Introduction
The Indian horse breeds are distinct not only because of their adaptation to different agro-climatic conditions prevailing in the country, but also because they have unique traits such as sturdiness, stiffness, endurance potential, relative disease resistance etc. However, the changed scenario after development of the road network and mechanisation combined with indiscriminate breeding with exotic or nondescript animals has led to drastic decline in the populations of these breeds. Since, presently only a few thousand true breeding horses of each of these breeds are available (Singhvi, Reference Singhvi2001; Yadav et al., Reference Yadav, Ghei and Tandon2001), it is necessary to evolve strategies for their conservation. The evaluation of genetic diversity/relationships among livestock breeds is an important prerequisite for developing cost-effective and meaningful breed conservation/improvement programmes. The microsatellite DNA markers, due to their highly polymorphic nature, have been extensively employed in the analysis of genetic diversity amongst breeds of various livestock species including horses (Bjornstad et al., Reference Bjornstad, Gunby and Roed2000; Cañón et al., Reference Cañón, Checa, Carleos, Vega-Pla, Vallejo and Dunner2000; Kelly et al., Reference Kelly, Postiglioni, DeAndres, Vega-Pla, Gagliardi, Biagetti and Franco2002; Tozaki et al., Reference Tozaki, Takezaki, Hasegawa, Ishida, Kurusawa, Saitou and Mukoyama2003; Achmann et al., Reference Achmann, Curik, Dovc, Kavar, Bodo, Habe, Marti, Solkner and Brem2004; Aberle et al., Reference Aberle, Hamann, Drogemuller and Distl2004; Solis et al., Reference Solis, Jugo, Meriaux, Iriondo, Mazon, Aguirre, Vicario and Estomba2005; Glowatzki-Mullis et al., Reference Glowatzki-Mullis, Muntwyler, Pfister, Marti, Rieder, Poncet and Gaillard2006). The present study was undertaken to characterise five Indian horse breeds for genetic variation and to establish relationships amongst them using a set of 25 microsatellite markers. The Thoroughbred horses, which are most common exotic horses in India, were also included in our study as an outgroup.
Material and methods
Samples
The blood samples were collected from 189 horses of five Indian horse breeds from their respective areas of distribution (Figure 1). The breeds involved and their sample sizes were: Marwari (42), Spiti (32), Bhutia (26), Manipuri (47) and Zanskari (42). The blood samples were also collected from Thoroughbred (24) horses from Haryana state. The Marwari horses are native to Marwar region of Rajsthan province and are supposed to have been evolved to fulfil the needs of erstwhile local princely state. The present population of Marwari horses is estimated to be less than 3000 (Singhvi, Reference Singhvi2001; Singh et al., Reference Singh, Yadav and Mehta2002). The other four Indian breeds included in the study are small sized and classified as ponies (Bhat et al., Reference Bhat, Bhat, Khan, Goswami and Singh1981). These pony breeds have close resemblance with Tibetan ponies. The Zanskari ponies are found in Zanskar and Ladakh areas of Jammu and Kashmir. They are small-sized animals with compact bodies and strong legs. They are known for their hardiness and well adapted to work at these high altitude areas of the Himalayas located between 3000 to 5000 m altitude. They are mainly used for transportation and agricultural operations. The Zanskari breed is at the verge of extinction as only a few hundred horses of this breed exist now (Yadav et al., Reference Yadav, Ghei and Tandon2001). The Spiti and Bhutia ponies with similar characteristics are also found in same kind of agro-climatic conditions. Though, the Spiti horses are distributed in Lahaul/Spiti, Kinnour and Pangi areas of Himachal Pradesh, they are mainly bred in a few hamlets of Pin valley using traditional selection practices for identifying males for breeding. There present population is estimated to be less than 3000 (Katoch et al., Reference Katoch, Dogra, Thakur and Gupta2004; Behl et al., Reference Behl, Pundir, Behl, Gupta, Gupta, Singh, Katoch, Dogra and Ahlawat2005). The Bhutia ponies with their estimated population of less than 5000 are distributed in the Middle/Eastern Himalayas all along the Tibet border reared by the Bhutia tribe (Bhat et al., Reference Bhat, Bhat, Khan, Goswami and Singh1981; Yadav et al., Reference Yadav, Ghei and Tandon2001). The Manipuri ponies with a present population of 2327, are found in Manipur province in north-east India. The Manipuri ponies are referred to as original polo ponies. They are evolved from ponies brought from Tibet around 1200 years ago (Anonymous, 2006). All these Indian horse breeds have been listed as threatened breeds.
Figure 1 The areas of distribution of five Indian horse breeds.
The genomic DNA was isolated from collected samples by standard procedure of digestion with proteinase-K, separation with phenol/chloroform/isoamylalcohol and precipitation with ethanol. The isolated DNA samples were stored at − 20°C and working dilutions were stored at 4°C.
PCR amplification
The genomic DNA was amplified by polymerase chain reaction (PCR) using 25 equine microsatellite loci (Table 1) using the protocol described in Crawford et al. (Reference Crawford, Dodds, Ede, Pierson, Montgomery, Garmonswa, Beattie, Davies, Maddox and Kappes1995). The amplified DNA fragments were analysed on 7% denaturing polyacrylamide gel and detected by silver staining (Bassam et al., Reference Bassam, Caetano-Anolles and Gresshoff1991). Alleles were scored manually against DNA size markers and known samples used as standards on every gel.
Table 1 PCR product size range (bp), observed number of alleles, observed heterozygosity, polymorphism information content and FST for 25 microsatellite loci in five Indian horse breeds
Statistical analysis
The allele frequencies, observed/effective number of alleles and observed/expected heterozygosities for each locus were calculated using POPGENE computer program (Yeh et al., Reference Yeh, Yang and Boyle1999). The polymorphism information content (PIC) was calculated as described by Botstein et al. (Reference Botstein, White, Skolnick and Davis1980). The tests for departure from Hardy-Weinberg equilibrium (HWE) and linkage disequilibrium between loci were performed using exact probability tests provided in GENEPOP version 3.4 (Raymond and Rousset, Reference Raymond and Rousset1995). Monte Carlo method (Gou and Thompson, Reference Gou and Thompson1992) was applied to compute unbiased estimate of the exact probabilities (P value). Length of chain was set to be 50 000 iterations with critical P value adjusted to 0.05 on population level.
Using variance based method of Weir and Cockerham (Reference Weir and Cockerham1984), population differentiation by F statistics was computed using FSTAT version 2.9.3 computer program (Goudet, Reference Goudet2001). Mean and standard deviation of F statistics parameters, θ, F, f, that are analogous to Wright's (Reference Wright1969) FST FIT and FIS, respectively, were obtained across breeds by jackknifing procedure over loci (Weir, Reference Weir1990). The level of significance (P < 0.05) was determined from permutation test with sequential Bonferroni procedures applied over all loci. The Nei's DA genetic distances (DA, Nei et al., Reference Nei, Tajima and Tateno1983) and Reynolds' genetic distances (DR, Reynolds et al., Reference Reynolds, Weir and Cockerham1983) between pairs of populations and neighbour joining tree between breeds were generated with POPULATIONS package (Langella, Reference Langella2002). The phylogenetic tree was visualised using TREEVIEW computer program (Page, Reference Page1996).
Principal component analysis was performed for each population from allele frequency data according to the procedures described by Cavalli-Sforza et al. (Reference Cavalli-Sforza, Mennozi and Piazza1994). The data for individual genotypes were prepared by scoring a ‘0’ if a particular allele was absent, ‘1’ if it was present in one copy and ‘2’ if it was homozygous. To further decipher the question that how many breeds are actually present, the data was analysed using STRUCTURE computer program (version 2, Pritchard et al., Reference Pritchard, Stephens and Donnelly2000) applying Markov chain Monte Carlo method without admixture using a burn in period of 30 000 iterations and data was collected after 106 iterations assuming number of breeds (K) between 1 and 6.
Results and discussion
All the loci reported in the study amplified successfully and produced unambiguous banding patterns from which individual genotypes could be assessed. Estimated parameters pertaining to genetic variation viz. observed/effective number of alleles and observed/expected heterozygosity, polymorphism information content (PIC) in the studied Indian horse breeds are summarised in Tables 1 and 2. A reasonable amount of polymorphism in all the five breeds is discernible from allele frequency data. A total of 183 alleles were detected across the 25 loci with mean number of alleles varying from 5.40 ± 1.04 in Spiti ponies to 5.80 ± 1.32 in Zanskari ponies. The total number of alleles across all five breeds ranged from 135 in Spiti ponies to 145 in Zanskari ponies. The overall PIC values varied from 0.71 at locus HTG14 to 0.86 at loci HMS7 and VHL20 across all five Indian horse breeds. The observed number of alleles and fairy high PIC values demonstrated that almost all the microsatellite loci utilised in the present study were sufficiently polymorphic suggesting their suitability in evaluation of Indian horse breeds. The PCR product size range varied from 78 to 102 bp at locus HTG6 to 238 to 250 bp at locus UCDEQ425. The allele sizes obtained at each locus across the studied Indian breeds were in agreement with the data published for Asian and European horse breeds (Cañón et al., Reference Cañón, Checa, Carleos, Vega-Pla, Vallejo and Dunner2000; Bjornstad and Roed, Reference Bjornstad and Roed2001; Bjornstad et al., Reference Bjornstad, Nilsen and Roed2003; Curik et al., Reference Curik, Zechner, Solkner, Achmann, Bodo, Dovc, Kavar, Marti and Brem2003; Tozaki et al., Reference Tozaki, Takezaki, Hasegawa, Ishida, Kurusawa, Saitou and Mukoyama2003; Aberle et al., Reference Aberle, Hamann, Drogemuller and Distl2004; Achmann et al., Reference Achmann, Curik, Dovc, Kavar, Bodo, Habe, Marti, Solkner and Brem2004). The effective number of alleles were distinctly less than the observed values across all loci in all the five breeds with mean values ranging from 4.52 ± 0.85 in Manipuri horses to 4.94 ± 1.18 in Zanskari horses.
Table 2 The population genetic variability in five Indian horse breeds evaluated using 25 microsatellite loci
The mean observed heterozygosity values ranged from 0.55 ± 0.07 in Bhutia and Manipuri horses to 0.61 ± 0.06 in Zanskari horses. The obsereved heterozygosity was lower than the expected heterozygosity in all the five breeds. The mean expected heterozygosity did not vary much between the studied breeds varying in a narrow range of 0.78 (Marwari, Manipuri and Zanskari) to 0.79 (Spiti and Bhutia). Heterozygote deficiency analysis revealed that all the five populations exhibited significant deviation from HWE (P < 0.05) at many loci. It is though difficult to explain the exact basis of this departure; however, this may be attributed to the lower population of size varying from a few hundred to a few thousand for all these breeds. The presence of low frequency null alleles segregating at these loci may be other possible reason. This deviation could also be linked to positive FIS (within population inbreeding estimates) values obtained in all the breeds.
Mean estimates of F statistics obtained from jackknifing over loci (Weir, Reference Weir1990) were: f (FIS) = 0.206 ± 0.033, θ (FST) = 0.065 ± 0.021 and F (FIT) = 0.245 ± 0.041. The overall estimates of F statistics were significantly (P < 0.01) different from zero. There was significant deficit of heterozygotes in all the breeds, ranging from 14.7% in Marwari to 27.9% in Manipuri. The average FIS values for of these breeds were significantly different from zero. Global analysis indicated that the studied breeds had a 20.6% deficit of heterozygotes (P < 0.01), whereas the total population had 24.5% deficit of heterozygotes (P < 0.01) (Table 3). The main cause for shortage of heterozygotes and excess of homozygotes (FIS>0) seems to be the inbreeding/non-random mating arising from small population sizes and extensive use of only a few breeding studs in these breeds. The locus under selection (genetic hitchhiking), null alleles (non-amplifying alleles) or presence of population sub-structure (Wahlund effect) may be the other possible reason for lack of heterozygotes in a population (Nei, Reference Nei1987).
Table 3 Within population inbreeding estimates (FIS) in five Indian horse breeds
The DA and DR genetic distances between pairs of populations are given Table 4. None of the five Indian horse breeds was found to be closely associated with Thoroughbreds with an average DA and DR of 0.217 and 0.067, respectively. Within Indian breeds the Marwari and Manipuri with a DA and DR of 0.186 and 0.058 were most distant apart. In fact, the genetic distances suggest that the Marwari breed was most distinguishable within the studied Indian horse breeds. The Marwari horses are medium sized with an average height of 154.19 ± 0.32 cm (Singh et al., Reference Singh, Yadav and Mehta2002). The Marwari breed was primarily developed for survivability and endurance in desert type environment by crossbreeding the local stock with Arabian horses. They can be expected to be fairly distant from other four pony breeds on the basis of physical characteristics and adaptability to environment.
Table 4 Nei's DA genetic distances (lower triangle) and Reynolds genetic distances (upper triangle) between five Indian horse breeds and Thoroughbred horses outgroup using 25 microsatellite loci
Within pony breeds, the maximum DA and DR found were only of the order of 0.133 and 0.038 between Zanskari and Manipuri. The least DA and DR were found to be 0.071 and 0.012, between Spiti and Zanskari indicating their close genetic relatedness. These results were also reflected in neighbour-joining tree, based on DA genetic distances, developed after 1000 bootstraps of the data where all the pony breeds joined first then with the Marwari with good statistical support (Figure 2). All the Indian breeds clustered independently from Thoroughbreds. The Spiti and Zanskari ponies joined first then with other two pony breeds of Bhutia and Manipuri with high statistical support. The lower genetic distances found between the pony breeds can be expected as animals of Zanskari, Spiti, Bhutia and Manipuri are small sized (less than 12 hands) ponies with similar physical characteristics (Bhat et al., Reference Bhat, Bhat, Khan, Goswami and Singh1981; Katoch et al., Reference Katoch, Dogra, Thakur and Gupta2004). All these pony breeds are supposed to be evolved from the Tibetan ponies. Moreover, principal component analysis showed a tight cluster of Zanskari, Spiti, Bhutia and Manipuri pony breeds well separated from Marwari horses. All the Indian breeds were clearly distinguishable from the Thoroughbred horses (Figure 3).
Figure 2 The neighbour joining dendrogram showing the genetic relationships among the five Indian horse breeds and Thoroughbred horses based on Nei's unbiased DA distances (Nei et al., 1983) using microsatellite markers. The numbers at nodes are values for 1000 bootstrap resampling of the data.
Figure 3 PCA of the transformed allele frequencies from 25 microsatellite loci typed in five Indian horse breeds and Thoroughbred horse outgroup. The first PC accounted for 69.8% of the underlying variation and the second PC condenses 8.8% of the variation.
Further, to study the population structure of Indian horse breeds, the data was analysed using STRUCTURE computer program. The models with assumed number of breeds, K = 1, 2 or 3 gave insufficient posterior probabilities, Pr(K/X) and the model with K = 4 was substantially better than models with even larger K. At K = 4, ln Pr(K/X) also stabilised to about minimum values (Table 5). When individual horses were clustered assuming number of breeds to be four, about 90% of individuals belonging to Zanskari, Spiti and Bhutia were assigned to one cluster (Table 6), whereas, majority of the Marwari and Thoroughbred individuals were assigned to their respective clusters. Though, Manipuri horses formed a separate cluster 12% of the Manipuri horses clustered with common cluster of Zanskari, Spiti and Bhutia horse. These results point towards the genetic closeness of pony breeds of India. These findings contribute to the knowledge of genetic structure of these endangered breeds and should aid in evolving efficient conservation/breeding strategies for the Indian horse breeds.
Table 5 Estimated posterior probabilities of K number of assumed breeds for sampled individuals with genotype X
Table 6 Proportion of membership of each of five Indian breeds and Thoroughbred horses in each of the four clusters
Acknowledgements
We gratefully acknowledge the valuable help received from following agencies or persons in obtaining samples: (1) Director and staff, Animal Husbandry Department, Manipur; (2) Marwari Horse Society and Marwar Horse Breeding and Research Institute, Jodhpur, Rajsthan; (3) Director and staff, National Research Centre on Yak, Dirang, Arunachal Pradesh; (4) Incharge, Network Project, NBAGR, Karnal, Haryana; (5) Dr Sanjeet Katoch, Department of Animal Breeding, Genetics and Biostatistics, College of Veterinary and Animal Sciences, Palampur, Himachal Pradesh; (6) Dr M. Ragnekar and Dr Z. Ahmed, Field Research Laboratory, Defense Research and Development Organisation, Leh, Jammu and Kashmir.
