Chiang Mai J. Sci. 2017; 44(4) 1279

Chiang Mai J. Sci. 2017; 44(4) : 1279-1294http://epg.science.cmu.ac.th/ejournal/Contributed Paper

Genetic Divergence of Austroasiatic SpeakingGroups in the Northeast of Thailand: A Case Studyon Northern Khmer and KuyPanthipa Chantakot Ψ Ψ Ψ Ψ Ψ [a], Pittayawat PittayapornΨΨΨΨΨ [b], Kanokpohn Srithongdaeng [c],

Suparat Srithawong [c] and Wibhu Kutanan* [c]

[a] Forensic Science Program, Faculty of Science, Khon Kaen University, Khon Kaen, T40002, Thailand.

[b] Department of Linguistics, Faculty of Arts, Chulalongkorn University, Bangkok, 10330, Thailand.

[c] Department of Biology, Faculty of Science, Khon Kaen University, Khon Kaen, 40002, Thailand.

* Author for correspondence; e-mail: wibhu@kku.ac.thΨ These authors contributed equally.

Received: 5 October 2015

Accepted: 21 April 2016

ABSTRACT

The Austroasiatic speaking people are regarded as the native inhabitants in Thailand,predating the coming of Kra-Dai-speaking groups from Southwestern China. Two of thelargest Austroasiatic speaking groups in Northeastern Thailand are the Northern Khmer andthe Kuy. Previous genetic surveys of these two ethnic groups mostly used mitochondrialDNA polymorphisms, therefore, the present study intended to utilize the variability of autosomalmicrosatellites to deeply investigate the genetic structure of the Northern Khmer and the Kuy.Thirty-one and forty-seven samples of unrelated Northern Khmer and Kuy, respectively,were genotyped for 15 microsatellites. Distance based and model based clustering methodswere employed to elucidate the evolutionary relationship between the studied populations andten other linguistically and geographically diverse comparable populations in Thailand. Analysisof Molecular Variance and Factorial Correspondence Analysis revealed a genetic heterogeneityin Austroasiatic populations but an opposite trend was observed in the genetically homogeneousKra-Dai populations. STRUCTURE result shows that the Northern Khmer contributesapproximately 31% of their genes to the gene pool of their neighbor, Lao Isan, indicating agenetic exchange among them. The extreme genetic divergence of the Kuy from otherpopulations seems to be much higher than the Northern Khmer. A genetic admixture of theNorthern Khmer and with their neighboring Lao Isan was detected and this is consistent witharchaeological evidence.

Keywords: autosomal microsatellites, northeastern Thailand, northern Khmer, Kuy,genetic divergence

1. INTRODUCTION

Northeastern Thailand or Isan isgeographically located on the KhoratPlateau and shares the border with Laos and

Cambodia. The ethnic and linguistic diversityamong the populations that inhabit thisregion is a result of historical migrations

1280 Chiang Mai J. Sci. 2017; 44(4)

and the resettlement of populations. The 18ethno-linguistic groups of the region speaklanguages that belong to two major languagefamilies, namely Austroasiatic and Kra-Dai(also known as Tai-Kadai). While theAustroasiatic-speaking groups are consideredto be indigenous, the arrival of Kra-Dai-speaking groups into the area is due tocontinuous Tai migrations from SouthernChina that started in the later centuries of thefirst millennium [1-2]. The lower parts of theregion, which is comprised of the provincesof Surin, Sri Saket, and Buriram is of specialinterest to population genetics. In particular,interesting questions regard whether geneticdistinction or genetic admixture might haveoccurred, as the three major ethnicpopulations have co-existed for centuries.

The Lao Isan are the largest of allethnolinguistic groups of northeasternThailand, numbering around 14 millionpeople [3]. They first came from the territoryof present-day Lao People’s DemocraticRepublic (Lao PDR) to occupy northeasternThailand the mid 14th century A.D. Historicalrecords show that they had reached thelower Isan region by the late18th century A.D.[4-5]. Prior to the arrival of the Lao, the Isanregion was primarily inhabited by peoplewho were part of the ancient Khmercivilization. Numerous archeological sitesclearly attest to the presence of Khmer culturesince around the 6th century A.D. [6-7].Separated from Cambodia to the South bynatural border called the Dongrak escarpment(Figure 1), the lower part of the Isan region ishome to 1.4 million Khmer ethnics [3].These “Northern Khmer” populationsspeak a dialect quite divergent from the“Southern Khmer” in Cambodia [8-10].Another Austroasiatic-speaking group inlower part of Isan is the Kuy, often calledby the Thai as “Suay” [11]. About 400,000Kuy speakers now reside in the provinces of

Surin and Srisaket [3]. Their first exodusfrom southern Laos took place in the laterpart of the 17th century A.D. This wasfollowed by sporadic migrations until the later18th century A.D. when a mass re-settlementoccurred [5, 12]. The present-day Kuy inThailand are trilingual, speaking both Laoand Northern Khmer in addition to theirnative language [9, 13].

Our previous study on maternal geneticvariation among ten northeastern Thaiethnicities, including the Northern Khmerand Kuy, indicated that geography was aninfluential factor [14]. The two groupsexhibited maternally genetic differentiationfrom other populations whereas a certaindegree of maternally genetic resemblancewas detected. However, the previous studyonly employed mitochondrial DNA(mtDNA) as the genetic marker, thusproviding a picture of their genetic relationshipon the maternal side only. In the current study,we therefore expanded our investigation toinclude bi-parentally inherited geneticmarkers. This set of markers is composed of15 autosomal microsatellites or short tandemrepeats (STRs) which have been proven to bepowerful genetic markers for inferring thegenetic relationship of populations withina regional scale [15-18]. In particular, weexamined the statistical distribution of allelefrequency at 15 STRs of the Northern Khmerand Kuy from the Northeast of Thailand.In addition, linguistically and geographicallydiverse populations in Thailand from earlierstudies were included in multiple geneticanalyses to reconstruct their genetic structureand genetic relationships, as well as to considervariance components contributed by linguisticand geographic differences.

2. MATERIAL AND METHODS

2.1 SamplesEach genotyped individual was

Chiang Mai J. Sci. 2017; 44(4) 1281

interviewed to ensure that they wereunrelated by at least three generations andthat all four grandparents had been bornwithin the village. Informed consent wasobtained after an interview and a buccal swabwas collected using a brush embedded in aGentra Puregene Buccal Cell Kit (Qiagen,Hilden, Germany). The use of human subjectsfor this study was approved by EthicsCommittee for Human Research of KhonKaen University, Thailand. Both the NorthernKhmer and Kuy (n=31 and n=47, respectively)samples reside in Sangkla District andSamrongtap District, respectively, in the areaof Surin Province, Thailand. In order toexecute detailed comparative statistics, thereference data from the surrounding

populations were selected based onhistorical evidence which stated that thenative Austroasiatic people inhabited theThai territory before the arrival of theKra-Dai speaking populations. In addition,because the statistical analyses in this studyincluded both distance based and modelbased clustering methods that need the rawgenotypic data, the selected representativeswere from previous studies in which theraw genotypic data of fifteen STRs wereavailable. Therefore, in total ten neighboringpopulations from our previous studies[19-20], i.e., Lawa1, Lawa2, Mon, Lue1, Lue2,Yuan, Yong, Khuen, Shan, and Lao Isanwere utilized in the population comparison(Figure 1 and Table 1).

Figure 1. Map of Thailand showing the studied populations and comparable populations.Filled circles and empty circles represent populations speaking Austroasiatic and Kra-Dailanguage, respectively.

1282 Chiang Mai J. Sci. 2017; 44(4)

2.2 DNA Extraction and STR TypingAccording to the manufacturer’s

specifications, genomic DNAs were isolatedfrom the buccal samples using the same kitas buccal swab collection. Multiplex PCRreaction was performed using a commercialAmpF!STR Identifiler kit (Applied Biosystem,Foster City, CA, USA) according tomanufacturer’s protocol but using a totalreaction volume per sample of 12.5 μl.Fifteen autosomal STR loci: D8S1179,D21S11, D7S820, CSF1PO, D3S1358, TH01,D13S317, D16S539, vWA, TPOX, D18S51,D5S818, FGA, D19S433 and D2S1338were amplified. Amplicons were genotypedby multicapillary electrophoresis in an ABI3100 DNA sequencer (Applied Biosystem).The results were further analyzed by GeneMapper software v.3.2.1 (Applied Biosystem).

2.3 Statistical AnalysesSeveral statistics on genetic variation

within population, that is, STR allele frequenciesestimated by gene counting following a testof Hardy-Weinberg equilibrium, observedheterozygosity (H

O) and a number of alleles,

and gene diversity (GD) were calculated byARLEQUIN 3.5 software [21]. Significancelevel for Hardy-Weinberg P-values wasadjusted according to the sequential Bonferronicorrection (α = 0.05/15 or 0.0033) [22].Since this set of markers is commonly usedin forensic genetics, a forensic parametersthat is a power of discrimination, matchingprobability, polymorphic informationcontent, power of exclusion and typicalpaternity index were determined byPowerstats program (www.promega.com/geneticidtools/powerstats).

Table 1. General information and genetic diversities of the studied and comparable populations.

Population

NorthernKhmerKuy

Lao IsanLawa1Lawa2

MonYuan

Lue1Lue2YongKheun

Shan

Samplesize (n)

31

47

455047

3687

51415548

44

TotalAlleles

116

112

123114103

113126

112104125114

117

Ho

0.7849

0.7611

0.76860.78

0.7643

0.77030.7662

0.78170.77560.76840.7402

0.7484

Gene diversity +/- S.D.

0.7730 +/- 0.3966

0.7539 +/- 0.3820

0.7600 +/- 0.38500.7669 +/- 0.38790.7510 +/- 0.3806

0.7900 +/- 0.40040.7806 +/- 0.3929

0.7652 +/- 0.38710.7618 +/- 0.38630.7757 +/- 0.39180.7585 +/- 0.3841

0.7829 +/- 0.3961

Location(District, Province)

Sangkha, Surin

Samrongtap, Surin

Kaset Wisai, Roi EtHod, Chiang MaiMae La Noi,Mae Hong SonPa Sang, LamphunMae Taeng andSanSai, Chiang MaiPua, NanTha Wang Pha, NanPa Sang, LamphunMae Wang andSan Pa Tong,Chiang MaiPang Ma Pa,Mae Hong Son

LinguisticclassificationAustroasiatic

Austroasiatic

Kra-DaiAustroasiaticAustroasiatic

AustroasiaticKra-Dai

Kra-DaiKra-DaiKra-DaiKra-Dai

Kra-Dai

References

Presentstudy

Presentstudy[20][19][19]

[19][19]

[19][19][19][19]

[19]

Chiang Mai J. Sci. 2017; 44(4) 1283

To investigate the amount of geneticvariation due to differences at threehierarchical subdivisions i.e. within individualsof a population, among populations withina group, and among groups of populationsaccording to linguistic classification(Austroasiatic and Kra-Dai groups)and geographic region (Northern andNortheastern Thailand), analysis ofmolecular variance (AMOVA) [23] asimplemented in ARLEQUIN 3.5 wasemployed. Spatial analysis of molecularvariance (SAMOVA) algorithm was appliedto reveal in more detail the genetic structureof the populations. SAMOVA maximizes thedifferentiation between geographicallyhomogeneous groups of populations onthe basis of autosomal genotyping data(SAMOVA v.1.0; [24]).

The relationship between populationswas investigated by pairwise F

st distances

based on the number of different alleleas well as the statistical significance using1,000 permutations by ARLEQUIN 3.5.To further detect population relatedness,factorial correspondence analysis (FCA)implemented in GENETIX v.4.05.2 wasemployed [25]. Further investigation ofpopulation substructure was performedusing the Bayesian clustering methodimplemented in STRUCTURE 2.3.2 [26-28]using the admixture with correlation betweenallele frequencies across cluster andLOCPRIOR model [28]. The number ofcluster (K) was set from 1 to 12, and foreach K, five independent replicates wereperformed with a MCMC chain burn-inlength of 50,000 iterations followed by100,000 iterations for estimate clustering.STRUCTURE Harvester [29] was used tocalculate a posterior probability ((lnPr(X|K))[24] and a second order rate of changelogarithmic probability between subsequentK values (delta K) [30], to identify the optimal

K in the data. Outputs from STRUCTUREwere graphically modified by DISTRUCT[31].

A Mantel test was utilized to examine thecorrelations and partial correlations betweentwo pairs of matrices of genetic andgeographic distances as well as genetic andlinguistic distances [32-33]. A matrix ofgenetic distance (F

st) was calculated by

ARLEQUIN while geographic distancesin kilometers between the approximatelocations of each population were computedas great-circle distances calculated fromtheir latitudinal and longitudinal coordinates.Pairwise linguistic distances were definedas following [34]. All matrices used for theMantel test are shown in Table 5.

The Mantel test was also performed tocorrelate between the matrices of Fst

fromautosomal STRs and mtDNA sequences.Mitochondrial DNA (mtDNA) hypervariableregion I (HVS-I) sequences were retrievedfrom previous literatures [14, 34]. Pairwisegenetic distances among populations basedon pairwise differences of mtDNA sequencewere calculated by ARLEQUIN (data notshown).

3. RESULTS

3.1 Genetic Variation within PopulationAmong all twelve populations, the

average Ho was greatest in the studied

Northern Khmer population (0.7849), whereasthe values of GD (0.7730), and total allele(116) were intermediate (Table 1), indicatinga rather high diversity in the Northern Khmergroup. The Kuy also exhibited rather highvalues of genetic diversity parameters, i.e., H

O

(0.7611), GD (0.7539) and total allele (112).After applying Bonferroni correction,deviation from Hardy-Weinberg equilibriumwas not detected. The overall combinedmatching probability in the Northern Khmerand Kuy was 3.822 × 10-15 and 2.236 × 10-15,

1284 Chiang Mai J. Sci. 2017; 44(4)

respectively while the combined power ofexclusion was 0.9998596 in the NorthernKhmer and 0.999996 in the Kuy. The mostpolymorphic locus for the Northern Khmerwas FGA, reflected by the highest H

o

(Table 2 and Table 3). As expected, the most

polymorphic locus was highly discriminating,as evidenced by the relatively high PD values(Table 2 and Table 3). All forensic parametersshowed that this set of loci was useful forforensic identification.

Table 2. Allele frequencies and statistical parameters of genetic and forensic interests for theNorthern Khmer.

Ho, observed heterozygosity. He , expected heterozygosity. HWE, Hardy-Weinberg p values. MP, matching probability.PD, power of discrimination. PIC, polymorphic information content. PE, power of exclusion. TPI, paternity index.Het, percent heterozygote

Allele6789

9.310111213

13.214

14.215

15.216

16.217181920

20.22122

22.22324

24.225

25.2262730

30.231

31.232

32.233.2H

o

He

HWEMPPDPICPETPIHet

D8S1179

16.109.708.1016.10

16.10

16.10

14.50

3.20

0.870.830.740.060.940.840.743.8887.10

D21S11

2.402.40

28.602.4023.809.504.8019.007.100.770.840.690.080.920.780.622.6381.00

D7S820

8.608.60

21.4040.0017.102.90

1.40

0.770.760.700.100.900.710.451.7571.40

CSF1PO

1.50

13.6042.4034.806.10

1.50

0.680.690.170.200.800.620.421.6569.70

D3S1358

1.60

25.801.6035.501.6024.208.101.60

0.800.720.070.140.860.700.612.5880.60

TH017.8026.6015.6029.7014.106.30

0.770.810.110.100.900.760.562.2978.10

D13S317

38.207.40

10.3020.6020.601.50

1.50

0.810.770.060.150.850.720.642.8382.40

D16S539

1.5013.20

7.4035.3019.1023.50

0.680.780.650.100.900.720.391.5567.60

D2S1338

9.403.1021.9014.10

21.004.70

10.9020.30

4.70

0.900.870.220.070.930.840.815.3390.60

D19S433

1.601.608.103.2045.206.506.5027.40

0.650.740.000.170.830.660.351.4164.50

vWA

18.30

16.70

33.3015.0013.303.30

0.720.800.560.0890.9110.760.4821.8873.30

TPOX

68.8014.10

3.1014.10

0.550.491.000.2950.7050.450.2481.1456.30

D18S513.10

10.906.301.6025.00

21.90

26.60

1.60

1.601.60

0.930.800.730.1090.8910.770.8085.3390.60

D5S8181.609.40

4.70

18.8026.6018.8018.80

0.930.830.330.0960.9040.790.8728.0093.80

FGA

10.001.701.7020.0016.703.3011.7016.701.708.301.705.001.70

0.930.880.590.0580.9420.860.7955.0090.00

Chiang Mai J. Sci. 2017; 44(4) 1285

Table 3. Allele frequencies and statistical parameters of genetic and forensic interests for theNorthern Kuy.

Ho, observed heterozygosity. He, expected heterozygosity. HWE, Hardy-Weinberg p values. MP, matching probability.PD, power of discrimination. PIC, polymorphic information content. PE, power of exclusion. TPI, paternity index.Het, percent heterozygote

3.2 Genetic Variation Among PopulationGeography and language have often been

reported as influential factors which can limitgene flow and maintain genetic distinctionamong populations in both local andregional geographic frameworks [35-37].In Thailand, several studies indicate geography[14] but some report language [34] depending

on the set of studied populations. One methodto quantify the effect of language andgeography on the genetic relationship amongpopulations is AMOVA. This analysis assessesthe proportion of genetic variation within andbetween geographically and linguisticallygroups (Table 4). Non-significant Fct

statistics(P > 0.01) were observed, indicating that

Allele6789

9.310111213

13.214

14.215

15.216

16.217181920212223242526283031

31.232

32.233.2HoHe

HWEMPPDPICPETPIHet

D8S1179

8.3013.506.3014.60

10.40

33.30

11.50

2.10

0.870.820.580.070.930.790.744.0087.50

D21S11

5.0043.8018.805.001.3015.006.300.830.770.660.120.890.710.602.5080.00

D7S820

21.9010.40

38.5024.004.201.00

0.700.740.280.120.880.690.441.7170.80

CSF1PO

10.6043.6040.401.10

4.30

0.550.650.400.180.820.560.221.0753.20

D3S1358

1.00

4.20

18.80

37.50

26.0011.501.00

0.830.750.080.140.860.700.663.0083.30

TH0121.9020.803.1046.906.301.00

0.640.690.660.140.860.640.351.4164.60

D13S317

53.107.30

8.306.3013.5011.50

0.640.690.050.160.840.640.321.3362.50

D16S539

1.0015.60

16.7020.8024.0014.60

7.30

0.870.830.350.080.920.800.744.00

87.5060

D2S1338

1.10

13.804.3022.307.405.301.1029.8010.603.201.10

0.780.830.950.050.950.800.582.3578.70

D19S433

6.3021.905.2037.506.3010.409.40

3.10

0.740.790.860.080.920.750.512.0075.00

vWA

46.90

6.30

6.30

15.6013.507.304.20

0.770.740.510.110.890.700.512.0075.00

TPOX

55.2010.40

10.4024.00

0.640.630.820.190.810.570.321.3362.50

D18S51

3.106.3020.80

18.80

21.90

18.80

2.10

1.00

3.104.20

0.830.840.360.070.930.810.622.6781.30

D5S8182.10

3.20

16.0030.9030.9013.80

1.10

2.10

0.850.770.940.110.890.730.662.9483.00

FGA

4.202.103.101.0026.0022.909.4019.807.304.20

0.870.830.900.070.930.800.744.0087.50

1286 Chiang Mai J. Sci. 2017; 44(4)

the genetic structure was neither influencedby geographic separation nor linguisticdiversification. AMOVA also specified thathigher genetic divergence between populationsin the Austroasiatic group than the Kra-Daigroup was observed, indicating that theAustroasiatic group was more geneticallystructured than the more homogeneous Taigroup. In addition, the Northern Thaipopulations exhibited higher genetic variationthan the Northeastern Thai populationswhich reflected genetic heterogeneity.

However, it should be noted that theAMOVA analysis might be not completely

straightforward, because differences ofgenetic differentiation in both linguistic andgeographic groups were detected. Anothermethod, the Mantel test, was performed toclarify geographic and linguistic factors inshaping genetic variation patterns. The resultsfrom the Mantel test indicated that genetic versuslinguistic distances were correlated (r=0.2759,P<0.01) and partially correlated (r=0.25179,P<0.01) whereas absent statistical significanceof correlation coefficients of genetic versusgeographic distances was detected (r=0.3063,P>0.01 for correlation; r=0.2852, P>0.01for partial correlation).

3.3 Genetic Differentiation andPopulation Structure

Among 66 pairwise Fst comparisons to

test for genetic differences between thestudied populations and other comparedpopulations, fifty-seven pairwise differences(86.36%) were statistically significant afterapplying Bonferroni correction (P<0.000758)(Table 5). The Northern Khmer and Kuyexhibited significant genetic differences to allpopulations reflecting a high geneticdivergence (Table 5). This pattern was also

noticed in other Austroasiatic speaking groupsfrom Northern Thailand (Lawa1, Lawa2, andMon). Nine non-significant genetic differenceswere observed between the closely relatedKra-Dai speaking populations (Table 5).

Population relationships were visualizedby FCA as shown in Figure 2. The NorthernKhmer and Kuy was segregated fromall populations on Axis 1, which explained15.69% of the distance matrix while theLawa2 was separated in an upward positionon Axis 2, which explained 14.27 % of the

Table 4. Analysis of molecular variance (AMOVA) results according to geographic and linguisticclassification.

Bold letter indicated statistically significant at P < 0.01

All samplesGeographyNortheastNorthNortheast/NorthLanguageAustroasiaticKra-DaiAustroasiatic/Kra-Dai

No. ofgroups

1

112

112

No. ofpopulations

12

3912

5712

Withinpopulations Fst

98.34 (0.0166)

97.97 (0.0203)98.56 (0.0144)98.13 (0.0187)

97.33 (0.0267)99.08 (0.0092)98.22 (0.0178)

% of variance (Fixation indices)

Withingroups F

sc

1.66

2.031.44

1.53 (0.0154)

2.670.92

1.52 (0.0153)

Amonggroups F

ct

0.33 (0.0033)

0.26 (0.0026)

Chiang Mai J. Sci. 2017; 44(4) 1287

variation. Lawa1, Lue2 and Khuen weredispersed further in an upward positionon Axis 2 and Axis 3 (11.45 % variation).The segregated populations, i.e., NorthernKhmer, Kuy, Lawa1, Lawa2, Lue2 and

Khuen mirrored genetic differentiation ofthese populations from the other sixpopulations (Mon, Lue1, Yuan, Shan, Yongand Lao) which occupied an intermediateposition in the plot.

Table 5. Genetic distance (Fst) between population based on number of different allele (below

the diagonal) and geographic distance matrix as well as linguistic distance matrix as expressedin parentheses (above the diagonal).

Bold letter indicated P<0.000758 (after applying Bonferroni correction)

Northern

Khmer

Lao Isan

Kuy

Lawa1

Lawa2

Mon

Yuan

Lue1

Lue2

Yong

Khuen

Shan

Northern

Khmer

0.01521

0.02618

0.02513

0.02349

0.01892

0.01365

0.01849

0.02905

0.01821

0.02646

0.01963

Lao

Isan

122

(3)

0.0195

0.0146

0.0259

0.0119

0.0033

0.0103

0.0180

0.0049

0.0172

0.0086

Kuy

48

(2)

82

(3)

0.0303

0.0325

0.0318

0.0176

0.0279

0.0326

0.0241

0.0361

0.0193

Lawa1

785

(2)

695

(3)

768

(2)

0.0243

0.0229

0.0115

0.0189

0.0173

0.0168

0.0206

0.0150

Lawa2

719

(2)

626

(3)

700

(2)

69

(1)

0.0261

0.0202

0.0235

0.0286

0.0267

0.0257

0.0239

Mon

681

(2)

580

(3)

658

(2)

143

(2)

83

(2)

0.0099

0.0164

0.0165

0.0139

0.0148

0.0140

Yuan

746

(3)

642

(2)

721

(3)

138

(3)

110

(3)

71

(3)

0.0032

0.0109

0.0028

0.0071

0.0051

Lue1

613

(3)

496

(2)

579

(3)

330

(3)

272

(3)

189

(3)

207

(2)

0.0114

0.0084

0.0057

0.0167

Lue2

614

(3)

499

(2)

581

(3)

319

(3)

262

(3)

179

(3)

197

(2)

10

(1)

0.0192

0.0099

0.0206

Yong

697

(3)

601

(2)

676

(3)

105

(3)

41

(3)

41

(3)

84

(2)

230

(2)

220

(2)

0.0100

0.0091

Khuen

710

(3)

613

(2)

689

(3)

101

(3)

46

(3)

41

(3)

69

(2)

229

(2)

219

(1)

16

(2)

0.0165

Shan

809

(3)

705

(2)

784

(3)

140

(3)

143

(3)

131

(3)

62

(2)

256

(2)

247

(2)

133

(2)

116

(2)

1288 Chiang Mai J. Sci. 2017; 44(4)

A model-based clustering method wasalso performed to clarify populationsubstructures. Due to the low level ofgenetic divergence among the populations inwhich F

st ranged from 0.00283 to 0.03606,

STRUCTURE analysis was executed usinga sampling location as the priori, whichprovided more information [38]. The highestoutput posterior probability averagedover five replicates [26] as well as ad hoc statisticdelta K which was calculated from the rateof change in the log probability of databetween successive K values [30] wereused to determine the most appropriateconfiguration. This was observed atK=8 (Supplementary 1). The populationsubstructure of representative runs at K=2to 8 are shown in Figure 3. No clearseparation of distinct population wasdetected at K=2, while the red-coloredcomponent initially emerged in the Lawa2at K=3. When K was continuously releasedto 4, 5, 6 and 7, the green, orange, purple,

and white components were predominantlyin Kuy, Lawa1, Mon, and Northern Khmer,respectively. Until K=8 which was the bestvalue of number of clusters that representthe structure of the data, weakly distinguishedcluster was formed in the Lue1 representedby black color. Therefore, at K=8, theNorthern Khmer, Kuy, Lawa1, Lawa2formed different genetic clusters from eachother and from the rest. The fifth clustermainly formed by Lao, Yuan, Yong andShan although each population show someintermixture with different clusters, forexamples, the influence of Northern Khmercluster in the Lao and the introgression ofLue cluster in the Yuan. The Lue1, Lue2,and Khuen were assigned to the sixth cluster.The seventh cluster mainly existed in theMon (purple) albeit the Mon was mixed bythe yellow and blue clusters. The last extremelyweak cluster showed in Lue1 where around21.7% consisted in this cluster.

Figure 2. The Factorial Correspondence Analysis (FCA) representing population relatedness.Each population was plotted into 3-dimensional plot. Axis 1, 2 and 3 accounts for 15.69%,14.27%, and 11.45%, respectively.

Chiang Mai J. Sci. 2017; 44(4) 1289

Figure 3. Clustering assignments depending on Bayesian method under an admixture withcorrelation between allele frequencies and LOCPRIOR models obtained by STRUCTURE.Each individual is represented by a single column that is divided into segments whose size andcolor correspond to the relative proportion of a particular cluster. Populations are separatedby black lines. The r value to the right of each DISTRUCT plot is lower than 1 indicating theinformativeness of sampling location in these analyses.

According to SAMOVA analysis whichhas been used to elucidate further detailedgenetic structure and differentiation (Table 6),when the putative number of populations wasincreasing from 2-groups until 7-groupscategory, the Kuy, Lawa2, Northern Khmer,Lawa1, Lue2, Mon were partitioned fromthe other populations, respectively (Table 6).The former separated population in lowernumber of populations, the higher genetic

differentiation. Although the maximization ofthe genetic differentiation among groups wasachieved for ten population clusters (1.408%,P<0.01), the major increase on percentvariation among group occurred for 8 groups,with value only increasing slightly thereafterSAMOVA result were broadly consistentwith STRUCTURE and FCA which reflectthe largely genetic divergence of the Kuy andLawa2 than the Northern Khmer.

1290 Chiang Mai J. Sci. 2017; 44(4)

Table 6. Groups of populations and fixation indices as inferred by SAMOVA.

Bold letter indicated P < 0.01

Fct = Fixation indices among groups of population

No. of

group

2

3

4

5

6

7

8

9

10

11

Population member

Kuy

Kuy

Kuy

Kuy

Kuy

Kuy

Kuy

Kuy

Kuy

Kuy

Northern Khmer, Lao, Lawa1, Lawa2, Mon, Yuan, Lue1, Lue2, Yong, Khuen, Shan

Fct

0.01151

0.01267

0.01249

0.01214

0.01222

0.01263

0.01324

0.01368

0.01408

0.01371

Lawa2

Lawa2

Lawa2

Lawa2

Lawa2

Lawa2

Lawa2

Lawa2

Lawa2

Northern Khmer, Lao, Lawa1, Mon, Yuan, Lue1, Lue2, Yong, Khuen, Shan

Northern

Khmer

Northern

Khmer

Northern

Khmer

Northern

Khmer

Northern

Khmer

Northern

Khmer

Northern

Khmer

Northern

Khmer

Lao, Lawa1, Mon, Yuan, Lue1, Lue2, Yong, Khuen, Shan

Lawa1

Lawa1

Lawa1

Lawa1

Lawa1

Lawa1

Lawa1

Lao, Mon, Yuan, Lue1, Lue2, Yong, Khuen, Shan

Lue2

Lue2

Lue2

Lue2

Lue2

Lue2

Lao, Mon, Yuan, Lue1, Yong, Khuen, Shan

Mon

Mon

Mon

Mon

Mon

Lao, Yuan, Lue1, Yong, Khuen, Shan

Lue1,

Khuen

Lue1,

Khuen

Lue1

Lue1,

Yuan

Lao, Yuan, Yong, Shan

Shan

Khuen

Khuen

Lao, Yuan, Yong

Shan

Shan

Lao, Yuan, Yong

Lao Yong

4. DISCUSSION

The results from all our analyses agree inrevealing a genetic divergence of the Kuyand the Northern Khmer populations fromthe other ethnolinguistic groups in northeasternThailand. This finding is highly compatiblewith the linguistic classification of theAustroasiatic and Kra-Dai language family.As shown by Axis 1 of the FCA (Figure 2),the Kuy and the Northern Khmer populationsin this study show a close genetic relationshipto each other but are genetically quitedistant from the other Austroasiatic groups.The clustering of Kuy and Northern Khmerneatly reflects the fact that the languages ofthe two groups are often classified togetherin the same major branch of the Austroasiaticfamily [39-40]. The genetic distance betweenthe Kuy and the Northern Khmer on one

hand, and the other populations studiedon the other is in sharp contrast to thehomogeneity among the Kra-Dai groups.Again, this close genetic relationship isexpected given that the languages of allpopulations included in the current studyall belong to the same subgroup within theTai branch of the Kra-Dai family [41-43].

In addition to the agreement betweengenetic and linguistic groupings, this studyalso reveals a genetic admixture coherentwith the history of our populations. Whencryptic population substructure was exploredby STRUCTURE, the Northern Khmerseems to share some genetic componentswith the Lao Isan, the majority in northeasternThailand. The admixed components in LaoIsan, composing of green, blue, yellow,and white components belong to the Kuy,

Chiang Mai J. Sci. 2017; 44(4) 1291

(Yuan-Yong-Shan), (Lue2-Khuen), andNorthern Khmer, respectively. The resultindicated that the Northern Khmer was oneof the parental populations who contributedgenes (approximately 31%) to the Lao Isanfrom Roi-Ed Province. Gene flow betweenthese two ethnicities has also been previouslyreported [44]. The admixture between thesetwo populations is consistent with the historyof the region. More specifically, biculturalismcomprising the Northern Khmer and theLao Isan cultures has existed in the lowerpart of Northeastern Thailand for a long time[10]. Moreover, the numerous monuments ofthe Khmer civilization in the area suggest apresence of a Khmer population prior tothe mass migration of the Lao from Laosduring from the 14th to 18th century A.D. [4].This suggests a long-lasting co-existence ofthe native Khmer and the immigrant Laothat resulted in a high degree of populationmixing.

The genetic drift which accounted forlosing maternal genetic diversity in thestudied Northern Khmer from SurinProvince has been previously reported [45].Such circumstance was erased by this studywith autosomal marker. A high value of allgenetic variation indices was observed(Table 1). Since mtDNA markers have 1/4the effective sample size of autosomal markerscarrying the information from both parents,mtDNA has been particularly subject to theeffects of random genetic drift. Therefore,we prove that the panel of STRs in the forensicstudy was an informative genetic marker indistinguishing Austroasiatic populations inThailand. This agrees with previous study[46] which reported that this marker set ishighly informative in separating more recentlydiverged Asian minority populations.However, other modes of inheritance areneeded to investigate genetic structure andreconstruct population history. Again, to get

more insight into Northeastern Thaipopulations, the paternal genetic marker isrequired for investigation.

One puzzle is the lack of admixturebetween Kuy and either Lao Isan orNorthern Khmer, which indicates a lack ofgene flow from or into the studied Kuypopulation. This finding is unexpectedgiven the extensive language contactbetween the two groups reported in theliteratures [9, 13]. Theoretically speaking,both languages and genetics can be transmittedfrom parents to their offspring. As such,concordance between genetics and languageis expected for cases in which genes andlanguage have the same evolutionaryprocess. To understand the lack of gene flowin the face of extensive language contact,at least two explanations are possible.The first one has to do with history oflanguage shift and maintenance in the lowerpart of northeastern Thailand. Although theKuy language in Thailand is still relativelyvibrant, its territory has shrunk significantlyas many communities have been shifting tospeak Lao or Northern Khmer [9, 11, 47].If intermarriage accelerates language shiftas it has often been reported [48, 49], a Kuypopulation that has maintained their originallanguage despite a widespread shift ispossibly less open to marriage to individualsfrom other ethnolinguistic groups. If thiswas indeed the case, we would not expectto see clear traces of gene flow in the Kuypopulation in our study. To test thishypothesis, data from other Kuy-speakingpopulations from neighboring provincesfrom Thailand and LPRD are needed.

Another possible explanation involvespatterns of residence after marriage.The Kuy are one of the few remainingmatrilocal residence societies in Thailandand Laos. Matrilocal residence means thatafter the marriage the females remain in their

1292 Chiang Mai J. Sci. 2017; 44(4)

natal villages and the males move to their wife’svillage. Marital residence pattern has beendemonstrated as one of the cultural factorsaffecting genetic variation in severalpopulations both outside and inside Thailand[50-51]. Large genetic distances betweengroups exist in matrilocal groups. The geneticuniqueness of the Kuy revealed by autosomewas similar to our previous studies whichused mitochondrial DNA as genetic markers[14]. We also tested the correlation betweenthe matrices of F

st from the autosomal STRs

and mtDNA sequences which showed astrong correlation between distance matricesof these two different markers in theirmode of inheritance (r = 0.5718, P<0.01).The tremendous genetic differentiation ofthe Kuy exhibited by our past and presentstudies might be affected by matrilocalresidence, however, future studies on Ychromosomal variation is needed to confirmour assumption.

In conclusion, our results demonstrateextreme genetic divergence among theKuy and moderate genetic differentiationamong the Northern Khmer. Both studiedpopulations also have genetic distinctionthough their geographic proximity. The Manteltest seems to support language as thepredominant factor in shaping geneticvariation in overall populations includingthe studied and comparable populations,however, AMOVA exhibited contrary resultsindicating that neither language nor geographywas concerned in determining patternsof genetic variation. This trend might bethe result of the very heterogeneous geneticstructure of Austroasiatic groups. However,genetic homogeneity existed in the Kra-Daipopulations, therefore, inconclusive addresswas proposed for whether geography orlanguage was more important in this study.However, to our knowledge, it seems likelythat genetic divergence of the Kuy might be

due to their history of language shift andmaintenance in the area or the matrilocaltradition practiced in Kuy society. Languageadoption but without genetic introgressionfrom the Northern Khmer to the Kuy orvice versa might be the possible reason.Little fraction of the Kuy componentprevailed in the Lao Isan genes, reflectingminor contribution from the Kuy to Lao Isan.On the contrary, the Northern Khmer havecontributed a high genetic fraction to theLao Isan gene pool, indicating a geneticexchange between these two linguisticallydifferent populations concordant witharchaeological and anthropological records.

ACKNOWLEDGEMENTS

We would like to thank all volunteerswho donated their buccal cells and villagechiefs for their participation. PP wasfunded by the RatchadaphiseksomphotEndowment Fund of ChulalongkornUniversity (RES560530179-HS). WK wassupported by Khon Kaen University(Grant number 590043).

DECLARATION OF INTEREST

The authors report no declarations ofinterest.

REFERENCES

[1] O’Connor R.A., J. Asian Stud., 1995; 54:968-996.

[2] Pittayaporn P., J. Humanities, 2014; 20:47-68.

[3] Premsrirat S., Deepadung S., BuasuangA., Suwanket E., Choosri I., Srijampa S.,et al., Ethnolinguistic Maps of Thailand.Institute of Language and Culture forRural, Development: Mahidol University,2004.

[4] Keyes C., Isan: Regionalism in NortheasternThailand, Department of Asian Studies,

Chiang Mai J. Sci. 2017; 44(4) 1293

Southeast Asia Program, CornellUniversity, New York, 1967.

[5] Wiphakphotchanakit T., Isan History,The Social Science Association ofThailand, Bangkok, 1999.

[6] Coedes G., The Indianized States of SoutheastAsia, East-West Center, Honolulu, 1968.

[7] Vallibhotama S., ‘Srisaket: the area of the“backward Cambodians”’, RuangBoran,1989; 15(4): 27-50.

[8] Thomas D., J. Lang. Cult., 1990; 9:98-106.

[9] Smalley W., Linguistic Diversity and NationalUnity: Language Ecolog y in Thailand,University of California Press, Berkeley,1994.

[10] Vail P., Asian Ethnicity, 2007; 8: 111-130.

[11] Van Der Haak F. and Woykos B.,Mon-Khmer Studies, 1990; 16-17: 109-142.

[12] Sa-ard O., Phrase to Sentence in Kuay (Surin),Master Thesis, Mahidol University,Thailand, 1984.

[13] Premsrirat S., Mon-Khmer Stud., 1997; 27:129-136.

[14] Kutanan W., Ghirotto S., Bertorelle G.,Srithawong S., Srithongdaeng K.,Pontham N. and Kangwanpong D.,J. Hum. Genet., 2014b; 59: 512-520.

[15] Crossetti S.G., Demarchi D.A.,Raimann P.E., Salzano F.M., Hutz M.H.and Callegari-Jacques S.M., Am. J. Hum.Biol., 2008; 20: 704-711.

[16] Gaibar M., Esteban E., Moral P.,Gomez-Gallego F., Santiago C., et al.,Ann. Hum. Biol., 2010; 37: 253-266.

[17] Babiker H.M.A., Schlebusch C.M.,Hassan H.Y. and Jakobsson M., Invest.Gen., 2011; 2: 12.

[18] Kraaijenbrink T., Van der Gaag K.J.,Zuniga S.B., Xue Y., Carvalho-Silva D.R.,et al., PLoS ONE., 2014; 9: e91534.

[19] Kutanan W., Kampuansai J., Colonna V.,Nakbunlung S., Lertvicha P., Seielstad M.,Bertorelle G. and Kangwanpong D.,J. Hum. Genet., 2011b; 56: 130-137.

[20] Srithawong S., Srikummool M.,Pittayaporn P., Ghirotto S.,Chantawannakul P., Sun J., Eisenberg A.,Chakraborty R. and Kutanan W., J. Hum.Genet., 2015, 60: 371-380.

[21] Excoffier L. and Lischer H.E., Mol. Ecol.Resour., 2010; 10: 564-567.

[22] Rice W.R., Evolution, 1989; 43: 223-225.

[23] Excoffier L., Smouse P.E. and QuattroJ.M., Genetics, 1992; 131: 479-491.

[24] Dupanloup I., Schneider S. andExcoffier L., Mol. Ecol., 2002; 11:2571-2581.

[25] Belkhir K., Borsa P., Chikhi L.,Goudet J. and Bonhomme F., Genetix4.00 WindowsTM software for sample genetics,Laboratoire G nome, Populations,Interactions, University of Montpellier,France, 2004.

[26] Pritchard J., Stephens M. and DonnellyP., Genetics, 2000; 155: 945-959.

[27] Falush D., Stephens M. and PritchardJ.K., Genetics, 2003; 164: 1567-1587.

[28] Hubisz M., Falush D., Stephens M. andPritchard J., Mol. Ecol. Resour., 2009; 9:1322-1332.

[29] Earl D.A. and VonHoldt B.M., Conserv.Genet. Resour., 2011; 4: 359-361.

[30] Evanno G., Regnaut S. and Goudet J.,Mol. Ecol., 2005; 14: 2611-2620.

[31] Rosenberg N.A., Mol. Ecol. Notes, 2003;4: 137-138.

[32] Mantel N., Cancer Res., 1967; 27: 209-220.

[33] Smouse P.E. and Long J.C., Yearbook Phys.Anthropol., 1992; 35: 187-213.

[34] Kutanan W., Kampuansai J., Fuselli S.,Nakbunlung S., Seielstad M., Bertorelle

1294 Chiang Mai J. Sci. 2017; 44(4)

G. and Kangwanpong D., BMC Genet.,2011a; 12: 56.

[35] Zerjal T., Beckman L., Beckman G.,Mikelsaar A.V., Krumina A., KucinskasV., Hurles M.E. and Tyler-Smith C.,Mol. Biol. Evol., 2001; 8: 1077-1087.

[36] Helgason A., Yngvadottir B.,Hrafnkelsson B., Gulcher J. andStef nsson K., Nat. Genet., 2004; 37:90-95.

[37] Pardi as A.F, Roca A., Garc a-VazquezE. and L pez B., PLoS ONE, 2012; 7:e50206.

[38] Latch E., Dharmarajan G., Glaubitz J.and Rhodes O.Jr., Conserv. Genet., 2006;7: 295-302.

[39] Diff loth G., The Contribution ofLinguistic Palaeontology to theHomeland of Austro-asiatic, ThePeopling of East Asia: Putting TogetherArchaeology, Linguistics and Genetics; inSagart L., Blench R. and Sanchez-MazasA., eds., Curzon: Routledge, 2005.

[40] Sidwell P., Issues in AustroasiaticClassification. Language and LinguisticsCompass, 2013; 7: 437-457.

[41] Li F.K., Handbook of Comparative Tai,University of Hawai’i Press, Honolulu,1977.

[42] Edmondson J.A. and Solnit D.B.,Introduction; In Edmondson J.A andSolnit D.B., eds., Comparative Kadai: the

Tai branch. Arlington, TX: The SummerInstitute of Linguistics and the Universityof Texas at Arlington, 1997: 1-26.

[43] Pittayaporn P., Phonology of Proto-Tai,Ph.D, Thesis, Cornell University, NewYork, 2009.

[44] Kutanan W., Srithawong S., Kamlao A.and Kampuansai J., Adv. Anthropol.,2014; 4: 7-12.

[45] Boonsoda P., Srithawong S., Srikuka S.and Kutanan W., Thai J. Genet., 2013; 6:40-48 (in Thai).

[46] Listman J.B., Biases in Study Design Affectingthe Inference of Evolutionary Events andPopulation Structure in Closely-related HumanPopulations, New York University, 2009.

[47] Seidenfaden E., J. Siam Society, 1952; 39:144-180.

[48] Castonguay C., Can. J. Sociol., 1982; 7:263-277.

[49] Stevens G., Am. Sociological Rev., 1985;50(1): 74-83.

[50] Kumar V., Langstieh B.T., Madhavi K.V.,Naidu V.M., Singh H.P., Bisward S.,Thangaraj K., Singh L. and Reddy M.,PLoS Genet., 2006; 2: e53.

[51] Besaggio D., Fuselli S., Srikummool M.,Kampuansai J., Castr L., Tyler-Smith C.,Seielstad M., Kangwanpong D. andBertorelle G., BMC Evol. Biol., 2007; 7(2):S12.