AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 65:291-304 (1984) Anthropometric and Genetic Distance Between Gonds of Central India URMILA PINGLE Indian Statistical Institute, Calcutta-700 035, India KEY WORDS Anthropometric measurements, Genetic polymorphisms, Genetic distance, Cluster analysis ABSTRACT A morphological and genetic study was undertaken on five Gondi-speaking populations of Central India (Andhra Pradesh and Maharashtra States). There has been no systematic biological study on this large Dravidian-speaking tribal group, 4 million in number, amounting to 13% of the total tribal population of India. Data was collected on 16 anthropometric measurements and seven genetic markers (blood groups, hemoglobin, G6PD and plasma protein polymorphisms) on the Raj Gonds, Kolams, Manne, Koyas and Plains Maria Gonds. Various genetic distance measures such as Mahalanobis’s D2 and Nei’s and Sanghvi’s measures and cluster analysis techniques were used to determine the relationship between these groups based on anthropometrics and genetic variables. The statistical analysis revealed the Gonds to be a heterogenous group in both morphology and genetic characteristics. The morphological and genetic distances between these five groups when projected graphically revealed that the spatial distribution of these Gonds generally corresponds to their present geographical distribution. However, the actual relationships among each of the Gond populations show differences when based on these two biological variables, the possible reasons for this being discussed in the paper. The emphasis of this study is on the importance of geographical proximity in producing morphological and genetic similarity between populations, brought about by a short distance a s well as similar geographical factors (such as soil, terrain, flora, etc.) drawing these populations together under a common ecocultural umbrella. Gonds are one of the largest tribal populations of Central India, roughly 4 million in number, amounting to about 13% of the total tribal populations of India. “Gondwana” the area of their distribution stretches into Madhya Pradesh, Maharashtra, and Andhra Pradesh States. The Gondi language belongs to the Dravidian group of languages. The Gonds had kingdoms in Deogarh on the foothills of Satpuras at Mandla (Madhya Pradesh) and Chanda (Maharashtra) from the 14th to the 18th centuries AD until they were defeated by the Marathas. The social and anthropological aspects of the Gonds of Andhra Pradesh have been studied over a period of 40 years by Furer-Haimendorf (1979) who considers the Gonds neither racially, culturally, nor linguistically a homogenous group. However, there has been no systematic biological 0 1984 ALAN R. LISS, INC. study on Gonds and the relationship betwen various subsections of this large linguistic tribal group. Hence, the present study was undertaken on five major Gondi-speaking populations of Andhra Pradesh and Maharashtra. AREA OF DISTRIBUTION AND BACKGROUND OF GOND POPULATIONS The area of study lies between 18”40’and 19”40’ north and 78’45’ and 80’45’ east, approximately (Fig. 1). The geology and hence the physical characteristics of the soil differ from west to east; the western portion Received July 7, 1982; accepted June 27,1984. Urmila Pingle’s present address is 1-69, Habshiguda, Hyderabad-500 007, Andhra F’radesh, India. -7- -Rivers 2 4 6 8 Toluk vllloge and s t r e o m s R o i i w o y line Plains Moria Gond a n d H i l l Marro Gand village M A Koyo v i l \ a g e ( M i x e d ) Plains M o r i a Gond v i l l a g e M a n n e a n d Koya v i i l o g e [ M i x e d ) K o y o willoge Manne RojGond o n d K o l o m v i l l a g e ( M i x e d ) RojGond villoge Kolom village headquarlers K E Y MI LES L l l i l 0 Taluk Tolu k Adllobob U t nur H R S T U A 2 4 6 8 MILES LLLU 0 Lokshetf pet Toluk A S A T E Sultona b o d T Sirpur Monthint Toluk ond Chcnnur Toiukr M AMARASTRA Fig. 1. Map showing distribution of all Gond villages (tribewise). A 4s1fobod Toluk Sironcha Toluk SO0 ulLC1 u 293 ANTHROPOMETRIC AND GENETIC DISTANCE AMONG GONDS of the field area is mountainous and covered by black cotton soil of Deccan trap. The tribal populations selected here are the Raj Gonds and Kolams of Utnoor and Asifabad taluks* (Andhra Pradesh). Though these two populations live in close proximity to each other they have distinct geomorphological territories. The Kolams, who until recently were shifting cultivators, live on the hilltops where the soil is thin and suitable only for cultivation of primitive millets. The Raj Gonds however cultivate the valley bottoms near streams where there is a thick layer of black cotton soil suitable for jowar and cotton cultivation with a plough. The total Kolam population in Adilabad district is about 9,386 whereas the Raj Gonds have a total population of 89,150. East of these trap hills, the plains of Gondwana sedimentary soils form the valley of Pranitha and Godavari rivers. The predominant tribal populations of these soils, the Koyas and Manne, belonging to Chinnur taluk (Andhra Pradesh), were selected for the study. The Koyas and Manne are approximately 7,000 each in both taluks. East of the river Prahnitha and mainly in the valley of the Indravati river live the main subgroup of Gonds, the Marias, Mwias, and Bisonhorn Marias. This is a country of dense forests and plains and hills. The sample population selected are the Maria Gonds living mainly in the Sironcha taluk of Chanda district (Maharashtra). This group is divided into two subgroups, the Plains Maria and Hill Maria Gonds. The Hill Marias have a low population density of four persons per km2 compared to the Plains Maria Gonds who are more numerous, their population density reaching 14 persons per km2 (Census of India, Maharashtra State, 1961),having a total population of 33,189. BACKGROUND OF THE GONDS Raj Gonds The basis of social structure of the Raj Gonds is a system of four exogamous phratries (saga) each of which is subdivided into a number of named clans (pari) (F'urer-Haimendorf, 1979). The most distinctive feature of the Gond religion is the cult of the phratry and clan deities, described by the generic term Persa pen or Great God. The frequency of consanguineous marriages is high as referred to in Table 1(Pingle, 1981). Kolams Most Kolams speak Kolami, which is also a Dravidian language closely related to Gondi. In their own language the Kolams refer to themselves as Kolavar but in Gondi they are called Pujari Purer-Haimendorf, 1979).The Kolams also speak in Gondi to the Gonds and Pardhans (bards to the Gonds). The Kolams, like the Gonds, are organized into exogamous clans and most of these have names identical with those of certain Gond clans. These clans are grouped in phratries corresponding to the Gond system. Kolams and Gonds consider themselves related communities and Kolams eat freely in the houses of Gonds and many Gonds partake without hesitation of the food of the Kolams (FurerHaimendorf, 1979). The data on consanguineous marriages and marriage distance is given in Table I. Manne The Manne are a Telugu-speaking subgroup of the Kolams and have forgotten Kolami under the dominance of Telugu culture and have now become a separate endogamous group, not intermarrying with the Kolams. They have a marriage system similar to that practiced by the rest of the Gonds. Koyas The Koyas speak Koye, a language described by Grierson (1906) as a dialect of Gondi. Most of them however have now forgotten the language and speak Telugu, which they have adopted from their neighbors. They have a phratry system similar to that of the TABLE 1. Consanguinity and marriage distance pattern among gonds Tribes Kolam Koya Manne Plains Maria Gonds Raj Gonds Total marriages Consanguineous marriages (%) Average inbreeding coeffkient Marriage distance (km) Village endogomy Mean f SE Median (%I 129 180 136 116 88 (68.2) 52 (28.9) 39 (28.7) 32 (27.6) 0.043 0.018 0.018 0.017 17.65 f 1.47 9.62 f 0.77 8.62 k 0.91 13.66 1.36 13.42 6.86 4.40 9.44 29.8 36.7 45.1 27.2 131 39 (29.8) 0.019 14.42 +_ 1.97 5.29 39.1 294 U. PINGLE Gonds and some of their clan Gods, which are still in Bastar, are brought down on clan festivals to the East Godavari district. Maria Gonds The Maria Gonds speak Gondi and are distributed in Sironcha taluk of Chanda district (Maharashtra). A supplement on the Maria Gonds of Chanda has been written by Grigson (1949).Their Bastar neighbors, called the Marias, are a separate endogamous group. Most of the Marias were practicing shifting cultivation some 50 years ago and some of them still do so on top of Abujhmar hills (Hill Maria Gonds). They had a Gotul system, which is a youth club for both unmarried girls and boys now more or less disappeared in Chanda but is still practiced among the Marias of Bastar (Elwin, 1947). The Godgrouping appears similar to that of the Raj Gonds and may be a recent importation to the Maria Gonds of Chanda (Grigson, 1949). However, the Bastar Marias and Murias have neither God-grouping nor Chuddur Penk. MATERIALS AND METHODS Data on anthropometric measurements and blood samples were collected on 100 individuals consisting of 50 males and 50 females from each of five Gond populations. In addition, a n extra 50 male samples for each of the above groups was collected for the purpose of screening for G6PD deficiency and hemoglobin and transferrin variants. Villages were selected to represent as far as possible the different areas of distribution of the tribe. Since most of the villages were situated along the main streams, villages were selected so as to represent each of the main streams thus covering a representative proportion of the tribal groups in the area selected for study. All available adults in the villages were covered avoiding as far as possible (as consanguinity was high in these groups) relatives and rejecting members of the same family. Information on age, marital status, relatedness of spouse, place of birth of spouse, etc., were also collected on each of the individuals. For each selected individual, after the measurements were taken, blood samples were collected by means of finger prick into heparnized microcapillary tubes. Biochemical methods Microtechniques were followed for blood grouping in ABO, MN, and RkO,Dsystems. The antisera were obtained from HaMtins Institute, Bombay, and the techniques followed were those advocated by this institute. Screening for G6PD deficiency was done by the brilliant cresyl blue (BLB) dye decolorization technique of Motulsky and CampbellKrant (1960). Separation of serum proteins like haptoglobin, transferrin, albumin, and hemoglobin were achieved by polyacrylamide electrophoresis. The method of Clark (1964)was adopted for separation of serum proteins on a 7% acrylamide gel. A 5.5% polyacrylamide gel was used with the buffer system of Dietz et a1 (1971) for hemoglobin separation. Sickle cell test was carried out on a heterozygote Hb AS sample using 2% sodium metabisulphite solution and then observing the red cells under a microscope for sickling after 15-20 minutes. Anthropometric techniques Sixteen anthropometric measurements were carried out and the methods followed were according to Human biology, I.B.P. Handbook No. 9, Guide to field methods (Weiner and Lourie, 1969).The following were the measurements carried out: 1)stature, 2) upper arm length, 3) fore arm length, 4) bigonial diameter, 5) morphological face height (nasion-gnathion), 6) nose height or length, 7) nose breadth, 8)head length, 9) head breadth, 10)biocromial diameter, 11)bi-iliac diameter, 12) right upper arm circumference, 13)right triceps skinfold, 14) right subscapular skinfold, 15)weight, and 16)bizygomatic breadth. Bizygomatic breadth, however, was measured as the minimum distance between zygomatic arches rather than the maximum distance as in I.B.P. methods. STATISTICAL METHODS Scrutiny of data The scrutiny of raw data for the presence of “outliers,” that is, observations that appear to be inconsistent with the remainder of that set of data, is a n important and essential step to valid analysis of the data. Since the presence of these outliers is likely to distort the inference process, as a first step, in order to detect the outlier, a “descriptive analysis” of the data was carried out. This involves 1) frequency distribution in narrow class intervals and drawing histograms; 2) computing measures of locations like mean, mode, median, and measures of dispersion; 3) recording maximum and minimum values; and 4) ANTHROPOMETRIC AND GENETIC DISTANCE AMONG GONDS computing measures of skewness and kurtosis. Outliers were suspected if 1)there are gaps in the histogram and the extreme observations are far removed from the nearest neighbors; 2) if the variation in some samples is larger than in the others for the same character; 3) if there is a general consistency in measures of skewness and kurtosis but in some cases the values are discordant. Transformation of a variable to make the transformed variable more approximately normal is generally advocated when there are strong reasons to believe that the original variable is not inherently a normally distributed one. However, in this case, the variables all represent standard physical measurements that are known to be reasonably normally distributed and the observed deviations from normality could therefore be attributed to a small number of extreme values, i.e., outliers mostly due to recording errors (21 in a total of 10,000). This view is further supported by the fact that, after deletion of these outliers, the variables under consideration exhibited very closely the structure of normal distribution. In the case of measurements such as skinfold thickness measurements and weight, the distribution to start with is not normal, the skewed distribution being due to inherent variability. These measurements were not used for multivariate analysis. PRELIMINARY ANALYSIS Anthropometric data Out of the 16 measurements only 10 were selected for further analysis. Of the six that were rejected, the upper arm and forearm length are highly correlated with stature and hence redundant, whereas the rest, consisting of soft tissue measurements like midarm circumference, triceps skinfold, subcapular skinfold, and weight showed a skewed distribution and therefore were not suitable for multivariate analysis. Also, they are highly affected by environmental factors and not suitable for comparative analysis. The following tests were carried out on all measurements: 1)variances within tribes and tests of homogeneity for both females and males; 2) tests for homogeneity of a set of correlation coefficients (Rao, 1965); 3) equality of variances and covariances for males and females by likelihood ratio test and using x2 approximation according to Box (1949). Only the ten measurements mentioned above were found 295 to have nearly the same variances for all the groups. A similar correlation matrix and the covariance matrices were found to be equal within both male and female groups. Hence only these measurements were used for multivariate analysis. Analysis of variance was carried out on each of the selected measurements between female groups and male groups and for sexual dimorphism. Genetic data The gene frequency and tests of homogeneity and consistency were applied for seven biochemical markers. Albumin protein was found to be invariant and was not used for comparative analysis. The gene frequency for the ABO system was estimated by maximum likelihood method, and tests for departure from Hardy-Weinberg equilibrium and heterogeneity were carried out according to Rao (1965). The test for consistency and homogeneity was also carried out on MN system according to Rao (1965). x2 test for heterogeneity of gene frequencies between the five Gond populations was carried out in the other genetic markers. Genetic distance measures Mahalanobis’s D2:This is the best measure for metric data provided that some assumptions are satisfied, the two most important being homogeneity of the variances, and correlations and normality of the multivariate distribution. This is given by (Xj-Xk) w-’(Xj-Xk) = Dj2k (Mahalanobis’s D2) where W is the pooled estimate of the within population dispersion matrix common to the jth and kth populations. D2 given above is a measure of generalized “sum of squares of differences” among the mean values, weighed approximately to take into account the internal correlations among the variables, and is a measure of overlap between populations and is thus more appropriate to study affinities or dissimilarities between individuals in different populations. Nei’s distance: Nei (1972) taking into account the effect of genetic polymorphism within populations defined the normalized identity of genes between populations as equivalent to protein identity. He then related it to the accumulated number of gene 296 U. PINGLE differences per locus, now called genetic distance. This measure of genetic distance, according to Nei, has several advantages over those proposed by Cavalli-Sforza and Edwards (1967), Balakrishnan and Sanghvi (1968),and others: 1)It is related to Malecot’s coefficient of kinship in a simple way; 2) it measures the accumulated number of gene substitutions per locus; 3) if the rate of gene substitution per year is constant it is linearly related to evolutionary time. G2c of Balakrishnan and Sanghvi: The same fundamental assumption in Mahalanobis’s D2 is followed, that is, though the populations may differ in their mean gene frequencies all of them have the same dispersion matrices. H-clust, a hierarchial clustering program written by Stephen C. Johnson of the Bell Telephone Laboratories, New Jersey, was used to give clustering based on both these techniques. The representations of these analyses in the form of dendograms or trees were made as they have a n advantge of being readily interpretable as conventional taxonomic hierarchies. Subgraphs were projected at different levels of clustering based on the complete linkage method. This helps in visualizing the spatial distribution of the populations. RESULTS Morphology The mean values for various measurements in male and female groups are given Cluster analysis in Tables 2 and 3. Analysis of variance In many empirical fields, there is a n in- showed F values were high for eight of the creasing interest in identifying the group- 11 measurements showing significant differings or clusterings of the object under study ences between the five tribal groups in both that best represent certain empirically mea- sexes (Table 4).There was also a significant sured relations of similarity. The problem difference between sexes within each tribal then is discovering if there is any structure group for each of these 11 measurements. (i.e., natural arrangement of the object into Except for head breadth, no other measurehomogenous groups) inherent in the data. ment showed a difference in sexual dimorThe two clustering techniques used in this phism between the five tribal groups. study are 1) single linkage by Sneath and x2 test for homogeneity on D2 matrices for Sokal(l973)or minimum method by Johnson males and female Gonds (Table 5) showed (1967) or 2) complete linkage or maximum that D2 values were significantly different method (Johnson, 1967). between the five groups in both sexes. Single linkage method: In this method a n Two clustering techniques, the single linkoperational taxonomic unit (OTU) that is a age and complete linkage methods were carcandidate for a n extant cluster has similarity ried out on D2 values and showed a near to that cluster equal to its similarity to the correspondence in relationship between the closest member within the cluster. These five Gond groups in both sexes (Figs. 2 and connections between OTUs and clusters and 3). However, the relationship between males between two clusters are established by sin- and females did not show close correspondgle links between pairs of OTUs. This proce- ence or consistency when either of the clusdure frequently leads to long, straggly tering techniques was followed. In the males, the Koyas cluster with Manne and Raj clusters. Complete linkage method: This is the di- Gonds, the Kolams and Plains Maria Gonds rect antithesis of the single linkage tech- joining separately in that order, whereas in nique. An OTU that is a candidate for the females two clusters are formed, one conadmission to a n extant cluster has similarity sisting of Kolams and Manne and the other to that cluster equal to its similarity to the of Koyas with Plains Maria Gonds. After furtherest member within the cluster. Where these two clusters join each other, the Raj two clusters join, their similarity is that ex- Gonds join separately. In both sexes the isting between the furtherest pair or mem- Manne (MN) lie intermediate, connecting the bers one in each cluster. The method will Kolams (KL1) and Raj Gonds (RG1) with generally lead to tight, hyperspherical, dis- Koyas (KY1)and Plains Maria Gonds (PM). crete clusters that join others only with diffi- The position of the Koyas (KY1) however is culty and at related low overall similarity different in the two sexes shifting nearer to values. Clusters by this method are compact Manne (MN) in the males and away from and more structured, showing more taxa and Manne towards Plains Maria Gonds (PM) in the females. more ranks. 137.308 184.173 118.481 103.314 (-1) 109.442 49.288 1613.038 315.058 262.510 (-1) 351.615 252.471 (-1) 226.308 46.731 75.481 96.771 39.308 0.781 0.726 0.691 0.709 0.318 0.620 7.719 2.257 1.746 2.307 1.722 2.350 0.970 1.977 1.612 0.442 SE Manne (56) X 141.073 (- 1) 184.839 120.018 103.115 ( - 1) 111.673 49.607 1607.607 311.018 264.500 353.607 258.036 228.071 49.286 70.625 94.852 38.054 0.606 0.793 0.787 0.871 0.815 0.571 8.554 2.354 2.317 2.313 1.677 1.978 1.109 1.983 1.364 0.364 - SE Koya (51) 141.843 186.922 118.647 101.900 108.843 47.529 1593.118 306.608 255.961 357.078 256.569 236.000 49.118 75.588 98.708 38.196 X - 0.602 0.889 0.767 0.729 0.819 0.624 7.633 . .~~ 2.835 2.226 2.127 1.934 2.065 1.542 1.978 1.362 0.324 SE 142.185 186.241 118.019 104.906 110.963 47.944 1606.648 305.778 256.778 351.111 264.481 243.741 58.113 (-1) 86.226 (-1) 105.509 39.185 0.664 0.834 0.747 0.698 0.917 0.419 7.873 2.185 1.799 2.115 2.011 2.577 2.061 3.230 1.795 0.401 Plains Maria Gonds (54) X SE - 139.481 187.148 119.259 102.096 110.736 47.815 1628.278 313.630 265.241 355.759 254.333 238.444 48.148 78.333 103.170 37.481 X 0.491 0.647 0.726 0.750 0.799 0.510 8.757 2.126 1.944 2.067 1.699 2.282 1.615 2.368 1.828 0.415 SE Raj Gonds (54) (xand SE) of anthropometric measurements after omitting outliers (number indicated in bracket) 135.260 177.340 113.280 96.043 103,120 44.560 36.580 1490.500 287.938 (-2) 235.959 (-1) 311.380 245.420 210.660 66.400 70.400 83.653 X - 'Abbreviations as in footnote to Table 2. UAL LAL BAB BIB MAC TFF SFF Wt St HL BzB BgB TFL NL NB HB Character Kolam (50) 1.811 1.803 2.195 2.914 2.511 1.310 SE 0.763 0.884 0.682 0.696 0.707 0.476 0.301 6.646 2.102 1.880 X 137.160 180.660 111.440 94.042 101.553 42.760 35.160 1505.540 293.840 240.040 321.000 247.000 215,000 62.200 69.260 82.531 - 136.479 (-1) 177.167 (-1) 113.204 95.596 (-1) 105.522 45.878 35.388 1503.551 289.479 (-1) 283.306 320.490 250.755 23.347 58.673 69.898 81.449 X - SE 0.650 0.745 0.662 0.688 1.014 0.499 0.342 6.614 1.848 2.080 1.751 2.117 2.281 2.019 2.140 1.483 X 135.125 180.042 111.396 96.721 (-1) 104.128 43.271 35.875 1500.833 286.813 233.958 317.478 (-2) 254.688 225.146 80.000 87.500 88.479 ~ SE 0.725 0.892 0.725 0.728 0.910 0.471 0.407 9.201 1,883 1,804 2.135 1.870 2.535 3.332 2.600 1.680 134.459 181.054 115.297 94.118 104.056 44.946 36.027 1515.216 287.784 238.432 312.081 247.838 222.622 75.676 78.514 87.943 X - SE 0.618 0.954 0.872 0.936 0.883 0.520 0.424 7.164 2.479 1.684 2.923 2.168 2.920 3.520 4.132 1.669 anthropometric measurements after omitting outliers (number indicated in bracket) Raj Gonds (37) Plains Maria Gonds (48) Manne (49) SE 0.609 0.779 0.855 0.862 0.817 0.436 0.369 8.521 2.113 2.073 1.727 1.929 2.141 2.997 2.815 1.580 of Koya (50) T A B L E 3. Female tribals: mean and standard error 'HB, head breadth; HL, head length; BzB, bizogmatic breadth; BgB, higonial diameter; TFL, morphological face height; NL, nose length; St, stature; UAL, upper arm length; LAL, forearm length; BAB, biacromial diameter; BIB, bi-iliac diameter; MAC, midarm circumference; TFF, triceps skinfold; SFF, subscapular skinfold; Wt, weight; NB, nose breadth. HB' HL BzB BgB TFL NL St UAL LAL BAB BIB MAC TFF SFF Wt NB C h ilri rt.er Kolam (52) X TABLE 2. Male tribals: 8 -J co N z: 8 z c: 5z ?- 3m 2 =! n m 5 z0 ?- E2 w 0 !s ?- 298 U. PINGLE TABLE 4. Analvsis o f variance findiuidual anthrooometric characters): F statistics Characters HB HL BzB %? NL NB St BAB BIB Wt Between male tribals F (4,482) Between female tribals F (4,482) Sex difference within tribals F (5,482) Sexual dimorphism between tribes F (4,482) 9.3106* * 2.5922* 1.0165 2.8285* 1.6866 3.4251** 4.6603** 2.6680* 1.4555 6.0459** 7.7744 * * 2.5778* 4.8178** 3.8753** 2.0975 2.4057* 5.3792** 2.0819 1.0683 4.2209** 3.3935** 4.3188** 29.0532** 33.4196* * 34.4140** 50.4718** 31.3000** 31.6999** 25.7829** 89.6933** 181.9257** 9.6177** 46.0932** 3.8416 * * 0.2814 1.6832 0.2658 0.1042 1.2634 1.6448 1.3948 2.0438 0.2774 0.3535 'Abbreviations as in Table 2 * P i.05. ** P < .01. TABLE 5. D 2matrix based on anthropometrics in Gonds (males and females) Kolam Kolam Koya Manne Plains Maria Gonds Rai Gonds SINGLE - KL1 PY Kova Plains Manne Maria Gonds Rai Gond 1.8831 0.8991 2.5188 1.6306 1.5878 0.8756 3.0084 1.5379 1.3564 1.5741 1.3567 1.5132 - 1.2068 2.2054 2.2286 2.3750 1.4313 2.7960 - 1.3700 1.2776 LINKAGE KYl YN COMPLETE R G t - 1.0 - t.2 - 1.2 - 1.4 0.2 0.4 0.6 0.8 1.4 i d 6 1.8 2 4 0 2'2 - 0.2 0.4 0.6 0.0 1.0 PY KL1 - - 1.6 - 1.0 - 2'0 2.2 - 2.4 - 2.6 - 2.0 - 3.0 3.2 3.4 L INKACE K 1 1 - Fig. 2. D2 among Gond males (cluster analysis). MN RGI 299 ANTHROPOMETRIC AND GENETIC DISTANCE AMONG GONDS SINGLE - KY4 PY LINKAGE KL4 YN COMPLETE RO4 0.4 - 0.6 - 0.6 0.0 - 0.8 0.2 0.2 0.4 1.0 - 1.2 - 1.2 I. 4 - 1.4 1.6 - 1.6 1.8 - 1.8 1.0 . 2 0 - 2.0 2 2 - 2.2 - - - 2.6 2.8 - 2.8 - 3.0 - - 3.4 R01 - 2.6 3.4 PY - - 3.2 LINKAGE - 2.4 - KY4 - - 3.2 YN - 2.4 3 0 - KL4 - - Fig. 3. D2 among female Gonds (cluster analysis). Subgraphs at various levels of clustering based on complete linkage method were drawn for both males and females separately (Figs. 4 and 5). These help in showing the multivariate spatial distributions of these groups, which in both sexeg generally correspond with their geographical distribution. The Manne lie intermediate linking the Raj Gonds and Kolams with the Koyas and Plains Maria Gonds. Genetic analysis All the groups are significantly different from each other in the ABO gene frequency distribution. The B gene frequency is highest in Kolams, Raj Gonds, and Manne (Table 6). x2 test was not carried out in the distribution of the Rh blood group types because of the low frequency of Rh negative allele in some of the groups. The Rh negative allele is highest in Raj Gonds and Kolams (Table 6) and lowest in Koyas. All groups show heterogeneity among each other for MN system (P < 0.05). The N gene frequency is lowest in the Manne and Plains Maria Gonds (Table 6). The Gond populations were significantly heterogenous for both hemoglobin and G6PD deficiency. The hemoglobin S gene frequency was highest in Plains Maria Gonds and Raj Gonds. The highest G6PD deficiency is found among the Raj Gonds and Plains Maria Gonds (Table 6). These groups also have a high hemoglobin S gene frequency, which is consistent with the fact that the grou s were and are in areas hyperendemic for jaalciparum malarial infection. These groups are however homogenous for haptoglobin protein. The haptoglobin 1 gene frequency is low in these populations ranging from 0.06 to 0.13, comparable to most Indian populations (in whom it is very low) (Table 6). The x2 test showed significant heterogeneity for transferrin types between all Gond populations at the P < 0.05 level. The slow variant D transferrin (subtyping for D variant not done), which is rare in other populations in India, is highest in Kolams, Manne, Raj Gonds, and Koyas (Table 6) and low in Plains Maria Gonds, showing a genetic cline from west to east with a decrease in D transferrin in the eastern groups. Another Dravidian-speaking group, the Oraons, have also been found to have a high D gene frequency (Kirk et al., 1962; Mukherjee et al., 1975). This D gene may prove a useful marker in differentiating the Dravidian-speaking from U.PINGLE 300 Level 1.5 Level < 1.0 KL1 0L K l t MN Level 2.0 RG1 PM Level 2 2.5 RG1 PM Fig. 4. D2 among male Gonds (complete subgraphs) forming separate endogamous entities. In addition, this tribal endogomy has a territorial basis, each of these populations being located in distinct geomorphological areas (Fig. 1). This is further substantiated by linguistic and cultural differences between them. When projected graphically, the morphological and genetic distances between these groups correspond generally to their present geographical distribution. The Manne lie intermediate, connecting the other four populations. The Raj Gonds and Plains Maria Gonds lie farthest away from each other a t two opposite poles. However, the morphological and genetic relationships between the Gond populations differ among themselves. The morphological relationship is the less consistent, the cluster patterns differing by sex. The Koya females are nearer the Manne and Raj Gond and further from the Plains Maria Gond females. The Kolams also vary by sex; the males are removed from all the groups, but the females are very close to the Manne females. This variability by sex in morphological relationship is difficult to inDISCUSSION terpret and may be in part due to differences Statistical analysis showed that the Gonds in sex dimorphism and the greater degree of are a heterogenous group, morphologically environmental influence on morphological and genetically different from each other, characteristics. the Munda-speaking tribals, as well as from South Indian tribals in whom the frequency of D gene is found to be very lwo. Nei’s and Sanghvi’s distance measures were computed based on gene frequencies of seven biochemical characters (Table 7). The two methods of clustering based on Nei’s and Sanghvi’s genetic distance matrices showed close correspondence in the relationships among the five tribal groups (Figs. 5 and 6). The Raj Gonds (RG1) and Kolams (KL1) cluster together, whereas the other three, the Plains Maria Gonds (PM), Manne (MN) and Koya (KYl), form a separate cluster. Complete subgraphs based on both these distance measures also showed a close correspondence in the relationship and spatial distribution of the five groups. The Manne are a n intermediate group connecting the Koyas and Plains Maria Gonds with Kolams and Raj Gonds (Fig. 7). The Manne are closer to Koyas and Plains Maria Gonds than to Kolams and Raj Gonds. HP 1 2.1 2.2 Transferr i n C CD D G6PD Normal Deficient S 0.0352 S 1.0000 0.0000 0.0403 D 0 0.9597 C Hp2 0.9270 Hpl 0.0730 0.9648 0.2143 n A 0.7857 m 149 137 12 0 0 13 76 132 10 0 62 30 6 MN M MN N Hb A AS 0.7575 0.2425 D 96 6 d 0.1748 0.3527 0.4725 p q r 17 44 25 16 ABO A B 0 AB Rh D d Obs. freq. 125 6 187 4 0 0 20 135 213 15 0 75 29 5 168 1 56 39 57 17 Obs. freq. 0.1789 0.8211 0.9231 0.0769 0.2452 0.1814 0.5734 0.9924 0.0076 0.0105 0.9895 Hp2 0.9355 H p l 0.0645 Hbs 0.0329 HbA 0.9671 n m D d p q r Koyas Gene frequency 142 0 136 6 0 0 15 83 0.0130 0.9870 0.1188 0.8812 0.8620 0.1380 0.2099 0.2347 0.5554 0.0211 D 1.0000 0.0000 0.9789 C Hp2 0.9427 H p l 0.0573 S A n 151 4 0 m 18 3 d D p q r Manne Gene frequency 80 103 2 7 32 36 30 Obs. freq. 7 106 159 3 0 0 18 78 0 140 27 86 13 4 101 2 28 26 40 9 0.0808 0.9191 0.1091 0.8981 0.8606 0.1393 0.1985 0.1865 0.6150 0.0093 D 0.9381 0.0619 0.9907 C Hp2 0.9063 Hpl 0.0931 S A n m d D p q r Plains Maria Gonds Obs. Gene freq. frequency TABLE 6. PhenotvDe and gene freauencies for genetic markers in the Gond DoDulations Kolams Gene frequency 92 13 135 8 0 0 14 63 IL20 20 0 61 27 3 83 7 24 34 21 12 Obs. freq. 0.0714 0.9286 0.1813 0.8187 0.7211 0.2789 0.2226 0.2968 0.4806 D C 0.9341 0.0659 0.0280 0.9720 Hp2 0.9091 Hpl 0.0909 S A n m D d p q r Raj Gonds Gene frequency s w 2 s 0 0 2, $5 u b m 0 2 2 r (I) El c, z2 22 5c3 E> 3 8 9 * 5 302 U. PINGLE which is predominantly influenced by Telugu culture. Unlike the morphologic, the genetic characteristics are purely heritable and, not being influenced by environment, provide a finer tuning to population differences. The genetic relationship also reflects more closely the effect of geographical proximity in producing a greater genetic similarity within these two clusters of Gond populations. The emphasis of the present study based on quantitative techniques is on the importance of geographical proximity in producing morphological and genetic similarity between populations, which is brought about The relationship based on genetic characters however is different from the morphological relationship and corresponds closely to the degree of geographical contiguity between the populations. The Raj Gonds and Kolams form a close cluster, with Manne, Koyas, and Plains Maria Gonds forming a second cluster. The Raj Gonds and Kolams live in close proximity to each other on the highlands consisting of black cotton soil, which is contiguous with Marathwada, whereas the Manne, Koyas, and Plains Maria Gonds lie on the east in the valley of the Pranitha-Godavari and Indravathi rivers, TABLE Z Genetic distance in Gonds (Nei and SanehuiJ Kolam Koya Manne Plains Maria Gonds Raj Gonds Kolam Koya Manne Plains Maria Gonds Raj Gonds - 0.5375 0.2837 0.4559 0.0889 0.0641 0.0365 0.0548 - 0.0956 0.0119 0.0182 0.1877 0.1457 0.0088 - 0.6238 0.3536 0.3537 0.0604 0.0325 0.0389 - 0.0095 Nei’s distance COMPLETE ‘001- 101 KLL PY - LINKAGE YN 101 .Wl- - - - .ooa- .005 - *005 -003 - - -009.O2 - -007 *04 ‘06 .08 - - cI SINGLE LINKAGE 111 .007 - .w1.02 - .04 - - .08 .06 Sanghvi’s distance KLL PY - Fig. 5. Nei’s genetic distance among Gonds (cluster analysis). YN KYI 303 ANTHROPOMETRIC AND GENETIC DISTANCE AMONG GONDS COMPLETE RG1 KL1 PY LINKAGE MN SINGLE LINKAGE KV1 RG1 KL1 PM MN L T Fig. 6. Sanghvi's genetic distance among Gonds (cluster analysis). 0.01 L.v.1 MN<Kvl 0 02 L.V.1 CY 0 . 0 6 4 Llr.1 0 . 0 4 Lev.1 .009 RG1 Fig. 7. Nei's genetic distance among Gonds (complete subgraphs). KY! U. PINGLE 304 by a closer distance a s well a s similar geographical factors (such as soil, terrain, flora, etc.), drawing these populations together under a common ecocultural umbrella. The closer the geographical distance between populations, the closer the genetic as well as morphological similarity between them. This is more so in populations such as these who, due to having adopted a settled agricultural occupation, are less mobile. This low mobility in turn produces a smaller marriage distance (distance between place of birth of spouses) thus restricting the spacial distribution of genes to a smaller area. ACKNOWLEDGMENTS I am deeply grateful to Professor C.R. Rao, F.R.S., Jawaharlal Nehru Professor, Indian Statistical Institute and at present Visiting Professor, Department of Mathematics and Statistics, University of Pittsburgh, for his guidance throughout the work and for having given me all facilities for analysis of the data in his department a t the University of Pittsburgh. I am indebted to Mr. N.V. Raja Reddy, Bheeemaram (Andhra Pradesh), for his constant and sustained encouragement and advice throughout the field work. I am thankful to Professor C. von FurerHaimendorf, Emeritus Professor, School of Oriental and African Studies, University of London, for his valuable comments on the conclusions of my study. LITERATURE CITED Balakrishnan, V, and Sanghvi, LD (1968) Distance between populations on the basis of attribute data. Biometrics 245359-865. Box, GEP (1949)A general distribution theory for a class of likelihood criteria. Biometrika 36t317-346. Cavalli-Sforza, LL, and Edwards, AWF (1967) Phylogenetic analysis: Models and estimation procedures. Evolution, 21,550-570. Census of India (1961) District Census Handbook, Chanda, Director, Government Printing and Stationery, Maharashtra State, Bombay. Clark, JT (1964) Simplified ‘disc’ polyacrylamide gel electrophoresis. Ann N.Y. Acad. Sci. 121t305. Dietz, AA, Lubrano, T, and Rubinstein, HM (1971) Haemoglobin and haptoglobin determination by disc electrophoresis. Clin. Biochem., 4.59. Elwin, V (1947) Muria and their Ghotul. Bombay: Oxford University Press. Furer-Haimendorf, C yon (1979) The Gonds of Andhra Pradesh, Vikas Publishing House Private Ltd., New Delhi. Grierson, G (1906) Linguistic Survey of India, Vol. IV, Munda and Dravidian Languages. Calcutta. Grigson, WV (1949)The Maria Gonds of Bastar. London: Oxford University Press. Johnson, SC (1967)Hierarchial clustering schemes. Psychometrika, 32241-254. Kirk, RL, Lai, LYL, Vyas, GH, and Vidyarthi, LP (1962) A genetic study of the Oraons of the Chota Nagpur Plateau (Bibar),India. Am. J. Phys. Anthropol. 20:375385. Motulsky, AG, and Campbell-Krant, JH (1960) Population genetics of GGPD deficiency of the red cell. In BS Blumberg (ed): Proceedings Conference Genetic Polymorphism and Geographic Variations in Disease. New York: Grune and Stratton. Mukherjee, BN, Das, SK, and Sharma, DP (1975)Serum protein and red cell enzyme polymorphism in Oraon tribe, India. Ann. Hum. Biol. 2t201-204. Nei, M. (1972) Genetic distances between populations. Am. Naturalist 106t283-292. Pingle U (1981) Ph.D. thesis: “Morphological and Genetic composition of Gonds of Central India: A Statistical Study” (unpublished) submitted to Indian Statistical Institute, Calcutta. Rao, CR (1965) Linear Statistical Inference and Its Applications, 2nd edition. New York: John Wiley and Sons. Sneatb, PHA, and Sokal, RR (1973) Numerical Taxonomy. San Francisco: W.H. Freeman and Co. Weiner, JS, and Lourie, JA (1969) Human Biology: A Guide to Field Methods, IBP Handbook No. 9. Oxford: Blackwell Scientific Publications.