Abstracts

Genetics of the Pantaneiro horse of the Pantanal region of Brazil

E. Gus Cothran¹, Sandra Aparecida Santos², Maria Cristina Medeiros Mazza³, Tery L. Lear¹ and J. Robson Bezerra Sereno²

¹Department of Veterinary Science, University of Kentucky, Lexington, KY 40546, USA.

²Embrapa, Centro de Pesquisa Agropecuária do Pantanal, Caixa Postal 109, Rua 21 de setembro, 1880, 79320-900 Corumbá, MS, Brasil. Send correspondence to S.A.S.

³Embrapa, Centro Nacional de Pesquisa de Floresta, Caixa Postal 019, Colombo, PR, Brasil

ABSTRACT

Genetic variation at seven red blood cell alloantigen, blood group loci and 10 biochemical genetic loci was examined in 102 Pantaneiro horses from the Pantanal region of Brazil and compared to that of other domestic horse breeds. Measures of both individual genic variation and populational genetic diversity within the Pantaneiro were near the average for domestic horse breeds. From the standpoint of genetic conservation there is no immediate concern for loss of variation within the Pantaneiro. Analysis of genetic relationship showed that the Pantaneiro was clearly of Iberian Peninsula descent and that it was most closely related to other Brazilian breeds.

INTRODUCTION

The two primary uses of genetic markers in population analyses are to assess levels of genetic variation within a population and to determine relationships with other populations. Here we report the results of genetic analysis of the Pantaneiro horse of Brazil. The exact origin of the Pantaneiro horse is not known but it is believed to have originated from horses brought into what is now Brazil by Spanish conquerors and settlers in the 16th and 17th centuries (Santos et al., 1992). The Guaicuru Indians appear to have played a key role in bringing the horses into the Pantanal. During their early history, the horses were largely wild and adapted to the conditions of the Pantanal through natural selection (Balieiro, 1971). The Pantaneiro became economically important to the people of the region with the development of cattle breeding in the Pantanal region. At the end of the 19th century there was a serious reduction in population size of the breed due to disease (Trypanosomiasis) and in recent times equine infectious anemia has threatened the breed (Santos et al., 1995). There are now efforts to preserve this breed including the creation of the Brazilian Pantaneiro Horse Breeders Association and incentives from the Brazilian government. In the present study, the Pantaneiro breed was analyzed for overall levels of variability and for variation within the four separate populations that were sampled. Genetic relationships of the Pantaneiro horses to other horse breeds, both from South America and other parts of the world, were also analyzed. The results of this study will be incorporated into efforts to conserve genetic variation within the uniquely adapted horse breed.

MATERIAL AND METHODS

Blood samples from a total of 102 Pantaneiro horses were collected in October 1992. Samples were taken from four separate farms with 31 horses from Nhumirim, 14 from the University of Mato Grosso do Sul, 28 from São Bento do Marajoara and 29 from Rancharia. Blood samples were sent to the University of Kentucky for analysis. The samples were separated into red blood cells (rbc), rbc lysate, serum and white blood cells (wbc).

Standard immunological procedures involving hemagglutination and complement-mediated hemolysis (Stormont and Suzuki, 1964; Stormont et al., 1964) were used to detect variation of red cell alloantigens at seven blood group loci. Starch and polyacrylamide gel electrophoresis and isoelectric focusing were used to detect variation at 10 serum and rbc lysate protein loci (Braend, 1973; Sandlberg, 1974; Juneja et al., 1978; Braend and Johansen, 1983; Pollitt and Bell, 1980; Henney et al., 1994).

The horse blood group loci examined were A, C, D, K, P, Q and U, and the biochemical protein loci were alpha-1-beta glycoprotein (Alb), albumin (Al), serum esterase (Est), vitamin D-binding protein (Gc), glucosephosphate isomerase (GPI), alpha-hemoglobin (Hb), 6-phosphogluconate dehydrogenase (6-PGD), phosphoglucomutase (PGM), protease inhibitor (Pi) and transferrin (Tfr). Nomenclature for variants at all 17 loci was in accordance with internationally standardized usage for horses (Bowling and Clark, 1985; Bowling and Ryder, 1987) except for variants at some loci which have not yet received international recognition.

Gene frequencies for biochemical loci were calculated by direct count. Frequencies of alleles at blood group loci were calculated by the allocation method (Andersson, 1985). Genetic variation was measured as observed heterozygosity (Ho), Hardy-Weinberg expected heterozygosity (He), unbiased expected heterozygosity (Hu; Nei, 1978), effective number of alleles (Ae), and the total number of variants found in each population (Na). Ho was calculated for biochemical loci only because of the presence of recessive alleles and/or ambiguous genotypes at blood group loci. Therefore, for direct comparison, He and Hu were calculated only for biochemical loci (in an ideal population, He (Hu) should equal Ho), for blood group loci, and for all 17 loci. In addition, populational inbreeding level was estimated by Wright's Fis = 1-(Ho/He). Values of genetic variation of the Pantaneiro were compared to those of 102 domestic horse populations that have been tested at the University of Kentucky (Cothran, G., unpublished results). Genetic relationship of the Pantaneiro to these other domestic breeds were investigated using Rogers' (1972) genetic similarity coefficient (S) and Nei's modified genetic distance (Da; Nei et al., 1983). Restricted maximum likelihood analysis (RML; Felsenstein, 1989) was used to construct the dendograms of Figures 1-³ .

Figure 3 - Partial consensus tree from 30 RML trees comparing the four Pantaneiro populations to the Iberian-derived breeds as in Figure 1.

RESULTS

Measures of genetic variation for the complete Pantaneiro population are given in Table I. There were no statistically significant deviations from expected levels of heterozygosity (based upon Hardy-Weinberg equilibrium theory) for any of the biochemical loci where Ho could be calculated. Table II shows Ho, He, Fis for biochemical loci and He and Ae for all loci for each of the four locations. The only significant deviation in expected heterozygozity was at the Est locus in São Bento population (data not shown); however, considering the number of tests conducted it is likely that this result could have occurred by chance.

Genetic similarity and distance measures of the Pantaneiro to selected domestic horse breeds are given in Table III. Figure 1 shows the consensus tree from 25 RML trees comparing the Pantaneiro to 52 domestic breeds and the Przewalski horse (used as the outgroup). Figure 2 shows the consensus tree for 20 RML trees using all Iberian horse breeds tested at the University of Kentucky and ^{Figure 3} shows a partial consensus tree from 30 RML trees showing the relationship of the four Pantaneiro localities to selecte Iberian-derived breeds. The tree in ^{Figure 3} was taken from an analysis that included all the breeds that were used for Figure 1 but only the branch that includes the Iberian breeds is shown here as the rest of the tree is the same as Figure 1. For all trees, the numbers at branching points represent how many times that branching pattern occurred out of the total number of replicate trees.

DISCUSSION

Efforts to conserve rare breeds of domestic animals must take conservation of genetic variability into account. It also must be recognized that genetic variation within a population can be divided into two components. The first component is individual genetic variation as measured by observed heterozygosity. Because Ho should be negatively correlated with inbreeding level, Ho provides an estimate of the overall genetic health of individuals within a population. In addition, Ho is the most robust estimate of genetic variation as it is not correlated with sample size (Cothran, G., unpublished results). The second component of genetic variation of populations is the population genetic diversity. There are a number of ways to estimate the population component of variability including He, Ae, and Na. Population genetic diversity is a measure of the overall genetic variation within populations and should be related to the long term adaptability of a population. Population variation is strongly associated with sample size as would be expected.

Individual genetic variation within the Pantaneiro (Ho) was slightly greater than mean Ho for 102 domestic horse populations (0.387 and 0.375, respectively). Thus, from a genetic conservation standpoint, there is no immediate concern about reduced genic variation within the breed. Similarly, the populational variation measures of the Pantaneiro were slightly greater than the means for domestic breeds with values of He at 0.369 compared to 0.365, Ae of 2.68 compared to 2.397 and Na of 72 compared to 64.99. Both individual and populational estimates of genetic variation of the Pantaneiro indicate that variation levels are normal for horses and that there is no indication of inbreeding. Indeed, there was a slight excess of Ho relative to He (Fis = -0.047). This Fis value was not statistically different from zero. None of the single locus Fis values were statistically significant, thus the Pantaneiro population showed no significant deviation from Hardy-Weinberg equilibrium.

Although variation within the Pantaneiro was slightly above the mean for domestic horse breeds, it was lower than that observed for three out of four other Brazilian breeds examined. Only the Mangalarga had lower Ho (0.328). The Campolina, Mangalarga Marchador and Brazilian Criollo had considerably higher Ho (0.410, 0.411 and 0.420, respectively). This could reflect greater outcrossing of these three breeds relative to the Pantaneiro or more diverse and/or more recent origins of these breeds. However, it should be noted that the breed with the lowest variation, the Mangalarga, is known to be derived from Mangalarga Marchador stock crossed to English Thoroughbreds and Anglo-Arabians (Hendricks, 1995).

Within the four Pantaneiro farms, patterns of variation were not the same (Table II). The highest Ho was for the Rancharia population (0.418) while the lowest was for the São Bento horses (0.358). For all four populations there was an excess of observed heterozygosity, although no overall Fis values were statistically significant. The greatest excess of Ho was for the São Bento population which had the lowest variation. This population also had the highest Ae value. Within populations there was only a single statistically significant deviation from Hardy-Weinberg expectations. This was at the Est locus in the São Bento population (Fis = -0.949). It is difficult to interpret this result. Considering that it was the only significant deviation out of 40 tests, it may represent a Type 1 error, even though this is quite a high Fis value and is highly significant (c² = 24.62, d.f. = 1, P< 0.001). As only two alleles were present at the Est locus in the São Bento population (Est-G and Est-I), such a result could have occurred if a stallion that was homozygous for the Est-G allele was mated to mares that were primarily homozygous for Est-I and we tested the progeny of these matings.

In summary, levels of genetic variability within the Pantaneiro breed were at about the normal level for horses. There should be little concern for maintaining genetic variation within the breed as long as population size is maintained and there is no drastic change in breeding practices. For example, overuse of particular stallions would likely result in a loss of variation, especially if such a practice was continued for many years.

Patterns of genetic similarity and distance were very similar (Table III) so the discussion here will focus on S. Highest mean S was the Iberian Peninsula-derived breeds (0.871). The next highest mean S was with the Oriental or Arabian type breeds, with a mean value of 0.840. The difference was only 0.031 but for comparisons of breeds of horses, which are all closely related, this is a substantial difference. The close relationship to the Iberian breeds is further emphasized by the individual breed comparisons. Of the 14 Iberian-derived breeds, 8 had S values with the Pantaneiro of above 0.87 and 2 had S of above 0.9. There were only six other breed comparisons where the S value exceeded 0.87 and for five of these the values were between 0.87 and 0.88. The two highest S values among the Iberian breeds were 0.904 for the American Paso Fino and 0.902 for the Lusitano of Portugal. Interestingly, the two breeds with the lowest Da comparisons were the Brazilian breeds Mangalarga Marchador and Campolina (0.035 and 0.036, respectively). The only non-Iberian-derived breed with an S value greater than 0.88 was the Moroccan Barb (0.904). This breed shows very close relationship with the Iberian breeds, although it most consistently groups with the Oriental breeds. From a historical standpoint, it makes sense that the Moroccan Barb would show a close relationship to Iberian breeds.

The Iberian-derived breeds that did not show a relatively close relationship to the Pantaneiro were the Chilote, Argentine Criollo and Venezuelan Spanish Horse. The Chilote is an isolated, island breed from Chile. Recent population size is small so the low similarity is not unexpected. The sample size for the Argentine Criollo in this comparison was rather small. This may well account for the low S, but until a larger sample can be analyzed, it is not possible to determine whether the S observed here is accurate. The Venezuelan Spanish Horse is a subpopulation of the Andalusian. It has rather limited variation compared to its parent breed and this likely is the reason for the low S value. Horse breeds with relatively low genetic variation tend to show relatively low genetic similarity to other breeds.

Figure 1 shows the genetic relationships among horse breeds based upon RML analysis of the gene frequency data. There are three major clusters. One consists of the heavy draft horse breeds and the pony breeds of, primarily, northern European origins. The second major grouping consists of the European and North American saddle and light harness horses and the Oriental and Arabian horse breeds. The third cluster is the horse breeds of Iberian Peninsula descent. The Pantaneiro fits in this third major cluster of Iberian horses as expected based on the genetic similarity and distance results.

Figure 1 only shows a selected group of 12 Iberian-derived breeds. Figure 2 includes an additional four Iberian horse breeds to take a more detailed look at the relationship of the Pantaneiro to horse breeds of Iberian ancestry. The first branch of this tree includes the Andalusian (and the Venezuelan Spanish horse which is a subpopulation of Andalusians) and the Lusitano. Based upon RML these breeds were the most ancestral of those analyzed. The next branch included the three Paso breeds. These New World breeds are closely related to their Iberian ancestors and have remained relatively pure bred. The third branch included the Garrano, a fairly primitive breed from Portugal, and the Chilote, a small horse breed from the island of Chiloe, Chile, which may be related to the Garrano (Cothran et al., 1993). The interior of the tree contains the remaining South American breeds and the Sorraia from Portugal. The pairing of the Sorraia with the Argentine Criollo was probably due to the combination of low variability of the Sorraia, small sample size of the Argentine Criollo and the fact that an Argentine Criollo stallion was used in the regeneration of the Sorraia.

The Pantaneiro paired with the Mangalarga in a cluster that also included the Brazilian Campolina and Mangalarga Marchador breeds. The Pantaneiro also clustered with the Mangalarga in Figure 1. This was a surprising result considering that the genetic similarity of the Mangalarga to the Pantaneiro was relatively low and certainly lower than that with the other Brazilian breeds (Table III). The explanation is that the RML analysis compares relationships among all the breeds, not just the Pantaneiro to the other breeds. The similarity of the Mangalarga Marchador and the Campolina to each other was greater than to the Pantaneiro and less to the Mangalarga than was the Mangalarga to the Pantaneiro which, essentially, left the Pantaneiro and Mangalarga to pair with each other. The main conclusion from this analysis is not that the Pantaneiro is most closely related to the Mangalarga but that all the Brazilian breeds are closely related to each other. The South American horse breeds sampled here form two main groups, the Criollo types and the other Brazilian breeds. These other Brazilian breeds are more specialized than are the Criollo types and the Pantaneiro fits best among the more narrowly defined breed types.

^{Figure 3} shows the RML comparison of the four Pantaneiro subpopulations with the main Iberian-derived breeds. Not including the Sorraia, Chilote and Garrano changed the tree topology somewhat but not significantly. The main point here is that all four Pantaneiro populations cluster together on the branch that includes the Mangalarga. The Nhumirim and University of Mato Grosso do Sul pair together and the São Bento and Rancharia populations are on the same branch. However, the overall degree of similarity of the four populations is high enough so that they should not be considered to be different.

The genetic data clearly show that the Pantaneiro horses have the Iberian origin indicated by their known history. The clustering of the Pantaneiro with the other Brazilian breeds and the high S with the Lusitano suggest that the breed has been influenced by the breeds of Portuguese origin. This assumes that the original horses lost in the Pantanal were lost by the Spanish expeditions (Santos et al., 1992) and that the Spanish horses of that time were different from the Portuguese horses of today. The modern American breeds of direct Spanish descent are slightly different than those of Portuguese origin but there may have been less difference in the 16th century. The genetic variability data do not suggest a great deal of recent crossing of the Pantaneiro with other breeds. However, it is certainly conceivable that there has not been complete isolation of the Pantaneiro throughout its history. If there had been isolation, it is likely that the genetic variability would have been lower. On the other hand, the high degree of adaptation of the Pantaneiro to the unusual environment of the Pantanal does strongly support some significant degree of isolation. It would only require a relatively small amount of input from outside the region to maintain the variation and increase the similarity to the Portuguese-derived horses of Brazil. The combination of the genetic and physical characteristics of the Pantaneiro shows that it represents a unique horse population that must be preserved.

ACKNOWLEDGMENTS

We thank the staff of the Equine Blood Typing Research Laboratory of the University of Kentucky for their expertise in the genetic typing of the horses. The investigation reported here is in connection with a project of the Kentucky Agricultural Experiment Station, paper 97-14-82. Thanks are also due to Pantaneiro horse breeders for permission in the collection of material.

RESUMO

Examinou-se a variação genética de 102 cavalos Pantaneiros da região do Pantanal, Brasil, usando sete locos de grupos sanguíneos e 10 locos genéticos bioquímicos. As medidas de variação genética individual e diversidade genética populacional dentro da população de cavalos Pantaneiros foram próximas da média encontrada para as raças de cavalos domésticos. Do ponto de vista de conservação genética, não há preocupação imediata com relação à perda de variação dentro da raça. Análises das relações genéticas mostraram claramente que o cavalo Pantaneiro descende de cavalos da península Ibérica e que ele é estreitamente relacionado com as demais raças brasileiras.

(Received September 29, 1997)

Publication Dates

History