You are on page 1of 4

Letters to the Editor 1243

Ober C, Elias S, Kostyu DD, Hauck WW (1992) Decreased Address for reprints and correspondence: Dr. Thomas J. Gill III, Department
fecundability in Hutterite couples sharing HLA-DR. Am J of Pathology, University of Pittsburgh School of Medicine, Scaife Hall, Room
716, Pittsburgh, PA 15261. E-mail: gill@med.pitt.edu
Hum Genet 50:6-14 © 1995 by The American Society of Human Genetics. All rights reserved.
Oksenberg JR, Persitz E, Amar A, Schenker J, Segal S, Nelken 0002-9297/95/5705-0033$02.00
D, Brautbar C (1983) Mixed lymphocyte reactivity nonre-
sponsiveness in couples with multiple spontaneous abor-
tions. Fertil Steril 39:525-529
Schacter B, Weitkamp LR, Johnson WE (1984) Parental HLA
compatibility in parents of offspring with neural-tube de- Am. J. Hum. Genet. 57:1243-1246, 1995
fects: evidence for a Tit-like locus in humans. Am J Hum
Genet 36:1082-1091
Our Tangled Heritage: Demographic History and
Address for reprints and correspondence: Dr. Carole Ober, Department of the Indian Caste System
Obstetrics and Gynecology, The University of Chicago, 5841 Maryland Avenue,
MC2050, Chicago, IL 60637. E-mail: carole@babies.bsd.uchicago.edu To the Editor:
© 1995 by The American Society of Human Genetics. All rights reserved. Mountain et al. (1995) have used mtDNA sequence vari-
0002-9297/95/5705-0032$02.00
ation to examine the genetic relationship among three
castes of India, the Havik, Mukhri, and Kadars, who
belong to the Brahmin (highest), Pancham (lowest), and
aboriginal castes (apparently ancient inhabitants of In-
dia), respectively. Phylogenetic analysis of the data re-
vealed an intermixing of lineages, which Mountain et
Am. J. Hum. Genet. 57:1243, 1995 al. (1995, p. 984) attributed to either "extensive mixing"
of the three groups or the possibility that the "mtDNA
lineages . . . arose prior to the split" of the three castes.
Reply to Ober They suggested that the cultural identification of the
populations included in the analyses were "inconsistent
To the Editor: with the genetic grouping." Mountain et al. found that,
Our citing of the reference by Kostyu et al. (1993) was in one case, five Mukhri sequences clustered with two
to give the readers an introduction to the work of Dr. Havik sequences and that the Havik and Mukhri se-
Ober and her colleagues, since it is important for the quences were scattered "throughout the tree," indicating
general reader to identify the contributions of others some level of gene flow between the two populations.
working in the field and to have one of their most recent Although the magnitude of gene flow (low, medium, or
references. It was not our intent, nor was it appropriate high; sensu Slatkin 1985) among these populations is
in this paper, to review this work in detail, since we not provided by Mountain et al., it must be high enough
have done so in a variety of different contexts. Our to produce the homogeneity that obscured the branching
reviews and the evolution of the ideas presented in Jin pattern of the lineages among the castes examined. The
et al. (1995) are given in the reference section of that results of the phylogenetic analysis appear to have sur-
paper, beginning with my 1983 review. In these publica- prised Mountain et al., in the face of the sociocultural
tions, due credit is given to all in greater detail. background of the three castes.
Mountain et al. (1995, p. 988) do not provide an
THOMAS J. GILL III approximate date as to when the split among these pop-
Department of Pathology ulations might have occurred but suggest that the caste
University of Pittsburgh School of Medicine society "crystallized fairly recently, -2,000 years ago."
Pittsburgh This period coincides approximately with the time of
the codification of several of the dharma shastras (moral
codes), notably the laws of Manu (Basham 1989), which
References have strongly influenced the social structure in Hindu
Gill TJ III (1983) Immunogenetics of spontaneous abortions
society. Mountain et al. seem to presume that such a
in humans. Transplantation 35:1-6 crystallized social structure with strict religious codes
Jin K, Ho H-N, Speed TP, Gill TJ III (1995) Reproductive may have not favored a significant amount of gene flow
failure and the major histocompatibility complex. Am J among caste groups within the past 2,000 years. There-
Hum Genet 56:1456-1467 fore, they inferred, on the basis of the phylogenetic anal-
Kostyu DD, Dawson DV, Elias S, Ober C (1993) Deficit of ysis, that the lack of clustering of mitochondrial lineages
HLA homozygotes in a Caucasian isolate. Hum Immunol according to caste affiliation may be the result of mixing,
37:135-142 which predates the separation of populations. Because
1244 Letters to the Editor

mtDNA is maternally inherited in humans, it is im- aboriginal castes of Kerala. Interestingly, it has been
portant to examine plausible cultural settings that might noted that the region of Karnataka studied by Mountain
have contributed to the exchange of maternal lineages et al. was ruled by an untouchable (Paraiah) king, was
among the three castes, particularly since the migration later conquered by Brahmin kings, and seems to have
of Haviks into the northwestern region of Karnataka. exchanged hands at least twice between the Kadambas
According to Slatkin (1987), gene flow is the genetic and the rulers of the untouchable class (Nanjundaiah
change in populations that is due to movement of ga- and Iyer 1928). Incidentally, many of the Havik Brah-
metes, individuals, or groups of individuals from one mins worship the same local female village deities (a
place to another. Gene flow via recognized marriage is practice uncommon among Brahmins) that the agricul-
a major component of the total gene flow observed in tural and other caste communities worship in this region
modern human societies. However, both the union of (Nanjudaiah and Iyer 1928), suggesting some degree of
gametes through extramarital relationships and the relationship between the Haviks and the local inhabit-
movement of individuals from one group into another ants. Furthermore, the "gotra" (exogamous clan; liter-
will also contribute to total gene flow. Such exchange ally,"cow pen") system practiced particularly by Brah-
of genes among populations should yield the same ef- mins enforces exogamous marriage among various clans
fects on mitochondrial lineages as do marriages. The of the Brahmin caste (Basham 1954). Marriages within
Indian caste system is no exception to this rule. It is also and among certain gotras are prohibited, but many
necessary to interpret the results in the context of the Brahmin clans have mixed gotras (Kosambi 1950).
cultural history of Karnataka in particular and southern Members of certain gotras have been known to admit
India in general. In the following, I have attempted to many local inhabitants freely into the Brahminic fold.
show that the genetic and demographic patterns ob- For instance, the Kashyapa gotra is "notorious for being
served by Mountain et al. may be due to (a) what I call, able from early days to absorb . . . aborigines who
"silent gene flow," which has occurred among the three wanted to become Brahmins" (Kosambi 1956, p. 110).
castes since the migration of Haviks into Karnataka, Interestingly, in Maharastra and Karnataka, the highest
and (b) the general influence of the agrarian system that number (82) of clans of Dravira Brahmins (of which
has prevailed in southern India during the past 1,000 Haviks are a part) belong to the Kashyapa gotra (Sher-
years. ring 1872).
Mountain et al. suggest that the offspring of unions
Sociocultural Patterns and Gene Flow between higher and lower castes move to the caste of
The Haviks migrated to the northwestern part of Kar- higher status; thus the progeny will always remain in
nataka -1,300 years ago and settled in the Havik and the higher caste, which may have also added to the pop-
the Haiga provinces, which were part of the Kadamba ulation growth of Brahmins. The religious rules among
dynasty (Moraes 1931). The founders of the Kadamba the Karnataka Brahmins allow this pattern of mobility
dynasty, who apparently allowed the Havik migration only in very rare and exceptional circumstances, such
into Karnataka, generally have been considered Brah- as rape and abduction (Nanjundaiah and Iyer 1928).
mins, but it is not unlikely that some of them belonged Instead, the Manu's laws were upheld, in principle, by
to the local aboriginal group, the "Nagas." After the all the dynasties that ruled over this region (Stein 1980).
migration of Brahmins, however, even Dravidians, who Outside the gotra system, however, the religious codes,
constitute the bulk of the ethnic groups in southern In- including the laws of Manu, advocated two other sys-
dia, were received into the Brahminic fold (Moraes tems of marriage for the Brahmin communities; they are
1931). The Brahmins of medieval southern India moved (1) hypergamy (anuloma, meaning "in the direction of
from place to place but were "not always descendants hair") and (2) hypogamy (pratiloma, meaning "against
of pure Brahmins" (Stein 1980). Conflicts between the the direction of hair"). In the former system, progeny
kings of the Kadamba dynasty and the kings who de- resulting from the union between a father from the
scended from the local aboriginal castes are described higher caste and a mother from the lower caste assumed
in the literature. For instance, around 1007-1008 A.D. a the mother's caste. For example, until recently, the Nam-
major southern Indian ethnic group called the "Cholas" budiri Brahmins of Kerala married the Nayar women,
invaded the Kadamba country and "slaughtered even who belong to the fourth varna (Shudra caste group),
women, children, and Brahmans and taking their girls and they also maintained a matrilineal system. By con-
to wife destroyed their castes" (Moraes 1931, p. 95). trast, in hypogamous marriage, children resulting from
The words "Cheras" (Kerala; Stein 1980) and "Cholas" the union of a Brahmin mother and a Shudra father
are the Dravidian equivalent of "Nagas" (Nanjundaiah produced children whose status was lower than that
and Iyer 1928). The Cheras inhabited the western parts of either parent-a Chandala, who is an untouchable
of medieval southern India (Pillai 1969). Note that the (Basham 1954; Sagar 1975). A majority of the villages
Kadar caste included in the study represents one of the donated to Brahmins during the medieval times, known
Letters to the Editor 1245

as "Brahmadeya villages" (or land grants to Brahmins), ers of the Brahmadeya lands, the actual cultivation was
were multiethnic settlements consisting of cultivators done by the lower castes (Kosambi 1956; Stein 1980;
and members of other castes, including the untouchables Shivanna 1983; Madhava 1991), including the untouch-
(the ideal Brahmadeya village had 18 castes). In these ables. This practice changed gradually after the rebellion
communities, members of multiple castes lived together, of farmers, beginning in the 16th century, perhaps due
albeit in different quarters of the village complex (Stein in part to the demand created by the European spice
1980; Madhava 1991). Thus, in every way the Brahma- traders (Madhava 1991). Therefore, Stein (1980, p. 68)
deya villages were "woven into the texture of rural soci- observed that, although Brahmins effectively dominated
ety," in which Brahmins, in general, were not only ethni- the locals without cultivating skill or experience, "peas-
cally indistinguishable from their peasant contem- ant agriculture in south India was responsible for the
poraries but were their neighbors (Stein 1980, p. 345)- overwhelmingly large source of wealth to the society at
a pattern still found intact in many Karnataka villages. all levels." This agrarian society, supported by various
In some medieval villages, however, a section was exclu- feudal systems that prevailed in medieval southern India,
sively reserved for individuals resulting from mixed may have favored a disproportionate distribution of re-
unions (Stein 1980). Nanjunadaiah and Iyer (1928) sources to the landowners, which in turn may have
stated that, although rules set out by various dharma added to their population growth. Clearly, population
shastras for the Brahmins prohibited illegitimate unions, growth of the Haviks may be interpreted as an overall
"polygamy and prostitution continued to exist.... outcome of a more complex agrarian system that pre-
The wife guilty of adultery was generally repudiated, vailed in southern India, rather than as the result of
and expulsion from caste usually followed" (Nanjun- efforts of any particular caste under consideration.
daiah and Iyer 1928, p. 356). These patterns suggest Availability of excess resources is widely understood to
that, although Brahmins and the untouchables are sepa- result in population growth in agrarian societies, as also
rated by several caste levels in the hierarchical society, was noted by Kosambi (1956, p. 106), with special refer-
as Mountain et al. have correctly noted, opportunities ence to Indian society: "as soon as people change from
existed, in the Brahmadeya villages and elsewhere in food gathering to production, they breed rapidly, and
medieval south India, for the collapse of these levels. improved food supply leads to more children and more
This may have facilitated the movement of individuals maturity and longevity."
from one caste to the other, through the narrow corri- In conclusion, I believe that the genetic grouping
dors that connected the Indian caste society. found by Mountain et al. appears to be more consistent
If 20 years is assumed to be generation length, as with the sociocultural history of the three castes than
suggested by Mountain et al., -50-70 generations have Mountain et al. suggest. The nuclear DNA sequences
elapsed since the immigration and subsequent expansion among Indian castes, which they wish to examine in the
of the Haviks in this region of Karnataka. As outlined future, should reveal patterns at least as homogenizing
above, the prevailing social, political, and cultural con- as those reported in their paper. In general, agrarian
ditions during the past 1,000-1,250 years may have systems that prevailed in southern India may have influ-
provided enough opportunities for the three castes to enced the pattern of population growth of the members
experience the magnitude of gene flow that the genetic of various caste groups.
analysis suggests. Although the magnitude of gene flow Finally, it has been remarked that "India is a learning
appears to be less than the suggested 3%, perhaps it laboratory for linguists, a museum for ethnographers
continued to occur in various forms and in variable and anthropologists, a treasure trove for archeologists,
quantities over time, as suggested by Slatkin (1987). a nightmare for epidemiologists . . . India pulsates, vi-
Such occasional yet ongoing gene flow must be contrib- brates, scintillates with such a plethora of human, ani-
uting toward the observed homogeneity among these mal, botanical, insect and divine life that no camera or
populations. Indeed, the results obtained by Mountain . . . pen, or cassette can fully capture" (Wolpert 1991,
et al. appear to be compatible with the observations of p. 3). Is India a perpetual puzzle for evolutionary geneti-
Stein (1980, p. 5), who in his extensive work on southern cists?
Indian agrarian societies, stated that "there is no validity DIDDAHALLY R. GOVINDARAJU
to the claim of a primordial ethnic separateness of Brah- Department of Genetics
mans from most others of the southern peninsula." Yale University School of Medicine
Population Growth and the Agrarian System New Haven
Mountain et al. also imply that the Haviks may have
been engaging in agriculture for the past 2,000 years, References
which may have contributed to their population growth. Basham AL (1954; reprint 1991) The wonder that was India.
On the contrary, although many Haviks were the own- Rupa, Calcutta
1246 Letters to the Editor

(1989) (Zysk KG [ed/ann]) The origin and develop- ancestral polymorphism (mtDNA lineages that arose
ment of classical Hinduism. Beacon, Boston prior to the separation of populations). Govindaraju ac-
Kosambi DD (1950) On the origin of Brahmin gotras. J Asiatic knowledges these two possibilities, then proceeds to ig-
Soc (Bombay) 26:21-80 nore the latter possibility, stating that our data indicate,
(1956) An introduction to the study of Indian history. "some level of gene flow between the two populations."
Popular Book Depot, Bombay
Madhava KGV (1991) Western Karnataka-its agrarian rela- We believe, however, that gene flow is only one of sev-
tions 1500-1800 A.D. Navrang, New Delhi eral possible interpretations of the observed data. While
Moraes GM (1931) The Kadamba Kula. BX Furtado & Sons, our data do not eliminate the possibility of gene flow,
Bombay particularly in the more distant past, they do suggest
Mountain JL, Hebert JM, Bhattacharyya S, Underhill PA, Ot- that high levels of recent gene flow are unlikely: we
tolenghi C, Gadgil M, Cavalli-Sforza LL (1995) Demo- observe little sharing of similar mtDNA types by Havik
graphic history of India and mtDNA-sequence diversity. Am and Mukri individuals (Mountain et al. 1995).
J Hum Genet 56:979-992 In our article we stated that the Havik have engaged
Nanjundaiah HV, Iyer LKA (1928) The Mysore tribes and in agricultural practices for hundreds or even thousands
castes. Mysore University, Mysore of years. We certainly did not imply that the Havik
Pillai KK (1969) A social history of the Tamils. Part 1. Univer- always worked the land themselves, as Govindaraju sug-
sity of Madras, Madras
Sagar SL (1975) Hindu culture and caste system in India. Up- gests. Such activity would make it difficult or impossible
pal Book Store, Delhi for the Havik to benefit maximally from an agrarian
Sherring MA (1872; reprinted 1974) Hindu tribes and castes, system. The Havik would be likely to gain more from
as represented in Banares. Vol 1. Cosmo, Delhi the practice of agriculture if they employed members of
Shivanna KS (1983) The agrarian system of Karnataka (1336- lower-ranking caste groups to carry out the cultivation.
1761). Prasaranga, Mysore University, Mysore This has been the practice, especially for crops such as
Slatkin M (1985) Rare alleles as indicators of gene flow. Evolu- rice (Campbell 1883). There is evidence, however, that
tion 39:53-65 the Havik themselves, both priests and laity, did carry
(1987) Gene flow and geographical structure of natural out the cultivation of a few special crops, including
populations. Science 236:787-792 spices and betel nuts. Buchanan (1870) noted such activ-
Stein B (1980) Peasant state and society in medieval south ity in 1801, as did Campbell (1883) later in the century.
India. Oxford University Press, Delhi We concluded (Mountain et al. 1995) that the
Wolpert AS (1991) India. University of California Press, Berke-
ley mtDNA sequences of two Indian caste groups suggest
that these groups have undergone dramatically different
Address for correspondence and reprints: Dr. Diddahally R. Govindaraju, demographic histories. While there may have been low
Department of Genetics, 333 Cedar Street, I-310 SHM, Yale University, New levels of gene flow between the two groups, in particular
Haven, CT 06510-8055. through extramarital relationships, such gene flow is
© 1995 by The American Society of Human Genetics. All rights reserved.
0002-9297/95/5705-0034$02.00 unlikely to have generated the observed pattern of pair-
wise nucleotide differences. It is more likely that the
differential allocation of resources during the agricul-
tural era enabled the Havik population to increase in
size while the Mukri population remained at a roughly
Am. J. Hum. Genet. 57:1246-1247, 1995 constant size. The observed patterns of mtDNA se-
quence diversity are consistent with such demographic
Reply to Govindaraju histories.
JOANNA L. MOUNTAIN,1 JOAN M. HEBERT,2
To the Editor: SILANJAN BHATTACHARYYA,3 PETER A. UNDERHILL,2
Govindaraju mentions a number of reasons why the CHRIS OTTOLENGHI,4 MADHAV GADGIL,5
magnitude of gene flow among Indian caste groups AND L. LUCA CAVALLI-SFORZA2
might be higher than we discussed in our recently pub- 'Department of Integrative Biology, University of
lished article (Mountain et al. 1995). As we stated California, Berkeley; 2Department of Genetics,
therein, some gene flow among caste groups may have Stanford University, Stanford; 3Calcutta; 4Paris; and
occurred. Without invoking several unrealistic assump- 5Centre for Ecological Sciences, Indian Institute of
tions, however, we cannot estimate the level of gene flow Science, Bangalore
on the basis of the available genetic data.
Govindaraju misinterprets our article at two points. References
We stated that the intermixing, in an inferred tree, of Buchanan F (1870) Journey from Madras through the coun-
mtDNA lineages from different caste groups can be at- tries of Mysore, Canara and Malabar. Higginbothams, Ma-
tributed either to gene flow among the groups or to dras

You might also like