Content uploaded by Siobhán M. Mattison
Author content
All content in this area was uploaded by Siobhán M. Mattison
Content may be subject to copyright.
Evolutionary Contributions to Solving
the “Matrilineal Puzzle”
A Test of Holden, Sear, and Mace’s Model
Siobhán M. Mattison
Received: 20 April 2010 /Accepted: 1 November 2010 /Published online: 31 May 2011
#Springer Science+Business Media, LLC 2011
Abstract Matriliny has long been debated by anthropologists positing either its
primitive or its puzzling nature. More recently, evolutionary anthropologists have
attempted to recast matriliny as an adaptive solution to modern social and ecological
environments, tying together much of what was known to be associated with
matriliny. This paper briefly reviews the major anthropological currents in studies of
matriliny and discusses the contribution of evolutionary anthropology to this body of
literature. It discusses the utility of an evolutionary framework in the context of the
first independent test of Holden et al.’s 2003 model of matriliny as daughter-biased
investment. It finds that historical daughter-biased transmission of land among the
Mosuo is consistent with the model, whereas current income transmission is not. In
both cases, resources had equivalent impacts on male and female reproduction, a
result which predicts daughter-biased resource transmission given any nonzero level
of paternity uncertainty. However, whereas land was transmitted traditionally to
daughters, income today is invested in both sexes. Possible reasons for this
discrepancy are discussed.
Keywords Mosuo .Evolution .Kinship .Matriliny .Sex-biased parental investment .
Intergenerational transmission of wealth
In matrilineal kinship systems, descent and inheritance are directed toward kin
related through females. Postmarital residence is variable but is often with the bride’s
mother (uxorilocal), the groom’s mother’s brother (avunculocal), or, somewhat less
commonly, involves separate residences for the bride and groom (nata- or duolocal;
Driver and Schuessler 1967; Gough 1961a; Murdock 1949). Furthermore, because
certain forms of altruism are directed toward and received from kin related through
females in matrilineal kinship systems, the role of the father and other affinal
Hum Nat (2011) 22:64–88
DOI 10.1007/s12110-011-9107-7
S. M. Mattison (*)
Department of Anthropology and Morrison Institute for Population and Resource Studies,
Stanford University, Main Quad, Building 50, 450 Serra Mall, Stanford, CA 94305–2034, USA
e-mail: siobhanm@stanford.edu
relations is often diminished relative to their role in patrilineal or bilateral societies;
instead, the mother’s brother often assumes the highest status in the family.
The special position of the mother’s brother has been the subject of much
puzzlement and debate in anthropology: its occurrence in patrilineal societies has
been said to defy normative unilineal principles (e.g., see Bloch and Sperber 2002
for a review) and its emphasis in matrilineal societies to undermine the principle of a
male’s authority over his offspring (e.g., Richards 1950; Schneider 1961; Schneider
and Gough 1961). Explanations for this have been sought on a variety of theoretical
grounds, from unilineal evolutionist (e.g., Bachofen 1967; Morgan 1964)to
functional (e.g., Malinowski 1930; Radcliffe-Brown 1924), to structural-functional
(e.g., Fox 1983; Murdock1949; Richards 1950; Schneider and Gough 1961), to
evolutionary (e.g., Alexander 1974,1977; Flinn 1981; Hartung 1976,1981,1985;
Holden and Mace 2003; Holden et al. 2003). This paper briefly reviews the rationale
underlying various statements of what has become known as the “matrilineal puzzle”
(Richards 1950) and the attempts to “solve”it. In line with the purpose of this
special issue, it focuses on the contributions of quantitative evolutionary anthro-
pologists by way of testing a recent model explaining the evolution of matriliny as
daughter-biased investment (Holden et al. 2003) among the matrilineal Mosuo of
Southwest China. As the first independent test of this model, this paper verifies the
model’s main predictions in a new setting, while highlighting some nuances in its
application in contemporary contexts, thereby adding to the empirical foundations of
our understanding of matrilineal kinship.
The “Matrilineal Puzzle”: Inception and Conception
The study of matriliny has a long history in anthropology and yet one that may be
largely unfamiliar to students of evolutionary anthropology (Knight 2008). With the
publication of Das Mutterrecht (Mother Right) in 1861, Bachofen (1967) was the
first anthropologist to theorize about the nature of matrilineal kinship (Divale 1974).
In 1877, Lewis Henry Morgan (1964), struck by similarities in classificatory kinship
terminology among matrilineal Native American tribes, pioneered (along with E. B.
Tylor, and followed by McLennan, Engels, and others) the school of “evolutionism,”
which described kinship systems as evolving in a unilineal fashion whereby
matriliny, which was thought to be associated with group marriage, was seen as a
primitive stage of evolution experienced by all societies on their route toward
civilized monogamy. In so doing, Morgan helped to establish social anthropology as
its own discipline (Knight 2008) and incited generations of subsequent debate about
the primacy of matrilineal kinship and the universality of human kinship elements
and structure.
The arguments of early evolutionists undoubtedly impacted the framing of the
matrilineal puzzle as it was conceived in the mid-twentieth century. According to the
unilineal evolutionists, when kinship elements of a given society mirrored those
expected in matrilineal societies, for example, they were deemed “survivals”—
vestiges of previous matriliny and evidence of unilineal evolution. Analogously, the
relative rarity of matrilineal kinship, which was found in only 17% of a worldwide
sample of societies (Murdock and White 1969), has been used as evidence of its
Hum Nat (2011) 22:64–88 6565
impending doom (see Douglas 1969) and contributed to the notion that matriliny is
inconsistent with modernization (Gough 1961b). Moreover, the presumed universality
of stages in human kinship gave way to the search for elementary kinship structures
(e.g., Lévi-Strauss 1969) and universal kinship principles (e.g., Radcliffe-Brown
1924). Both of these notions—the modern incongruence of matriliny and the search
for kinship universals—shaped the initial framing of the matrilineal puzzle, arguably
hindering our ability to understand systematically the adaptive features of matriliny
until modern evolutionary anthropology provided a theory embracing the variation
inherent in human kinship systems.
Though matriliny had already been central to the study of anthropology for more
than half a century, the first modern systematic attempts to understand its functions
were made by functionalists such as Malinowski (1932) and structural-functionalists
pursuing cross-cultural study, beginning with Murdock (1949) and culminating with
the publication of Schneider and Gough’s(1961) edited volume, Matrilineal
Kinship, before the study of kinship fell out of favor among anthropologists during
the latter part of the twentieth century. These attempts were associated with what has
been dubbed the “matrilineal puzzle”:“the difficulty of combining recognition of
descent through the woman with the rule of exogamous marriage”(Richards
1950:246). Put differently, matrilineal kinship, by vesting authority in men and
tracing descent through women, splits a man’s allegiance between his own natal kin,
with whom he is reared, and those of his wife and children, whom he desires to
control (e.g., Schneider 1961).
Lacking any particular unifying theory about the nature of matriliny, the
structural-functionalists embarked on numerous comparative studies to yield insights
into the associations between matriliny and other social and ecological variables.
Though rare, matriliny (or its corollary, matrilocal residence) was consistently found
in association with horticulture (Aberle 1961; Keesing 1975) or where agricultural
yields were low (Douglas 1969); in the presence of warfare, especially external
warfare (Divale 1974; Ember and Ember 1971; Jones 2011; see also Ember 1974);
and when men were otherwise absent (Keegan and Maclachlan 1989). Matriliny was
rarely found in association with plow agriculture or with significant animal
husbandry or pastoralism (Aberle 1961) and was thought to erode under conditions
of economic prosperity (e.g., Goody 1962; Gough 1961b; Murdock 1949). Finally,
matriliny was associated with high frequencies of divorce (e.g., Gluckman 1950;
Poewe 1978) and low levels of paternity certainty (Aberle 1961; Murdock 1949).
In tying together these associations, mid-twentieth-century anthropologists
problematized matriliny. In its original conception, the matrilineal puzzle empha-
sized the difficulty inherent to male members of matrilineal systems. Men were
expected to cope with matriliny only under conditions of poverty or low productivity
or when their absence owing to warfare or other reasons prevented them from
governing their households. If conditions changed such that men acquired more
resources or otherwise improved their lots, they would desire to regain authority
over their wife and children and would push the kinship system away from matriliny.
The emphasis on men and universal male authority overlooked to some extent
women’s contributions to kinship governance (cf. Murdock’s1949 argument that the
sexual division of labor was responsible for kinship arrangements; see also Nolin
2011), however, and assumptions of universal male authority absent theoretical
66 Hum Nat (2011) 22:64–88
justification made understanding the benefits of matriliny elusive. The rarity of
matriliny and high frequencies of divorce were cited as evidence of the problematic
nature of matriliny against a background in which nuclear families were held to be
the universal building blocks of kinship systems.
The Adaptive Value of Matriliny
Modern evolutionary anthropologists picked up the question of matriliny more or
less where the structural-functionalists left off, adding to what was known in at least
three important ways: (1) by tying together existing particulate associations under a
common theoretical framework; (2) in formalizing predictions based on theory
through quantification; and (3) by asking whether matriliny could be understood as
an adaptation, rather than as a tenuous, problematic solution, to certain socio-
ecological circumstances. In attempting to understand matriliny as an adaptation,
evolutionary anthropologists invoked the theory of natural selection, predicting that
matriliny would evolve under circumstances where it benefitted individual
reproductive success (i.e., genetic representation in future generations). Rather than
assuming a universal desire by males to control their reproductive partners and
biological children, evolutionary anthropologists asked when it would benefit men to
invest in their matrilateral nieces and nephews.
Paternity certainty formed the basis of initial attempts by evolutionary anthro-
pologists to explain the adaptive functions of matriliny. Kin selection theory
(Hamilton 1964; Maynard Smith 1964) predicts that, for a given net benefit, costly
investment in others will be proportionate to their level of genetic relatedness, r.
Under conditions of certain paternity, a man’s maternal nieces and nephews are only
half as closely related to him (r=0.25) as are his own offspring (r= 0.50). Whereas a
mother’s parentage is virtually certain, a father is rarely entirely sure of his paternity.
Thus, if paternity certainty were low enough, it could be in a man’s best interest to
invest in matrilateral nieces and nephews, to whom his relatedness is assured, rather
than to raise offspring to whom he might be unrelated (Alexander 1974,1977;
Anderson 2006; Flinn 1981; Gaulin and Schlegel 1980; Greene 1978; Kurland 1979;
Lancaster and Kaplan 2000; Trivers 1972).
Though several empirical studies indicate that paternity confidence is associated
with the level of paternal investment (e.g., Anderson 2006; Anderson et al. 2006;
Gaulin and Schlegel 1980; Huber and Breedlove 2007; Flinn 1981; Lancaster and
Kaplan 2000; Marlowe 1999),
1
the level of paternity certainty necessary to produce
conditions under which men are likely to be more related to their matrilateral nieces
and nephews is probably unrealistically low (Flinn 1981; Holden et al. 2003),
ranging from a probability of paternity (P) of 0.268 (Greene 1978) to 0.33
(Alexander 1974,1977:320; Kurland 1979) in the short-term, to 0.46 if the
1
Paternity confidence refers to a man’s assessment of the likelihood that he is the genetic father of a given
child. The studies referred to here may include other people’s assessments of likely parentage rather than
the putative father’s per se. Paternity certainty as used in this paper refers to the actual probability of
paternity, which may differ from paternity confidence (e.g., Anderson 2006).
Hum Nat (2011) 22:64–88 6767
compounding geometric effects of paternity on relatedness over several generations
are included (Hartung 1985). These levels are well below certainty rates of 0.9 cited
by most researchers (usually cited in terms of uncertainty at 10%; e.g., Alfred 2002;
Cervino and Hill 2000; Stewart 1989; all cited in Anderson 2006), and the observed
certainty levels both in “high paternity confidence”societies (0.981) and “low
paternity confidence”societies (0.702; all figures from Anderson 2006).
Given that unrealistically low levels of paternity certainty are necessary to favor
men who choose to invest in their nieces and nephews over their own putative
children, paternity certainty cannot fully explain the evolution of matriliny. Recently,
evolutionary anthropologists have modified hypotheses concerned solely with
paternity certainty to incorporate another variable known to be associated with
matriliny and to affect investment in daughters versus sons: wealth (Holden et al.
2003). The matriliny-as-daughter-biased-investment hypothesis (MDBI) recognizes
and incorporates the variable effects of wealth on men’s and women’s reproductive
success. Following the logic of the Trivers-Willard hypothesis, the model recognizes
that it should benefit parents to invest wealth in the sex whose reproduction stands to
gain most from such investments (e.g., Cronk 1989; Hartung 1976,1982; Holden et
al. 2003; Mace 1996; Trivers and Willard 1973). Many forms of heritable wealth are
thought to impact male reproductive success more than female: both livestock and
productive land are usually more beneficial to males than to females, for example,
owing to their greater impacts on males’ability to acquire partners (Holden et al.
2003). Thus, there are reasonable premises for incorporating wealth into a model that
attempts to explain the evolution of matriliny.
It is worth noting here that the MDBI hypothesis, while adequately incorporating
the effects of wealth on the reproductive success of women versus men, is perhaps
misnamed and/or not applicable in its narrowest form (described below) to all
matrilineal societies. Importantly, whereas in a number of matrilineal societies,
parents together confer inheritance on their offspring, in others, inheritance is
transferred from mother’s brother to sister’s son or, as in the Mosuo case,
collectively from one generation of matrilineally related relatives to the next.
Acknowledging variation in transmission pathways, Holden et al. speculate that
daughter-biased investment of resources is still key:
In other matrilineal societies, property is transferred from the mother’sbrotherto
his sister’s son (Schneider and Gough 1961). For grandparents, this is equivalent
to inheritance by their daughters’offspring. This type of inheritance allows sons
to use inherited resources during their lifetime, while ensuring that those
resources are ultimately transferred to the daughters’children (Holden et al.
2003:102).
A simple diagram clarifies the mistake in this reasoning (Fig. 1): In societies
where men’s property is controlled separately from women’s property, there is no
direct transmission of men’s property to their daughters, nor to their daughters’
offspring. Either men inherit from their maternal uncles (panel a) or one generation
of matrilineal relatives inherits collectively from another (panel b). Indeed, the only
case in which matrilineal transmission can be considered daughter-biased or
granddaughter-biased is in societies that practice pooling of parental resources
(panel c).
68 Hum Nat (2011) 22:64–88
How, then, do we reconcile the model’s main predictions and empirical
correspondence to at least one matrilineal society with variance in the rules of
inheritance that seem to undermine the process by which the model operates? The
key lies in the relative impacts of the resources on male and female reproduction. If
resources have equal or lower impacts on male reproduction relative to female, the
nominal transmission mechanisms are moot because the household’s resources
primarily support reproduction by (resident) daughters. Among the Mosuo, for
example, though men nominally are partial stewards of economic resources, and are
said to play a role in decisions related to financial expenditures (e.g., Weng 1993),
they are rarely in complete control of household resources. This point is critical
because it means that men in such situations effectively do not use a household’s
resources to promote their own reproductive interests. If substantial household
resources support men’s reproduction, and men devote these to their sisters’
offspring, then Holden et al.’s model cannot account for transmission dynamics
because the model is based on parental decision-making (i.e., the certainty parents
have over sons’offspring). If men are effectively without property, on the other
AB
C
Fig. 1 Diagrammatic representations of the flow of resources according to the type of matrilineal
transmission. In A (upper left), mother’s brother transmits to sister’s son such that, while women are never
stewards of property, their daughters nonetheless effectively stand to inherit the resources from her
matrilineal household. Men in A do not practice daughter-biased investment because their resources are
transferred to sisters’children, not their own. Panel B (upper right) corresponds to transmission of
resources as traditionally occurs among the Mosuo. The situation is similar to A, except that women are
also stewards over property and men temporarily are able to use property even though their sisters’
children inherit over the long-term. Panel C (lower left) shows the only type of inheritance structure that
corresponds in name to daughter-biased or granddaughter-biased inheritance. In this case, a child’s
biological father must also confer property to his own children, rather than to those of his sisters. NB:
Triangles represent males; circles, females; equal signs, marriages or reproductive unions; and overarching
bars, sibships. Shading indicates who transmits property in each generation
Hum Nat (2011) 22:64–88 6969
hand, then the question of who inherits is still a question of daughters or sons—only,
as theorized by Hartung (1985) and formalized by Holden et al. (2003), one faced by
mothers but not by fathers.
The Model
The MDBI hypothesis examines the effects of the probability of paternity (P) and the
inheritance of wealth on the benefits to parents of investing wealth in daughters
versus sons.
2
It begins with the straightforward premise that parents should invest
wealth equally in daughters and sons when the inclusive fitness benefits of doing so
are equal, or when BS = BD (where BS is the benefit of wealth to sons’reproductive
success and BD to daughters’). Because paternity in sons’offspring is not assured,
the benefit of investing in sons must be devalued by P: PBS = BD. Rearranging, the
model predicts that the benefits of investing in sons and daughters are equal when
BS=BD ¼1=P:
This relationship is depicted in Fig. 2: when BS/BD > 1/P, it is more beneficial to
invest in sons; when this benefit ratio falls below 1/P, it is more beneficial to invest
in daughters.
Though simple, this model potentially explains many of the features associated
with matriliny once used to support the premise that matrilineal inheritance was
somehow “puzzling”or a “cumbersome dinosaur”(Douglas 1969:123) doomed for
extinction. In particular, the model’s fundamental prediction is that matriliny evolves
when the increased impact of resources on men’s reproduction does not outweigh the
risk of non-paternity in sons’offspring. Regardless of whether inheritance is
transmitted primarily from women to their daughters (with men as temporary
stewards) or from parents in matrilocal marriages to their daughters, the prediction is
the same and thus the model is applicable to the forms of matrilineal wealth
transmission considered in this paper and by Holden et al. The inclusion of paternity
certainty explains links between matriliny and high rates of divorce and protracted
absences, including warfare, while the term reflecting differential impacts of wealth
on men’s and women’s reproduction explains links to resource scarcity and
horticulture. Moreover, in contrast to the viewpoints of unilineal evolutionists (and
currently favored by certain former Soviet countries and China; e.g., see Divale
1974; Pusey 2009), this hypothesis predicts transitions from other forms of kinship
to matrilineal if conditions change to make daughter-biased investment beneficial to
parents. Quantitative model specification makes sense of variation in the factors
associated with matrilineal kinship, contributing to our understanding of matriliny
without appealing to unfounded and unexplained universals in human kinship. I turn
2
I retain Holden et al.’s explanation here, but throughout the paper, when I say “parents,”“mothers”
would be more appropriate to the Mosuo context, as explained above. Note that in their test of MDBI,
Holden et al. compare the matrilineal Chewa to the patrilineal Gabbra. According to Holden et al., among
the Chewa, 75% of land is inherited directly from mother to daughter. Moreover, marriage among the
Chewa was historically uxorilocal (Phiri 2009). Thus, the Chewa correspond to panel C of Fig. 1,
satisfying the conditions of Holden et al.’s transmission process.
70 Hum Nat (2011) 22:64–88
now to testing this model to evaluate whether it can explain matrilineal inheritance
among the Mosuo of Southwest China.
Study Population
The Mosuo are a population of approximately 40,000 minority Chinese living on the
border of Sichuan and Yunnan provinces in the Himalayan Mountains (Walsh 2005). The
Mosuo may be subdivided into two distinct subpopulations according to kinship
practices: patrilineal and matrilineal (see Shih 1993). Discussion in this paper will focus
on the matrilineal Mosuo residing near Yongning, their cultural center and township
seat, and Lugu Lake, the center of tourism, both in Yunnan Province (Fig. 3).
The majority of Mosuo until recently were subsistence agriculturalists, raising such
crops as buckwheat, corn, wheat, potatoes, and garden vegetables for their own
subsistence, and engaging in animal husbandry of livestock, including cattle,
3
as a
significant sideline (Cai 2001;Shih1993,2010). Beginning in the 1980s and
increasingly since the mid-1990s, a subset of the Mosuo residing primarily along Lugu
Lake have earned their living from profits driven by a thriving tourism industry
(Mattison 2010a,2010b;Walsh2001,2005). These profits are distributed to some
3
Cattle are rare among matrilineal societies. Though the Mosuo keep cattle, they do so in insignificant
numbers, such that one head of cattle is often shared among multiple households. Moreover, cattle
apparently are not used as bridewealth payments or for consumption, but as draft animals.
Fig. 2 The MDBI model depicting the direction of sex-biased investment. The y-axis depicts the ratio of the
benefit of investing wealth in sons versus the benefit of investing the same wealth in daughters (BS/BD). The
x-axis is the probability of paternity (P) in sons’offspring. The line (BS/BD=1/P) represents the values of
the benefit ratio and the probability of paternity for which equal investment in sons and daughters is
predicted. Above this line, parents are expected to invest in sons because the benefit of wealth to sons
outweighs the risk of non-paternity in their offspring. Below the line, parents are expected to invest in
daughters. (Modified from Holden et al. 2003)
Hum Nat (2011) 22:64–88 7171
extent communally, with each family sending one representative to assist in community-
based tourism ventures such as public dance displays and boating excursions to
destinations inside the lake (e.g., Xing et al. 2009). Family-owned hotels and shops have
nonetheless resulted in substantial income variation among households (Mattison
Fig. 3 Map of study site. The matrilineal Mosuo reside along the border of Sichuan and Yunnan
provinces in southwestern China (from Mattison 2010a)
72 Hum Nat (2011) 22:64–88
2010a), and incomes are such that most individuals in areas where tourism is prevalent
live entirely off of associated profits (Mattison 2006). Most families residing in areas
away from the lake have retained agricultural traditions as the major source of
subsistence, though individuals in many of these families have salaried occupations
ranging from wage laborers engaged in physical activities to television anchors.
Among traditional Mosuo, women effectively transmit property to their daughters
and their daughters’offspring while Mosuo men act as temporary co-stewards of
property, which is transmitted to their sisters’offspring (Fig. 1b). All offspring of
matrilineally related women in a household have usufruct access to household
property, but only offspring of female descendants stand to pass property to
subsequent generations. Labor is supposedly dedicated to a man’s natal household
rather than to his partner’s household, but historical participation in caravans would
have led to prolonged absences and, correspondingly, low male contributions to any
household labor. Whatever rights men have to property, nominal or real, are
transferred to their sisters’children. Lineage affiliation among the Mosuo is also
matrilineal: children of both sexes belong to their mother’s lineage and normally
reside with her throughout their lives. The most important inherited resource shared
by a household until recently was land, but money and other durable goods have
now become more important, especially in areas where tourism is prevalent.
Men’s authoritative roles among the Mosuo were traditionally relegated to their natal
lineages (Cai 2001;Shih1993,2000,2010). Practicing a system of pairing known as
“walking marriage”(sese), most men traditionally visited their lovers at night, retaining
separate residences throughout the duration of their unions. According to ethnographers
of the Mosuo (Cai 2001;Shih1993,2000,2010), walking marriages involve no
contract between lovers, paternity is not assured and is unimportant, and multiple
concurrent unions are possible and do not incite jealously. Men engaging in sese are
thus expected to refrain from active participation in their partners’lineages under most
circumstances, as this may cause tension between a woman’s affinal and consanguineal
relatives. Anecdotal field evidence and reports from ethnographers of the Mosuo (e.g.,
Cai 2001;Shih2010) indirectly support the nominal nature of men’s historical
participation in decision-making: when men were in control of resources (e.g., they were
paid individual incomes), they were likely to expend them as individuals (e.g., as gifts to
lovers) rather than through normative communal mechanisms.
The relationship between various forms of wealth and childbearing has not been
examined systematically among the Mosuo. Given that land historically was the
most important source of subsistence until recently, it is reasonable to expect that it
might be associated positively with fertility among postreproductive individuals
whose reproduction took place prior to major economic changes. Moreover,
although land was a valuable asset to the Mosuo, it has not been particularly scarce
in matrilineal areas (Shih 1993,2010). Land is also not particularly productive,
requiring intensive manual labor to work. The plow is used for agriculture, but most
other activities are carried out by hand. Thus I anticipate that land will have roughly
similar effects on male and female reproduction among postreproductive individuals:
where land is neither scarce nor highly productive, men cannot easily translate it into
higher marginal reproductive success (Holden et al. 2003).
With recent infrastructural and economic changes, land arguably has ceased to be
the most important asset to families reproducing in contemporary contexts.
Hum Nat (2011) 22:64–88 7373
Increasingly, parents value education for their children and seek reproductive
partners with similar values. Indeed, many young people have abandoned an
agricultural lifestyle in pursuit of steady employment, whether as tourism
entrepreneurs or as salaried professionals (Mattison 2010a,2010b). As subsistence
and reproduction become increasingly tied to income, the opportunities for each
gender to translate resources into reproduction may change. Given that income is
fungible and easily defensible, it could result in higher reproductive returns for men
relative to women, particularly if men are able to attract more sese partners in
connection with higher incomes. Recent evidence (Mattison 2010a,2010b) points to
more marriage than indicated in previous reports, particularly among the wealthy.
Given recent tendencies toward monogamy, paternity certainty likely is higher for
men reproducing in contemporary contexts compared with postreproductive men. At
the same time, the benefits of wealth to sons’and daughters’reproduction may be
relatively equal in the context of monogamy, favoring daughter-biased investment if
paternity is not assured. Whereas land historically was transmitted to female heirs,
ethnographic evidence suggests that income is inherited by both male and female
heirs (Mattison2010a, b). In this paper, I examine whether wealth transmission—
land among postreproductive Mosuo and income among currently reproductive
Mosuo—conforms to predictions based on the MDBI hypothesis.
Methods
Data Collection
The data analyzed in this paper were collected from January through October of
2008 (for a description of the complete methods, see Mattison 2010b). Household
demographic surveys were conducted in 12 villages in the geographic area between
Yongning and Luoshui villages, consisting of 177 unique households. Villages were
chosen in order to obtain an accurate representation of current lifestyles, from
normatively conservative to progressive and from subsistence-based to income-
based. All households claiming Mosuo ethnicity were chosen for several villages; in
other cases, households were chosen based on convenience sampling (i.e., an adult
member of the household was home during the first attempted contact).
For each demographic survey, an adult member of the household (and often
several other interested members of the household) acted as respondent and
provided information on the household’s and each household member’s
characteristics. In particular, household monthly income and information on
land, property, and assets were provided. Each respondent also provided
information on household members (including all members who were born in
the household, regardless of their current residence) including their roles in the
household (i.e., their relation to the household head); their individual incomes, if
any; and their approximate age, sex, level of education, occupation, marital
status, and number of living children. Household residents who were present
during the time of the survey were weighed and their heights measured; these
data were not consistently available and are not included in these analyses. All
surveys were administered orally to respondents by a member of the research
74 Hum Nat (2011) 22:64–88
staff in Mandarin Chinese or the local Sichuan dialect, when possible, and
otherwise translated into Naru, the Mosuo language, by a local assistant.
Data Analysis
This paper examines whether MDBI predictions hold for two different forms of
wealth in two age-based cohorts of adults: a cohort of postreproductive individuals
whose fertility decisions were made under putatively “traditional”circumstances and
a cohort of currently reproductive individuals whose fertility decisions have been
impacted by fertility policies and whose subsistence has been altered by a recent
transition to a market economy. Among postreproductive individuals, a threshold of
58 years old (i.e., birth during or prior to 1950) was used to examine MDBI. This
age cutoff aims to limit exploration of MDBI to only those Mosuo whose fertility
decisions were relatively unconstrained by the various historical events that have
been shown to affect fertility and wealth. Among these events were the incorporation
of the Mosuo into the Chinese communist system, which began in 1950, when the
People’s Government of Ninglang County was established in Yongning (Shih 1993,
2010); the Great Leap Forward, which impacted fertility through famine and its
effects on marital practices (Shih and Jenike 2002); and the various birth planning
policies that were implemented in China beginning in the early 1970s. The precise
moment at which such policies actually began to affect fertility decisions in rural
China is highly variable (e.g., Harrell et al. 2011; Lavely and Freedman 1990;
Skinner et al. 2000); thus, examination of actual fertility data among the Mosuo may
be insightful. Figure 4shows the distribution of number of living children reported
Fig. 4 Average number of live children reported by Mosuo individuals in each age cohort. The horizontal
line at an average of 3 children shows a point of transition between relatively low completed fertility (for
those aged 45 and over) of fewer than 3 children and a relatively high completed fertility of more than 3
children. The cutoff age of 58 was chosen as the midpoint of the transitional cohort in order to determine
fertility behavior of individuals in the absence of external influence
Hum Nat (2011) 22:64–88 7575
by Mosuo of given age cohorts. There is a clear delineation between those who
reported an average of more than three children and those who reported fewer.
The transition seems to occur among individuals aged 56–60; 58 was the
midpoint of this age range and thus deemed a suitable threshold for depicting
changes in fertility decisions (see also Table 5 in Zhang 1990:255, which
indicates a similar age threshold).
I also assessed the impacts of monetary income on female versus male
reproduction among currently reproductive individuals aged 17–45. These age
cutoffs were chosen to establish the pattern of parental decision-making in the
context of the contemporary economy, where income has become increasingly
important to reproduction. A 45-year-old would have been 15 years old in 1978
when the family planning policy was implemented but would not have experienced
the effects of a market economy as acutely as a 30-year-old, whose reproduction
would have commenced after income from tourism would have become an important
resource. To accommodate the nuances in timing of both fertility policies and
economic change, I ran the same set of analyses on cohorts of individuals from
17 years of age to anywhere between 30 and 45 years of age, resulting in 16 different
regressions (summarized below).
The statistical methods used to analyze these predictions closely followed the
procedures used by Holden et al. (2003), with some minor modifications. In
particular, the impact of wealth (land or income) on reproduction is assessed as the
coefficient of the slope of the regression of the number of living children on a given
type of wealth, computed separately for males and females via an interaction term
that allows for varying effects of wealth on reproductive output based on gender. A
significant interaction term is interpreted as evidence that the effects of wealth on
reproduction are different for each gender; non-significance indicates that the effects
are similar. Thus the first model predicts that the ratio of the benefits of land to sons’
reproduction versus daughters’is outweighed by the risk of non-paternity in sons’
offspring, such that it benefits parents to invest land in daughters, whereas the
second model predicts that this ratio balances such a risk (or is roughly equivalent to
the reciprocal of the probability of paternity) such that it benefits parents to invest
income in both daughters and sons. The probability of paternity, P, is simply the
inverse of the ratio of the male to female regression coefficients.
Some features of the variables used in this analysis are important to consider
when interpreting the results. First, I am using current estimates of landholdings to
estimate prior impacts on reproduction within the postreproductive cohort. There is
no way to know definitively how directly this measure corresponds to a measure that
might have been made during the time these individuals were reproducing. On the
one hand, communalization of land and property undoubtedly altered the amount of
land held by a given individual over the course of the reproductive period; on the
other, many individuals reported that they had not changed residences since birth
and that elite families had been able to maintain their status and, eventually, wealth
throughout tumultuous periods. Acquisition and maintenance of wealth through this
period will be considered in future fieldwork; for now, the current size of land is the
best estimate available. Second, both income and land are shared at the household
level, though their effects on individuals are considered in the analysis. As described
above, wealth is largely shared among household members, and relatively few
76 Hum Nat (2011) 22:64–88
individuals report earning individual salaries. To assess the impacts on reproduction
of the wealth attributed to individuals, a control for the number of total adults in the
household with whom wealth might be shared is included in both models.
Households reporting zero income were excluded in Model 2, following Holden et
al. and to improve model fit. A control for age is included in the second set of
models, but not the first. It had no substantive effect on the first model. A control for
land is included in the second set of models to allow for the possibility that it might
affect reproduction where individuals continue to pursue an agricultural lifestyle.
Finally, education is included in the model of currently reproductive individuals to
control for delayed childbearing that may be associated with prolonged education.
Data were log transformed as necessary to improve model fit and accommodate
assumptions of normality in linear regressions. All analyses were conducted in R
(version 2.11.1; R Core Development Team 2010).
Summary of Predictions
In summary, the analysis is partitioned into two models. In the first model (M1), I
examine the effects of land on postreproductive individuals and predict:
M1.1 The effect of land on reproduction is significant for both sexes; and
M1.2 The ratio of the benefits to sons of land wealth relative to the benefit to
daughters is outweighed by hypothetical values of non-paternity in sons’
offspring.
The second model (M2) focuses on the effects of earned income on currently
reproductive individuals and predicts:
M2.1 The effect of income on reproduction is significant for both sexes; and
M2.2 The ratio of the benefits to sons of income relative to the benefit to daughters
balances hypothetical values of non-paternity in sons’offspring.
Results
Descriptive Statistics
The demographic surveys resulted in information on 1,156 individuals of known
age: 893 adults (over age 17) and 263 children. Table 1shows descriptive
statistics of interest for surveyed individuals according to status of reproduction.
There is considerable variation in all variables considered in these analyses.
Household-level variables (e.g., wealth) do not differ significantly based on
reproductive category (nor would we expect them to), but there are notable
differences in the proportion of included individuals who were male (fewer in the
older generation), average completed level of education (grade 1 among
postreproductives and grade 6.3 among current reproductives), and in the average
number of surviving children reported (fewer among individuals currently
reproducing).
Hum Nat (2011) 22:64–88 7777
Model 1: Land Effects on Postreproductives
Table 2and Fig. 5show regression results for the effect of land on the number of
surviving offspring among postreproductive adults. As predicted (M1.1), land has
a positive effect on reproduction for both sexes after controlling for the number
of adults in the household. The effect of land on reproduction does not differ
between males and females, as the interaction term is not significant. The ratio of
the benefits to sons’reproduction relative to the benefit to daughters’is
calculated from the regression coefficients. The slope for females (the reference
category) is simply the slope for land: 0.9674. The slope for males is calculated
by adding the slope for the interaction term (where male=1) to the slope for
females (where male= 0) and is 0.4489. The ratio of the benefits of land to sons’
reproduction is thus 0.4489/0.9674 or 0.4640. Statistically equivalent slopes
indicate that the effects of land on reproduction do not differ for males and
females. If Pis less than 1, then it benefits parentstoinvestlandindaughters.
Thus, these results are consistent with the MDBI model, and specifically with
prediction M1.2: the relative benefit of land to sons versus daughters is outweighed
by the risk of non-paternity for any level of paternity lower than absolute certainty,
and it benefits parents to invest land in daughters.
Table 1 Descriptive statistics for populations and variables of interest, by reproductive status. Means are
reported for continuous variables, with standard deviations in parentheses
Postreproductive Reproductive
Number of children
a
3.81 (2.28) 0.94 (0.93)
Log of monthly income (RMB) 6.06 (2.56) 6.17 (2.38)
Log of land (mu) 3.01 (0.69) 2.81 (0.88)
Highest educational grade (years) 1.12 (2.88) 6.30 (6.99)
% male 35.8 46.3
Number of adults in household 5.13 (1.89) 4.85 (1.97)
N122 656
a
Live children only
Table 2 Regression output for the effect of land on reproduction (number of surviving offspring) for
postreproductive individuals over the age of 58 in 2008
Estimate SE p
Intercept 0.4558 1.2916 0.7248
Log of land 0.9674 0.3902 0.0147*
Number of adults in household 0.1567 0.1069 0.1455
Male 0.5490 1.7917 0.7599
Interaction (Male * Log of land) −0.5185 0.5918 0.3828
N=119
Adjusted R
2
=0.1238
78 Hum Nat (2011) 22:64–88
Importantly, because there was a relative scarcity of men at older ages in the
dataset (Table 1), it is possible that, owing to either mortality or migration, men who
had reproduced with women in the sample were not, themselves, included in results
analyzed here, and men’s fertility was consequently biased downward. Results
shown elsewhere (Mattison 2010b) have not revealed higher levels of out-migration
for men compared with women. It is possible that men suffered higher mortality at
older ages, however. To control for this possibility, I performed an additional
regression in which individuals over the age of 70, where the female advantage
appeared most significant among postreproductives, were excluded. The results (not
shown) were not substantively altered; thus, the similarity in male and female
reproduction in Model 2 is probably not due to sampling bias.
Model 2: Income Effects on Current Reproductives
Table 3summarizes the results from 16 linear regressions of the effects of earned
income on reproduction among young adults, aged 17–x, where the xvaried from 30
to 45. Controlling for covariates, the effect of income on reproduction was positive,
regardless of maximum age, and significant or near significant in most regressions,
conforming to prediction M2.1. Interestingly, land seems to have a negative effect on
reproduction for individuals reproducing in contemporary contexts, even though the
ethnographic evidence suggests that some families have maintained an agrarian
lifestyle. Education was negatively associated with reproduction and age was
positively associated with reproduction in all regressions. As in M1, the intercepts
and slopes for the effect of income on male reproduction were never significant,
indicating similar effects of earned income on male and female reproduction.
To exemplify these regressions, Table 4and Fig. 6show the effects of various
covariates on the number of living children reported by currently reproductive
individuals, aged 17–40. Again, differences between the sexes in terms of the effect
Fig. 5 The effect of land (log mu)
on the number of surviving off-
spring among postreproductive
individuals (more than 58 years old
in 2008). The thicker gray dashed
line represents the predicted effect
based on regression output for
females (i.e., controlling for mod-
eled covariates) and the black solid
line, for males; thin dashed lines
represent 95% confidence intervals
for the female regression and the
thin solid lines, for the male
regression
Hum Nat (2011) 22:64–88 7979
of earned income on reproduction are not significant. The scale of the effects is
lower than in M1, however, both owing to lower reproduction overall in the younger
cohort and as a result of the scales on which land and income were tabulated. The
slope for females of the effect of income on reproduction was 0.0537, and for males,
0.0179 (0.0537 + −0.0358). The ratio of BS/BD is thus 0.333, and the probability of
paternity necessary to invest equally in sons and daughters is 3. Again, because the
difference in slopes is not significant, the probability of paternity necessary to invest
equally in sons and daughters is 1. The above illustrate how Pis calculated for each
regression.
Table 3 Summary statistics of 16 regressions designed to assess the impact of covariates on the number
of surviving children among individuals currently reproducing, aged 17–x. Age range for cutoff: x=
{30,45}
Coefficient Mean Range Number of Regressions in which
Variable Was Significant
a
(%)
Intercept −0.885 −1.101–−0.596 16 (100) | 16 (100)
Log of monthly income 0.048 0.037–0.064 3 (18.8) | 12 (75)
Log of land −0.100 −0.126–−0.064 12 (75) | 14 (87.5)
Number adults in household −0.001 −0.013–−0.010 0 (0) | 0 (0)
Highest educational grade −0.026 −0.041–−0.013 16 (100) | 16 (100)
Age 0.065 0.056–0.072 16 (100) | 16 (100)
Male
b
−0.040 −0.157–0.052 0 (0) | 0 (0)
Interaction (Male*Income) −0.040 −0.055–−0.023 0 (0) | 0 (0)
Nranged from 250 to 600 (increasing with increasing age threshold) across all regressions with a mean of
434.
a
The number of times a variable was significantly associated with the residual number of children over all
regressions; significance is defined as p≤0.05 and p≤0.1.
b
The reference category is female.
Table 4 Regression output for the effect of income and covariates on reproduction (number of surviving
offspring) for currently reproductive individuals aged 17–40 in 2008
Estimate SE p
Intercept −0.9630 0.2629 0.003***
Log of monthly income 0.0537 0.0270 0.0470*
Log of land −0.1019 0.0396 0.0104*
Number of adults in household −0.0078 0.0164 0.6350
Highest educational grade −0.0168 0.0050 0.009***
Age 0.0656 0.0045 <0.001***
Male −0.0760 0.2678 0.7764
Interaction (Male*Log of income) −0.0358 0.0384 0.3517
N= 499
Adjusted R
2
=0.3909
80 Hum Nat (2011) 22:64–88
Discussion
The MDBI hypothesis explains matriliny as an outcome of daughter-biased wealth
investment by parents seeking to optimize allocation of resources in terms of inclusive
fitness (Holden et al. 2003). In their paper, Holden and colleagues show evidence that
supports their hypothesis in two different societies, one matrilineal and one patrilineal.
The patrilineal Gabbra derive nearly three times as much benefit from wealth invested
in sons relative to daughters, whereas the benefits of investing wealth do not differ by
sex among the matrilineal Chewa. The levels of paternity certainty necessary to make
investing in sons beneficial were 0.36 and 0.94, respectively, indicating that paternity
certainty can be quite low among the Gabbra before it makes sense to invest in
daughters, because the impacts of wealth on male reproduction were much higher than
on female reproduction, but not among the Chewa, where it would rarely be beneficial
to invest wealth in sons. It is important to note here that according to this model,
matriliny derives causally from the type of wealth available to support reproduction in
different societies. If the wealth benefits men to such a degree that it outweighs
existing levels of non-paternity in their offspring, then mothers and their partners will
invest wealth in sons. Patriliny results not only as mothers invest resources in sons but
also because sons become freer to pursue their own reproductive interests and as a
result begin to invest their resources in their own children. If my explanation of how
this model applies to matrilineal societies in which wealth is transferred from mother’s
brother to sister’s son is correct, then men in such societies effectively invest very little
in any descendants. When they possess the means to impact reproduction (e.g.,
because prolonged absences are no longer necessary or the source of wealth changes),
men choose to invest in their own children and/or romantic partners and the kinship
system shifts away from matriliny.
Applying this framework to the Mosuo revealed that the MDBI hypothesis is
generally consistent with inheritance transmission in this population, while
Fig. 6 The effect of monthly
income (log RMB) on the number
of surviving offspring among
young adults aged 17–38 in 2008.
The thicker gray dashed line ?
represents the predicted effect
based on regression output for
females (i.e., controlling for
modeled covariates) and the black
solid line, for males; thin dashed
lines represent 95% confidence
intervals for the female regres-
sion and the thin solid lines, for
the male regression
Hum Nat (2011) 22:64–88 8181
highlighting some interesting nuances of this data set. Among postreproductive
individuals whose childbearing occurred prior to the implementation of the birth
planning policy, the impacts of land on reproduction were similar for men and
women, such that parents would need complete certainty over paternity in sons’
offspring in order to invest land equally in sons and daughters. Given that paternity
certainty was likely to have been relatively low among postreproductive individuals
engaging in sese, the model predicts daughter-biased investment of land, which was
indeed the norm among traditional Mosuo.
The effects of earned income on reproduction among young adults currently
reproducing must be interpreted within the socioecological context in which
reproductive decision-making occurs. As with the first model, the second set of
models showed that the relationships between earned income and reproduction were
similar for men and women. Here, too, parents would need complete certainty in
sons’offspring in order to invest income equally in sons and daughters. Given the
increasing prevalence of marriage and monogamy among young Mosuo (Mattison
2010a), it is likely that paternity certainty is higher among recent age cohorts relative
to previous generations. Opportunities for employment close to home may also lead
to higher paternity certainty as men can spend more time in close proximity to their
affines. At the same time, limits on childbearing for rural and ethnic Chinese of two
or three children leaves relatively little variation in reproductive outcomes,
decreasing the relative advantage of investing wealth in sons that might normally
be expected with respect to income. When parents can be assured of paternity, it may
indeed benefit them to invest income in both sons and daughters. If actual paternity
is lower than 100%, however, parents are expected to bias investment of income
toward daughters. Without knowing prevailing levels of paternity confidence, it is
impossible to say whether income is distributed in ways consistent with the MDBI
model. Future studies of MDBI could incorporate internal variation in inheritance
practices to examine whether, for example, married individuals are more likely to
invest equally in sons and daughters relative to individuals engaged in non-marital
unions.
The MDBI model can be used to analyze variable practices among societies
(e.g., Holden et al. 2003), but it is also flexible enough to analyze transmission of
different forms of wealth within the same society, as has been done here (see e.g.,
Keegan and Maclachlan 1989; Pelto and Pelto 1975;Poewe1978 for discussion of
the importance of intrasocietal variation). This flexibility is important because
sex-biased parental investment rarely involves binary decisions of whether to
invest in just daughters or just sons, but rather responds to the variable effects of
each type and increment of investment on a child’s reproductive success (Sieff
1990). Indeed, the model’s ability to capture variation in sex-biased investment
could be expanded to incorporate the differential returns of other types of
investment to parental reproductive success, such as those accruing directly to
parents through their children’s help (e.g., Borgerhoff Mulder 1998;Smithand
Smith 1994;Turke1988).
Another interesting modification of the model could incorporate the synergistic
effects of paternity certainty and wealth. Evolutionary theory predicts that men will
seek fidelity from reproductive partners as a requirement for substantial paternal
investment (e.g., Fortunato and Archetti 2009; Marlowe 2000). If this is the case,
82 Hum Nat (2011) 22:64–88
increased levels of wealth might be positively correlated with increased paternity
certainty in offspring when paternal investment is significant and men exert control
over resources important to reproduction. The empirical significance of paternity
certainty or confidence in affecting paternal investment in offspring (e.g., Anderson
et al. 2006; Flinn 1981; Gaulin and Schlegel 1980; Greene 1978; Kurland 1979) and
wealth in impacting sex-biased transmission of resources (e.g., Cronk 1989; Mace
1996; Trivers and Willard 1973) is well-established. Because these are currently
modeled as separate, independent effects on sex-biased transmission of wealth, the
MDBI model probably underestimates the synergistic effects of these two variables
when acting together.
Finally, when considering the rules and causes of sex-biased intergenerational
wealth transmission in places like China, it is important to acknowledge the very real
possibility that women’s and men’s reproduction may not be impacted equally by
resources; rather, women’s reproduction may be impacted more than men’s. Given
large female deficits in China’s population and corresponding impacts on the
marriage market (e.g., Coale and Banister 1994), it is possible that in many locales,
women achieve higher average reproductive success than do men. This is because
excess males from populations not counted in surveys such as mine may remove a
portion of reproductive opportunities from local men. Under such circumstances,
parents of local girls may always derive higher benefits from investing certain types
of wealth in daughters. In these cases, Ptakes on values greater than 1 and is no
longer interpretable strictly as a probability. Modifications to the model that allow
for higher impacts of resources on female reproduction may be called for, and
researchers employing the model must take into account the dynamics of wealth and
reproduction specific to each population under study.
I began this article by reviewing the historical anthropological arguments
associated with matriliny and by arguing that the evolutionary, quantitative
perspective added to our understanding of the forces producing matriliny by
providing a singular rationale for a variable outcome in kinship behavior. I hope to
have shown the utility of an evolutionary framework in tying together what was
already known about matrilineal kinship, including the factors that lower P, such as
warfare and prolonged male absences, and the connections between matriliny,
unstable reproductive unions, and its nuanced relationship with resource availability
and subsistence base. Because unilineal evolutionism is still the dominant
perspective in the geographic area where this work was conducted, it is worth
mentioning that the evolutionary framework employed here is at odds with the
theoretical rationale of the Chinese state (e.g., Pusey 2009). Instead of viewing
matriliny as a fossil-like vestige of some primitive stage in human evolution (Yan
1984), I analyze it as an adaptation to contemporary environments. This perspective
is consistent with recent evidence that associates matrifocality with a lack of male
contributions to their households (e.g., Quinlan 2006; Stack 1974) and transitions
away from matriliny with economic development (e.g., Holden and Mace 2003;
Mattison 2010a; Sear 2008), illustrating once again the relative flexibility in
responses envisioned by some modern evolutionary perspectives.
Our understanding of matriliny is important not only to kinship studies but
also to our understanding of human evolution in general (Knight 2008). Because
the human life history is so different than those of our ape ancestors, (e.g., Hill
Hum Nat (2011) 22:64–88 8383
1993; Mace 2000;Voland1998), one of the central questions of evolutionary
anthropology involves the extent to which different family members contributed to
women’s fertility, allowing for our unique life history (e.g., Hawkes et al. 1998;
Hrdy 1999;Kaplanetal.2000;Kramer2005;Leonettietal.2005; Sear and Mace
2008). The MDBI hypothesis provides some insights into the types of environ-
ments that might have been conducive to different classes of kin in assisting with a
woman’s reproduction. In order for paternal care to become critical, either
paternity certainty must have been relatively assured or the benefits of wealth to
sons must have exceeded its benefits to daughters.
Conclusion
The MDBI hypothesis adds to our understanding of matriliny by tying together
variation in inheritance practices among and within societies under a simple yet
elegant and precise mathematical model. It explains various features previously
known to be associated with matriliny, such as its incompatibility with economic
development (Douglas 1969; Goody 1962; Gough 1961b;Holdenetal.2003;
Murdock 1949), its association with warfare (Divale 1974; Ember and Ember 1971;
Holden et al. 2003), and its association with low paternity certainty (Alexander
1974,1977;Flinn1981; Greene 1978;Holdenetal.2003;Kurland1979)and
divorce (Gluckman 1950;Poewe1978), providing an ultimate reason for variation
in inheritance via evolutionary theory. My data support the utility of this model in
explaining the sex-biased transmission of two types of wealth, land and income,
among two reproductive cohorts of the matrilineal Mosuo of Southwest China,
while highlighting the importance of local socioecological contexts in shaping
transmission dynamics. It is the first independent empirical test of Holden et al.’s
model, as well as an application of evolutionary theory to a geographic area in
which unilineal evolutionism still drives theoretical understanding of kinship
practices.
Acknowledgments Thanks to co-editor Mary Shenk for collaborating on the AAA session that led to
this special issue and to general editor Jane Lancaster for all her assistance and encouragement. This
research was supported by a National Science Foundation Doctoral Dissertation Improvement Grant (BCS
0717918) and a China Studies Program (Fritz Endowment) grant from the University of Washington. Pilot
studies were supported by the American Philosophical Society and the Department of Anthropology at the
University of Washington. Software and administrative assistance was provided by the Center for Studies
in Demography and Ecology at the University of Washington. Eric A. Smith, Donna Leonetti, Stevan
Harrell, Brian Wood, David Nolin, and four anonymous reviewers provided useful comments and
criticisms on drafts of this paper.
References
Aberle, D. (1961). Matrilineal descent in cross-cultural perspective. In D. M. Schneider & K. Gough
(Eds.), Matrilineal kinship (pp. 655–727). Berkeley: University of California Press.
Alexander, R. D. (1974). The evolution of social behavior. Annual Review of Ecology and Systematics, 5,
325–383.
84 Hum Nat (2011) 22:64–88
Alexander, R. D. (1977). Natural selection and the analysis of human sociality. In C. E. Goulden (Ed.),
Changing scenes in the natural sciences (Vol. 12, pp. 283–337). Philadelphia: Academy of the
Natural Sciences, Special Publication.
Alfred, J. (2002). Flagging non-paternity. Nature Reviews Genetics, 3, 161.
Anderson, K. G. (2006). How well does paternity confidence match actual paternity? Current
Anthropology, 47(3), 513–520.
Anderson, K. G., Kaplan, H., & Lancaster, J. B. (2006). Demographic correlates of paternity confidence
and pregnancy outcomes among Albuquerque men. American Journal of Physical Anthropology, 131,
560–571.
Bachofen, J. J. (1967). Myth, religion, and mother right: selected writings of J. J. Bachofen. Princeton:
Princeton University Press. Originally published in German in 1861.
Bloch, M., & Sperber, D. (2002). Kinship and evolved psychological dispositions. Current Anthropology,
43(5), 723–748.
Borgerhoff Mulder, M. (1998). Brothers and sisters: how sibling interactions affect optimal parental
allocations. Human Nature, 9,119–162.
Cai, H. (2001). A society without fathers or husbands: the Na of China. New York: Zone Books.
Cervino, A. C. L., & Hill, A. V. S. (2000). Comparison of tests for association and linkage in incomplete
families. American Journal of Human Genetics, 67, 120–132.
Coale, A. J., & Banister, J. (1994). Five decades of missing females in China. Demography, 31(3),
459–479.
Cronk, L. (1989). Low socioeconomic status and female-biased parental investment: the Mukogodo
example. American Anthropologist, 91(2), 414–429.
Divale, W. T. (1974). Migration, external warfare and matrilocal residence. Behavior Science Research, 9
(2), 75–133.
Douglas, M. (1969). Is matriliny doomed in Africa? In M. Douglas, P. M. Kaberry, & C. D. Forde (Eds.),
Man in Africa (pp. 121–135). London: Tavistock Publications.
Driver, H. E., & Schuessler, K. F. (1967). Correlational analysis of Murdock’s 1957 ethnographic sample.
American Anthropologist, 69(3/4), 332–352.
Ember, M. (1974). The conditions that may favor avunculocal residence. Behavior Science Research, 9(3),
203–209.
Ember, M., & Ember, C. E. (1971). The conditions favoring matrilocal versus patrilocal residence.
American Anthropologist, 73(3), 571–594.
Flinn, M. (1981). Uterine versus agnatic kinship variability and associated cross-cousin marriage
preferences: an evolutionary biological analysis. In R. D. Alexander & D. W. Tinkle (Eds.), Natural
selection and social behavior: Recent research and new theory (pp. 439–475). New York & Concord:
Chiron Press.
Fortunato, L., & Archetti, M. (2009). Evolution of monogamous marriage by maximization of inclusive
fitness. Journal of Evolutionary Biology, 23, 149–156.
Fox, R. (1983). Kinship and marriage: an anthropological perspective. Cambridge: Cambridge University
Press.
Gaulin, S. J. C., & Schlegel, A. (1980). Paternal confidence and paternal investment: a cross-cultural test
of a sociobiological hypothesis. Ethology and Sociobiology, 1, 301–309.
Gluckman, M. (1950). Kinship and marriage among the Lozi of Northern Rhodesia and the Zulu of Natal.
In A. R. Radcliffe-Brown & D. Forde (Eds.), African systems of kinship and marriage (pp. 166–206).
London: Oxford University Press.
Goody, J. (1962). Death, property and the ancestors: A study of the mortuary customs of the LoDagaa of
West Africa. Palo Alto: Stanford University Press.
Gough, K. (1961a). Variation in residence. In D. M. Schneider & K. Gough (Eds.), Matrilineal kinship
(pp. 545–576). Berkeley: University of California Press.
Gough, K. (1961b). The modern disintegration of matrilineal descent groups. In D. M. Schneider & K.
Gough (Eds.), Matrilineal kinship (pp. 631–654). Berkeley: University of California Press.
Greene, P. (1978). Promiscuity, paternity, and culture. American Ethnologist, 5(1), 151–159.
Hamilton, W. D. (1964). The genetical evolution of social behaviour, parts I and II. Journal of Theoretical
Biology, 7,1–52.
Harrell, S., Wang, Y., Han, H., Zhou, Y., & Duro dos Santos, G. (2011). Fertility decline in rural China: a
comparative analysis. Journal of Family History, 36,15–36.
Hartung, J. (1976). On natural selection and the inheritance of wealth. Current Anthropology, 17(4),
607–622.
Hartung, J. (1981). Paternity and inheritance of wealth. Nature, 291, 652–654.
Hum Nat (2011) 22:64–88 8585
Hartung, J. (1982). Polygyny and inheritance of wealth. Current Anthropology, 23(1), 1–12.
Hartung, J. (1985). Matrilneal inheritance: new theory and analysis. Behavioural and Brain Sciences, 8,
661–688.
Hawkes, K., et al. (1998). Grandmothering, menopause, and the evolution of life histories. Proceedings of
the National Academy of Sciences, 95, 1336–1339.
Hill, K. (1993). Life history theory and evolutionary anthropology. Evolutionary Anthropology, 2,78–88.
Holden, C. J., & Mace, R. (2003). Spread of cattle led to the loss of matrilineal descent in Africa: a
coevolutionary analysis. Philosophical Transactions of the Royal Society of London, B: Biological
Sciences, 270(1532), 2425–2433.
Holden, C. J., Sear, R., & Mace, R. (2003). Matriliny as daughter-biased investment. Evolution and
Human Behavior, 24(2), 99–112.
Hrdy, S. B. (1999). Mother nature: A history of mothers, infants, and natural selection. New York:
Pantheon Books.
Huber, B. R., & Breedlove, W. L. (2007). Evolutionary theory, kinship, and childbirth in cross-cultural
perspective. Cross-Cultural Research, 41(2), 196–219.
Jones, D. (2011). The matrilineal tribe: An organization of demic expansion. Human Nature, 22 (1&2).
doi:10.1007/s12110-011-9108-6.
Kaplan, H., Hill, K., Lancaster, J., & Hurtado, A. M. (2000). A theory of human life history evolution:
diet, intelligence, and longevity. Evolutionary Anthropology, 9, 156–185.
Keegan, W. F., & Maclachlan, M. D. (1989). The evolution of avunculocal chiefdoms: a reconstruction of
Taino kinship and politics. American Anthropologist, 91(3), 613–630.
Keesing, R. M. (1975). Kin groups and social structure. New York: Holt, Rinehart & Winston.
Knight, C. (2008). Early human kinship was matrilineal. In N. Allen, H. Callan, R. I. M. Dunbar, & W.
James (Eds.), Early human kinship (pp. 61–82). Oxford: Blackwell.
Kramer, K. L. (2005). Children's help and the pace of reproduction: cooperative breeding in humans.
Evolutionary Anthropology, 14, 224–237.
Kurland, J. A. (1979). Paternity, mother’s brother, and human sociality. In N. A. Chagnon & W. Irons (Eds.),
Evolutionary biology and human social behavior (pp. 145–180). North Scituate: Duxbury Press.
Lancaster, J. B., & Kaplan, H. S. (2000). Parenting other men’s children: costs, benefits and consequences.
In L. Cronk, N. Chagnon, & W. Irons (Eds.), Adaptation and human behavior: An anthropological
perspective (pp. 179–201). New York: Aldine de Gruyter.
Lavely, W., & Freedman, R. (1990). The origins of the Chinese fertility decline. Demography, 27(3), 357–
367.
Leonetti, D. L., Nath, D., Hemam, N., & Neill, D. B. (2005). Kinship organization and grandmother’s
impact on reproductive success among the matrilineal Khasi and patrilineal Bengali of N.E. India. In
E. Voland (Ed.), Grandparenthood: The second half of life (pp. 194–214). Piscataway: Rutgers
University Press.
Lévi-Strauss, C. (1969). The elementary structures of kinship, revised ed.. Boston: Beacon.
Mace, R. (1996). Biased parental investment and reproductive success in Gabbra pastoralists. Behavioral
Ecology and Sociobiology, 38,75–81.
Mace, R. (2000). Evolutionary ecology of human life history. Animal Behaviour, 59,1–10
Malinowski, B. (1930). Kinship. Man, 30,19–29.
Malinowski, B. (1932). The sexual life of savages in north-western Melanesia: An ethnographic account
of courtship, marriage, and family life among the natives of the Trobriand Islands, British New
Guinea. London: Routledge. Originally published in 1929.
Marlowe, F. W. (1999). Male care and mating effort among Hadza foragers. Behavioral Ecology and
Sociobiology, 46,57–64.
Marlowe, F. W. (2000). Paternal investment and the human mating system. Behavioral Processes, 51,45–61.
Mattison, S. M. (2006). Pilot project: the behavioral ecology of sex-biased parental investment among the
ethnic Na of Southwest China. Unpublished data in the author’s possession.
Mattison, S. M. (2010a). The economic impacts of tourism and erosion of the visiting system among the
Mosuo of Lugu Lake. Asia Pacific Journal of Anthropology, 11, 157–174.
Mattison, S. M. (2010b). Demystifying the Mosuo: The behavioral ecology of kinship and reproduction of
China’s“last matriarchal society.”PhD dissertation, University of Washington, Seattle.
Maynard Smith, J. (1964). Group selection and kin selection. Nature, 201, 1145–1147.
Morgan, L. H. (1964). Ancient society. Cambridge: Belknap Press of Harvard University Press. Originally
published in 1877.
Murdock, G. P. (1949). Social structure. New York: Macmillian.
Murdock, G. P., & White, D. R. (1969). Standard cross-cultural sample. Ethnology, 8(4), 329–369.
86 Hum Nat (2011) 22:64–88
Nolin, D. (2011). Kin preference and partner choice: Patrilineal descent and biological kinship in
Lamaleran cooperative relationships. Human Nature, 22 (1&2). doi:10.1007/s12110-011-9113-9.
Pelto, P. J., & Pelto, G. H. (1975). Intra-cultural diversity: some theoretical issues. American Ethnologist,
2(1), 1–18.
Phiri, K. (2009). Some changes in the matrilineal family system of the Chewa of Malawi since the
nineteenth century. Journal of African History, 24, 257–274.
Poewe, K. O. (1978). Matriliny in the throes of change: Kinship, descent and marriage in Luapula,
Zambia, part one. Journal of the International African Institute, 48(3), 205–218.
Pusey, J. (2009). Global Darwin: revolutionary road. Nature, 462, 162–163.
Quinlan, R. J. (2006). Gender and risk in a matrifocal Caribbean community: a view from Behavioral
ecology. American Anthropologist, 108(3), 464–479.
R Core Development Team (RCDT). (2010). R: A language and environment for statistical computing
(Version 2.9.0). Vienna, Austria: R Foundation for Statistical Computing.
Radcliffe-Brown, A. (1924). The mother’s brother in South Africa. South African Journal of Science, 21,
542–555.
Richards, A. I. (1950). Some types of family structure amongst the Central Bantu. In A. R. Radcliffe-
Brown & C. D. Forde (Eds.), African systems of kinship and marriage (pp. 83–120). London: Oxford
University Press.
Schneider, D. M. (1961). The distinctive features of matrilineal descent groups. In D. M. Schneider & K.
Gough (Eds.), Matrilineal kinship (pp. 1–29). Berkeley: University of California Press.
Schneider, D. M., & Gough, K. (Eds.). (1961). Matrilineal kinship. Berkeley: University of California Press.
Sear, R. (2008). Kin and child survival in rural Malawi: Are matrilineal kin always beneficial in a
matrilineal society? Human Nature, 19, 277–293.
Sear, R., & Mace, R. (2008). Who keeps children alive? A review of the effects of kin on child survival.
Evolution and Human Behavior, 29,1–18.
Shih, C.-K. (1993). The Yongning Moso: sexual union, household organization, gender and ethnicity in a
matrilineal duolocal society in Southwest China. Unpublished PhD dissertation, Stanford University,
Palo Alto, CA.
Shih, C.-K. (2000). Tisese and its anthropological significance: issues around the visiting sexual system
among the Moso. L’homme, 154–155, 697–712.
Shih, C.-K. (2010). Quest for harmony: The Moso traditions of sexual union & family life. Palo Alto:
Stanford University Press.
Shih, C.-K., & Jenike, M. R. (2002). A cultural-historical perspective on the depressed fertility among the
matrilineal Moso in Southwest China. Human Ecology, 30(1), 21–47.
Sieff, D. F. (1990). Explaining biased sex ratios in human populations: a critique of recent studies. Current
Anthropology, 31(1), 25–48.
Skinner, W. J., Henderson, M., & Yuan, J. (2000). China’s fertility transition through regional space: using
GIS and census data for a spatial analysis of historical demography. Social Science History, 24(3),
613–652.
Smith, E. A., & Smith, S. A. (1994). Inuit sex ratio variation: population control, ethnographic artifact, or
parental manipulation? Current Anthropology, 35, 595–624.
Stack, C. (1974). All our kin: strategies for survival in a black community. New York: Harper & Row.
Stewart, A. D. (1989). Screening for cystic fibrosis. Nature, 341, 696.
Trivers, R. (1972). Parental investment and sexual selection. Nature, 112, 164–190.
Trivers, R. L., & Willard, D. E. (1973). Natural selection of parental ability to vary the sex ratio of
offspring. Science, 179,90–92.
Turke, P. W. (1988). Helpers at the nest: Childcare networks on Ifaluk. In L. L. Betzig, M. Borgerhoff
Mulder, & P. Turke (Eds.), Human reproductive behavior: A Darwinian perspective (pp. 173–188).
New York: Cambridge University Press.
Voland, E. (1998). Evolutionary ecology of human reproduction. Annual Review of Anthropology, 27,
347–374.
Walsh, E. R. (2001). The Mosuo—beyond the myths of matriarchy: gender transformation and economic
development. Unpublished PhD dissertation, Temple University, Philadelphia.
Walsh, E. R. (2005). From Nü to Nü’er Guo: negotiating desire in the land of the Mosuo. Modern China,
31(4), 448–486.
Weng, N. 1993. The mother house: The symbolism and practice of gender among the Naze in Southwest
China. PhD Dissertation, University of Rochester, New York.
Hum Nat (2011) 22:64–88 8787
Xing, Y., Xia, Z., & Dai, J. (2009). Conservation and tourism development of house settlements in
Mosuo matriarchate in Lugu Lake. Frontiers of Architecture and Civil Engineering in China, 3(2),
204–210.
Yan, R. (1984). A living fossil of the family: a study of the family structure of the Naxi nationality in the
Lugu Lake region. Social Sciences in China, 3(4), 60–83.
Zhang, K. (1990). Family marriage and fertilityin a matriarchal society: Social survey of the Naxi nationality in
Ninglang County, Yunnan Province. Chinese Journal of Population Science, 2(3), 247–256.
Siobhán M. Mattison obtained her Ph.D. from the Biocultural program of the Department of
Anthropology at the University of Washington. Her dissertation explored the behavioral ecology of
kinship and reproduction among the ethnic Mosuo of Southwest China. She is currently a Mellon
Foundation John E. Sawyer postdoctoral scholar at Stanford University and the Morrison Institute
studying the bases of sex-biased parental investment in China and India. Her research interests also
include demography, statistical modeling, and the emergence of health and income disparities as they
relate to transitions in economic markets.
88 Hum Nat (2011) 22:64–88
