Consanguinity, human evolution,

and complex diseases

A. H. Bittlesa,b,1 and M. L. Blacka
a
Centre for Comparative Genomics, Murdoch University, South Street, Perth WA 6150, Australia; and bCentre for Human Genetics, Edith Cowan University,
Joondalup Drive, Perth WA 6027, Australia

Edited by Diddahally R. Govindaraju, Boston University School of Medicine, Boston, MA, and accepted by the Editorial Board August 27, 2009
(recevied for review june 25, 2009)

There is little information on inbreeding during the critical early have been a mistake from a biological perspective. However,
years of human existence. However, given the small founding studies conducted by his son George (8) into the prevalence and
group sizes and restricted mate choices it seems inevitable that basic health outcomes of contemporary first-cousin marriage in
intrafamilial reproduction occurred and the resultant levels of Great Britain helped to convince Darwin to the contrary, on the
inbreeding would have been substantial. Currently, couples re- grounds that “the widely different habits of life of men and
lated as second cousins or closer (F ≥ 0.0156) and their progeny women in civilized nations, especially among the upper classes,
account for an estimated 10.4% of the global population. The would tend to counterbalance any evil from marriages between
highest rates of consanguineous marriage occur in north and healthy and somewhat closely related persons” (9). But by that
sub-Saharan Africa, the Middle East, and west, central, and south stage the topic of cousin marriage had become a matter of often
Asia. In these regions even couples who regard themselves as acrimonious public debate on both sides of the Atlantic, and by
unrelated may exhibit high levels of homozygosity, because mar- the end of the 19th century legislation banning first-cousin
riage within clan, tribe, caste, or biraderi boundaries has been a unions had been enacted by 12 state legislatures in the United
long-established tradition. Mortality in first-cousin progeny is States (5).
≈3.5% higher than in nonconsanguineous offspring, although As indicated in the title of this review, a central aim is to
demographic, social, and economic factors can significantly influ- consider the influence of consanguinity on complex genetic
ence the outcome. Improving socioeconomic conditions and better disorders. As a starting point, the historical background to
access to health care will impact the effects of consanguinity, with Western and other world attitudes toward consanguinity will be
a shift from infant and childhood mortality to extended morbidity. briefly examined, followed by discussion of the relationship
At the same time, a range of primarily social factors, including between consanguinity and community endogamy in determin-
urbanization, improved female education, and smaller family sizes ing population profiles of genetic disease, the current global
indicate that the global prevalence of consanguineous unions will prevalence of consanguineous unions, and the overall impact of
decline. This shift in marriage patterns will initially result in first-cousin marriage on survival and health.
decreased homozygosity, accompanied by a reduction in the ex-
pression of recessive single-gene disorders. Although the roles of Civil and Religious Regulation of Consanguineous Marriage
common and rare gene variants in the etiology of complex disease The roots of negative Western attitudes toward consanguinity
remain contentious, it would be expected that declining consan- extend back over 1,500 years. In the Eastern Roman Empire
guinity would also be reflected in reduced prevalence of complex the legality of first-cousin marriage had been confirmed by the
diseases, especially in population isolates. Emperor Arcadius in 400 AD (10), possibly in acceptance of the
marriage regulations defined in the Old Testament Book of
community genetics | inbreeding | reproduction | health | social structure Leviticus 18:7–18. But according to the Venerable Bede writing
in the early 8th century (11), in 597 AD Augustine the first

I t is generally accepted that the founding population size of

Homo sapiens was small, with effective population estimates
ranging downward from ≈10,000 to 1,900–2,800 and ≈1,000 to
Archbishop of Canterbury was advised by Pope Gregory I that
first-cousin marriage was banned by sacred law, a somewhat
overly enthusiastic interpretation of Leviticus 18:6, “none of you
≈700 (1–4). With such limited total numbers and population shall approach to any that is near kin to him, to uncover their
dispersal caused by a hunter–gatherer existence, a substantial nakedness.” Depending on the translation of Bede consulted,
level of inbreeding would have been inevitable, almost certainly Gregory I further advised that first-cousin unions “do not result
involving multiple loops of kin relationships. Close-kin unions in children” (11), an opinion that is factually incorrect (12), or
continued during the subsequent slow population growth of that “the offspring of such marriages cannot thrive” (10), which
human groups living mainly in scattered rural settlements, with also is at best an overstatement.
bottlenecks caused by periodic epidemics, famines, and warfare. Until 1917 the Roman Catholic Church required dispensation
Even in mid 19th-century Europe and North America first- for unions between couples related as first, second, or third
cousin marriage remained both socially accepted and quite cousins (equivalent to a coefficient of inbreeding, F ≥ 0.0039),
widely favored, especially among the more privileged classes (5,
6). Against this background it is puzzling that in recent gener- This paper results from the Arthur M. Sackler Colloquium of the National Academy of Sciences,
ations human inbreeding has been subject to widespread nega- “Evolution in Health and Medicine” held April 2–3, 2009, at the National Academy of Sciences
in Washington, DC. The complete program and audio files of most presentations are available
tive opinion and prejudice in Western societies. on the NAS web site at www.nasonline.org/Sackler_Evolution_Health_Medicine.
Ironically, suspicion as to the advisability of first-cousin
Author contributions: A.H.B. designed research; A.H.B. and M.L.B. performed research;
marriage had been raised by Charles Darwin (7) in the improb- M.L.B. analyzed data; and A.H.B. and M.L.B. wrote the paper.
abl context of a book on self-fertilization in orchids. In keeping The authors declare no conflict of interest.
with family tradition Darwin married his first cousin Emma This article is a PNAS Direct Submission. D.R.G. is a guest editor invited by the Editorial
Wedgwood in 1839, with 10 children born over the next 17 years. Board.
Although happily married, after the death of three of their 1
To whom correspondence should be addressed. E-mail: [email protected].
children, including his favorite daughter Annie in 1851 probably This article contains supporting information online at www.pnas.org/cgi/content/full/
of tuberculosis, Darwin became concerned that their union may 0906079106/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.0906079106 PNAS | January 26, 2010 | vol. 107 | suppl. 1 | 1779–1786

Consanguineous marriage (%)
Unknown
<1
1-4
5-9
10-19
20-29
30-39
40-49
50+

Fig. 1. Global distribution of marriages between couples related as second cousins or closer (F ≥ 0.0156).

with a wide range of reasons accepted as grounds for consan- side to preclude a consanguineous union (16). Whereas in
guinity dispensation, e.g., the small size of the local population, Dravidian south India, cross first-cousin marriage (between a
advanced bridal age, or lack of dowry (13). As a result of man and his mother’s brother’s daughter) and more especially
misunderstanding after the switch from the Roman to the uncle–niece marriages are favored across all castes. Because of
Germanic system for calculating degrees of consanguinity, dur- their customary nature, cross-cousin marriages were recognized
ing the late 11th to the early 13th centuries the requirement for by the government of India in the Hindu Marriage Act of 1955
dispensation expanded to include fourth-, fifth-, and sixth-cousin and the legality of uncle–niece marriages was confirmed in the
marriages (F ≥ 0.00006), a level of regulation that rapidly proved Hindu Code Bill of 1984 (17).
impractical at local level (10). Because Luther had attacked the
dispensation requirements for consanguineous unions as repre- The Current Global Prevalence of Consanguineous Marriage
senting the rules of the church rather than of divine intention, As illustrated in Fig. 1, based on detailed information accessible
and as a revenue-raising device (10), after the Reformation the at the Global Consanguinity website (www.consang.net), close-
Protestant denominations largely accepted the Levitical mar- kin marriage continues to be preferential in many major popu-
riage proscriptions with no restriction on first-cousin unions. lations, with the influence of religion apparent in the major
The Levitical guidelines also permit uncle–niece marriage regional differences in consanguinity prevalence across the globe
(F = 0.125), which along with first-cousin marriages are still (18). Despite anthropological reports indicating consanguineous
practiced in many Sephardi Jewish communities. Marriage reg- marriage throughout sub-Saharan Africa, and in populous Asian
ulations in Islam permit first-cousin and double first-cousin (F = countries including Bangladesh and Indonesia, little quantitative
0.125) marriages, but uncle–niece unions are prohibited by the information on consanguinity is available from these regions.
Quran. Contrary to common belief there is no encouragement Nevertheless, current data indicate that some 10.4% of the 6.7
of consanguinity within Islam, and although the Prophet Mu- billion global population are related as second cousins or closer
hammad married his daughter Fatima to his ward and first cousin (F ≥ 0.0156). Although the overall prevalence of consanguine-
Ali, several hadith (sayings of the Prophet) endorse marriage ous marriage seems to be declining, in some countries the
between nonrelatives (14). It therefore seems that the strong present-day rates of consanguinity exceed those of the preceding
preference for first-cousin marriage in most Muslim countries, generation, possibly reflecting greater overall survival to adult-
principally the parallel paternal subtype, i.e., between a man and hood that in turn increases the numbers of marriageable bio-
his father’s brother’s daughter, reflect both pre-Islamic Arab logical relatives (19).
tradition and the rules introduced in the Quran enabling female Large-scale emigration of people from countries where con-
inheritance of wealth (15). sanguinity is preferential to North America, Europe, and Oce-
First-cousin marriage is generally permitted within Buddhism, ania was an important demographic feature of the latter half of
but the marriage regulations in Hinduism are more complex. the 20th century. As previously indicated, first-cousin marriages
According to the north Indian tradition believed to date back to (F = 0.0625) have the potential to cause legal problems for
200 BC, pedigrees are examined over an average of seven migrants and state law enforcement authorities in the United
generations on the male side and five generations on the female States because these unions are now either illegal or a criminal

1780 | www.pnas.org/cgi/doi/10.1073/pnas.0906079106 Bittles and Black

offense in 31 of 50 states (5, 6, 20), despite a unanimous or biraderi (19). For this reason, in many populations the clan or
recommendation in 1970 that all such state laws should be its hereditary social/occupational equivalent may be the most
rescinded (21). In Western Europe there are at least 10 million logical unit for genetic screening and genetic counseling pro-
resident migrants from regions where consanguinity is prefer- grams, as exemplified by the distribution pattern of β-thalasse-
ential, and it is the possibility that the progeny of consanguineous mia in Oman where >50% of cases were diagnosed in just one
unions are more likely to be affected by recessive genetic of the 185 major tribes and subtribes (37, 38).
disorders that has aroused greater controversy, for example, with
calls by some legislators for a ban on first-cousin marriages in the Consanguinity and Health
United Kingdom’s Pakistani community (19, 22). Although a Within genetics, contemporary attention on consanguineous
decline in first-cousin marriage has been observed in the Nor- marriage continues to be largely focused on the expression and
wegian Pakistani community (23), no similar trend seems to have identification of rare autosomal recessive alleles, a recent ex-
occurred in the United Kingdom’s Pakistani population (24) or ample being a comparative study in Norway of progressive
in the Turkish or Moroccan communities in Belgium (25), and encephalopathy in Pakistani migrants and the indigenous pop-
a rapid reduction in the preference for consanguineous unions ulation (39). But as indicated in Fig. 2, from an overall health
by first- and second-generation migrant families in Europe perspective consanguinity is a much wider and more complex
appears improbable. topic involving major social, economic, and demographic influ-
ences, differential reproductive behavior, and early- and late-
The Comparative Roles of Consanguinity and Endogamy in onset morbidity and mortality. A thorough appreciation of the
Genetic Studies salient nongenetic variables is therefore essential in addressing
Intracommunity marriage is the norm in regions where consan- the concerns of individuals, families, and communities with
guineous marriage is favored, usually contracted within long- regard to reproductive choices, and in designing genetic education
established male lineages, e.g., within the clan (hamula) and and genetic counseling programs for consanguineous couples.
tribe in Arab societies, within caste in India, and intrabiraderi in The highest overall prevalence of consanguineous unions is in
Pakistan. Because gene flow between communities is highly poor rural communities, which are typified by low levels of
restricted in most traditional societies, adjacent villages or even maternal education, early age at marriage and first birth, short
coresident subcommunities may exhibit very different inherited birth intervals, and longer reproductive spans (15, 40–42). Each
disease profiles, reflecting local founder mutations and genetic of these factors is independently associated with larger family
drift (18). These characteristics have been demonstrated in sizes and higher rates of infant and early childhood mortality,
tribe-specific single gene disorders in Saudi Arabia (26–28), the with reproductive compensation for early losses a further com-
differential origins and expansion patterns of β-globin mutations plicating issue in assessing the overall health outcomes of
in an Israeli Arab village (29), and village- and lineage-specific consanguinity (12). Comprehensive genetic education and pre-
predisposing genes for visceral leishmaniasis in Sudan (30). marital genetic counseling programs can help to lessen the
Under these circumstances and whether or not the parents are burden of genetic diseases in such communities, as reported in
known to be consanguineous, a recessive founder or de novo Israeli Arab and Bedouin villages (43–45). While in Middle
mutation of chronic effect can rapidly increase in frequency Eastern countries such as Bahrain educational programs aimed
within a particular community or subcommunity, resulting in the at high school children, and through them their parents and
birth of an affected child. In communities with a high level of relatives, have had a marked beneficial effect in reducing the
consanguineous marriage, the diagnosis of a recessive disorder incidence of sickle cell disease (46). There are, however, current
in one or more members of the same family is generally limitations to the success of these initiatives in many low-income
indicative of a recent mutation, whereas the presence of a rare countries, in particular the lack of clinicians, genetic counselors,
disorder in several families suggests an older mutational event or nurses, and scientific support staff with appropriate specialist
previous admixture through marriage with a person from an- training (47). Patients referred for genetic counseling may also
other community (31). expect directive advice as to whether or not to proceed with a
Population substructure, whether caused by ethnic, geograph- pregnancy, with failure to provide an opinion interpreted as a
ical, religious, or social divisions, often results in variant marker lack of knowledge on the part of the clinician (48), and even
allele frequencies in different subpopulations. The occurrence of when specific rulings have been provided by religious authorities
type 1 errors, i.e., false positive results, is of major importance permitting prenatal diagnosis of genetic diseases and selective
in case-control studies, association studies, and clinical trials (32, termination of a pregnancy, this option may remain unaccept-
33). Conflicting opinions have been expressed as to the impact of able to individual couples (15).
population stratification on genomewide studies with, for
example, the claim that in the United Kingdom if persons of Consanguinity, Mortality, and Morbidity
non-European ancestry are excluded “the extent of population To investigate the impact of consanguinity on deaths from ≈6
stratification in the British population is generally modest” (34). months gestation to an average of 10 years of age, a metaanalysis
Conversely, in the more homogenous Icelandic population it was was conducted directly comparing prereproductive mortality in
believed that population substructure had to be considered in the first-cousin versus nonconsanguineous progeny within specific
sampling strategy, with the implication that it would be of much populations. The study sample comprised 69 populations resi-
greater importance in larger populations with more diverse dent in 15 countries located across four continents, with a total
genetic origins (35). Because genomic studies consistently report sample size of 2.14 million (Table S1). An unweighted linear
that a large majority (93–95%) of genetic variation is within- regression comparing mean mortality in first-cousin versus
population (36), the latter opinion is unsurprising and highlights nonconsanguineous progeny in each population was plotted
the need for vigilance in case-control studies to preclude spu- according to the standard equation y = a + bx. The results are
rious associations. presented in Fig. 3 as a scatter diagram and show a mean excess
As discussed in the following sections, population stratifica- mortality at first-cousin level of 3.5% (r2 = 0.70; P < 0.00001)
tion may also be of critical importance in the investigation of that is consistent across the range of control mortalities, i.e., the
consanguinity-associated morbidity and mortality, with straight- level of excess consanguinity-associated mortality is independent
forward comparisons drawn between the progeny of first cousins of the basal (nonconsanguineous) death rate in each study
versus unrelated parents of dubious validity unless both sets of population. The estimate of 3.5% excess deaths among first-
parents are known to be members of the same clan, tribe, caste, cousin progeny compares with an earlier global estimate of 4.4%

Bittles and Black PNAS | January 26, 2010 | vol. 107 | suppl. 1 | 1781
 Marriage arrangements and dowry

Family solidarity and compatibility with in-laws

Social and economic Female autonomy

effects Marital violence
Divorce rates

Age at marriage

Net fertility Reproductive span

Maternal/fetal compatibility
Reproductive compensation

CONSANGUINITY
Maternal and perinatal factors
Congenital defects
Single gene disorders, including
sensoneural defects
Morbidity
Intellectual disability, behavioral and
psychiatric disorders
Specific infections
Adult-onset diseases

Prenatal losses
Mortality Infant and childhood deaths
Adulthood deaths

Fig. 2. Influences and outcomes of consanguineous marriage.

excess mortality (49) calculated from 38 studies each of which to further downward revision as data from better-designed
was included in the present analysis, and it matches the 3.5% studies become available.
excess mortality derived for Italian data of the early to mid 20th The influence of first-cousin marriage on the prevalence of
century (13). autosomal recessive single-gene disorders was examined as part
Initial estimates of the adverse effects of consanguineous of an investigation into consanguinity-associated morbidity in a
marriage, expressed as lethal gene equivalents, had produced Pakistani community in the United Kingdom (51). From the
significantly higher values for consanguinity-associated mortal- results of this 5-year prospective study it was calculated that there
ity, mainly because of lack of control for the negative correlation would be a ≈7/1,000 increase in autosomal recessive disorders
between consanguinity and socioeconomic status (50). Although per 0.01 increase in the mean coefficient of inbreeding (52).
control for the effects of nongenetic variables was improved in Thus, in a national population such as Pakistan where ≈50% of
the present study, the mean value of 3.5% excess mortality at the marriages were between first cousins (F = 0.0625) (53) some
first-cousin level is an upper-level estimate that may be subject 22/1,000 extra single-gene disorders would be expected.
Unfortunately, the original study omitted control for popula-
tion subdivision, which has been shown to be a notable feature
0.5
52
of indigenous and migrant Pakistani populations (54–56), and as
70
21
55
previously noted is typical of many more traditional populations.
Wahlund effect predicts that subdivided populations character-
0.4 istically exhibit higher than predicted levels of homozygosity.
56
20
72
22
Given the known levels of population substructure associated
30 13
14
54
with biraderi membership in Pakistan and the Pakistani com-
Deaths In 1C Progeny

0.3 65
63

57 53
munity in the United Kingdom, nonconsanguineous couples are
11
31
73
at higher risk of sharing the same recessive disease mutation than
64
59 71
34
32
16 counterparts in populations where limited or no substructure
33

0.2
40
35
28
36
29
exists. The consequent random consanguinity effect on the
38
7 15
75 37 78
distribution and expression patterns of recessive disease genes
4774 568
79 12 43
6 76
means that in populations with significant subdivision the ben-
44

0.1
2
41 58 39 60 3
48
eficial health outcomes that have been claimed through simply
66 67
69 77
4
18
avoiding consanguineous marriage are almost certainly exagger-
19
49 25
1
50 26
61 8
27
ated and require reassessment (19, 57).
0.0
0.0 0.1 0.2 0.3 0.4 0.5
Consanguinity and Complex Diseases
There has been extended debate on the nature of the genetic
Deaths In NC Progeny
contribution to complex diseases, i.e., whether the common
Fig. 3. Comparative mortality in first cousin (1C; F = 0.0625: y axis) versus disease/common variant or the common disease/rare variant
nonconsanguineous progeny (NC; F = 0: x axis) in 69 study populations. hypothesis is more applicable (58), with the role of copy number

1782 | www.pnas.org/cgi/doi/10.1073/pnas.0906079106 Bittles and Black

variants also proposed (59, 60). Consanguinity would be ex- the comparative levels of inbreeding are genetically quite im-
pected to exert a greater influence on the etiology of complex precise and principally reflect village endogamy rather than
diseases if rare autosomal recessive alleles were causally impli- consanguinity per se. As previously discussed, until the early 20th
cated, whereas if disease alleles that are common in the gene century church dispensation would have been required for
pool are involved then intrafamilial marriage would have a marriages between spouses related as third cousins or closer
proportionately lesser effect. However, because both gene–gene (F ≥ 0.0039) in these devoutly Roman Catholic communities
interactions and numerous nongenetic factors in prenatal and (13). In the absence of church records indicating dispensation for
postnatal life also contribute to the disease phenotype, a single marriages contracted within the prevailing consanguinity regu-
all-embracing solution to the genetics of complex diseases is lations, the consanguineous relationships examined may princi-
highly improbable. pally have been random rather than preferential in nature and
Major genomewide analyses of diseases with onset primarily reflected restricted marriage partner choices. The analysis of
in childhood and adulthood have identified associations with genealogical data covering four to five generations showed
specific chromosomal regions, e.g., for type 1 and type 2 diabetes substantial levels of consanguinity in some communities, with
(61, 62), although these studies have emphasized the large mean coefficients of inbreeding ranging from α = 0.002 to 0.049
numbers of genes involved and the small increased risk that calculated at village level, indicating major variations in local
appears to be associated with most individual variants. Concern marriage patterns driven by both the history and the geograph-
also has been expressed that concentration on the identification ical location of each settlement (80).
of gene variants via patients with the disease under study rather Pedigree-based estimates of consanguinity and the resultant
than full genome sequencing of randomly ascertained samples levels of homozygosity have several limitations; in particular,
could lead to significantly inflated rates of false positives (63). they do not provide information on close-kin marriages that have
Investigations into the effects of consanguinity on congenital occurred in distant generations and thus underestimate cumu-
defects have produced quite varied results, in large part because lative inbreeding effects, and with rare exceptions incorrectly
of a lack of standardized assessment protocols and the different ascribed paternity is not recorded. To complement the pedigree-
environmental and socioeconomic circumstances of the study based approaches previously adopted and avoid these difficul-
populations. Using nonconsanguineous progeny as controls, ties, high-density genome scans were used to estimate individual
estimates of the excess level of congenital defects in first-cousin autozygosity (Froh) from uninterrupted runs of homozygosity
offspring have ranged from 0.7% to 7.5% (64–68), but the Latin (ROH). An appropriate length threshold was empirically derived
American Collaborative Study of Congenital Malformation for ROH and the method was applied to data derived from
based on 34,1902 newborns found a significant association with residents of the Dalmatian islands, the Orkney islands off the
consanguinity only for hydrocephalus, postaxial polydactyly, and north coast of Scotland, mainland Scotland, and the state of
bilateral oral and facial clefts (69). Utah (84). Initial comparisons of Froh values ranging from 0.5,
A different picture emerges from the large literature on i.e., with a minimum length threshold of 0.5 Mb, to 1 (length
congenital heart defects, which are conservatively estimated to threshold 1 Mb) and 5 (length threshold 5Mb) with pedigree data
have an incidence of 50/1,000 live births (70). Although a from the Orkneys indicated good correlation with pedigree-
consistent positive association between consanguinity and dis- based mean coefficients of inbreeding and so confirmed the
orders such as ventricular septal defect and atrial septal defect applicability of the method for the direct assessment of autozy-
has been demonstrated, indicating the involvement of common gosity. The method has been further applied to investigate
variants, both positive and negative associations with patent changes in autozygosity through time in two American study
ductus arteriosus, atrioventricular septal defect, pulmonary atre- populations. The steady decreases observed in the size and
sia, and tetralogy of Fallot have been reported in different frequency of ROH > 1 Mb in length in these populations were
populations (71–74), suggestive of community-specific founder ascribed to expanded marriage pools and larger effective pop-
mutations. It is, however, also possible that nonstandardized ulation sizes and interpreted as indicating future ongoing reduc-
diagnostic protocols may have contributed to the variant findings tions in the frequency of rare recessive disorders (85).
reported by different study centers. When applied to behavioral disorders genomewide analysis
As yet relationships between consanguinity and complex has indicated the potential contribution of thousands of alleles
diseases of adulthood have been significantly underinvestigated, of very small effect in schizophrenia and bipolar disorder, with
and the few studies published have relied mainly on rudimentary significant genetic overlap between the two disease states (86,
sampling strategies, with simple consanguineous versus noncon- 87). At the same time, homozygosity mapping in autism (88) and
sanguineous comparisons in disease prevalence and inadequate a case-control study of bipolar disorder type 1 in consanguineous
attention paid to possible genetic or demographic subdivisions. progeny (89) both implicated the causal expression of rare
Accordingly, the results obtained often are contradictory, e.g., recessive genes. ROH similarly have been shown to be signifi-
with both positive and negative associations reported between cantly more common in patients with schizophrenia spectrum
consanguinity and breast cancer (75–77), and consanguinity and disorders, suggesting the involvement of recessive alleles in the
heart disease (75, 78, 79). Long-term studies conducted on the etiology of the disorder (90). Reverting to earlier comments on
Dalmatian islands in the Adriatic Sea have indicated a positive the relationship between endogamy and consanguinity, an as-
association between inbreeding and a very wide range of com- sociation between consanguinity and Alzheimer disease was
mon adulthood disorders, including hypertension, coronary demonstrated in a genealogical study of the Saguenay region in
heart disease, stroke, cancer, uni/bipolar depression, asthma, Québec (91), and multiple loci for Alzheimer disease were
gout, peptic ulcer, and osteoporosis (80–82). The data thus identified in a highly endogamous and consanguineous Israeli
suggest virtually ubiquitous causal involvement of rare autoso- Arab kindred (92), in both cases indicative of founder mutations.
mal recessive genes in adult-onset disease in this population, with Thus, from a more general perspective these results strengthen
the more general corollary that increasing genomewide het- the argument that all association studies on complex diseases
erozygosity after a decline in consanguineous marriage should would benefit from a sound prior knowledge of community
lead to a widespread reduction in the burden of common genetic demographic and genetic structure.
diseases (83).
The Dalmatian studies have the very considerable advantage Discussion
of demographically well-characterized populations with known Although consanguinity is a highly complex and multifaceted
ethnic origins, although the actual definitions used in assessing topic (Fig. 2), the claimed social and cultural advantages, such as

Bittles and Black PNAS | January 26, 2010 | vol. 107 | suppl. 1 | 1783
Biological disadvantages Social benefits of because of rapidly declining family sizes, future global reductions
vs
of consanguinity consanguinity in the prevalence of consanguinity appear to be inevitable (19).
What effect will this predicted reduction in consanguinity have
in terms of human evolution and on the prevalence of genetic
disease? Recent studies have identified the ongoing role of
positive natural selection during an extended period when
Rural effective population sizes were small and consanguinity would
Biological Social have been high (101–103), and the very rapid increases in global
population numbers over the course of the last 150 years would
suggest even greater acceleration in the pace of current and
future human adaptive evolution (104). Although the mixing of
previously separated breeding groups should lead to a marked
Urban initial reduction in the global prevalence of rare autosomal
Biological Social recessive disorders (85), the subsequent dispersal of phenotyp-
ically normal heterozygotes through newly agglomerated breed-
ing pools will in time result in the “random” mating of noncon-
sanguineous carriers of recessive mutations. But the rate at
which these changes in mating patterns occur will necessarily be
more rapid in increasingly panmictic urbanized populations than
Fig. 4. Contrasting biological and social outcomes of consanguineous mar-
in endogamous ethnic, religious, geographical, or social isolates.
riage in traditional rural and modern urban settings.
Whether similar predictions are possible for complex diseases
will very much depend on the proportional contribution of
ease of marriage arrangements, enhanced female autonomy, recessive genes, and more especially rare recessive genes, to
more stable marital relationships, greater compatibility with individual diseases in different populations. For the moment the
in-laws, lower domestic violence, lower divorce rates, and the greatest promise in identifying genes of major effect for complex
economic benefits of reduced dowry and the maintenance of any diseases continues to reside in endogamous communities with
landholdings (15, 41, 42, 47, 93–95) have received much less extensive genealogical records (105). Convincing support for this
attention than studies into adverse genetic outcomes. It there- approach is provided by the high frequencies of autosomal
fore is not surprising that the prevailing Western public and recessive disease genes diagnosed in numerically small, highly
medical opinion with regard to consanguinity is largely negative. endogamous Arab Israeli communities (106). Yet, surprisingly,
There is the additional problem that in many societies that favor in these communities and other isolates where consanguinity is
consanguineous unions marriages are usually arranged by and/or much less common, multiple mutations in specific disease genes
meet with prior parental approval, a practice frequently misrep- have been identified where a single founder mutation would
resented and criticized as “forced marriage” (15). more usually have been expected (29, 107). Because limited
For families living in impoverished rural areas with limited or genetic diversity and restricted allelic heterogeneity are gener-
no formal education or access to medical services, young age at ally expected in isolated founder populations, it also is salutary
marriage and first pregnancy, short birth intervals, and high that a genomewide association analysis of obesity and other
infant and childhood mortality rates primarily caused by infec- metabolic disorders in a Pacific island community, in which
tious and nutritional disorders, the social and economic advan- reduced haplotype diversity and extended linkage disequilibrium
tages offered by consanguineous marriage and the strengthening had already been demonstrated, failed to detect major contrib-
of family relationships often outweigh the biological disadvan- utory alleles and instead indicated the presence of common
tages of close-kin marriage for a majority of families (96, 97). The variants of small effect (108, 109).
current scenario in urban populations is quite different, espe- Having largely been ignored for many years, the specific roles
cially in developed countries with better living and public health of population bottlenecks and consanguinity in influencing
conditions, low levels of infectious disease, and ready access to variation between and within populations are now receiving due
modern health facilities. Newborns with a genetic disorder that attention, with special focus on homozygosity in identifying
in previous generations may have died in infancy of no known recent common ancestry via ROH analysis (110). The potential
cause are now referred to specialist centers for diagnosis, and complexity of the interrelationships between consanguinity and
they and their families can anticipate a lifespan that will extend human health and disease was highlighted by the reported
at least into adolescence and more probably into mid to late association between consanguinity and predisposition to major
adulthood, usually requiring ongoing medical care. infectious diseases (111). If these findings are substantiated, by
Unless a de novo mutation has been identified the diagnosis ameliorating the risk of exposure to infectious agents a global
will effectively involve other family members as potential or decline in consanguinity could also providentially reduce the risk
obligate carriers and so could become a negative factor in all of inflammatory disease and hence the development of coronary
disease in middle and old age (112).
future family marriage arrangements (19, 98). For this reason, in
Time will tell whether these as yet tenuous epidemiological
disorders with a very adverse clinical outcome and involving
connections can be sustained. In the interim, it is important to
multiple affected family members, such as progressive retinop-
emphasize that in assessing the impact of consanguinity on any
athy and amelogenesis (99) and severe intellectual disability
aspect of health a clear causal relationship needs to be estab-
(100), marriage to a nonrelative may not be a realistic option,
lished, rather than reliance on speculation driven solely by the
resulting either in celibacy or continued intrafamilial marriage. presence of a close kin union in the family pedigree. At the same
Within the wider community, greater understanding and accep- time, rigorous control for population stratification should be a
tance of genetic explanations for familial patterns of disease and prerequisite in the many populations where community subdi-
the unfavorable medical outcomes experienced by some con- visions exist if confused and confusing conclusions are to be
sanguineous families can significantly influence the perceived avoided.
balance of advantage and disadvantage associated with intrafa-
milial marriage (Fig. 4). Therefore, in conjunction with increas- ACKNOWLEDGMENTS. We thank the referees for constructive comments.
ing difficulty in finding a marriageable cousin of acceptable age A.H.B. is supported by National Science Foundation Grant 0527751.

1784 | www.pnas.org/cgi/doi/10.1073/pnas.0906079106 Bittles and Black

 1. Harpending HC, et al. (1998) Genetic traces of ancient demography. Proc Natl Acad 37. Rajab A, Patton MA (1997) Major factors determining the frequencies of hemoglo-
Sci USA 95:1961–1967. binopathies in Oman. Am J Med Genet 71:240–242.
2. Tenesa A, et al. (2007) Recent human effective population size estimated from 38. Rajab A, Patton MA (1999) Analysis of the population structure in Oman. Commun
linkage disequilibrium. Genome Res 17:520–526. Genet 2:23–25.
3. Liu H, Prugnolle F, Manica A, Balloux F (2006) A geographically explicit genetic map 39. Strømme P, et al. (2009) Parental consanguinity is associated with a 7-fold increased
of worldwide human-settlement history. Am J Hum Genet 79:230–237. risk of progressive encephalopathy: A cohort study from Oslo, Norway. Eur J Paed
4. Zhivotovsky LA, Rosenberg NA, Feldman MW (2003) Features of evolution and Neurol, in press.
expansion of modern humans, inferred from genomewide microsatellite markers. 40. Bittles AH, Mason WH, Greene J, Appaji Rao N (1991) Reproductive behavior and
Am J Hum Genet 72:1171–1186. health in consanguineous marriages. Science 252:789–794.
5. Ottenheimer M (1996) Forbidden Relatives: The American Myth of Cousin Marriage 41. Bittles AH (1994) The role and significance of consanguinity as a demographic
(Univ Illinois Press, Urbana), pp 19–41. variable. Pop Dev Rev 20:561–584.
6. Bittles AH (2003) The bases of Western attitudes to consanguineous marriage. Dev 42. Khlat M (1997) Endogamy in Arab countries. Genetic Disorders Among Arab Popu-
Med Child Neurol 45:135–138. lations, eds Teebi A, Farag TI (Oxford Univ Press, New York), pp 63–80.
7. Darwin C (1862) On the Various Contrivances by Which British and Foreign Orchids 43. Shiloh S, Reznik H, Bat-Miriam-Katznelson M, Goldman B (1995) Premarital genetic
Are Fertilized by Insects, and on the Good Effects of Intercrossing (John Murray, counseling to consanguineous couples: Attitudes, beliefs, and decisions among coun-
London), pp 359–360. seled, noncounseled, and unrelated couples in Israel. Soc Sci Med 41:1301–1310.
8. Darwin GH (1875) Marriages between first cousins in England and Wales and their 44. Raz AE, Atar M (2004) Cousin marriage and premarital carrier matching in a Bedouin
effects. J Stat Soc 38:153–184. community in Israel: Attitudes, service development, and educational intervention. J
9. Darwin C (1876) The Effects of Cross and Self-Fertilization in the Vegetable Kingdom Fam Planning Reprod Health Care 30:49–51.
(John Murray, London), pp 460–461. 45. Zlotogora J, Carmi R, Lev B, Shalev SA (2009) A targeted population carrier screening
10. Goody J (1985) The Development of the Family and Marriage in Europe (Cambridge program for severe and frequent genetic diseases in Israel. Eur J Hum Genet 17:591–
Univ Press, Cambridge, UK), pp 48–60 and 134–146. 597.
11. Bede (1991) The Ecclesiastical History of the English People (Penguin Books, London), 46. Al Arrayed S (2005) Campaign to control genetic blood diseases in Bahrain. Commun
Revised Ed, pp 79–81. Genet 8:52–55.
12. Bittles AH, Grant JC, Sullivan SG, Hussain R (2002) Does inbreeding lead to increased 47. Hamamy H, Bittles AH (2009) Genetic clinics in Arab communities: Meeting individual,
human fertility? Ann Hum Biol 29:111–131. family, and community needs. Pub Health Genomics 12:30–40.
13. Cavalli-Sforza LL, Moroni A, Zei G (2004) Consanguinity, Inbreeding, and Genetic Drift 48. Eldahdah L, et al. (2007) Outcome of chromosomally abnormal pregnancies in Leb-
in Italy (Princeton Univ Press, Princeton). anon: Obstetricians’ roles during and after prenatal diagnosis. Prenat Diagn 27:525–
14. Hussain R (1999) Community perceptions of reasons for preference for consanguin- 534.
eous marriages in Pakistan. J Biosoc Sci 31:449–461. 49. Bittles AH, Neel JV (1994) The costs of human inbreeding and their implications for
15. Bittles AH, Hamamy HA (2009) Consanguinity and endogamy in Arab countries.
variations at the DNA level. Nat Genet 8:117–121.
Genetic Disorders among Arab Populations, ed Teebi A (Springer, Heidelberg), 2nd 50. Bittles AH, Makov E (1988) Inbreeding in human populations: An assessment of the
Ed, in press.
costs. Human Mating Patterns, eds Mascie-Tyalor CGN, Boyce AJ (Cambridge Univ
16. Kapadia KM (1958) Marriage and the Family in India (Oxford Univ Press, Calcutta),
Press, Cambridge, UK), pp 153–167.
2nd Ed, pp 117–137.
51. Bundey S, Alam H (1993) A 5-year prospective study of the health of children in
17. Bittles AH (2002) Endogamy, consanguinity, and community genetics. J Genet 81:91–
different ethnic groups, with particular reference to the effect on inbreeding. Eur J
98.
Hum Genet 1:206–219.
18. Bittles AH (2009) Consanguinity, genetic drift, and genetic diseases in populations
52. Christianson A, Howson CP, Modell B (2006) Global Report on Birth Defects (March
with reduced numbers of founders. Human Genetics: Principles and Approaches, eds
of Dimes, White Plains, NY), pp 83–84.
Vogel F, Motulsky AG, Antonarakis SE, Speicher M (Springer, Heidelberg), 4th Ed, in
53. Hussain R, Bittles AH (1998) The prevalence and demographic characteristics of
press.
consanguineous marriages in Pakistan. J Biosoc Sci 30:261–275.
19. Bittles AH (2008) A community genetics perspective on consanguineous marriage.
54. Wang W, et al. (2000) A genome-based study of consanguinity in three coresident
Commun Genet 11:324–330.
endogamous Pakistan communities. Ann Hum Genet 64:41–49.
20. Paul DB, Spencer HG (2008) “It’s OK, we’re not cousin by blood”: The cousin marriage
55. Qamar R, et al. (2002) Y-chromosomal DNA variation in Pakistan. Am J Hum Genet
controversy in historical perspective. PLoS Biol 6:e320.
70:1107–1124.
21. National Conference of Commissioners (1970) Handbook on Uniform State Laws and
56. Overall ADJ, Ahmad M, Thomas MG, Nichols RA (2003) An analysis of consanguinity
Proceedings of the Annual Conference Meeting in its Seventy-Ninth Year (Port City
and social structure within the U.K. Asian population using microsatellite data. Ann
Press, Baltimore).
Hum Genet 67:525–537.
22. Dyer O (2005) MP is criticized for saying that marriage of first cousins is a health risk.
57. Overall ADJ (2009) The influence of the Wahlund effect on the consanguinity hy-
Br Med J 331:1292.
pothesis: Consequences for recessive disease incidence in a socially structured Paki-
23. Grjibovski AM, Magnus P, Stoltenberg C (2009) Decrease in consanguinity among
parents of children born in Norway to women of Pakistani origin: A registry-based stani population. Hum Hered 67:140–144.
58. Schork NJ, Murray SS, Frazer KA, Topol EJ (2009) Common vs. rare allele hypotheses
study. Scand J Pub Health 37:232–238.
24. Shaw A (2000) Kinship, cultural preference, and immigration: Consanguineous mar- for complex diseases. Curr Opin Genet Dev 19:212–219.
riage among British Pakistanis. J R Anthropol Inst 7:315–334. 59. Estevill X, Armengol L (2007) Copy number variants and common disorders: Filling the
25. Reniers G (1998) Postmigration Survival of Traditional Marriage Patterns: Consan- gaps and exploring complexity in genomewide association studies. PLoS Genet
guineous Marriage Among Turkish and Moroccan Immigrants in Belgium (Depart- 3:e190.
ment of Population Studies, University of Gent, Gent, Belgium), Interuniversity Papers 60. Need AC, et al. (2009) A genomewide investigation of SNPs and CNVs in schizophre-
in Demography, PPD-1 Working Paper 1998-1. nia. PLoS Genet 5:e1000373.
26. Ozand PT, et al. (1990) Prevalence of different types of lysosomal storage diseases in 61. Todd JA, et al. (2007) Robust associations of four new chromosome regions from
Saudi Arabia. J Inherit Metab Dis 13:849–861. genomewide analyses of type 1 diabetes. Nat Genet 39:813–815.
27. Ozand PT, Devol EB, Gascon GG (1992) Neuro-metabolic diseases at a national referral 62. Lyssenko V, et al. (2008) Clinical risk factors, DNA variants, and the development of
center: Five years experience at the King Faisal Specialist Hospital and Research type 2 diabetes. N Engl J Med 359:2220–2232.
Centre. J Child Neurol 7 (Suppl): S4–S11. 63. Li B, Leal SM (2009) Discovery of rare variants via sequencing: Implications for the
28. Rashed M, Ozand PT, Al Aqeel A, Gascon GG (1994) Experience of King Faisal Specialist design of complex trait association studies. PLoS Genet 5:e1000481.
Hospital and Research Center with Saudi organic acid disorders. Brain Dev 64. Schull WJ (1958) Empirical risks in consanguineous marriages: Sex ratio, malforma-
16 (Suppl): 1–6. tion, and viability. Am J Hum Genet 10:294–343.
29. Zlotogora J, et al. (2005) The origin and expansion of four different β-globin muta- 65. Jaber L, Merlob P, Bu X, Rotter JI, Shohat M (1992) Marked parental consanguinity as
tions in a single Arab village. Am J Hum Biol 17:659–661. a cause for increased major malformations in an Israeli Arab community. Am J Med
30. Miller EN, et al. (2007) Y chromosome lineage- and village-specific genes on chromo- Genet 44:1–6.
somes 1p22 and 6q27 control visceral leishmaniasis in Sudan. PLoS Genet 3:e71. 66. Stoltenberg C, Magnus P, Lie TR, Daltveit AK, Irgens LM (1997) Birth defects and
31. Zlotogora J, Hujerat Y, Barges S, Shalev SA, Chakravarti A (2006) The fate of 12 parental consanguinity in Norway. Am J Epidemiol 145:439–448.
recessive mutations in a single village. Ann Hum Genet 71:202–208. 67. Zlotogora J (2002) What is the birth defect risk associated with consanguineous
32. Heiman GA, Hodge SE, Gorroochurn P, Zhang J, Greenberg DA (2004) Effect of marriage? Am J Med Genet 109:70–71.
population stratification on case-control association studies. Hum Hered 58:30–39. 68. Bromiker R, Glam-Baruch M, Gofin R, Hammerman C, Amitai Y (2004) Association of
33. Wang K (2009) Testing for genetic association in the presence of population strati- parental consanguinity with congenital malformations among Arab newborns in
fication in genomewide association studies. Genet Epidemiol, in press. Jerusalem. Clin Genet 66:63–66.
34. The Wellcome Trust Case Control Consortium (2007) Genomewide association study 69. Rittler M, Liascovich R, López-Camelo J, Castilla EE (2001) Parental consanguinity in
of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 447:661– specific types of congenital anomalies. Am J Med Genet 102:36–43.
678. 70. Pierpont ME, et al. (2007) Genetic basis for congenital heart defects: Current knowl-
35. Helgason A, Yngvadóttir B, Hrafnkelsson B, Gulcher J, Stefánsson K (2005) An Icelan- edge. Circulation 115:3015–3038.
dic example of the impact of population structure on association studies. Nat Genet 71. Gnanalingham MG, Gnanalingham KK, Singh A (1999) Congenital heart disease and
37:90–95. parental consanguinity in South India. Acta Paediatr 88:473–474.
36. Rosenberg NA, et al. (2002) Genetic structure of human population. Science 72. Becker SM, Al Halees Z, Molina C, Paterson RM (2001) Consanguinity and congenital
298:2381–2385. heart disease in Saudi Arabia. Am J Med Genet 99:8–13.

Bittles and Black PNAS | January 26, 2010 | vol. 107 | suppl. 1 | 1785
73. Nabulsi MM, et al. (2003) Parental consanguinity and congenital heart malformations 94. Assaf S, Khawaja M (2009) Consanguinity trends and correlates in the Palestinian
in a developing country. Am J Med Genet 116A:342–347. Territories. J Biosoc Sci 41:107–124.
74. Khalid Y, et al. (2006) Consanguineous marriage and congenital defects: A case- 95. Clark CJ, Hill A, Jabber K, Silverman JG (2009) Violence during pregnancy in Jordan:
control study in the neonatal period. Am J Med Genet 140:1524–1530. Its prevalence and associated risk and protective factors. Violence Against Women
75. Shami SA, Qaisar R, Bittles AH (1991) Consanguinity and adult morbidity in Pakistan. 15:720–735.
Lancet 338:954–955. 96. Bittles AH (2001) Consanguinity and its relevance to clinical genetics. Clin Genet
76. Liede A, et al. (2002) Contribution of BRAC1 and BRAC2 mutations of breast and 60:89–98.
ovarian cancer in Pakistan. Am J Hum Genet 71:595–606. 97. Bittles AH (2005) Endogamy, consanguinity, and community disease profiles. Com-
77. Denic S, Bener A (2001) Consanguinity decreases risk of breast cancer: Cervical cancer mun Genet 8:17–20.
unaffected. Br J Cancer 85:1675–1679. 98. Shaw A, Hurst JA (2009) “I don’t see any point in telling them”: Attitudes to sharing
78. Ismail J, et al. (2004) Risk factors for nonfatal myocardial infarction in young South genetic information in the family and carrier testing of relatives among British
Asian adults. Heart 90:259–263. Pakistani adults referred to a genetics clinic. Ethn Health 14:205–224.
79. Jaber L, Shohat T, Rotter JI, Shohat M (1997) Consanguinity and common adult 99. Jalili IK, Smith NJD (1988) A progressive cone-rod dystrophy and amelogenesis im-
diseases in Israeli Arab communities. Am J Med Genet 70:346–348. perfecta: A new syndrome. J Med Genet 25:738–740.
80. Rudan I, et al. (2003) Inbreeding and the genetic complexity of human hypertension. 100. Basel-Vanagaite L, et al. (2007) Genetic screening for autosomal recessive nonsyn-
Genetics 163:1011–1021. dromic mental retardation in an isolated population in Israel. Eur J Hum Genet
81. Rudan I, et al. (2003) Inbreeding and risk of late-onset complex disease. J Med Genet 15:250–253.
40:925–932. 101. Voight BF, Kudaravalli S, Wen X, Pritchard JK (2006) A map of recent positive selection
82. Rudan I, et al. (2004) Inbreeding and susceptibility to osteoporosis in Croatian island in the human genome. PLoS Biol 4:e72.
isolates. Coll Antropol 28:585–602. 102. Sabeti PC, et al. (2007) Genomewide detection and characterization of positive
83. Campbell H, et al. (2007) Effects of genomewide heterozygosity on a range of selection in human populations. Nature 449:913–918.
biomedically relevant human quantitative traits. Hum Mol Genet 16:233–241. 103. Williamson SH, et al. (2007) Localizing recent adaptive evolution in the human
84. McQuillan R, et al. (2008) Runs of homozygosity in European populations. Am J Hum genome. PLoS Genet 3:e90.
Genet 83:359–372. 104. Hawks J, Wang ET, Cochran GM, Harpending HC, Moyzis RK (2007) Recent accelera-
85. Nalls MA, et al. (2009) Measures of autozygosity in decline: Globalization, urbaniza- tion of human adaptive evolution. Proc Natl Acad Sci USA 104:20753–20758.
tion, and its implications for medical genetics. PLoS Genet 5:e1000415. 105. Varilo T, Peltonen L (2004) Isolates and their potential use in complex gene mapping
86. Lichenstein P, et al. (2009) Common genetic determinants of schizophrenia and efforts. Curr Opin Genet Dev 14:316–323.
bipolar disorder in Swedish families: A population-based study. Lancet 373:234– 106. Zlotogora J, Shalev S, Habiballah H, Barjes S (2000) Genetic disorders among Pales-
239. tinian Arabs: 3. Autosomal recessive disorders in a single village. Am J Med Genet
87. The International Schizophrenia Consortium (2009) Common polygenic variation 92:343–345.
contributes to risk of schizophrenia and bipolar disease. Nature 460:748–752. 107. Zlotogora J (2007) Multiple mutations responsible for frequent genetic diseases in
88. Morrow EM, et al. (2008) Identifying autism loci and genes by tracing recent shared isolated populations. Eur J Hum Genet 15:272–278.
ancestry. Science 321:218–223. 108. Bonnen PE, et al. (2006) Evaluating potential for whole-genome studies in Kosrae, an
89. Mansour H, et al. (2009) Consanguinity associated with increased risk for bipolar I isolated population in Micronesia. Nat Genet 38:214–217.
disorder in Egypt. Am J Med Genet B Neuropsychiatr Genet 150:879–885. 109. Lowe JK, et al. (2009) Genomewide association studies in an isolated founder popu-
90. Lencz T, et al. (2007) Runs of homozygosity reveal highly penetrant recessive loci in lation from the Pacific island of Kosrae. PLoS Genet 5:e10000365.
schizophrenia. Proc Natl Acad Sci USA 104:19942–19947. 110. Auton A, et al. (2009) Global distribution of genomic diversity underscores rich
91. Vézina H, et al. (1999) A genealogical study of Alzheimer disease in the Saguenay complex history of continental human populations. Genome Res 19:795–803.
region of Québec. Genet Epidemiol 16:412–425. 111. Lyons EJ, Frodsham AJ, Zhang L, Hill AVS, Amos W (2009) Consanguinity and suscep-
92. Farrer LA, et al. (2003) Identification of multiple loci for Alzheimer disease in a tibility to infectious diseases in humans. Biol Lett 5:574–576.
consanguineous Israeli–Arab community. Hum Mol Genet 12:415–422. 112. Crimmins EM, Finch CE (2006) Infection, height, and longevity. Proc Natl Acad Sci USA
93. Hussien FH (1971) Endogamy in Egyptian Nubia. J Biosoc Sci 3:251–257. 103:498–503.

1786 | www.pnas.org/cgi/doi/10.1073/pnas.0906079106 Bittles and Black