Multilocus Genotyping to Study Population Structure in Three Buffalo Populations of India

Three buffalo populations viz. Bhadawari, Tarai and local buffaloes of Kerala were genotyped using 24 heterologous polymorphic microsatellite loci. A total of 140 alleles were observed with an average observed heterozygosity of 0.63. All the loci were neutral and 18 out of the 24 loci were in Hardy Weinberg Equilibrium. The FIS values (estimate of inbreeding) for 16 loci in all the three populations were negative. This indicated lack of population structure in the three populations. The effective number of immigrants was 5.88 per generation between the Tarai and Bhadawari populations which was quite high suggesting substantial gene flow. The genetic distances revealed closeness between the Tarai and Bhadawari populations which was expected from geographical contiguity. The FST values were not significantly different from zero showing no population differentiation. The Correspondence Analysis based on the allelic frequency data clustered the majority of the Tarai and Bhadawari individuals as an admixture. (


INTRODUCTION
Microsatellite are short segments of DNA of a specific motif made up of 1-6 base repeats which may be repeated upto 50-60 times.Microsatellite markers have exceptional variability, are easy to score and thus are considered to be the most powerful genetic markers for population studies.It is typical for microsatellite loci to have 10 or more alleles with a heterozygosity values of 0.6 or more which can be detected even from relatively small samples (Bowcock et al., 1994;Primmer et al., 1996).Microsatellite are rapidly replacing other polymorphic markers for identifying relatives to infer various demographic parameters (Bloom et al., 1996;Goldstein et al., 1996;James and Lagoda, 1996;Fan et al., 2005).The primers are designed to amplify the genomic regions which include a well defined repeat structure responsible for the observed variations.The variation at the locus with the repeat structure allows for the development of inferential methods based on explicit models of microsatellite evolution.Microsatellites are the best markers for population genetic studies and demographic inferences.
India has 10 recognized breeds of buffaloes distributed in the various agro climatic regions of the country.Most of the buffaloes are reared for milk production and have their breeding tracts in the north and the northwestern regions of the country.In addition to the recognized breeds, there are several distinct populations of buffaloes in various parts of the country.Three populations were used for this study.Among these, the Bhadawari breed is an established breed which inhabits the borders of Uttar Pradesh and Madhya Pradesh.Tarai buffaloes are medium sized and found in the foot hills of Uttaranchal state, Bareilly and Pilibhit districts of Uttar Pradesh.The distributions of Tarai and Bhadawari buffaloes are in geographical contiguity.Another distinct population in the southernmost part of India is the Kerala buffalo locally known as Kuttanad buffalo.These buffaloes are mainly utilized as work animals in the paddy fields.
They are considered to be good swimmers (Anilkumar and Raghunandanan, 2003).The animals in Kuttanad area are of short stature with an average height of 109 cm.These animals yield 1-2 kg of milk and are reared as work animals.
The present study was undertaken to find out the genetic parameters of the three buffalo populations.

MATERIALS AND METHODS
Blood samples were collected from the breeding tract of three buffalo populations.Ten ml of whole blood was collected from jugular vein of each animal using heparinised vacutainer tubes and transported to the lab at 0-5°C.
The primer sequences, annealing temperature, MgCl 2 concentration, size range of PCR products along with the total number of alleles observed in the three populations are given in Table 1.The amplified PCR products from the heterologous markers were sequenced using ABI Avant 3100 to confirm the sequences of the microsatellite repeats.A total of 104 individuals were utilized for the study; 40 animals each of Bhadawari and Tarai, and 24 animals of Kerala buffalo.
The PCR conditions were standardized for all of the 24 primer pairs selected for the study.The variables, which required standardization, included annealing temperature, MgCl 2 concentration, quantity of primer, Taq polymerase and dNTP's.The PCR products were loaded on 6% denatured polyacrylamide gel with urea as the denaturing agent.The standard DNA markers were simultaneously run on the gel for sizing of the alleles.The polyacrylamide gel electrophoresis was run for a sufficiently long period for proper resolution of the alleles.The polyacrylamide gel was fixed in acetic acid (10%) and stained with silver nitrate following a standard protocol (Bassam et al., 1991).The size of the alleles were estimated by making a standard curve taking log 10 of the molecular weights of the standard markers on the X-axis and the mobilities of the DNA bands on the Y-axis.The sizes of the alleles were calculated from the standard curve.

Statistics
The data for the 24 microsatellite loci in the three populations were subjected to statistical analysis.The locus and population wide gene frequencies, the number of alleles, effective number of alleles, and observed and expected heterozygosities were calculated with the Popgene software (Yeh et al., 1999).The Ewens Watterson test of neutrality was also carried out using 10,000 permutations.
The Hardy-Weinberg equilibria of the loci in the three populations were tested against null hypothesis of random union of gametes.The U test with heterozygotic deficiency as the alternative hypothesis was carried out using the GENEPOP software (Raymond and Rousset, 2003).The software performed a probability test utilizing Markovs' chain (dememorization 5,000, batches 100, iterations per batch 1,000).Significance levels were calculated per locus, per population and over all loci and all populations combined.Two estimates of F IS (Weir and Cockerham 1984;Robertson and Hill, 1984) were estimated.
The F statistics values F IS , F IT and F ST were estimated using Jack-knifing over loci and the confidence interval were generated using 10,000 permutations with the GDA software (Lewis and Zaykin, 2002).The number of migrants (Nm) was estimated using Nm = 0.25 (1-F ST )/F ST.
The genotypic linkage disequilibrium was estimated between all the possible pairs.The significance of population differences was tested using the exact test of population differentiation in the GENEPOP software based on the allele frequency data.
The correspondence analysis which is a weighted principal component analysis was performed using the allele frequency data for the individuals of the three populations and the 24 loci using GENETIX software (Belkhir et al., 1998).
Two genetic distances viz.Nei's standard and Nei's D A were calculated using the DISPAN software (Ota, 1993) and (δµ) 2 was estimated with the MICROSAT software (Minch  , 1996).The dendrograms were constructed by using the PHYLIP software (Felsenstein, 1993).

RESULTS AND DISCUSSION
In total, 140 alleles were observed from the 24 loci analysed.The number of alleles per locus ranged from 3 (ILST 49, ILST 72, ILST 87) to 9 (CSRM 60, CSSM 47) with a mean of 5.83 alleles.The number of alleles was similar to Indian water buffaloes (Arora et al., 2004), but different from Sri Lankan water buffaloes (Barker et al., 1997).The comparative allele frequencies are given in Figure 1.Nine loci out of the 24 had one allele which was predominant in all the populations.The expected heterozygosity varied between 0.1350 for ILSTS 87 and 0.834 for CSRM 60 with a mean value of 0.6337 across the three populations (Table 2).There was no linkage disequilibrium among the 24 loci utilized in this study.18 loci in the three populations were in HW equilibrium.ILSTS 29 was the only locus which was not in HWE in all the three populations (Table 3).The population wide F IS (Robertson and Hill, 1984) values were negative for 16, 15 and 17 loci in Bhadawari, Kerala and Tarai populations, respectively.This indicated lack of population structure in all the three populations.The private alleles identified in the three populations are given in Table 4.
Heterozygosity is defined as the probability that a given individual randomly sampled from a population will be heterozygous at a given locus.The statistics F ST is an estimate of the variation due to differentiation among populations, which is the reduction in heterozygosity of a population due to genetic drift.The statistics F IS is an estimate of variation with in populations that measures the reduction in heterozygosity in an individual due to nonrandom mating within sub populations.F IT is the over all inbreeding coefficient of an individual relative to the total population.This includes the contribution due to nonrandom mating within sub populations (F IS ) and that due to population subdivision (F ST ).The mean values of F, θ (estimate of population differentiation-F IT ) and f (within population inbreeding estimate-F IS ) from Jackknifing over loci were 0.0092, 0.0648 and -0.0593, respectively (Table 5).All the three values did not deviate significantly from  zero and lay between upper and lower bound calculated on the basis of 10,000 replicates.The negative value of F IS indicated mating between individuals which are less closely related than the average relationship in the population.The F IS values of most of the loci were negative indicating outbreeding or mating with migrants.This could be explained by the indiscriminate use of Murrah bulls in all the three populations.This view was supported by the higher values of Nm obtained in the present study.Such possibilities have also been reported by Pundir et al. (1997) and Sethi (2001) on the basis of surveys of the areas.The pairwise F ST values estimated were 0.1024 (between Bhadawari and Kerala) 0.0408 (between Bhadawari and Tarai) and 0.0644 (between Kerala and Tarai) buffalo populations.These values were obtained using exact test for population differentiation.The F ST values were significant between Bhadawari and Kerala, while the values were nonsignificant between Bhadawari and Tarai, and Kerala and Tarai populations.The three populations were tested for genotypic differentiation using the null hypothesis that the genotypic distribution was identical across populations.The unbiased estimates of the p-value of the log likelihood (G) based exact test revealed that the null hypothesis was accepted in 17 of the 24 loci studied.There was no genotypic differentiation for the loci ILSTS 087, CSSM 029, CSSM 019, HAUT 027, ILSTS 030, BM 1818 and ILSTS 072.The overall X 2 value was infinity at df 48 and the genotypic distribution was significantly different across all the populations.The genic differentiation using the null hypothesis that the allelic distribution was identical across populations was not accepted, suggesting a significant difference in the allelic distributions among populations.
The three genetic distances, Nei's standard and D A , and Goldstein (δµ 2 ), also revealed closeness between Bhadawari and Tarai populations, while Kerala buffaloes were distant (Table 6).These were in agreement with the genetic distances reported by Arora et al. (2004) and Vijh et al. (2005).The tree of inter-individual genetic distance (Nei' D A ) using the Neighbour Joining algorithm revealed that all the 24 Kerala individuals clustered together while the Tarai and Bhadawari individuals grouped in distinct and mixed clusters (Figure 2).The genetic basis of the samples was quite wide since using Nei's D A 20% of the individuals did not cluster in their own populations.It has been used by several authors for clustering purposes (Cho, 2005).
The Correspondence Analysis (which is a weighted Principle Component Analysis) was performed on the three populations using allele frequencies of the 24 loci to summarize the breed relationships.Figure 3 shows close relationship among individual belonging to the Bhadawari and Tarai populations.The first four factors contributed a total of 17.3% with value of 6.08, 4.43, 3.67 and 3.11% respectively.The analysis indicated that all individuals of Kerala buffaloes clustered together while two clusters one each for Bhadawari and Tarai buffaloes were found.Some of the individuals of these latter two populations were clearly overlapping with one another and thus form another group.
The number of migrants per generation based on F ST values were lowest (2.19) between Kerala and Bhadawari.The values were highest (5.88) between the Bhadawari and Tarai populations.The estimated Nm value could be explained on the basis of contiguity of the breeding tract between these two populations.The large number of migrants detected in the Tarai and Bhadawari populations might be due to larger exchange of individuals between these two populations or use of Murrah buffaloes in both populations.The number of migrants, estimated by using the private alleles as implemented in GENEPOP (Barton and Slatkin, 1983) was 1.191.This could be because of few private alleles among the three populations studied.
The study indicated that there is an abundance of within population genetic variation in the three buffalo populations.This is further supported by the fact that the genetic distance among the three populations were quite low.The homogeneity among the breeds may be due to immigration as there is excess of heterozygosity in all the three  populations.The Nm values obtained from F ST also supported this.However, the explanation to such a situation is that since only three populations were taken for the study it is very possible that the migrants came from populations other than these three.It is well known that to improve milk production there has been indiscriminate use of Murrah bulls/ semen in the breeding tract of the three populations under investigation.

Figure 1 .
Figure 1.No of alleles and their frequencies in the three populations studied.

Figure 3 .
Figure 3. Clustering of individuals based on correspondence analysis.

Table 1 .
Details of the microsatellite primers used S. No.

Table 2 .
Heterozygosities in the three buffalo populations studied

Table 3 .
F IS values for microsatellite loci in the three buffalo populations studied Robertson and Hill (1984)eir and Cockerham (1984); 2Robertson and Hill (1984).

Table 4 .
Private alleles in the three buffalo populations studied

Table 5 .
F-Statistics over all the populations