Genetic Heterogeneity Among Three Adi Tribes of Arunachal Pradesh, India

Genetic Kinship Among an Isolated Adi Tribe of Arunachal Pradesh:Isonymy in the Adi Panggi

Maji, Suvendu.Krithika, S.Vasulu, T. S.

Human Biology, Volume 79, Number 3, June 2007, pp. 321-337 (Article)

Published by Wayne State University PressDOI: 10.1353/hub.2007.0044

For additional information about this article

Access Provided by University of Toronto Library at 03/02/11 3:32PM GMT

http://muse.jhu.edu/journals/hub/summary/v079/79.3maji.html

http://muse.jhu.edu/journals/hub/summary/v079/79.3maji.html

BOOKCOMP, Inc. — Wayne State University Press / Page 321 / Printer proof / Human Biology 79-3 / Journal

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[First Page]

[321], (1)

Lines: 0 to 29

———9.13582pt PgVar———Normal Page

PgEnds: TEX

[321], (1)

Genetic Kinship Among an Isolated Adi Tribe of ArunachalPradesh: Isonymy in the Adi Panggi

suvendu maji,1 s. krithika,1 and t. s. vasulu1

Abstract Isolated tribes in remote areas are important for genetic studies,and one such little known subtribe of the Adi tribe, namely, the Adi Panggi(Pangi) of the Upper Siang District of Arunachal Pradesh, India, was studiedfor surname distribution to deduce the deviation from random mating and ge-netic kinship between villages. The estimates of homonymy (homozygosity)vary between villages; husbands show wider variation (0.009 to 0.23) thanwives (0.005 to 0.054). The remote villages of Sumsing and Sibum and GekuTown show lower entropy among husbands’ surnames than among Panggiwives. The highest equivalent surname number was found among Sibum hus-bands (9.9), Panggi wives (12.6), and Panggi and non-Panggi wives (13.5).The estimates of unbiased random isonymy among husbands and wives to-gether show the smallest values in Sibum (0.05) and the highest values in Sum-sing and Ramku (0.16). The random and nonrandom components of the in-breeding coefficient show avoidance of inbreeding among the Panggi villages(−0.012 to −0.27) except in Sibum (0.012). Genetic kinship between villagesbased on the Mij distance shows different clusters of villages among husbandsand wives. Both the Panggi wives and the Panggi and non-Panggi wives showa similar pattern of clustering between villages. The wide homonymy varia-tion between villages among the patrilocal Adi Panggi indicates differentialgenetic kinetics among husbands and wives, avoidance of inbreeding, andfemale-oriented differential gene flow with little effect on the overall inter-village genetic kinship.

Surnames are a part of the cultural heritage of a community and provide insightsinto the biological aspects of the ethnic groups. Early attempts to study surnamesto investigate inbreeding (in England) and consanguinity (Spanish community)can be traced to George Darwin (1875) and Shaw (1960). The development of aformal method of isonymy analysis (Crow and Mange 1965) and of investi-gating intercommunity relationship (genetic kinship) by isonymy (Lasker 1978)has initiated rapid progress in the use of surnames for various purposes. Sincethese early investigations, isonymy has been used across a variety of populationsin different parts of the world, mostly in European and Latin American countries.

1Biological Anthropology Unit, Indian Statistical Institute, Kolkata, India.

Human Biology, June 2007, v. 79, no. 3, pp. 321–337.Copyright © 2007 Wayne State University Press, Detroit, Michigan 48201-1309

KEY WORDS: RANDOM ISONYMY, EXOGAMY INDEX, ENDOGAMY INDEX, HOMONYMY, HET-ERONYMY, TIBETO-BURMAN POPULATIONS, ADI TRIBES, INDIA.


322 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[322], (2)

Lines: 29 to 38


PgEnds: TEX

[322], (2)

The studies pertain to (1) genetic structure, especially the measurement of devia-tions from random mating, such as inbreeding, extinction of surnames, and effectsof migration; (2) disease prevalence; (3) population markers to measure maritalisonymy; (4) genetic affinity among populations; and (5) patterns of relationshipresulting from geography, language, and association with other genetic traits andfor the study of microevolution. Colantonio et al. (2003) presents a brief reviewon the topic.

In the Indian context there has been a debate over the application of sur-names for the study of inbreeding and genetic kinship. This debate is based onthe premise that some of the assumptions of the isonymy method are not satisfied,especially among several castes ( jati) and subcastes (Kashyap and Tiwari 1980;Kashyap 1980). However, there are several communities (e.g., tribes, some castegroups, and small isolated communities) where isonymy methods can be appliedto infer population structure and genetic kinship (Kashyap 1980). For example,the application of surname analysis has been validated among Ahamadiyyas, areligious community of Kashmir (Kashyap 1980; Kashyap and Tiwari 1980); theYanadi, an aboriginal tribe (Vasulu 1980); and the fishing caste Vedde (Reddy andMalhotra 1995). Surname analysis has also been used successfully to detect socialnetworks (reciprocity) among the Yanadi, Pattusali, and Vedde in Andhra Pradesh(Rao 1995).

Surname structure among Indian populations varies widely. Surnames sym-bolize the cultural and biological dimensions of a population, and their role, func-tion, and distribution differ widely among castes, tribes, and other ethnic groups.For example, among castes isonymous marriages are generally prohibited. In somesocieties the surnames follow the rules of village exogamy; in some others theyreflect clan exogamy. In tribes and in some castes surnames play an importantcultural role as group identity, as a way to trace common origin and descent, andas a way to identify marriage and kinship ties (Das 1953). In these populations sur-names satisfy the assumption of stability and commonality and therefore can befruitfully used as a genetic trait to study the population genetic aspects. In general,tribal populations reflect stable surname distribution, and surnames are inheritedthrough the paternal line and therefore are more suitable to investigate the geneticstructure and the microevolutionary process. Surnames can provide genetic in-sights into population structure, especially among remote tribes where hardly anydetailed cultural and biological studies are available or have been attempted.

The validity of isonymy methods has been considered in this study in a smallisolated tribe, namely, the Adi Panggi (Pangi), a little known subtribe of the Adi(Abor) tribal cluster in Arunachal Pradesh. In the absence of biological studiesamong the remote tribes, surname analysis can be used to investigate deviationsfrom random mating, especially inbreeding, and intervillage migration and to inferthe possible roles of gene flow and genetic drift in different villages.

Adi tribes consist of several subtribes. These subtribes inhabit different alti-tudes of the southeastern part of the Himalayan Mountains along the Siang Rivervalley in the central region of Arunachal Pradesh in northeastern India. In general,


Isonymy in the Adi Panggi / 323

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[323], (3)

Lines: 38 to 56

———-0.50598pt PgVar———Normal Page

PgEnds: TEX

[323], (3)

an Adi tribal village consists of a group of families belonging to a few specificclans living together at different locations on the mountain slopes and valleys.

The Adi Panggi are one of the smallest isolated subtribes located in the up-per mountain ranges of the river. They reside in seven villages and number about4,000 individuals. Although they share a common historical migration, possiblecommon origin, and linguistic and cultural affiliation with other Adi subtribes,they form a separate group and maintain their identity, which can be recognizedby their geographic isolation, high endogamy, and vivid surname and clan struc-ture, which is different from other Adi subtribes (Koley 2005). Like other tribesin India, they are patrilocal, patrilineal, and patriarchal. Sons tend to stay in thevillage and inherit their father’s surname, whereas daughters migrate to their hus-band’s village after marriage and adopt their husband’s surname. In the Adi Panggithe marital alliance between groups is generally maintained by female exchangeamong different clans with the same or different surnames, leading to the possibil-ity of isonymous marriages, that is, marriages with repeated occurrence of a pair ofthe same surname. Isonymous marriages are suitable for investigating populationstructure and its genetic implications.

There are hardly any studies related to culture, biology, or genetics of thisisolated subtribe of the Panggi, possibly because of their geographic isolation andinaccessibility, which pose problems for communication and contact (Singh 1994).The few studies on the Panggi are studies of ABO blood group polymorphism(Bhattacharjee 1954; Krithika et al. 2006), cultural aspects related to puberty, mar-riage, and childbirth (Sarma 1960), anthropometric variation (Roy 1966), and arecent ethnographic study (Koley 2005).

Some of the Panggi villages are remotely located, and a few villages arelocated near the town. As a result, intertribal marriages are more common amongthe villages near the town, and the remote villages have high endogamy. This situ-ation is expected to be reflected in the intervillage surname distribution; especially,the remote villages are expected to show more genetic homogeneity and the leastvariation in surname distribution, and the villages near the town are expected toshow more surname variability and greater migration, possibly through femalesrather than males. We have examined these expectations in the Panggi using theisonymy method to reflect the intervillage marital relationships, to detect the extentof inbreeding and the extent of genetic interrelationship between villages, and thusto validate the use of isonymy methods in the Adi Panggi tribe.

Materials and Methods

Subjects. Demographic data on and blood samples from the Adi Panggi tribalpopulation were collected from six villages in Geku Circle (Figure 1) for a molecu-lar population genetic study. The surname data consisted of 154 husbands and 130wives. Some villages are remote, and a few are near the newly established town,Geku Town. Of the six villages, Sumsing and Sibum are remotely located about15–30 km away from Geku Town, whereas the remaining three villages (Ramku,


324 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[324], (4)

Lines: 56 to 90


PgEnds: TEX

[324], (4)

Figure 1. Upper Siang District of Arunachal Pradesh showing areas of different Panggi villages (ar-rows) along Siang River valley.

Kumku, and Peram) are located close to Geku Town. The village-wise distributionof occurrence of surnames among husbands and wives and its frequency are shownin Table 1.

The marriages among the Panggi are contracted within the group more sothan in the remote villages. However, in recent years, because of improvementsin education, infrastructure, and developmental activities, the marriage patternamong the villages close to the town has changed. This change is prominent amongthe educated and employed residing in the newly developing towns. Thus somemarriages with other Adi subtribes of non-Panggi wives are observed in GekuTown. To take this into account, we considered three separate data sets: (1) hus-bands’ (Panggi) surnames (18), (2) Panggi wives’ surnames (22), and (3) Panggiand non-Panggi wives’ surnames (37).

Among the studied villages, Ramku and Kumku are small and located closeto one another; there is more frequent marital exchange with Geku Town than withother far-off settlements. Therefore we have reorganized the village data into twosets: (1) all six villages and (2) Sumsing and Sibum (two remotely located villages)and the Geku cluster (the three villages near Geku Town).

Isonymy Analysis. Random isonymy (I) was calculated from the frequencydistribution of the abundance of each surname separately in males (pi) and females(qi) and for both male and female surname pairs together in each (ith) village. If pi



1234567891011121314151617181920212223242526272829303132333435363738394041424344

[325], (5)

Lines: 90 to 90

———* 516.0pt PgVar

———Normal Page

PgEnds: TEX

[325], (5)

Tabl

e1.

Tota

lNum

ber

ofSu

rnam

esA

mon

gH

usba

nds

and

Pang

gian

dN

on-P

angg

iWiv

esin

Five

Adi

Pang

giV

illag

es

Hus

band

sW

ives

Non

-Pan

ggiW

ives

Hus

band

s/W

ives

Hus

band

s/N

on-P

angg

iWiv

es

Num

ber

ofN

umbe

rof

Num

ber

ofN

umbe

rof

Num

ber

ofVi

llag

eor

Tow

nSu

rnam

esF

requ

ency

Surn

ames

Fre

quen

cySu

rnam

esF

requ

ency

Surn

ames

Fre

quen

cySu

rnam

esF

requ

ency

Gek

uTo

wn

933

624

428

1257

1661

Pera

m8

2310

186

2410

4116

47R

amku

312

911

314

923

1226

Kum

ku3

175

151

167

328

33Su

msi

ng11

4111

352

3714

7616

78Si

bum

1228

1427

128

1755

1856

Tota

l15

413

014

723

427

5


326 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[326], (6)

Lines: 90 to 124


PgEnds: TEX

[326], (6)

and qi are the frequencies of a particular surname in males (husbands) and females(wives), respectively, then the random isonymy estimates for males, females, andboth males and females are

Im =∑

p2i , (1)

If =∑

q2i , (2)

Imf =∑

piqi, (3)

respectively.Because the random isonymy is biased for sample size variations, Releth-

ford suggested using unbiased random isonymy and its variance estimates (Barraiet al. 1989). We calculated both the unbiased random isonymy and the varianceestimates of random isonymy for the Panggi population, as suggested by Releth-ford (1988). If Ni and Nj are the number of individuals in populations i and j,respectively, then the unbiased random isonymy (Iij ) is defined as

Iij =∑

nisnjs

NiNj

, (4)

where nis and njs are the numbers of individuals with surname s in populations iand j, respectively, and summation is over all surnames.

An approximate variance of I was estimated as per Barrai et al. (1989).From the distributions of husbands’ and wives’ surnames we also estimated thehomonymy (homozygosity) and heteronymy (heterozygosity) within and betweenvillages.

Based on random isonymy, we estimated the Rij distance to investigate in-tervillage genetic affinities as a result of differential marital migration of husbandsand wives between the studied villages and for (the combined data of) the Gekucluster. The distance is defined as

Rij =∑

Si1Si2

2N1N2, (5)

where Si1 and Si2 are the number of surname frequencies in any two villages 1 and2 and N1 and N2 are the total number of surnames in the corresponding villages.

Based on the proportion of surnames (Pi), we estimated migration distancesMij and Rij between villages. We also estimated the Euclidean distance (Eij ) be-tween villages according to the method of Cavalli-Sforza and Edwards (1967).Based on the Mij , Rij , and Eij distances, we constructed two different phylo-genetic trees—the tree calculated using the unweighted pair group method witharithmetic mean (UPGMA) and the neighbor-joining tree—using Mega 3.0 soft-ware. Because the Rij and Eij distance measures show more consistent phyloge-



1234567891011121314151617181920212223242526272829303132333435363738394041424344

[327], (7)

Lines: 124 to 194


PgEnds: TEX

[327], (7)

netic trees of interrelationship between Adi Panggi villages, we discuss the resultsbased on the Rij distance measure only.

We also calculated entropy, the equivalent surname number (ESN), and re-dundancy as defined by information theory (Shannon and Weaver 1963). Theequivalent surname number, also known as the effective surname number, definesthe number of surnames in a sample with entropy H, where each surname is rep-resented once (Barrai et al. 1989, 1991).

Inbreeding Coefficient. The Crow and Mange model (1965) allow us to esti-mate the average coefficient of inbreeding of the population using the frequencyof isonymous unions:

P(r) =∑

piqi, (6)

where pi and qi are the frequencies of the ith surname in husbands and wives,respectively. The random component F(r), the structural consequence of the pre-vious genetic history of the population (such as genetic drift, migration), can beestimated using

F(r) = P(r)

4. (7)

The nonrandom component F(n), a contemporary event that is linked to the pres-ence of phenomena encouraging or not encouraging isonymy, is calculated as

F(n) = [P − P(r)]

4[1 − P(r)]. (8)

The total kinship coefficient of total inbreeding (F) corresponds to the averageprobability of a couple carrying two identical alleles by descent. The total inbreed-ing coefficient is defined as

F = F(r) + [1 − F(r)]F(n) (9)

(Dyke et al. 1983).

Results

Prevalence of Surnames Among the Panggi. The village-wise distribution ofoccurrence of surnames among husbands and wives (Panggi and non-Panggi) isshown in Table 1. In Sumsing there are 11 surnames shared by 41 husbands and 35wives (P), an average of 3.7 and 3.1, respectively, per surname. In both husbandsand wives Sumsing shows a maximum of 14 surnames shared by a maximum of76 spouses. Ramku and Kumku show the least number of surnames because thevalues are based on a sample of the village, whereas in other villages most of thecouples have been included. There are 15 non-Panggi surnames from marriages in-volving non-Panggi wives. The occurrence of different surnames among husbands


328 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[328], (8)

Lines: 194 to 216


PgEnds: TEX

[328], (8)

Table 2. Frequencies (%) of Different Surnames Among Husbands and Wives (Panggiand Non-Panggi) in Adi Panggi

Husbands Wives, only Panggi Wives, Panggi and Non-PanggiSurname (N = 154) (N = 130) (N = 147)

Paron 15.58 16.92 14.97Panyang 22.08 29.23 25.85Taku 5.84 10.00 8.84Panggeng 8.44 3.85 3.40Tagi 7.79 3.85 3.40Mongku 10.39 3.08 2.72Tosang 2.60 2.31 2.04Jopir 0.65 0.00 0.00Taying 1.95 2.31 2.04Tatung 1.30 1.54 1.36Ejing 8.44 3.85 3.40Paloh 3.90 3.08 2.72Padun 1.30 5.38 4.76Tateh 3.90 3.08 2.72Tayom 1.30 1.54 1.36Aje 1.95 0.00 0.00Tarang 1.95 2.31 2.04Taruk 0.65 3.85 3.40Teksin – 0.77 0.68Kirom – 0.77 0.68Gete – 1.54 1.36Tangu – 0.77 0.68Non-Panggi (15) – – 11.56

and wives are shown in Table 2. There are 18 different surnames among 154 hus-bands and 22 different surnames among 130 Panggi wives that are distributed in sixvillages. Husbands do not show non-Panggi surnames, whereas 15 surnames arerepresented among non-Panggi wives. Husbands have fewer surnames, whereaswives represent more diversity of surnames. On average, a single surname is sharedby about 8.5% of husbands, whereas the average among wives is about 6%. Thethree most common surnames among husbands are Paron, Panyong, and Mangku,which represent about 47% of the husbands. For wives Panyong, Paron, and Takuare the three most common surnames and occur among 57% of Panggi wives.

Homonymy and Heteronymy (Homozygosity and Heterozygosity). Village-wise estimates of the homozygosity of surnames are shown in Table 3. Ramkushows the highest homozygosity (0.236) among husbands, and the lowest ho-mozygosity (0.0098) is found in Sibum. Kumku shows the highest average ho-mozygosity for both Panggi (0.095) and non-Panggi (0.070) wives. The highestaverage surname frequency (5.67) among husbands is found in Kumku and thelowest average surname frequency (2.33) is found in Sibum. Because the samplesizes for Ramku and Kumku (located near Geku Town) are small, we considered



1234567891011121314151617181920212223242526272829303132333435363738394041424344

[329], (9)

Lines: 216 to 271


PgEnds: TEX

[329], (9)

Table 3. Estimation of Homozygosity (Homonymy) of Husbands’ and WivesSurnames’ in the Adi Panggi

Wives

Village Husbands Only Panggi Including Non-Panggi

Geku Town 0.0174 0.0544 0.0245Peram 0.0220 0.0130 0.0052Ramku 0.2361 0.0138 0.0077Kumku 0.1511 0.0951 0.0703Sumsing 0.0168 0.0151 0.0115Sibum 0.0098 0.0062 0.0054Geku cluster 0.0096 0.0140 0.0055

these two settlements together with Peram and Geku Town under the Geku cluster,along with the two remote villages, Sumsing and Sibum, and the homozygosityestimates were compared separately for males and females. Sumsing shows thehighest homozygosity for both husbands and wives.

Entropy and Equivalent Surname Number. The entropy of surname distri-bution for the three data sets is shown in Table 4. In Sibum and Sumsing bothhusband and wives (Panggi and non-Panggi) show higher entropy values (2.8–3.8) than in other villages. In general, the non-Panggi wife sample shows higherentropy (1.8–3.8) than Panggi wives (except in Sibum). The distribution of wives’surnames, which include Panggi and non-Panggi surnames, shows the highestlevel of entropy (3.8035) in Peram, followed by 3.7534 in Sibum. In the groupeddata the Geku cluster shows higher entropy than the village-wise data, and it isequal to the values observed in the two remote villages (Sibum and Sumsing). Theequivalent surname number, or effective surname number, in the Panggi sampleis also shown in Table 4. The two remote villages, Sibum and Sumsing, show a

Table 4. Entropy and Equivalent Surname Number of Husbands’ and Wives’ Surnamesin the Adi Panggi

Entropy Equivalent Surname Number

Wives, Wives, Including Wives, Wives, IncludingVillage Husbands Only Panggi Non-Panggi Husbands Only Panggi Non-Panggi

Geku Town 2.8769 0.5947 2.5706 7.3459 1.5101 5.9406Peram 2.7029 0.9416 3.8035 6.5113 1.9207 13.9627Ramku 0.8167 0.9319 3.2497 1.7614 1.9078 9.5116Kumku 1.2533 0.4693 1.7988 2.3839 1.3844 3.4793Sumsing 2.8555 2.9726 3.1693 7.2377 7.8495 8.9964Sibum 3.3154 3.6619 3.7534 9.9549 12.6576 13.4864Geku cluster 3.1733 2.9792 3.7312 9.0214 7.8856 13.2802


330 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[330], (10)

Lines: 271 to 324


PgEnds: TEX

[330], (10)

Table 5. Estimation of Isonymy (Imf ) Between Husbands and Wives and Its Variance inthe Adi Panggi

Isonymy Between Husbands and Wives

Village Isonymy (Imf ) Variance

Geku Town 0.1262 0.0007Peram 0.1304 0.0355Ramku 0.1666 0.0012Kumku 0.0549 0.0171Sumsing 0.1658 0.0514Sibum 0.0555 0.0082Geku cluster 0.1505 0.0508

higher equivalent surname number among husbands, Panggi wives, and Panggiand non-Panggi wives in all six locations except for Panggi and non-Panggi wivesin Peram. In the pooled data of the Geku cluster the values are about the same asthat of Sibum.

Random Isonymy Between Husband and Wife and Its Variance. Within-village random isonymy (between husbands and wives) Imf and its variance esti-mates in the Panggi villages are shown in Table 5. The values range from 0.05 to0.16. The low variance in the studied villages and Geku Town is due to the leastnumber of rare surnames, which include non-Panggi wives. The lowest varianceis observed in Geku Town (0.0007).

Exogamy and Endogamy Index. Based on the birthplaces of the studied cou-ples, we calculated the exogamy and endogamy index (Table 6). A higher exogamyindex indicates higher marital migration, which influences the surname distribu-tion as well. A higher marital exogamy index, as observed among wives as wellhusbands, is in accordance with expectation among such a patrilocal society asthe Panggi. Exogamy values for husbands, wives, and husbands and wives werehighest in Geku Town and lowest in Sibum. Peram also shows lower husbandexogamy.

Estimation of Inbreeding Coefficient. The observed isonymous marriagesand estimates of the random and nonrandom components of the inbreeding coef-ficient are shown in Table 7 and are depicted in Figure 2. The highest observedisonymous marriage frequencies are found in Sibum, and isonymous marriagewas absent in Geku Town and Kumku. The nonrandom component of inbreedingshows negative values among all the Panggi villages except Sibum. This suggestsavoidance of inbreeding within the community. The nonrandom component of in-breeding was highest in Sibum and Geku Town. The total inbreeding coefficientestimated using the isonymy method is low in all the villages except Sibum, where



1234567891011121314151617181920212223242526272829303132333435363738394041424344

[331], (11)

Lines: 324 to 365


PgEnds: TEX

[331], (11)

Table 6. Exogamy and Endogamy Index of the Adi Panggi

Both Husband from Wife from Both ExogamyHusband and Village and Village and Husband and and

Total Wife from Wife from Husband from Wife from EndogamyVillage Marriages Same Village Another Village Another Village Other Villages Index

Geku Town 33 42.42 39.39 6.06 12.12 0.5349Peram 23 78.26 13.04 8.70 0.00 0.1220Ramku 13 53.85 15.38 15.38 15.38 0.4444Kumku 17 82.35 5.88 11.76 0.00 0.0968Sumsing 42 76.19 9.52 11.90 2.38 0.1507Sibum 28 89.29 10.71 0.00 0.00 0.0566Geku cluster 86 61.63 22.09 9.30 6.98 0.2932

the higher inbreeding coefficient (0.0237) is contributed by positive values of thenonrandom component.

Genetic Affinity Between Villages Among Husbands and Wives. In patrilo-cal societies the pattern of marital migration differs among wives and husbands.This nonrandom pattern influences the genetic relationships and can be inferredfrom the surname distribution among females and males. The migration distanceanalysis (Mij ) based on the surname distribution reflects the genetic relationshipbetween villages among males and females. The patterns of the relationship basedon the Rij and Mij distances are shown in Figures 3 and 4. The patterns reflect thegenetic relationships as a result of marital migration between villages among hus-bands, Panggi wives, and Panggi and non-Panggi wives separately. The clusteringpattern obtained from the neighbor-joining tree differs from that obtained from theUPGMA tree. Ramku and Kumku are separated from the other villages. Sibum,

Table 7. Estimation of Random and Nonrandom Component of Inbreeding Based onIsonymy in the Adi Panggi

Random Isonymy

Inbreeding Coefficient

Isonymous Random Nonrandom TotalObserved Marriages by Component, Component, Inbreeding

Village Isonymy (P) Random Mating P(r) F(r) F(n) Coefficient (F)

Geku Town 0.0000 0.1263 0.0316 −0.0276 0.0049Peram 0.0455 0.1304 0.0326 −0.0185 0.0147Ramku 0.0769 0.1667 0.0417 −0.0187 0.0237Kumku 0.0000 0.0549 0.0137 −0.0130 0.0009Sumsing 0.0513 0.1659 0.0415 −0.0239 0.0186Sibum 0.1071 0.0556 0.0139 0.0122 0.0259Geku cluster 0.0238 0.1505 0.0376 −0.0269 0.0117


332 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[332], (12)

Lines: 365 to 373


* PgEnds: PageBreak

[332], (12)

Figure 2. Inbreeding coefficient: observed, random, and nonrandom isonymy. GT, Geku Town; PR,Peram; RK, Ramku; KM, Kumku; SS, Sumsing; SB, Sibum; G com, Geku cluster.

Geku Town, and Sumsing cluster together, indicating close genetic affinity amongmales, which is a result of intervillage migration of males. For the wives (Panggi,and Panggi and non-Panggi samples) the clustering pattern between villages isalso similar, with Kumku separated from the rest and with Geku Town, Sibum,and Sumsing clustering together.

DiscussionIn many human societies surnames play an important role in social organi-

zation, marriage regulation, and so on. Surnames are group specific, denote closerkinship, and are passed on through generations. As such, surnames provide po-tential information about the population structure and can be used to deduce thegenetic structure of populations (Crow and Mange 1965; Lasker 1978; Yasuda etal. 1974), provided that they satisfy two basic assumptions: (1) that they resem-ble a genetic trait or allele located on the Y chromosome (in patrilineal society)that is stable over generations and (2) that the surname distribution satisfies theexpectation of the neutral theory of evolution (Kimura 1983).

In this regard, tribal populations in India generally have a well-organizedsocial structure (e.g., Das 1953) in which clan, surname, and so on describe the



1234567891011121314151617181920212223242526272829303132333435363738394041424344

[333], (13)

Lines: 373 to 395


PgEnds: TEX

[333], (13)

Figure 3. Dendrograms based on the Rij distance showing the clustering of six villages. GT, GekuTown; PR, Peram; RK, Ramku; KM, Kumku; SS, Sumsing; SB, Sibum.

substructuring of the population. This substructure plays a vital role in marriageregulation (Reddy and Malhotra 1995), social organization [which may be relatedto network relationship (Rao 1995)], intertribal warfare [e.g., the formation ofsubtribes as a result of fission in the Adi (Lego 2006)], resource sharing, hunt-ing strategies, shifting cultivation or farming, and so on. The clan and surnamesare regulated through marriage or kin migration, warfare, and so on [e.g., amongthe Xavante and Yanamamo Amerindian tribes (Salzano et al. 1967; Neel andChagnon 1968)], and their genetic consequences can be investigated by isonymyor surname analysis.

In the Adi tribal cluster a clan consists of a set of surnames, and a set of clansrepresents a subtribe. The Adi subtribes are group specific, indicating a commonancestor, and can trace their descent from that common ancestor several genera-tions back. These ancestors can be inferred from the Adi subtribes’ oral folkloretradition (e.g., Blackburn 2004). They are also reflected in the nomenclature ofthe subtribes’ surnames, specifically, their unique tradition of naming the eldest


334 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[334], (14)

Lines: 395 to 406

———1.144pt PgVar———Short Page

PgEnds: TEX

[334], (14)

Figure 4. Clustering of six villages (neighbor-joining tree) based on the Mij distance (unbiased ran-dom isonymy for husbands and wives).

son after his grandparents, where some syllables (suffix or prefix) of the name arederived from the names of the grandparents. From this unique way of naming theeldest son, it is possible to navigate the linking of grandparents and to trace thecommon ancestor to more than 15 generations in the past.

However, in some societies the surnames are polyphyletic, and this violatesone of the assumptions of surname analysis. The complexity of caste structureposes serious problems for the application of isonymy analysis to population struc-ture studies (Kashyap 1980). In such cases genetic affinities between subgroupsbased on surname analysis may not correspond with the results obtained fromother genetic markers. This is not the case among the Adi; the subtribes share acommon origin and are differentiated as a result of their fission-fusion popula-tion structure, where genetic drift is supposed to play a major role in their geneticdifferentiation. Therefore the results of the isonymy analysis indicate the geneticaffinities between different villages and differential marital migration of husbandsand wives.

The Adi tribal cluster consists of a couple of little known small subtribesthat are remotely located; thus few anthropological studies have been attempted(Singh 1994; Lego 2005). Among the Adi Panggi, ABO blood groups, anthro-pometric variation, and a few cultural practices were studied 40 to 50 years ago(Bhattacharjee 1954; Sarma 1960; Roy 1966; Singh 1994). These studies do notprovide details about the population structure aspects of the tribe. In this regardsurnames do provide a glimpse of the population structure and the genetic struc-ture of these populations. Both Bhattacharjee’s (1954) study and Krithika et al.’s(2006) recent study on blood groups among the Adi Panggi showed consistentfrequencies of and significant differences in ABO allele frequencies among thesetribes, supporting the Hardy-Weinberg law of stability of gene frequencies. How-ever, the genotype frequencies from these two studies show significant variation



1234567891011121314151617181920212223242526272829303132333435363738394041424344

[335], (15)

Lines: 406 to 416

———* 24.0pt PgVar

———Short Page

PgEnds: TEX

[335], (15)

over a period of about 50 years, although such changes are expected in view ofthe population’s isolation and small size. The differences in random isonymy indifferent villages suggest genetic heterogeneity, which is also reflected in the ABOallele frequencies.

The clustering pattern based on the Mij distance shows wide differences be-tween villages among husbands and wives. These differences are due to differen-tial sex-specific marital migration in the past. To have an idea of recent trends, wecalculated the exogamy and endogamy index based on the birthplaces of couplesin each village. The recent trends also reflect wide heterogeneity between villages.The highest values (0.53 and 0.44) are found for Geku Town and Ramku, which isclose to Geku Town. These values are due to recent migrations of couples movingto the town as a result of occupational and developmental activities. In the remotevillages of Sibum and Sumsing, the values are lower (0.05 and 0.15) because ofhigher village endogamy rates.

One of the expectations of considering the Panggi wives and non-Panggiwives separately was to investigate the influence of sex-specific (female) geneflow on the overall genetic relationship between villages among wives. The twoneighbor-joining trees based on Panggi wives and Panggi and non-Panggi wivesbetween villages show more or less similar patterns. This indicates that overallthere is no significant effect of female-oriented gene flow on the genetic relation-ship between villages among wives. This could be because of low migration ratesof non-Panggi wives. But in due course of time, an increase in marriages involv-ing non-Panggi females is expected to significantly alter the genetic relationshipbetween villages.

The surname analysis among the Panggi tribe shows a negative componentof random inbreeding, which suggests avoidance of close kin marriages in all thevillages except Sibum, one of the oldest and most remotely located villages. Ingeneral, isonymous marriages are supposed to not be allowed in other tribes andcastes. The Panggi, however, show isonymous marriages, with the surnames be-longing to different clans. How they have the same surname in different clans isnot clearly known. However, the Panggi also show random isonymous marriages,although the frequency is low, indicating past events of genetic drift.

Recently, molecular genetic studies based on autosomal, mtDNA and Y-chromosome markers have been helping us to investigate the possible origin, routeof migration, and affinity with other groups. In this regard, the surnames also playan important role. Such molecular genetic studies based on surname analysis havebeen found to be more useful in unraveling past genetic history (e.g., Sykes andIrven 2000). Thus the results of this study will help us to draw samples for anmtDNA and Y-chromosome study to investigate the past genetic history and ori-gins of the Panggi, migration, and affinity with other Tibeto-Burman-speakingpopulations.


336 / maji et al.

1234567891011121314151617181920212223242526272829303132333435363738394041424344

[336], (16)

Lines: 416 to 467


PgEnds: TEX

[336], (16)

Acknowledgments We thank the Government of Arunachal Pradesh and all participantsfor their active cooperation during our fieldwork. Our sincere thanks go to Tozing Tagiand Oyam Panyang (Katan and Geku Circles) for their assistance in data collection. Weacknowledge the support and intellectual output about Adi tribes from R. N. Koley (ARO,Directorate of Research) and Kaling Borang, of Pasighat. We thank the director of the IndianStatistical Institute, Kolkata, for providing financial assistance for this study among thetribes of Arunachal Pradesh.

Received 16 June 2006; revision received 5 January 2007.

Literature Cited

Barrai, I., G. Formica, R. Barale et al. 1989. Isonymy and migration distance. Ann. Hum. Genet.53:249–262.

Barrai, I., C. Scapoli, G. Formica et al. 1991. Isonymy in records of births and deaths in Ferrara. Ann.Hum. Biol. 8:395–404.

Bhattacharjee, P. N. 1954. Blood group investigations in the Abor Tribe. Bull. Dept. Anthropol. 3:51–54.

Blackburn, S. 2004. Memories of migration: Notes on legends and beads in Arunachal Pradesh, India.Eur. Bull. Himalayan Res. 25/26:15–60.

Cavalli-Sforza, L. L., and A. W. P. Edwards. 1967. Phylogenetic analysis model and estimation proce-dures. Am. J. Hum. Genet. 19:233–257.

Colantonio, S. E., G. W. Lasker, B. A. Kaplan et al. 2003. Use of surname models in human populationbiology: A review of recent developments. Hum. Biol. 75:785–807.

Crow, J. F., and A. P. Mange. 1965. Measurements of inbreeding of the frequencies of marriages be-tween persons of the same surnames. Eugen. Q. 12:199–203.

Darwin, G. H. 1875. Marriages between first cousins in England and their effects. J. Stat. Soc. 38:153–184.

Das, T. C. 1953. Social organization of the tribal peoples. Ind. J. Soc. Work 14(3):245–262.Dyke, B., A. V. James, and W. T. Morrill. 1983. Estimation of random isonymy. Ann. Hum. Biol.

10:295–298.Kashyap, L. K. 1980. Trends of isonymy and inbreeding among the Ahmadiyyas of Kashmir. J. Biosoc.

Sci. 12:19–25.Kashyap, L. K., and S. C. Tiwari. 1980. Kinetics of genetic kinship as inferred by isonymy among the

Ahmadiyyas of Kashmir valley. Hum. Biol. 52:311–324.Kimura, M. 1983. The Neutral Theory of Molecular Evolution. Cambridge, UK: Cambridge University

Press.Koley, R. N. 2005. Sociocultural Aspects of Adi Panggi. Itanagar, India: Directorate of Research, Gov-

ernment of Arunachal Pradesh (in press).Krithika, S., S. Maji, and T. S. Vasulu. 2006. Genetic heterogeneity among three Adi tribes of Arunachal

Pradesh, India. Hum. Biol. 78(2):221–227.Lasker, G. W. 1978. Increments through migration to the coefficient of relationship between commu-

nities by isonymy. Hum. Biol. 50:235–240.Lego, N. 2005. History of the Adis of Arunachal Pradesh. Itanagar, India: Ponung Lego.Lego, N. 2006. Modern History of Arunachal Pradesh (1825–2006). Itanagar. India: Ponung Lego.Neel, J. V., and N. A. Chagnon. 1968. The demography of two tribes of primitive relatively unaccul-

turated American Indians. Proc. Natl. Acad. Sci. USA 59:680–689.Rao, A. R. 1995. Reciprocity in marital and social networks: Illustration with Indian data. Hum. Biol.

67:887–904.



1234567891011121314151617181920212223242526272829303132333435363738394041424344

[337], (17)

Lines: 467 to 489

———* 294.36pt PgVar

———Normal Page

* PgEnds: PageBreak

[337], (17)

Reddy, B. M., and K. C. Malhotra. 1995. Relevance of surname in population structure: A study amongthe Vadde, a fishing community of Kolleru Lake in Andhra Pradesh, India. J. Hum. Ecol. 6:161–168.

Relethford, J. H. 1988. Estimation of kinship and genetic distance from surnames. Hum. Biol. 60:475–492.

Roy, S. 1966. Anthropometry of the Adis of the Siang District. Shillong, India: North-East FrontierAgency.

Salzano, F. M., J. V. Neel, and D. Maybury Lewis. 1967. Further studies on Xavante Indians. I. Demo-graphic data on two additional villages and genetic structure of the tribe. Am. J. Hum. Genet.19:463–489.

Sarma, J. 1960. Puberty, marriage, and childbirth among the Pangi and the Minyong Abor women.Anthropos 55:97–113.

Shannon, C. E., and W. Weaver. 1963. The Mathematical Theory of Communication. Urbana: Univer-sity of Illinois Press.

Shaw, R. F. 1960. An index of consanguinity based on the use of Spanish-speaking surnames. J. Hered.51:221–230.

Singh, I. P. 1994. People of India, National Series, v. 3, The Scheduled Tribes. Anthropological Surveyof India. New Delhi: Oxford University Press.

Sykes, B., and C. Irven. 2000. Surnames and the Y chromosome. Am. J. Hum. Genet. 66:1417–1419.Vasulu, T. S. 1980. Demographic Study of a Genetic Isolate: Genetic Implications of Demography of

Yanadi. Mysore: Anthropological Society of India.Yasuda, N., L. L. Cavalli-Sforza, M. Skolnir et al. 1974. The evolution of surnames: An analysis of

their distribution and extinction. Theor. Popul. Biol. 5:123–142.

Documents

Genetic Heterogeneity Among Three Adi Tribes of Arunachal Pradesh, India