an extension of relatedness, have the potential to ameliorate this disadvantage in several ways. First, one of the main features of human kinship systems is the extension of kinship to affinal kin, meaning in-laws [19–21]. Many affines share a real genetic stake in descendent generations, and so their fitness is in this sense interdependent [22–27]. Another common feature of human kinship systems is prescriptions for preferred marriage partners. Such preferences often have the effect of bringing more distant relatives back into the family fold (e.g. cross-cousin marriage), and mean that dispersing individuals may nevertheless find themselves in the presence of consanguines. In these ways, human kinship systems may facilitate the development of cooperative ties both within and beyond consanguineal relatives, ties that should perhaps be particularly important for the dispersing sex. However, kinship systems also divide, creating distinctions between categories of kin (e.g. matrilateral versus patrilateral) and establishing normative obligations for particular types of kin relations. We expect that these cultural aspects of kinship which modify how people interact with all types of kin should have consequences for people’s cooperative relationships, and ultimately for their reproductive success.

To explore how these various aspects of cultural kinship systems may differentially influence cooperation, we focus on the case of Tamil kinship. Dravidian kinship has long been seen as a particularly distinctive and complex form of kinship [28,29], and one that has been argued to have important ramifications for regional demographic patterns. Tamils traditionally favoured close kin marriages, with preferential marriages commonly being between cross-cousins or between a maternal uncle and his niece [30]. Such marriages result in a tangling of kinship relations [31], and mean that many women continue to reside near their natal family after marriage. Consequently, many have argued that matrilateral relatives are often present and prominent providers of support for South Indian women, giving them greater autonomy and better health outcomes than women in other regions of India that have different marriage patterns [30,32–38]. Tamils, then, nominally have patrilocal residence, but many women actually continue to reside in their natal place, and have patrilineal inheritance, but place a simultaneous emphasis on matrilateral relatives. Here, we attempt to use this variability to investigate how these aspects of kinship which extend beyond simple relatedness are associated with the support that people are able to call upon.

(a). Hypotheses

Both the primate literature on sex-biased dispersal and the demographic work on marriage practices and village exogamy of India consistently emphasize that leaving the natal place entails leaving one’s consanguineal relatives. This leads us to hypothesize that: (Hypothesis 1) People residing in their natal village after marriage should (a) have more consanguineal relatives present and (b) name more people as providing them with support, than those who are not in their natal village after marriage.

The hypothesized supportive advantage for natal residents rests on the grounds that consanguineal relatives should be more likely to offer support to one another than to non-relatives, because of inclusive fitness. This leads us to hypothesize that: (Hypothesis 2) Greater genetic relatedness should be associated with an increased likelihood of a supportive relationship.

Kinship systems further incorporate ties to unrelated individuals connected through bonds of marriage that lead to shared social, economic and reproductive interests. Consequently: (Hypothesis 3) (a) Greater affinal relatedness should be associated with an increased likelihood of a supportive relationship. This should be especially so for non-natal residents if they have fewer consanguineal relatives (see Hypothesis 1a), in which case, (b) compared with natal residents, non-natal residents should name more of their spouse’s consanguineal relatives as providing them with support.

We next consider whether people differentiate between their consanguineal relatives based on laterality. Despite being patrilineal, Tamils see relations with matrilateral relatives as being particularly affectionate and those with patrilateral relatives as being potentially contentious, thanks in part to disputes over inheritance [39,40]. Both these local conceptions of kinship and evolutionary research on kin support lead to the hypothesis that: (Hypothesis 4) matrilateral relatives should be more likely to have supportive relationships with each other than patrilateral relatives.

Within the broad class of matrilateral relatives, the role of the mother’s brother (the tāymāmaṉ) is particularly culturally salient. He has specific responsibilities at life cycle rituals and is understood to be someone whom one can ask things of freely [40,41]. Tamil men are seen as having a duty to ensure the well-being of their sisters and their sisters’ children [42]. Because of the strong sentiment within Tamil culture (and because of parallels to research showing the importance of the mother’s brother in matrilineal [43] and in some other descent systems [44]), we hypothesize that: (Hypothesis 5) Men should be especially likely to provide support to their sisters and their sisters’ immediate family.

2. Material and methods

To answer these questions, we draw on data gathered by the first author as part of ethnographic fieldwork conducted in two neighbouring villages in the Indian state of Tamil Nadu, ‘Aḻakāpuram’ and ‘Teṉpaṯṯi’ (both pseudonyms). Each village has roughly 400 adult residents, representing a mix of different caste and religious denominations (see electronic supplementary material, §S1 for more details). Most engage in a mix of agricultural and wage labour, with a growing number seeking more skilled work, whether in a factory, shop or office. Residents assist each other in many ways: working together in the fields, sharing news and employment opportunities, watching each others’ children, sharing meals, etc. Whether kin, caste-mates or simply neighbours, their assistance is crucial to each others’ livelihoods [45].

(a). Kinship relations

Networks representing kinship relations between the residents of the villages are constructed based on data from a household census first conducted in 2011 and updated in 2017 (figure 1; see electronic supplementary material, §S2 for more details). While the resulting kinship network for Aṉakāpuram is dominated by one large component linking many of the Paḽḽar caste together, the distributions of kin are similar across the two villages. For example, the average coefficient of relatedness among survey respondents is quite comparable (0.0031 in Aṉakāpuram versus 0.0034 in Teṉpaṯṯi).

Figure 1.

The kinship networks of Aṉakāpuram ((a) N = 440) and Teṉpaṯṯi ((b) N = 344). Nodes (individuals) are coloured by caste. Edges are coloured by the nature of the kinship relationship.

For Hypothesis 1, we use records of each person’s marital status and conta ūr, their ‘natal place’, to create a variable denoting whether people are living in their natal village or not. Across the two villages, almost all unmarried people (95%) are living in their natal village, and among those who have ever been married, 30% of women and 87% of men are living in their natal village. Histograms showing the number of kin of varying types that residents (a) have residing in the same village and (b) named as a source of support are provided in figure 2.

Figure 2.

Distributions of the number of resident kin and of the number of kin named as providing support, for four groups of kin (consanguineal, affinal, matrilateral and patrilateral) broken out by gender and natal/non-natal residency, for all survey respondents in Aṉakāpuram and Teṉpaṯṯi. Numbers in the upper right of the plots give the mean and per cent of zeroes for the overall distribution in each plot.

For Hypothesis 2, we use records of people’s parentage from our kinship data to calculate the estimated genetic relatedness between residents, dividing them into three groups: those with an approximate relatedness of 0.5 (or greater), 0.25 (to less than 0.5) and 0.125 (to less than 0.25). Because of potential issues of completeness, we only consider relatedness up to the 0.125 level.

For Hypothesis 3, we examine relationships with affines, both the relatives of spouses and the spouses of relatives, using an ‘affinal relatedness’ coefficient, based on the estimated relatedness of the consanguineal relatives involved [23], divided into four groups: spouses (assigned an affinal relatedness of 1.0), followed by affines with an affinal relatedness of 0.5 (e.g. a brother- or daughter-in-law), 0.25 (e.g. a spouse’s aunt or uncle) and 0.125 (e.g. a full cousin’s spouse). Electronic supplementary material, table S2 provides a full description of the individuals included in each category.

For Hypothesis 4, we disaggregate the genetic relatedness matrices into five groups: one consisting of immediate family (parents, children and full siblings), and four composed of matrilateral and patrilateral relatives, with estimated relatedness of 0.25 and 0.125.

For Hypothesis 5, we include additional terms for the support provided by people to their siblings and their siblings’ immediate family. This includes not only a term for men’s support of their sister and sister’s immediate family, but for all combinations of sibling relationships.

(b). Social support networks

The social support network is drawn from a survey conducted with the adult residents of the villages (N = 440, 97% in Aṉakāpuram and 344, 94% in Teṉpaṯṯi) in September 2017 (see electronic supplementary material, §S3 for more details). All interviewees provided oral consent. The survey consisted of several questions asking who they would turn to for different kinds of help, including getting a loan, borrowing household items, getting help with physical tasks, having convivial conversations, discussing important matters and getting help finding work. While interviewees could name anyone, here we limit our focus to include only ties among survey respondents within each village, which constitute 65% of all nominations. For the questions studied here, interviewees named an average of seven other residents as providing them with support of some kind. These nominations are combined to create a network representing the flows of support between survey respondents (figure 3),which comprise 3266 ties in Aṉakāpuram and 2474 ties in Teṉpaṯṯi. We find that while kin are more likely to provide certain types of support, there are limited differences in which kin provide support of different types, so the aggregation of different support types should have a limited impact on our results (see electronic supplementary material, table S5).

Figure 3.

The social support networks of Aṉakāpuram ((a) N = 440) and Teṉpaṯṯi ((b) N = 344). Nodes (individuals) are coloured by caste. Edges are directed, with arrows pointing to the individual asked for support. Node position is determined by the Fruchterman–Reingold algorithm.

(c). Data analysis methods

We construct a series of exponential random graph models, or ERGMs [46,47], which allow us to model the probability of a support tie between two people, based on individual, dyadic and structural terms, among which are terms representing the kinship relations between individuals (see electronic supplementary material, §S4 for more details). To assess how residing in one’s natal village (or not) shapes the assortment of kin present and overall access to support, we conduct a series of simple Poisson regressions that model the number of kin present. To assess how residence patterns shape the propensity to call upon those kin, we conduct a series of simple binomial regressions that model the proportion of available kin named as a source of help (see electronic supplementary material, §S5 for more details). For these regressions, we include only people who have ever married, as we are interested in post-marital residence.

3. Results

(a). Do people who live in their natal village after marriage have more kin present and greater support?

We find that people residing in their natal village after marriage have substantially more consanguineal relatives present than those who moved into the village (electronic supplementary material, table S10), and that they also report having more support ties overall (electronic supplementary material, table S11). However, the difference between (ever-married) non-natal and natal residents in the number of support ties is slight compared with the difference in the number of co-resident relatives that they have (figure 4a,b). In Teṉpaṯṯi, for example, despite having substantially fewer consanguineal relatives (four fewer, on average), non-natal residents have only one fewer support tie in the village, on average.

Figure 4.

(a) Odds ratios (with 95% confidence intervals) for natal versus non-natal ever-married residents for having relatives of various types residing in the same village (‘co-resident’ kin type), for having named others as support partners (named support), and for having named relatives of various types as support partners (‘named’ kin type), for each village, based on simple Poisson and binomial regressions (see electronic supplementary material, tables S10 and S11). (b) Violin and dot plots showing the number of co-resident consanguineal relatives (top), named support partners (middle) and named affinal kin (bottom, specifically spouse’s consanguineal relatives) for natal and non-natal ever-married residents of Aṉakāpuram.

In ERGMs that include terms for the likelihood of both incoming and outgoing support ties among natal and non-natal residents, we find a similar, small effect of being from elsewhere (Model 1 in figure 5 and electronic supplementary material, tables S8 and S9). Overall, ever-married people (in their natal village or not) are more likely to be asked for help than unmarried people (who are almost exclusively living in their natal village). Predictions of the probability of a support tie between individuals based on this model show that, among ever-married people, non-natal residents have a lower probability of asking for support from natal residents than natal residents themselves. However, the magnitude of this difference is very slight (in both villages, a difference of roughly 0.5% in the probability of a tie: 4.0% versus 3.5% in Aṉakāpuram; 5.2% versus 4.7% in Teṉpaṯṯi, see electronic supplementary material, §S4 for more details).

Figure 5.

Odds ratios (with 95% confidence intervals) for the variables of interest across each of the main ERG models, for each village.

(b). Are kin in the village more likely to provide social support?

We first look solely at the role of genetic relatedness (Model 2 in figure 5 and electronic supplementary material, tables S8 and S9). As expected, ERGMs show that consanguineal relatives are more likely to report having supportive relationships with each other than with non-kin, however, support does not consistently decline with relatedness in this model. Binomial regressions show that non-natal residents (who have fewer consanguineal relatives present) more readily call upon those relatives, if they have them, than natal residents (figure 4a and electronic supplementary material, table S11).

We next add in model terms for affinal kin to the ERGMs, with terms for spouses and for affinal kin up to a relatedness of 0.125. Including these affine terms substantially improves model fit (Model 3 in figure 5 and electronic supplementary material, tables S8 and S9) relative to the model with relatedness only. In both villages, there is a lower likelihood of support with more distant relatives (both for affines, and now also for consanguineal relatives), but the differences are not statistically significant. In the binomial regressions, we find that, in Aṉakāpuram, non-natal residents are more likely to name affines (specifically, their spouses’ relatives) as providing them with support than natal residents (figure 4a and electronic supplementary material, table S11). While this does not hold for Teṉpaṯṯi, the ERGMs suggest that more distant affinal kin (those with an affinal relatedness of 0.125, such as a spouses’ cousins), are not more likely to be called upon than non-kin in this village. Accordingly, when only closer affines are considered, non-natal residents in both villages are more likely to call upon them for support than natal residents (electronic supplementary material, table S11).

(c). Are matrilateral or patrilateral kin more likely to provide social support?

When we break out the relatedness matrix to distinguish between relatives who are matrilateral or patrilateral, we find no clear evidence of either side being consistently favoured (Model 4 in figure 5 and electronic supplementary material, tables S8 and S9). Breaking out these groups results in almost no improvement to the model log-likelihood, so any difference between them explains relatively little variance in the data.

Binomial regressions show that, if they have them living in the village, non-natal residents are generally more likely than natal residents to call upon either matrilateral or patrilateral relatives (figure 4a), although in Teṉpaṯṯi, we do not find a significant difference between natal and non-natal residents for calling upon matrilateral relatives, likely owing to low power. The fitted model probabilities for these regressions in Aṉakāpuram indicate that the overall probability of naming a matrilateral relative as a source of support is slightly greater than the probability of naming a patrilateral relative, for both natal and non-natal residents: natal residents of Aṉakāpuram have a 16% chance of calling upon a matrilateral relative and a 12% chance of calling upon a patrilateral relative. For non-natal residents, their probabilities are 28% for matrilateral kin and 24% for patrilateral.

(d). Is the mother’s brother in particular more likely to provide social support?

To explore whether men are especially supportive of their sisters and their children, we run a final model (Model 5 in figure 5 and electronic supplementary material, tables S8 and S9), which includes additional terms capturing relationships between co-resident siblings and their siblings’ immediate families. Contrary to our expectations, we find that brothers are not more likely to be named as a source of support by their sisters and their sisters’ immediate family. Instead, in Teṉpaṯṯi, brothers are less likely to help their brothers and their brothers’ immediate family, while sisters are more likely to help their sisters and their sisters’ immediate family, and in Aṉakāpuram sisters are more likely to help their brothers and their brothers’ immediate family. However, we note that some of these relationships are relatively rare, especially those with sisters (e.g. there are only 11 women with co-residing sisters in Teṉpaṯṯi), so these results should be interpreted cautiously.

(e). Factors beyond kinship

Finally, we make a few observations about the other covariates in our models and how they interact with kinship. Caste-based homophily is strong in these networks: within-caste support ties are 1.9–2.2 times as likely in Teṉpaṯṯi and 2.4–3.0 times as likely in Aṉakāpuram (based on Model 3). Reciprocity and transitivity appear to be major features of the networks, and the coefficients for these terms remain relatively consistent across the models despite the addition of relatedness matrices that are symmetric and represent many transitive relationships. For this reason, the reciprocity observed in these networks appears not to be a simple side effect of more frequent interactions among kin (or even within castes).

4. Discussion

Our results suggest that whether consanguineal, affinal, matrilateral or patrilateral, when someone is seen as kin, they are more likely to be named as providing support. Among those who are seen as kin, the ordering of relatedness terms (whether genetic or affinal) does suggest a lessening of support with more distant kin, as has been seen in many primates and in humans [11,48], although the differences between the relatedness levels are generally not significant in our ERGMs. Instead, the clearer difference between the various categories of kin is in their availability, both generally, and for specific groups of people. Those residing in their natal village tend to have many more close consanguineal relatives present, and because of local marriage patterns, these tend to be patrilateral relatives. Those who are not residing natally—who are more often women—have many fewer consanguineal relatives to draw upon, and so rely more heavily on whichever kin are present, affinal or otherwise.

Through marriage and residence patterns, kinship systems structure which kin are most likely to be co-resident. Those residing in their natal village might seem to be at an obvious advantage with their greater reserve of relatives (as seen in many non-human primates), but we find that the flexibility facilitated by this extension of kinship means that even people with few relatives present are able to compensate in other ways: we find that affinal kin are an important source of support for residents, on par with consanguineal relatives. This is clearest for those who have left their natal village: even though they may have many fewer consanguineal relatives, the number of support ties they have within the village is quite similar to that of natal residents. This is at least in part because non-natal residents are more likely to call on whichever kin they have present, whether consanguineal or affinal. Practically, as they are less likely to have consanguineal relatives present, this means that non-natal residents are often more reliant on their affines.

Even the additional consanguineal relatives that natal residents have are not necessarily a boon; instead, they can be a source of potential conflict. If people are competing for scarce resources with their co-resident kin, they may be less inclined to assist one another [49,50]. Where wealth is heritable, siblings may be in competition over inheritance (such as land) [51,52], creating competition among some kin, but not others. We see some evidence of this with the suggestion of aversion between brothers in Teṉpaṯṯi, who, in this context, are the group of siblings most likely to be in conflict over inheritance. Further, people are often understood to have a duty (as well as an incentive) to help their kin. When the obligations that are created through kinship ties go unfulfilled, it can lead to substantial tension. This holds especially for those relationships with the greatest expectations, such as a Tamil man’s duty to look out for his sister’s family, and potentially even to form marital alliances with her daughters (through his own marriage, or that of his sons). If these obligations are met, the relationship may be an especially close one; if not, it may be strained or even severed entirely [31,40,53]. Beyond the fundamental issue of low statistical power, such ambivalence might contribute to the null finding for our hypothesis (following Tamil valuations) that brothers should be a particularly important source of support for sisters and their families.

These different expectations for particular types of kin are closely linked to the distinctions that kinship systems draw between groups of kin based on lines of inheritance and descent, most obviously between matrilineal and patrilineal kin. Despite being nominally patrilocal and patrilineal, we find matrilateral kin to be present relatively often (32% of people have at least one co-resident matrilateral relative). While we had expected that laterality might influence the propensity of kin to provide support, we instead find no strong evidence of greater solicitude from matrilateral than patrilateral relations in either village. Instead, in agreement with our findings for affinal kin, the particular category of kin involved does not seem to be especially important in structuring these dyadic support ties. Still, the fact that we observe essentially bilateral support is in some sense surprising, given that this is a male-biased kinship system. While in this setting there may be a bias towards males in inheritance and the reckoning of descent, supportive relationships with matrilateral relatives continue to be important. So, while these delineations of kin may help organize some cultural practices (such as inheritance), they do not imply that relationships with kin from either side are neglected or negligible [54].

While we have so far emphasized the distinctions between matrilateral and patrilateral relatives and between consanguineal and affinal relatives, it is important to note that these categories are often merged within persons. Close kin marriages, for example, inherently entail some relatives becoming affines as well as consanguines, and such entanglements regularly occur even without such prescriptions (as when two sisters marry two brothers). Such layering of kinship relations should presumably result in yet stronger relationships between multiply articulated kin.

Not only can these multiple categories of kinship strengthen bonds between relatives, but they may also broaden who counts as kin. In this setting, members of the same caste are often seen as an extended group of relations (contam), both colloquially and in actuality, as caste endogamy means that caste-mates are often distantly related by blood and/or by marriage. Not surprisingly, then, in all of our models, caste homophily has a large positive effect, with people being roughly twice as likely to have a supportive relationship if they are of the same caste than if they belong to different castes. In other settings (e.g. [54–56]), cultural kinship and extended kinship groups have been shown to be particularly helpful for coordination and cooperation. Overall, this suggests that kinship systems can substantially increase the pool of potential partners that a person has to call upon [19,20].

Finally, while kin are clearly important, we should not forget that the majority of supportive partners are neither close consanguineal nor close affinal kin (out to our relatedness thresholds of 0.125). As our ERGMs show, and as the large literature on cooperation also establishes [57], there are other foundations on which to build cooperative relationships. Some of these mechanisms, such as direct reciprocity, rest on direct, rather than indirect, fitness benefits and reciprocity is one of the largest predictors of a tie in our ERGMs. When people are of the same caste, have shared partners, or live near one another, they are again more likely to have a supportive relationship. Each of these factors may lead to recurrent interactions between people, such that supporting a person now may lead to future benefit, whether it is directly reciprocated or not. Importantly, all of these other mechanisms may also be at work among kin, who are almost always of the same caste, share many common connections and live close to one another. Kinship systems, then, may produce favourable conditions that promote cooperation not only through inclusive fitness, but also through these other mechanisms, the effects of which are difficult to disentangle.

Through these analyses, we have started to explore how kinship helps predict the presence or the absence of supportive relationships between residents, but future work could extend our analyses in a number of important ways. First, we have yet to explore how kinship may influence not just the existence of a supportive relationship, but also its strength, tenor and substance. It is particularly those relationships that are strong, stable and emotionally supportive that may be most crucial for well-being [58,59], so determining how kinship interacts with relationship quality may give a better sense of the various benefits of kinship. Future work could also focus on how kinship might differentially structure distinct types of supportive ties. Second, given the nature of our data, we have specifically modelled who people say they turn to for support. Data showing the actual provisioning of support or the decision to provide it might reveal important divergences from our findings, potentially highlighting conflicts between kin [60]. Third, we limited our analyses to relationships within the village, but it is clear that relationships—especially those with kin—extend far beyond its boundaries; indeed, the ability to maintain ties across time and space crucially distinguishes humans from other primates [19]. Future work should explore how people’s reliance on such relationships varies between individuals and between settings (e.g. with residence, across the age course, or in stable or stochastic environments). Fourth, we have truncated our measures of relatedness (both genetic and affinal) at 0.125, and found that kin with a relatedness of 0.125 are not consistently more likely to provide support than non-kin. To better establish how cooperative relationships may decrease with greater kinship distance, calculations of relatedness might need to be extended to include more distant relatives. Work could also move closer to a full evaluation of inclusive fitness by creating calculations of shared stake in future generations, which importantly erases some of the distinction between consanguines and affines [22,23,26]. Finally, more research remains to be done on how different mechanisms (e.g. inclusive fitness, direct and indirect reciprocity, mutualism) combine to enable cooperation among kin; and we have only begun to investigate how cultural notions of kinship may further complicate or facilitate these dynamics. Already, it is clear that understanding the importance of kin to cooperation in both male- and female-biased kin systems requires examining relationships with all types of kin, for different types of help, and at different scales of cooperation.

Ethics

The fieldwork was approved by the Human Subjects Institutional Review Boards of Stanford University and the University of Cincinnati.

Data accessibility

Those interested in accessing the anonymized data should contact the first author. R code is available through GitHub: https://github.com/eapower/Kinship.

Authors' contributions

E.A.P. collected the data, designed the research, analysed the data and wrote the paper. E.R. designed the research, analysed the data and wrote the paper.

Competing interests

We declare we have no competing interests.

Funding

Funding for fieldwork was provided by a National Science Foundation Doctoral Dissertation Improvement grant (no. BCS-1121326), a Fulbright–Nehru Student ResearcherAward, the Stanford Center for South Asia, Stanford University, the Santa Fe Institute, and a National Science Foundation Interdisciplinary Behavioral & Social Science Research grant (no. IBSS-1743019). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.