ANNALSOF HUMAN BIOLOGY, 2002, VOL. 29, NO. 5, 538±549

Populationstructure and genetic di€ erentiation among the substructured Vysyacaste population in comparison to theother populations of Andhra Pradesh,India

N. Lakshmi , D. A. Demarchi‡, P. Veerraju and T. V. Rao y y y Departmentof HumanGenetics, Andhra University, Visakapatnam, India ‡ Departamentoy de Bioquõ ÂmicaClõ Ânica,Facultad de CienciasQuõ Âmicas,Universidad Nacional de Co Âr- doba, Co rdoba,Argentina Received20 December2000; in revised form 25 September2001; accepted 16 November2001

Summary. Objectives:Thepresent paper focuses on the study of the patterns of genetic microdi€erentiation among one of thesubstructured caste populations of AndhraPradesh, namelyVysya, with referenceto 17 otherTelugu speaking populations from the same region of India. Subjectsand methods :Atotalof 302 individuals from the three Vysya subgroups (101 of AryaVysya, 100 from Kalinga Vysya and 101 from Thrivarnika) were typed in 17 blood groupsand protein polymorphisms. Nei’s gene diversity analysis, as well asneighbour- joiningtree and UPGMA cluster diagrams, derived from standard genetic distances, R- matrixanalysis and a regressionmodel for investigatingthe patterns of external gene ¯ ow andgenetic drift dueto isolation under the island model, were done at two levels:(1) consideringonly the three Vysya populations and (2) considering common loci among 20 populationsof AndhraPradesh. Results:Sevenof the 17 systemsinvestigated were found to be monomorphicamong all the threeVysya groups. The UPGMA tree and bidimensional scaling of the D2 distances derivedfrom R-matrix analysis show averydistinct cluster of Vysya populations. Applica- tionof the model of regressionof averageheterozygosity versus the distance of populations fromthe centroid shows thethree Vysya populations placed as clear outliers above the theoreticalregression line. Conclusions :Di€erent approaches employed in this studygive support to the hypothesis of di€erent origin and/ ordemographicstory for thethree Vysya groups compared with other populationsof AndhraPradesh.

1.Introduction India is aland of enormous human genetic, cultural and linguistic diversity (Roychoudhury, Roy, Dey et al. 2000).India issubdivided by20 major languages, hundreds of dialects, religion, almost innumerable castes and tribes, o€ ering a unique opportunity for the study of the process of subdivision and their evolutionary consequences (Puppala and Crawford 1996).The social structure of the Indian population is governed byhierarchical caste system. In the Hindu caste system, each caste belongs to one of the ®vemain classes/varnas,namely, Brahmins (priests), Kshatriya(warriors), Vyshya(business community), Sudra (rest of the castes) and Pancham (tribes). The caste structure is fairlyrigid, and each caste remains asan endogamous unit, although the levelsof endogamy varysubstantially (Malhotra and Vasulu1993). In India, there arenearly 5000 well-de® ned endogamous populations (Thangaraj,Ramana and LaljiSingh 1999).India is known for its unique population structure, which is characterized bysubdivision of its population into anumber of hierarchical caste, tribe and religious groups. These generalfeatures of population structure applyto each linguistic region of the country, whilelanguage forms a

Annalsof Human Biology ISSN 0301±4460 print/ ISSN 1464±5033 online # 2002Taylor & FrancisLtd http://www.tandf.co.uk/journals DOI:10.1080/ 03014460110114707 Populationstructure amongthe Vysya caste 539 strong barrier to gene ¯owbetween such regions even among the castes of same or similar hierarchy. The State of Andhra Pradesh with its 70million people, organized into several endogamous castes, tribes and religious groups, presents enormous varietyof popu- lations, sociocultural patterns and organization. These castes and tribes arealso known to present extreme variationin size and nature of sub-structuring. Some of the major castes show number of endogamous subdivisions within them but with non-overlapping geographicdistributions. Furthermore, populations of Andhra Pradesh areknown to practise consanguineous marriageswith high rates of village endogamy.This willlead to reduction in e€ ective population size, creating breeding isolates within apparently single endogamous castes and subtribes. This willprovide situations conducive for rapid genetic microdi€ erentiation among such populations. Therefore, study of such substructured populations alongwith the others mayhelp inunderstanding the mechanisms behind the pattern of genetic variationobserved in aparticular region. The present paper focuses on the study of genetic microdi€ er- entiation among one of the substructured caste populations of Andhra Pradesh, namelyVysya, with reference to 17other populations from this Telugu-speaking Andhra Pradesh.

1.1.Population background The Vysyasare third in rank among the four-fold Hindu caste hierarchy, next to Brahmins and Kshatriyas. There arethree endogamous subcastes within them: the AryaVysyas, the KalingaVysyas and the Thrivarnikas.These populations belong to acommon stock, despite their separate caste identities and present sociocultural di€ erentiation (Sherring 1881).Of the three, the AryaVysyas are popularly known asGavara Komatis in Andhra Pradesh and with di€ erent synonyms such asAgarwal, Bania, Gupta, Chetty, etc. in other provinces of the country (Gupta 1988).They are a community of traders and businessmen, strict vegetariansin food habits, practise consanguineous marriagesand aredistributed mostly in towns and cities. The GavaraKomati occupy the highest position in the social hierarchy of the trading communities, followed byThrivarnikas, and KalingaKomati, the latter being regarded asthe lowest in the social scale. The KalingaVysyas live in the old Kalingacountry which extended roughly from Visakapatnamin Andhra Pradesh to contiguous areasin Orissa state. Theyare the second largestsection of Vysyas,their number running into severalthousands. Like the AryaVysyas, Kalinga Vysyas are known for trade and business enterprise as their core activity.Interestingly, they arenon-vegetarian bydiet, but practise con- sanguineous marriagesand worship deities belonging to both Vaishnaviteand Saivitesects, like the AryaVysyas do (Census of India 1981). Thrivarnikascall themselves asThrivarnika Vaishnavites. They claim that their community cameinto existence during 11thcentury ADatthe time of greater Vaishnavareformer Sri Ramanujacharya.There is, however, acontroversy regard- ingtheir origin. Theyfollow Brahminicalcustoms more scrupulously than others do, except that they arenon-vegetarians bydiet. The major economic resource for the Thrivarnikasis trade. Their principal occupation isbusiness in gold, silverand glass. Theyconsider themselves only next to AryaVysyas in social hierarchy. Although economically a‚uent, most of them haveonly informal education. Theyare found mostly in bigtowns and cities inAndhra Pradesh, besides alargenumber of families in Chennai. Their number is said to be in the vicinityof about 15000individuals. 540 N. Lakshmi et al.

The Thrivarnikascan be identi®ed with their facialmarkings known as`namam’. Their namamis U-shaped with white and red markings. Theyare strictly anendo- gamous group. Theyattribute that their religious creed is distinct and in no way similar with Komati. Theydi€ er remarkably with the GavaraKomati inthe obser- vanceof various customary practices. The 17other Telugu-speaking populations for which comparative marker data were availableon seven common polymorphic loci, represent widesocio-economic network of the region. These populations represent upper castes like Brahmins; middle-ranking castes such asReddy, Kamma,Kapu; low-ranking castes such as Pattusali, Jalari,Bestha, Relli, etc.; traditionally untouchable scheduled castes such asMala and Madiga;and tribal group such asPardhi (Gopalam and Rao 1981, Naidu, Mohrenweiser and Neel 1985,K. P.Lakshmi, personal communication, Char,Lakshmi, Gopalam et al. 1989,S. Praveena,personal communication, Ramesh and Veerraju 1992,1996a, b, 2001).

2.Subjects and methods Atotal of 302blood samples werecollected from the unrelated adults of the three Vysyagroups (101of AryaVysya, 100 from KalingaVysya and 101from Thrivarnika) distributed in the three north coastaldistricts of Andhra Pradesh, namelyVisakapatnam, Vizianagaram and Srikakulam. Subjects werechosen only from the respective three municipalities of those three districts, except for Kalinga Vysyasfor whom the samples werealso collected from Bobbili municipality of Vizianagaramdistrict. About 3mL of blood wasdrawn byvenepuncture using disposable syringeinto sterile, labelled test tubes containing EDTA. The samples werestored in athermocol box containing ice and transported to the laboratory on the same day.Plasma and cells wereseparated. Plasmawas stored at 20°Cuntil use. One drop of packed cells wastaken to test A1A2BO and Rh (D) blood¡ groups on the same dayof collection. The remaining cells werehaemolised. The haemolysate wassubjected to electrophor- esis for G6PD and HBthe next dayas the G6PD enzyme isunstable. The remaining haemolysate wasstored at 20°C until use. Red cell enzymes weredetermined ¡ on starch and starch agarosegels. The methods opted for di€ erent systems wereas follows: 6PGD, G6PD, LDH and MDH (Harris and Hopkinson 1978);ACP and ESD(Karp and Sutton 1967);PGM2 and SOD (Spencer, Hopkinson and Harris 1964);glyoxalase (P¯ ugshaupt, Scherz and Butler 1978);haemoglobin (Bhatia1986). HP and CPweretyped bydisc gelelectrophoresis (Clark1964). Alb and Tf weretyped byvertical polyacrylamide gel electrophoresis (Kitchin and Bearn1966).

2.1.Statistical methods Gene frequencies werecalculated bygene counting, except for ABO, and Rh (D), for which maximum likelihood estimates werecalculated using the program MAXIM. Achi-square goodness-of-®t test wasused to test the ®tof the observed genotype frequencies to those expected under Hardy±Weinberg equilibrium. Nei’s gene diversity analysis(Nei 1973)as wellas neighbour-joining and UPGMA cluster diagrams,derived from standard genetic distances (Nei 1972),were accomplished by using NJBAFD program, supplied byDr N.Takezaki.These analyseswere done at two levels:(1) considering only the three Vysyapopulations and the 12polymorphic Populationstructure amongthe Vysya caste 541 genetic loci and (2) considering seven common loci among 20populations of Andhra Pradesh.

2.2.R-matrix analysisand the regression model of Harpending and Ward (1982) D2 distances derived from R-matrix analysis(Harpending and Jenkins 1973)were employed for multidimensional scaling(MDS) (Kruskal 1964),collapsing into two dimensions on which the 20populations areprojected to gaugethe genetic anities. MDSscalingwas accomplished using the NTSYS 1.8package (Rohlf 1993). For gaininginsights into the patterns of external gene ¯owand genetic drift due to isolation under the island model, weemployed the Harpending and Ward (1982) regression model. The model predicts that the averageheterozygosity of the popula- tion (Hi)should be equal to the overallmean heterozygosity of the entire population (in this case,a group of populations from Andhra Pradesh with common recent origin) HT,multiplied by(1 rii), where rii is the genetic distance from gene fre- quency centroid. If gene ¯ow¡from outside the region variesin amount from popula- tion to population, this linear relationship no longer holds. Isolated groups willbe less heterozygous than the linear prediction, hence liebelow the theoretical regres- sion line, whereas populations receivingmore gene ¯owfrom outside willbe more heterozygous and lieabove the line. Thus, the outliers aboveand below the theor- eticalregression line maygive some insights into the population structure.

3. Results 3.1.Genetic polymorphism and di€ erentiation in the substructured Vysya Out of the 17loci tested, 10were polymorphic in atleast one of the three Vysya groups. The allelicfrequency of the 10polymorphic loci is presented in table 1. LDH, MDH, PGM2, SOD, HB,CPand ALBwere found to be monomorphic among allthe three Vysyagroups. Severaldeviations from Hardy±Weinberg equi- librium reach statisticallysigni® cance because of anexcess of homozygotes: among the Aryafor the PGM1 system ( p < 0:05),among the Kalingafor the ACPand GLO ( p < 0:01),and among the Thrivarnikasfor the ESD ( p < 0:05).The South Indian populations, especiallythose from Andhra, areknown for the practice of consan- guineous marriages,amongst the highest in the world (Sanghvi1966, Cavalli-Sforza and Bodmer 1971).Therefore, the observed excess of homozygotes for several systems inthe present context maybe attributable to inbreeding. With reference to the polymorphic loci, the following salient features emerge in comparison to the other Telugu-speaking populations studied earlier (Lakshmi, unpublished results): (1) While alleleA could not be traced among the Thrivarnikas,it is observed with negligiblefrequency among the other two groups. (2) The 6PGD*Callele is completely absent in Thrivarnikas. (3) G6PD de®ciency isobserved only among the KalingaVysyas but with highest frequency (8%) compared with anyother population from Andhra Pradesh. The results of the gene diversity analysisfor the Vysyasare presented in table 2.It isapparent that the levelsof within group diversity arevery low for the studied loci, particularlyfor 6PGD, G6PD, and TF.The largestproportion of variation(about 0.50)is observed atthe ESD, PTC,GLOand ACP1loci. AryaVysyas tend to show slightlyhigher heterozygosity atmost of the loci compared with the other two populations, which is re¯ected in the averageheterozygosity aswell. 542 N. Lakshmi et al.

Table1. Allelic frequencies at 10polymorphicloci among the three Vysya groups fromcoastal Andhra Pradesh.

Locus Allele Arya Kalinga Thrivarnika

ABO A1 0.0303 0.0050 0.0000 A2 0.0104 0.0050 0.0000 B 0.1926 0.1690 0.2357 O 0.7667 0.8210 0.7643 Rh D 0.8593 0.9000 0.9005 d 0.1407 0.1000 0.0995 ACP ACP*A 0.2600 0.2228 0.3850 ACP*B 0.6800 0.7446 0.5700 ACP*C 0.0600 0.0326 0.0450 ESD ESD*1 0.4839 0.6605 0.5500 ESD*2 0.5161 0.3395 0.4500 6PGD 6PGD*A 0.9850 0.9900 1.0000 6PGD*C 0.0150 0.0100 0.0000 GLO GLO*1 0.4780 0.4350 0.3958 GLO*2 0.5220 0.5650 0.6042 PGM1 PGM1*1 0.6000 0.7806 0.7680 PGM1*2 0.4000 0.2194 0.2268 PGM1*3 0.0000 0.0000 0.0052 HP HP*1 0.1414 0.2525 0.2050 HP*2 0.8586 0.7474 0.7950 TF TF*B 0.0000 0.0050 0.0000 TF*C 0.9900 0.9800 1.0000 TF*D 0.0100 0.0150 0.0000 G6PD B 1.0000 0.9184 1.0000 b 0.0000 0.0816 0.0000

Table2. Locus-wise and population diversity estimates along with theaverage values.

Within-groupdiversity Nei’sdiversity estimates

Locus Arya Kalinga Thrivarnika HT HS GST

ABO 0.377 0.298 0.362 0.346 0.344 0.005 Rh 0.243 0.181 0.180 0.201 0.200 0.004 ACP 0.469 0.397 0.527 0.472 0.462 0.021 ESD 0.502 0.451 0.497 0.492 0.481 0.022 6PGD 0.030 0.020 0.000 0.017 0.016 0.005 GLO 0.502 0.494 0.481 0.492 0.490 0.005 PGM1 0.482 0.344 0.361 0.408 0.394 0.034 HP 0.244 0.379 0.328 0.320 0.315 0.013 TF 0.020 0.040 0.000 0.020 0.020 0.006 G6PD 0.000 0.151 0.000 0.053 0.050 0.056 Average 0.287 0.276 0.274 0.282 0.277 0.017

Decomposition of the total diversity ( HT)into within ( HS)and between ( HT HS) components (GST) suggests that only about 1.7%of the total diversity is due¡ to di€ erences between populations whilethe rest is between individuals within popula- tions. Nei’s (1972)standard genetic distances between pairs of Vysyapopulations sug- gest relativelygreater di€ erentiation of the KalingaVysyas from AryaVysyas Populationstructure amongthe Vysya caste 543

Figure1. Regression of average heterozygosity on the gene frequency centroid of the three Vysya populations,based on 10 polymorphic loci.

(0.011)as well as from Thrivarnikas(0.008), whereas the lowest distance is found between these two populations (0.006). The results of application of Harpending and Ward (1982)regression model to the genetic data(® gure 1)suggest that the AryaVysya with arelativelygreater hetero- zygosityis placed abovethe theoretical line asan outlier, suggesting aboveaverage gene ¯owfrom outside the Vysyaswhereas the other two populations liebelow but relativelycloser to the line. However, the distance from the centroid for the Kalinga Vysyasis somewhat higher than for the other two groups, suggesting probably larger e€ ect of stochastic processes.

3.2.Genetic di€ erentiation of Vysyas with reference toother Telugu-speaking populations The comparative dataon 7of the 12loci earlier studied for 17other populations from this region wereused for comparative analysisalong with the Vysyas.Nei’s (1973)gene diversity analysisperformed separatelyfor the three Vysyasand for all the 20groups suggest arelativelylower degree of di€ erentiation in the localVysya populations (GST =0.011 0.003)when compared with the populations of the region asa whole (GST =§0.018 0.002),consistent with the hierarchy of popula- tions involved. § 544 N. Lakshmi et al.

Figure2. Projection of 20 Telugu populations onto the two-dimensional space based on multidi- mensionalscaling of the D2 valuesobtained from R-matrix analysis of seven loci.

In order to gaugethe genetic anities of the Vysyasto other populations in the region R-matrix analysisof Harpending and Jenkins (1973)and the genetic distances D2 obtained from this analysiswere subjected to the multidimensional scaling (Kruskal 1964).After 49iterations, astress of 0.114was achieved which is, according to Kruskal’s criteria, good and interrelations among populations canadequately be reduced to two dimensions. The population relationships thus obtained aredepicted in®gure 2.From this diagramit canbe seen that the Vysyapopulations areclearly di€ erentiated from the rest of Telugu-speaking populations, most of whom are placed closely together with fewexceptions. While the Pardhi is distinctly placed, far removed from allother groups, the middle-ranking Yadavaand Kammaalso lie outside this compact cluster. However, the UPGMA (®gure 3)tree constructed based on the Nei’s standard distances among the 20groups suggest much more rational and clearclusters of populations that arebroadly consistent with known ethnohis- torical and social backgrounds. For example,not only arethe four Vysyagroups forming adistinct cluster, separated from allother groups, but also others like Pardhi and Brahmin, the scheduled castes Malaand Madiga,the lower castes Vanjara,Bestha and Kurma and others like Pattusale, Reddy and Jalari,Gowda, Munnuru Kapu and Mudiraj each form asrational subclusters. Figure4 depicts relationship between mean per locus heterozygosity and the distance from the centroid ( rii)of the 20populations of Andhra Pradesh. Here again,the three Vysyagroups areplaced asclear outliers abovethe theoretical regression line and quite distant from the centroid, suggesting both the e€ ects of gene ¯owfrom outside and of the stochastic processes due to subsequent isolation. On the other hand, Munnuru Kapu, Relli, Kurma, Kammaand Yadavacan be Populationstructure amongthe Vysya caste 545

Figure3. UPGMA tree based on Nei’s standard genetic distances between pairs of 20 Telugu populations. considered asoutliers in the increasing order of magnitude below the theoretical regression line. Most of these groups are,however, placed relativelycloser to the centroid, suggesting greaterisolation but probably recent history of these groups. Among the remaining groups placed abovethe line, the scheduled castes Madigaand Malaand the lower caste Jalariare placed themselves somewhat farther from the line, suggesting aboveaverage external gene ¯ow.However, these groups areplaced close to the centroid, suggesting relativelyless di€ erentiation.

4.Discussion India o€ ers innumerable but unique situations for pursuing research in popula- tion structure and micro-genetic di€ erentiation. The Vysyasof the present study are the localrepresentatives of one of the ®vevarnas of the Hindu caste system, but diverged and evolvedin the course of time into independent endogamous subcastes, the basic units within which evolution must havetaken place for anumber of gen- erations. The extent of di€ erentiation asrevealedby the GST valueis small(0.018) but nevertheless relativelyhigher than the magnitude observed for populations with similar hierarchy of substructure. For example,Reddy and Chopra (1999),studying 10traditional blood group, red cell enzyme and serum protein loci, observed GST valuessomewhat lower (about 1.1%)among the subgroups of ®shermen from coastal areasof Andhra Pradesh and Orissa. However, the GST valueis consider- ablyless when compared with that based on allthe Telugu-speaking groups (0.0203). This mayre¯ ect the history of subdivision of these clusters of groups. However, what 546 N. Lakshmi et al.

Figure4. Regression plot of average heterozygosity versus the distance from gene frequency centroid ofthe 20 Telugu populations and the theoretical regression line. ismore remarkable and worth noting is the fact that these three Vysyagroups form a verydistinct cluster and diverged from the other localgroups much earlier (®gure 3). This mayalso re¯ect probably di€ erent origin and settlement of these groups in this areaas compared with the other groups. It maybe pertinent to note here that in a separate study based on ®nger dermatoglyphic variablesamong the major caste groups of Andhra Pradesh, Reddy and Reddy (1992)observed Vysyas,both male and female samples independently, to haveclearly diverged from the other groups. There areother clusters in this tree that areconsistent with the socio-economic aliations of the groups which maylend further support to the genetic realityin the clustering pattern. The Vysyagroups depend mostly on trade and commerce thus restricted mostly to urban locales with compact group identities and solidarity and with the least possibilities for genetic interaction with other groups. This mayhave givenrise to smaller breeding isolates and greatergenetic isolation with possibilities for rapid genetic di€ erentiation, although their position abovethe theoretical line, as outliers, inthe regression plot of heterozygosity versus rii does not bear testimony to this. Thatthey arefar removed from the centroid in the regression plot nevertheless lends support to surmise if they did not reallyhave had adi€ erent origin or history compared with other populations of Andhra Pradesh. What would be the probable explanation for ahigher degree of heterozygosity observed among the Vysyas,given that their overallpopulation size is much smaller than manyother groups used in this plot? One possibility could be that these Populationstructure amongthe Vysya caste 547 populations aremostly based on urban habitats with greaterpossibilities for mobi- lityeven to distant areas,thus wereable to managelarger e€ ective population sizes compared with most other groups. Although largein the overallpopulation size, it is known that the marriagecontacts arelimited to verysmall distances. Therefore, at the operational levelthe e€ ective sizes must havebeen much smaller in most of these groups. This, instead of the routine interpretation of greaterthan averagelevels of external gene ¯owinto the Vysyas,might havebeen responsible for the observed position of Vysyasin the regression plot. The other plausible explanation could be that since they havebeen diverged considerably from other Telugu-speaking groups with apossibility of distinct origin and migration into this area,the founding popu- lations might havebeen from veryheterogeneous sources. This is not anunlikely proposition giventheir occupational pattern.

Acknowledgements Manythanks to B.M.Reddy for helpful comments and suggestions. D.A. Demarchi is member of the Investigator Careerof the Consejo Nacional de Investigaciones Cientõ®casy TeÂcnicas de laRepu ÂblicaArgentina (CONICET).

References Bhatia, H. M., 1986, Manualof Laboratory Methods in Human Genetics .Organisedby I. C.Vermaat AIIMS (NewDelhi: Indian Council of MedicalResearch Publications). Cavalli-Sforza, L. L., and Bodmer, W. F., 1971, TheGenetics of HumanPopulations (SanFrancisco: W.H.Freeman& Co.), pp.350± 351. Census of India, 1981,Part X-C series 2AndhraPradesh. Survey report on village Bhairavanitippa, Anantapuramdistrict (NewDelhi: Government of IndiaPublications). Char, K. S. N., Lakshmi, P., Gopalam, J., Sastry, J. G., and Rao, P. R., 1989,Genetic di€ erentiation amongsome endogamous populations of Andhra Pradesh, India. AmericanJournal of Physical Anthropology , 78, 421±429. Clark, J. T., 1964,Simpli® ed `disc’ polyacrylamide gel electrophoresis. Annalsof the New York Academy of Sciences, 121, 428±436. Gopalam, K. B., and Rao, P. R., 1981,Genetic studies on Vysyas of Andhra Pradesh, S. India:A1A2BO, Rh(D),transferrin, groupspeci® c component,haptoglobin and pseudocholinesterase types. Acta Anthropogenetica , 5, 175±180. Gupta, T. V. S., 1988, In Vaishyasin India ,editedby K.C.Gupta(Hyderabad: All India Vaish Samaj), pp. 9±10. Harpending, H., and Jenkins, T., 1973,Genetic distances among Southern African populations. In Methodsand Theories of Anthropological Genetics ,editedby M. H.Crawford andP. L. Workman(Alburqerque, NM: Universityof New Mexico Press), pp.177± 199. Harpending, H. C., and Ward, R. H., 1982,Chemical systematics and human populations. In BiochemicalAspects of Evolutionary Biology ,editedby M. H.Nitecki(Chicago: University of ChicagoPress), pp.213± 256. Harris, H., and Hopkinson, D. A., 1978, Handbookof Enzyme Electrophoresis in Human Genetics (Amsterdam:North Holland). Karp, G. W. Jr, and Sutton, H. E., 1967,Some new phenotypes of human red cell acid phosphatase. AmericanJournal of HumanGenetics , 19, 54±62. Kitchin, F. D., and Bearn, A. G., 1966,The electrophoretic patterns of normal and variant phenotypes ofthe Group Speci® c Components(Gc) in human serum. AmericanJournal of Human Genetics , 18, 201±214. Kruskal, J. B., 1964,Nonmetric multidimensional scaling: a numericalmethod. Psychometrika , 29, 115± 129. Malhotra, K. C., and Vasulu, T. S., 1993, In HumanPopulation Genetics ,editedby P. P.Majumder (NewYork: Plenum Press), pp.207± 233. Naidu, J. M., Mohrenweiser , H. W., and Neel, J. V., 1985,A serobiochemicalgenetic Study of Jalari andBrahmin caste populations of Andhra Pradesh, India. HumanHeredity , 35, 148±156. Nei, M., 1972,Genetic distances between populations. AmericanNaturalist , 106, 283±292. Nei, M., 1973,Analysis of gene diversity in subdivided populations. Proceedingsof theNational Academy ofSciences,USA , 70, 3321±3323. 548 N. Lakshmi et al.

Pflugshaupt, R., Scherz, R., and Butler, R., 1978,Human red cell glyoxalase I polymorphismin the Swiss population.Phenotype frequencies and a simpli®ed technique. HumanHeredity , 28, 235±237. Puppala, S., and Crawford, M. H., 1996,Genetic structure and phylogeny of populations of India: AndhraPradesh. Homo, 47, 73±84. Ramesh, M., and Veerraju, P., 1992,Red cell glyoxalase-I polymorphism among the Rellis of Visakhapatnam,Andhra Pradesh. IndianAnthropologist , 22, 9±14. Ramesh, M., and Veerraju, P., 1996a,Blood groups in two sub-populationsof Rellis of Coastal Andhra Pradesh. Journalof Human Ecology , 7, 67±69. Ramesh, M., and Veerraju, P., 1996b,Plasma protein markers among two sub-populationsof Rellis of Visakhapatnam,Andhra Pradesh. SouthAsian Anthropologist , 17, 7±9. Ramesh, M., and Veerraju, P., 2001,Red cell enzyme variation among two subpopulations of Rellisof AndhraPradesh. Anthropologist , 3, 57±60. Reddy, B. M., and Reddy, C. P., 1992,Dermatoglyphic a nities among the Telugu populations with contrastingethnohistorical backgrounds. AmericanJournal of HumanBiology , 4, 669±682. Reddy, B. M., and Chopra, V. P., 1999,Biological a nities between migrant and parental populations of ®shermenof the East Coast of India. HumanBiology , 71, 803±822. Rohlf, F. J., 1993, NTSYS-PC. NumericalTaxonomy and Multivariate Analysis System. Version 1.8 (New York:Exeter Publishing, LTD). Roychoudhury , S., Roy, S., Dey, B., Chakraborty, M., Roy, M., Roy, B., Ramesh, A., Prabhakaran, N., Usha Rani, M. V., Vishwanathan , H., Mitra, M., Sil, S. K., and Majumder, P. P., 2000,Fundamental genomic unity of ethnic India is revealedby analysis of mitochondrialDNA. CurrentScience , 79, 1182±1192. Sanghvi, L. D., 1966,Inbreeding in India. EugenicsQuarterly , 13, 291±301. Sherring, M. A., 1881, HinduTribes and Castes ±TheBrahminical Tribes of theMadras Presidency , vol. II(Delhi:Cosmos Publications),pp. 105± 107. Spencer, N., Hopkinson, D. A., and Harris, H., 1964,Phosphoglucomutase polymorphism in man. Nature, 204, 742±745. Thangaraj, K., Ramana, G. V., and Lalji Singh, 1999,Y chromosomeand mitochondrial DNA polymorphismsin Indian populations. Electrophoresis , 20, 1743±1747.

Addressfor correspondence:Dr DarõÂoA.Demarchi,Departmento de BioquõÂmica Clõ Ânica, Facultad de CienciasQuõ  micas,Ciudad Universitaria, Co  rdoba5000, Argentina. Email: [email protected]

Zusammenfassung. Ziel: Dervorliegende Artikel konzentriert sich auf die Untersuchung des Musters genetischerMikrodi€ erenzierunginnerhalb einer der substrukturierten Kastenbevo Èlkerungenvon Andhra Pradesh(der Vysya), im Vergleich mit 17 anderen Telugu sprechenden Bevo È lkerungender gleichen Region Indiens. Materialund Methode: Beiinsgesamt 302 Individuen der drei Vysya-Untergruppen (101 der Arya Vysya, 100der Kalinga Vysya und 101 der Thrivarnika) wurden 17 Blutgruppen und Proteinpolymorphismen typisiert.Nei’s Genverteilungsanalyse, Neighbour-Joining Tree und UPGMA Clusterdiagramme, erstellt ausdurchschnittlichen genetischen Absta Ènden,R-Matrix-Analysen und ein Regressionsmodell zur Unter- suchungder Muster desexternen Gen¯ usses unddes genetischen Drifts (zuruÈckzufuÈhrenauf Isolation durchInselmodell), wurden in zwei Stufen durchgefu È hrt: (1)unter Beru È cksichtigungder drei Vysya-Popu- lationenallein und (2) unter Beru È cksichtigungallgemeiner Loci innerhalb der 20 Populationen von AndhraPradesh. Ergebnisse: Siebender 17 untersuchten Systeme wurden als monomorph in allen drei Vysya-Gruppen festgestellt.Der UPGMA-Stammbaum und das bidimensionale Diagramm der D2 AbstaÈnde,abgeleitet ausder R-Matrix-Analyse, zeigen einen unterschiedlichen Cluster derVysya-Populationen. Die Anwen- dungdes Regressionsmodells der mittleren Heterozygotie im Vergleich zur Distanz der Bevo Èlkerungenaus denCentroiden zeigt eine deutliche Plazierung der drei Vysya-Bevo È lkerungenoberhalb der theoretischen Regressionsgeraden. Schluûfolgerung: Dieverschiedenen in dieser Studie angewandten Methoden unterstu È tzendie Hypothese vomunterschiedlichen Ursprung und/ oderder demographischen Geschichte der drei Vysya-Gruppen im Vergleichmit anderen Bevo Èlkerungenvon Andhra Pradesh.

Re sume . Objectif:Cetarticle e  tudiele processus de microdi€ e  renciationge  ne tiquedes Vysya, l’une des sous castesdes populations de l’Andhra Pradesh, avec re  fe rence aÁ 17autres populations de lame Ãmere  gion del’Inde,parlant le Telugu. Sujets et meÂthodes:302individus appartenant aux trois sous castesVysya (101 Arya Vysya, 100 Kalinga Vysyaet 101thrivamika) ont e ÂteÂtypeÂspour17 groupes sanguins et polymorphismesdes prote Âines.Une analysede diversite ÂgeÂneÂtiquede Nei ainsi que des con® gurations d’associations de voisinage et des diagrammesd’agre  gationUPGMA extraits de distances ge  ne tiquesstandard, des analyses de matrices - Retun mode Álede re Âgressionpour explorer les modalite Âsde¯ uxge Âniqueexterne et dede Ârivege ÂneÂtique provoqueÂeparl’isolement de type insulaire on e ÂteÂe€ectue ÂsaÁdeuxniveaux: (1) conside Ârantseulement les trois populationsVysya et (2)en englobanttous les loci pre Âsentsdans 20 populations de l’AndhraPradesh. Populationstructure amongthe Vysya caste 549

ReÂsultats: On a trouve que7 des17 syste Á mes explore s e taientmonomorphes dans les trois groupesVysya. LedendrogrammeUPGMA et l’e  chelonnementbidimensionnel des distances D2 extraitesde l’analyse de lamatrice-R montrent une agre Âgationtre ÁsmarqueÂedestrois populationsVysya. L’application d’un modeÁ le de re gressionde l’he  te rozygosite moyennesur ladistance des populations a Á partir ducentroõ Ède, indiqueque les trois populationsVysya sont nettement en-dehors et au-dessus de la ligne de re Âgression theÂorique. Conclusion:Lesdiverses approches utilise  esdanscette e  tudeappuient l’hypothe Á sed’une origine distinct et oud’unehistoire de Âmographiqueparticulie Áredes trois groupesVysya par rapport aux autres populations del’AndhraPradesh.