Multifactorial Inheritance of Common White Markings in the Arabian Horse

Transcription

1 Multifactorial Inheritance of Common White Markings in the Arabian Horse C. M. Woolf From the Department of Zoology, Arizona State University, Tempe. This study was supported by an institutional research grant (NIH Biomedical Subaward No. 98) awarded by the Office of the Vice President for Research and by a minigrant (MG-88-7) awarded by the College of Liberal Arts and Sciences, Arizona State University. The author thanks the officials of the Arabian Horse Registry of America, Inc., Zuni Street, Westminster, Colorado 8, for supplying computerized printouts of common white markings and for their cooperation throughout this study. Address reprint requests to C. M. Woolf, Department of Zoology, Arizona State University, Tempe, AZ 887. Journal of Heredity 99;8:-6; -/9/$. The results of a previous study were compatible with the hypothesis that common white facial markings in the Arabian horse have a multifactorial mode of inheritance. I expanded that study to () include the legs and therefore obtain insight into the heritability of common white markings in all peripheral regions (face and legs) of the Arabian horse and () investigate the influence of sex and the genotypes that produce the bay and chestnut phenotypes on the variation in common white markings. Both studies were based on computerized data obtained from the Arabian Horse Registry of America, Inc. Each leg of a horse was scored from to depending on the amount of whiteness present, and the four leg scores were added to obtain the total leg score for each horse. The facial region was divided into five areas, and each horse was given a score from to according to the number of areas with whiteness. Sire families were analyzed in which each sire family consisted of a sire, his foals, and the dams of those foals. There was a correlation between white facial scores and white leg scores, suggesting that both types of white markings are influenced by the same genetic mechanism. Sire-foal and dam-foal regression analyses were compatible with the hypothesis that common white leg markings also show multifactorial inheritance. Although the results support the model that additively acting genes (polygenes) influence the presence and extent of common white markings, the results also show that males are slightly more marked than are females and that chestnut horses are more heavily marked than are bay horses. I conclude that complex genetic systems and nongenetic factors determine the presence of common white markings in the Arabian horse. In a previous study (Woolf 989) based on computerized records obtained from the Arabian Horse Registry of America, Inc., I concluded that common white facial markings in the Arabian horse have a multifactorial mode of inheritance. My results were compatible with the hypothesis that the variation in white facial markings is influenced by both genetic and nongenetic factors. I demonstrated that a strong linear relationship exists between the scores for the amount of facial whiteness in a foal and the scores for each parent. For example, in a data base consisting of 7, dam-foal pairs, there was almost perfect agreement between foal means and the point on the regression line for each dam score. Such a relationship would not be expected if dominant and recessive inheritance accounted for specific white facial markings, as proposed by Blunn and Howell (96), or if only a single locus were responsible for the presence and extent of these markings. Furthermore, the results of matings between parents with various scores supported the multifactorial hypothesis. Nebe (98), in a study of 8,6 foals of Hessian saddle horses, supported the multifactorial mode of inheritance of white markings. Although the genetic component of a multifactorial model usually is assumed to consist of many genes (polygenes), each with an equal, additive small effect, this assumption is too simplistic for a complex trait such as white markings and perhaps for most other quantitative traits; although polygenes (perhaps with equal and unequal additive effects) play an important role, specific genes with major effects also influence variation. The complexity of the situation will be discussed here and in subsequent papers. It has been shown by previous investigators (Dreux 966, Nebe 98, Walther 9) that a correlation exists between white facial markings and white leg markings in other horse breeds, suggesting that they are part of the same genetic system. It has also been shown in other breeds Downloaded from at Penn State University (Paterno Lib) on March, 6

2 that chestnut horses have more white markings than do bay and black horses (Crew and Buchannan Smith 9, Dreux 966, Munckel 99, Munckel 9, Nebe 98, Walther 9) and that males are slightly more marked on the average than are females (Dreux 966). The influence of color and sex attests to the complexity of the genetic system. This study, which is also based on computerized records, was intended to expand the study of the Arabian horse to () include the legs in order to gain insight into the genetics of common white markings in all peripheral regions and () determine the influence of color and sex on variation in common white markings. Materials and Methods To register an Arabian horse, the owner must send information to the Arabian Horse Registry of America, Inc. (the Registry), including descriptions and drawings of white facial and leg markings. The Registry does not permit the release of the original registration forms with these drawings for studies of this kind; however, computerized records containing some of the information on the registration forms are made available by the Registry. The advantages and disadvantages of using records maintained by horse registries for genetic studies have been discussed elsewhere (Blunn and Howell 96, Woolf 989). Beginning in 967, the Registry initiated a program of computerizing the information on the registration forms. In 98, the Registry altered the information that was digitized, and the existing computerized records now appear in three different formats Format : no descriptions of white markings; Format : descriptions of white facial and leg markings; Format : descriptions of white facial markings but no descriptions of the extent of white leg markings. No white markings (Format ) are given for a horse registered before 967 unless the horse was brought to the attention of the Registry after 967. Descriptions of white facial and leg markings (Format ) are given for horses registered between 967 and 98 that were not brought to the attention of the Registry after 98. White facial markings are described by star, strip, snip, upper lip, lower lip, and chin, with appropriate adjectives. White leg markings are described as coronet, pastern, fetlock, sock, and stocking, with appropriate adjectives. Two examples are as follows: () star, thin strip, snip, left fore fetlock, right hind partial sock, left hind pastern and () no white markings. These records can be used to study both white facial markings and white leg markings. In 98, the Registry decided to discontinue digitizing the terms coronet, pastern, fetlock, sock, and stocking. As a consequence, the computerized records (Format ) for current registrations contain descriptions of white facial markings but do not describe the extent of white leg markings. They state which white facial markings are present, whether a hoof is dark or light, and which legs have white markings without describing the extent of whiteness on the legs. Although these records lend themselves to a study of white facial markings, they cannot be used to study the extent of white leg markings. The Registry has carried out a program of updating the format for horses brought to its attention because of a change of horse ownership or for another transaction such as the placement of a freeze mark. A horse born and registered before 967 may have white markings described by Format if a transaction for that horse occurred between 967 and 98; the descriptions may follow Format if the transaction occurred after 98. Similarly, a horse born and registered between 967 and 98 and described originally by Format may have the white marking descriptions changed to Format if a transaction occurred after 98. Because Arabian horses frequently change owners and because the Arabian population is increasing in the United States, most living Arabian horses are described according to Format, and the proportion described in this manner is on the increase. This situation is unfortunate for a geneticist who wishes to study white leg markings in the Arabian horse. At the initiation of this study, over, Arabian horses had been registered. Study of Common White Leg Markings Computerized printouts obtained from the Registry list the foals produced by a sire and the dams of those foals. The sire, dams, and foals constitute a sire family. No names are given for the dams and foals; only their registration numbers are listed. The damfoal pairs appear on the printout by the ascending registration numbers of the foals. The previous study of white facial markings was based on sire families consisting of 7, dam-foal pairs. These sires were screened to determine whether they had produced a relatively large number of foals that were registered during the period, when white leg markings were described according to Format. As noted in the previous paper (Woolf 989), the original sires were not selected randomly for common white markings. Although some were selected solely because they had produced a large number of foals, others were selected because they possessed specific markings. Because of certain hypotheses that were being tested, I made an effort to obtain several sires with no white facial markings or whiteness from the forehead to the lower lip or chin or with isolated markings low on the face. Beginning with the first dam-foal pair on the printout, I scrutinized the dams sequentially for the format used to describe the white leg markings. They were passed over quickly if Format or Format was used. If Format was used for the dam, attention was directed to her foal. If the foal was registered after 967, as shown by the registration number, and if Format was used to describe the foal's white leg markings, I placed a mark by that pair in the printout. If at least 8 dam-foal pairs were concordant for Format descriptions, the sire family was included in the study. I chose this number arbitrarily at the beginning of the study. Using this criterion, only of the sire families used in the study of white facial markings (Woolf 989) qualified for this study; three additional sire families were selected from a group obtained from the Registry for an unrelated study of coat color genetics. I selected the additional sire families on the basis of the criterion, not because of the white leg markings in the sires. No attention was paid to the specific white leg markings on the dams or foals during the screening process. A total of 6 sire families and,6 dam-foal pairs made up the data base. After a sire family had been selected according to the criterion, the concordant dams and foals were scored for white leg markings and white facial markings. Scoring of Dams and Foals For the purposes of registration by the Registry, white leg markings are defined as follows: ) coronet: white marking extending no more than an inch above the coronet band, ) pastern: white marking anywhere between the coronet marking and the bottom of the fetlock joint, ) fetlock: white marking extending above the pastern but below the top of the fetlock joint, ) sock: white marking extending above the top of the fetlock but below the Downloaded from at Penn State University (Paterno Lib) on March, 6 Woolf Common White Markings in the Arabian Horse

3 midpoint of the cannon, ) stocking: white marking extending above the midpoint of the cannon. Each leg of a dam and foal was scored separately, as follows:, no white marking;, coronet or whiteness on hoof;, pastern;, fetlock;, sock; and, stocking. A partial fetlock was scored as a fetlock, and a partial sock and a partial stocking were scored as a sock and a stocking, respectively. Whiteness on the hoof is often omitted in Format descriptions. A hoof is always considered white if there is whiteness above it, and there usually is whiteness above a white hoof or partially white hoof. However, the Format computerized records may occasionally state that a horse has a hoof that is partially white, completely white, or with a heel (white marking on the heel of the hoof) with no whiteness above it. In this study, any whiteness on the hoof with no whiteness above it was grouped with coronet, and that leg was given a score of. Summing the scores for each leg provided a total leg score. For example, a horse with four stockings was given a total leg score of. As one objective of this study was to gain insight into the heritability of white markings in the peripheral regions (face and four legs), the dams and foals also were scored for white facial markings using the system described previously (Woolf 989):, no white facial markings;, one area (such as a star or snip);, two areas (such as star-strip/blaze);, three areas (such as star-strip-snip);, four areas (such as star-strip/blaze-snip-upper lip); and, five areas (star-strip-snip-upper lip-lower lip/ chin). When the facial region is divided into five areas, the score gives the number of areas with whiteness. The sum of the facial score ( to ) and the total leg score ( to ) gives the total score ( to ) for each horse, which is a measure of total whiteness in the peripheral regions. Study of Common White Markings in and Dams Arabian horses are registered as bay, black, chestnut, gray, or roan. The roan color occurs infrequently among Arabian horses and is defined for registration purposes as the presence of about a - mixture of white hairs with bay, black, or chestnut hairs throughout the body. This phenotype is due to a dominant allele QR) that is homozygous lethal (Hintz and Van Vleck 979, Sponenberg and Beaver 98) and appears after the foal coat has been shed. If a dominant allele (G) causing premature grayness is present, gray hairs soon appear around the eyes and flank and below the elbow of a horse that is born bay, black, or chestnut. A few white hairs may be present on an eyelid at birth. Graying is progressive. A older horse with the G allele may be white or may possess tufts of yellowish-brown hairs that produce the "fleabitten" trait. Alleles at the A and E loci interact to produce the following relationships between genotypes and phenotypes: A/-;E/-, bay; a/a;e/-, black; A/-;e/e, chestnut; a/a;e/e, chestnut. The genotype e/e is epistatic to the alleles at the A locus in the environment of a hair follicle (Woolf 988). Registration records do not indicate whether a gray or roan horse was born bay, black, or chestnut; for this reason, gray and roan horses were not included in this study of dams. Black horses occur at a low frequency in the Arabian population (Woolf 988), and they also were ignored in this study. I sampled bay and chestnut dams whose common white leg markings were described by Format. These dams were part of different sire families. Each sample contained n =, dams, for a total of, dams. The markings in the facial region and legs were scored, and comparisons were made between the two different types of dams. White Facial Markings in Foals Classified by Sex and Color Computerized records obtained from the Registry during the course of this study varied in regard to content depending on the information requested and the information the Registry provided at the time of the request. The records used in the previous study (Woolf 989) and in the study of white leg markings in the present paper did not give the sex of the foals. To study the influence of this variable, I obtained seven additional sire families from the Registry giving the sex of the foals; the colors of the sires, dams, and foals; and the common white markings in the members of these sire families. The dams and foals were scored for white facial markings using the standard scores ranging from to. A total of,77 dam-foal pairs were included in this study. White leg markings were not studied because of the limited number of dams and foals described by Format in the seven sire families. Statistical Methods In the previous study (Woolf 989), the facial scores of over 7, dams and foals showed positive skewness (coefficient of Table. Leg scores in,6 dams Leg / P r Right foreleg Left foreleg Right hind leg Left hind leg Sum of foreleg scores Sum of hind leg scores skewness = +.). Similarly, the leg scores of the horses in the present study showed positive skewness (coefficient of skewness = +.7). It will be shown below, however, that the scores for chestnut horses are less skewed than are the scores for bay horses. Because selection should be effective in increasing or decreasing the frequency and extent of common white markings in an Arabian population, the observed positive skewness undoubtedly reflects the past actions of Arabian breeders who tended to select against common white markings in certain subpopulations. A polygenic model assumes that genetic variation can exist among horses that are devoid of common white markings. If a common white marking appears when a threshold number of genes is reached, horses that do not have a common white marking can vary in regard to the number of genes they possess below the threshold number; therefore, horses without white markings can vary in regard to the number of genes they can contribute to their foals. Data supporting this concept were presented in the previous paper (Woolf 989). It is acknowledged, however, that the facial and leg scores ranging from to used to measure the amount of whiteness, although useful for these studies, are an artificial construct and may not reflect an underlying polygenic continuum precisely. I used parametric methods in this study to analyze the scores. These methods included t tests, correlation and regression analysis, and analysis of covariance. In the previous study of white facial markings (Woolf 989), the use of parametric and nonparametric methods for various tests of hypotheses led to the same conclusions, and when analysis of covariance procedures were employed, the hypotheses of homogeneous variances and regression coefficients usually could be accepted. The robustness of analysis of variance and an analysis of covariance in regard to violations of some of the assumptions has been discussed thoroughly in the literature (Anderson 96, D'Agostino 97, Glass et al. 97, Lunney 97). Downloaded from at Penn State University (Paterno Lib) on March, 6 The Journal of Heredity 99:8()

4 Based on a review of this literature and the tests run in the previous study (Woolf 989), I conclude that parametric methods are robust in regard to the degree of nonnormality present in the distributions of scores observed in this study or any slight lack of homogeneity of variances or regression coefficients that may be present. Results Leg and Facial s in Dams leg scores for,6 dams, representing a sample of horses, are listed in Table. These dams are from the 6 sire families I used to study leg markings. A moderate correlation was present between foreleg scores (r =.8), hind leg scores (r =.9), and the sums of foreleg and hind leg scores (r =.6). The means of the scores support the conclusions of previous investigators (Crew and Buchannan Smith 9, Dreux 966), working with other breeds of horses, that white markings are more extensive on the average on the hind legs than they are on the forelegs (. versus.9). The means also show the presence of directional asymmetry; more whiteness was present on left side legs than on right side legs (fore:. versus.87; hind:.77 versus.6). This difference was noted by Dreux (966) in French trotting horses. Relationships between dam face scores (y>q and dam leg scores ( Yj) are shown in Table. A moderate correlation existed between the face scores and the individual leg scores (rvaries from.8 to.). The strength of the relationship increases if the ^variables are the sums of the foreleg scores (r =.6) or the sums of the hind leg scores (r =.6) and reaches a maximum if the Yi variables are the sums of the total leg scores (j =.). These results verify the conclusions of other investigators (Dreux 966, Nebe 98, Walther 9) that white facial markings are correlated with white leg markings, suggesting that both types of white markings are products of the same genetic mechanism. Influence of Sires on Leg s of Foals The total leg scores of the 6 sires I used in the study and the adjusted means of total leg scores in their foals are shown in Table. employed adjusted means rather than actual means because the sires were mated with different dams. I obtained the adjusted means from analysis of covari- Table. Relationship between face score (JQ and leg scores (FJ in,6 dams Leg Right foreleg Left foreleg Right hind leg Left hind leg Forelegs Hind legs All four legs r Range of face scores: - Range of each leg score: - Range of the sum of foreleg scores: - Range of the sum of hind leg scores: - Range of the sum of all four leg scores: - ance procedures, using the dam score as the covariate. A slight curvilinear regression is suggested by the points shown in Figure (F= 6.8 for a third-degree polynomial analysis and for and df, P <.). However, because of the limited number of sires and the presence of genetic heterogeneity among sires with the same score, the results seemingly are also compatible with a linear relationship model, as shown for 7, pairs of dam-foal facial scores in the previous study (Woolf 989). The best-fitting linear line drawn through the points has a slope of b =.6. The scatter of means around this regression line (Figure ) is minimal considering the limited number of sires and the variation in the number of foals produced by the sires. The observed relationship supports the hypothesis that white leg markings as well as white facial markings (Woolf 989) have a multifactorial mode of inheritance. Influence of Dams on Leg and Facial Markings in Foals If one defines the heritability of a metric character as the ratio of additive variance to phenotypic variance (/J = V A /Vp), procedures are described in the literature for estimating the heritability of a metric character when a stallion is mated with many different mares and each mating leads to the production of one foal (Becker 98, Falconer 98). For each sire family, the regression coefficient is calculated from the paired dam-foal measurements, and the individual sire family regression coefficients are pooled to obtain a common (weighted) regression coefficient. The estimate of heritability is obtained from the equation t? = b c, where b c is the common regression coefficient. Heritability estimates assume that the metric character has a Gaussian distribution, which is not the case for facial and Table. Total leg score in each sire and the adjusted mean of the total leg scores in his foals Sire Sire total leg score No. foals Adjusted mean of total leg scores in foals Relationship between sire total leg scores and adjusted foal means: Number of paired measurements: n = 6 Regression coefficient: 6 =.6 Correlation coefficient: r=.9 Coefficient of determination: i =.89 leg scores in the Arabian horse. The issue is complicated further because, as shown below, the scores for bay horses and chestnut horses had different distributions, suggesting that the alleles E and e influence phenotypic variance. This situation will be addressed in a subsequent publication dealing with bay and chestnut Arabian populations. It was considered by Nebe (98), who concluded that additive variance, dominant variance, and nongenetic variance contribute to phenotypic variance in Hessian saddle horses. Although the assumptions behind the equation W- = V A /V P are not fulfilled in the total Arabian population, the values derived from this equation yield comparative biological information that is useful for the study of white markings in the peripheral regions (face and legs). The heritability estimates (Table ) using individual leg scores ranged from. to.. They are higher when using the sums of the foreleg scores (/J =.6), the sums of the hind leg scores (Z? =.), and the sums of all leg scores (Z? =.68). The heritability values increase as additional leg areas are included. The observed increase also occurs if a correction is made for positive skewness by using the logarithmic transformation Log e (score + c), where c is a constant added to each score to optimize the goodness of fit to a Gaussian distribution. Because of different skewness values, the value of c differs between facial and leg scores. The heritability values Downloaded from at Penn State University (Paterno Lib) on March, 6 Woolf Common White Markings in the Arabian Horse

5 Table. Regression of foals on dams and heritability" Location of white characteristic Common regression coefficient (fee ± SE) Heritability (A ± SE) Right foreleg.77 ±.8 Left foreleg.7 ±.9 Right hind leg. ±.96 Left hind leg.76 ±. Forelegs Hindlegs All legs.88 ±..66 ±.9 Face.7 ±.8 Face and all legs.8 ±..8 ±.6. ±.8.8 ±.9.7 ±..66 ±.6. ±.8.99 ± ±..687 ±.6.77 ±. " For each of the 6 sire families a regression coefficient was calculated from the paired dam-foal measurements, and the individual sire family regression coefficients were pooled to obtain a common (6J regression coefficient. The estimate of heritability is obtained from the equation /i = 6 r The standard error (SE) of this estimate is (standard error of the common regression coefficient). based on transformed scores are lower than are those based on actual scores. The observation that white facial scores show a higher correlation with the sums of all leg scores (Table ) than they do with individual leg scores or the sums of the foreleg or hind leg scores (Table ) and that heritability increases as additional leg areas are included (Table ) leads to the model in which polygenes influence the potential for white markings in the peripheral parts of the body and the heritability of white markings in the total body is higher than the heritability of whiteness is any subdivision of the body. This model is supported by the data in Table. The heritability of facial scores was /i =.68. Combining the facial score and the total leg score for each horse produces a score, ranging from to, that reflects total whiteness in the body. The heritability value based on these scores is If =.77, which is higher than either the heritability of white facial markings or the heritability of total white leg markings. White Leg and Facial Markings in and Dams The results from the study of white leg and facial markings in, bay dams and, chestnut dams are summarized in Tables through 7. The data support other studies (Crew and Buchannan Smith 9, Dreux 966, Munckel 99, Munckel 9, Nebe 98, Walther 9) that found that common white markings are more extensive in chestnut horses than in bay horses. As shown by the mean scores, more whiteness was present in the legs of chestnut dams than in those of bay dams (Table ). < O SIRE TOTAL LEG SCORE Figure. Graphic representation of the data in Table. Adjusted foal means (total leg scores) are plotted against the total leg scores of 6 diflerent sires. The regression line has a slope of 6 =.6, and the coefficient of determination is r =.89. The differences in means are all highly significant. dams exhibit a quantitative shift upward of leg scores, relative to bay dams, resulting in fewer chestnut dams with no white leg markings and more chestnut dams with stockings. The distributions in bay dams are positively skewed, but in chestnut dams the distributions have a reversed J shape (Table 6). The study of white facial markings in these dams also shows that chestnuts are more heavily marked (Table 7). The mean score for chestnut dams is higher than is the mean score for bay dams (.9 versus.96, P <.), and the two distributions are remarkably different. There is a quantitative shift in the facial scores of the chestnut dams, relative to the bay dams, toward the higher values, resulting in fewer chestnut dams with scores of and more chestnut dams with scores of. The distribution of scores for bay dams is positively skewed (coefficient of skewness = +.); the distribution for chestnut dams is slightly negatively skewed (coefficient of skewness = -.). Combining the two distributions results in a distribution with a coefficient of skewness of +.8, which is similar to the value observed in the previous study (Woolf 989). Table 7 does not show additional quantitative differences in facial markings between bay and chestnut dams. A strip is any white marking below the eyeline and above the top of the nostrils that is contained within the nasal bones. A blaze is any white marking below the eyeline and above the top of the nostrils that extends outside both nasal bone lines. Among the, chestnut dams,.% had a blaze, compared with only.% of the bay dams. An Arabian horse with broad white facial markings extending from the forehead to the chin has the markings star, blaze, snip, upper lip, lower lip, and chin. The frequency of these phenotypes among the chestnut dams was.6%. Although such markings occur in Arabian bay horses, the frequency is very rare, and none was observed in the sample of, bay dams. Thus, when markings in a bay horse extend from the forehead to the chin, the marking below the eyeline and above the nostrils tends to be a strip. White Facial Markings in Male and Female Foals Produced by Seven Sires A total of,77 foals were present in the seven sire families obtained from the Registry, with each foal classified by sex. The Table. Comparison of white leg markings In, bay and, chestnut dams" Right foreleg Left foreleg Right hind leg Left hind leg f " Range of scores on each leg = <. <. <. <. Downloaded from at Penn State University (Paterno Lib) on March, 6 The Journal of Heredity 99:8()

6 Table 6. Percentages of bay and chestnut horses with leg scores Foreleg Right Left Hind leg Right Left Table 7. Percentages of white facial scores in, bay and, chestnut dams" " Comparison of means: t =.9, P <.. sex ratio favors females (,986 versus,787). An uneven sex ratio in computerized registration records is not surprising, because registration is not always complete. Mares are usually preferred, and a male without promise as a sire or gelding may not be registered. Facial scores were determined for the male and female foals in each sire family, and adjusted means were obtained from analysis of covariance, using dams as the covariate. The adjusted means are listed in Table 8. In each sire family the males have higher adjusted means than do the females, but the difference between means is significant at P =. in only three of the seven families. Pooling the results gives adjusted means of. and.78 for male and female foals, respectively. The difference, although small, is highly significant (P >.). These results for the Arabian horse support the observation made by Dreux (966) for French trotting horses that males are slightly more heavily marked than are females. The distributions of the scores in Table 9 suggest that sex invokes a slight quantitative shift resulting in males with fewer low scores ( and ) and more high scores ( and ) than females. A similar trend occurs in both bay and chestnut horses (Table ). Discussion In mammals, melanocytes in skin and hair follicles develop from melanoblasts that are derived from neural crest cells (Mintz 97). Common white facial and leg markings in Equus caballus are characterized by a lack of pigment in hairs and in the skin beneath the hairs. The developmental events leading to common white markings are unknown. Although the failure of certain melanoblasts to survive in peripheral locations may be involved, a likely explanation for common white facial and leg markings is the failure of melanoblasts to migrate and proliferate fully in presumptive limb and facial tissues (Seidel and Elsden 987). This explanation assumes that complex genetic systems and nongenetic factors influence the migration and clonal proliferation of melanoblasts in presumptive limb and facial tissues. A polygenic system, alleles at the E locus, and the genetic mechanisms associated with sex differentiation all seem to play a role. The difference between bay and chestnut horses suggests that melanoblasts with the /Tallele have migration or proliferation patterns different from those of melanoblasts with the genotype e/e or that peripheral tissues with different genotypes influence the fate of immigrant melanoblasts differentially. However, the possibility also must be considered that the effect is not due to contrasted alleles at the E locus but results from an important closely linked locus that is part of the polygenic system that influences variation in common white markings. Thus, a chromosome with e may tend tp contain a different closely linked allele than does a chromosome with E. Past selection by breeders for chestnut horses with "flashy" white markings, a phenotype favored by some fanciers, could have caused this type of linkage disequilibrium. The possibility of such a model is supported by the observation that the loci for the tobiano and roan phenotypes are closely linked with the locus (Andersson and Sandberg 98, Sponenberg et al. 98), suggesting that this region of the chromosome is important in regulating melanogenesis in E. caballus. Although the slight quantitative difference between males and females I observed can be attributed to basic mechanisms related to sex differentiation, it may merely reflect quantitative differences resulting from the presence of one or two X chromosomes. Even though I observed an overall significant sex difference in this study, variation existed among the seven sire families, suggesting that sires influence this phenomenon. If variation existed among X chromosomes for important alleles acting additively, sire families would be expected to show varying degrees of divergence between male and female foals depending on the X chromosomes in the sires. The role of nongenetic factors is implied by heritability values that are below.. Experimental evidence for the role of nongenetic factors comes from a study of monozygotic horse twins produced by embryo micromanipulation (Allen and Pashen 98, Slade et al. 98). Two pairs of monozygotic twins produced by Allen and Pashen (98) showed marked variation in white leg markings. One member of a pair of male twins had socks on all four legs; the other member had socks on only the left foreleg and left hind leg. One member of a pair of female twins had a sock on the left foreleg; the other member had no white leg markings. Prominent white facial markings in both sets of twins showed differences in outline between the members of each pair. The influence of genetic and nongenetic factors also can be assessed by observing white facial markings and white leg markings on both sides of the same horse. Different sides of the same bilateral organism are analogous to monozygotic twins. Although each side develops autonomously (Mintz 97), the frequent occurrence of similar white marking patterns on both sides of the same animal provides evidence that the migration, clonal proliferation, and survival of melanoblasts is under genetic control. Asymmetry results from variability in cell movements, clonal proliferation, and perhaps cell survival, termed "ordinary developmetnal noise" by Mintz (97). Aside from the disturbance caused by directional asymmetry observed in this study (Table ), each side has the same genetic potential for common white markings in the facial and leg regions. Random events in presumptive tissues result in fluctuating asymmetry, such as the pronounced extension of a snip Downloaded from at Penn State University (Paterno Lib) on March, 6 Woolf Common White Markings in the Arabian Horse

7 Table 8. White facial markings in,77 foals produced by seven sires" Sire 6 7 Pooled Male foals n ,787 score Female foals n ,986 " Adjusted means were obtained from analysis of covariance. into only one nostril or the presence of a stocking on one hind leg and a coronet on the other. In the presence of a given genetic potential for white markings, ordinary developmental noise may result in marked quantitative variation in the expression of white markings in any of the six peripheral regions (the four legs and both sides of the face); as a consequence, the magnitude of the potential is best assessed by considering the total organism rather than a single region. Because of the random effects in presumptive tissues, the heritability of whiteness in a single region would be lower than the heritability of whiteness in the total body (Table ). The relatively high heritability value of common white markings also indicates that selection can be effective in producing subpopulations in which the great majority of the members have no white markings or extensive white markings on the face and legs. The apparent effectiveness of selection in other breeds was discussed by Sponenberg and Beaver (98). Thus, variation in white markings in the Arabian horse is influenced by complex genetic systems and ordinary developmental noise, resulting in fluctuating bilateral asymmetry. A further analysis of the genetic systems, including differences between bay and chestnut horses, will be the subject of future publications. score References r P P < P<. P< > P>. > P >. > P >. > P >.. > P>.. Allen WR and Pashen RL, 98. Production of monozygotic (identical) horse twins by embryo micromanipulation. J Reprod Fert 7:67-6. Anderson NH, 96. Scales and statistics: parametric and non-parametric. Psych. Bull. 8:-6. Andersson L and Sandberg K, 98. A linkage group composed of three coat color genes and three serum protein loci in horses. J Hered 7:9-9. Becker W, 98. Manual of quantitative genetics, th ed. Pullman, Washington: Academic Enterprises; 88 p. Blunn CT and Howell CE, 96. The inheritance of white facial markings in Arabian horses. J Hered 7: Crew FAE and Buchannan Smith AD, 9. The genetics of the horse. Bibliographia Genetica 6:-7. D'Agostino RB, 97. A second look at analysis of variance on dichotomous data. J Ed Meas 8:7-. Dreux P, 966. Introduction statistique a la genetique des marques blanches limitees chez le cheval domestique. Ann Genet 9:66-7. Falconer DS, 98. Introduction to quantitative genetics, nd ed. New York: Longman; p. Glass GV, Peckham PD, and Sanders JR, 97. Consequences of failure to meet assumptions underlying the fixed effects analyses of variance and covariance. Rev Educ Res :7-88. Hintz HF and Van Vleck LD, 979. Lethal dominant roan in horses. J Hered 7:-6. Lunney GH, 97. Using analysis of variance with a dichotomous dependent variable. J Ed Meas 7:6-69. Mintz B, 97. Gene control of mammalian differentiation. Annu Rev Genet 8:-7. Munckel H, 99. Untersuchungen uber Farben und Abzeichen des Pferde und ihre Vererbung. Z Tierzucht Zuchtungbiol 6:-. Munckel H, 9. Erganzende Untersuchungen uber die Abziechen am Pferde und ihre Vererbung. Zucht :6-. Nebe HD, 98. Die Farbvererbung beim Pferd unter besonder Berucksichtigung der Vererbung weisser Abzeichen. Inaugural-Dissertation zur Erlangung des Doktorgrades bei dem Fachbereich Veterinarmedizin und Tierzucht der Justus-Liebig-Universitat Giessen; 8 p. Seidel GE Jr and Elsden RP, 987. Why identical twins may be different. Hoard's Dairyman :7. Slade NP, Williams TJ, Squires EL, and Seidel GE Jr, 98. Production of identical twin pregnancies by microsurgical bisection of equine embryos. In: th International Congress on Animal Reproduction and Artificial Insemination, Urbana-Champaign, Illinois, 98. Urbana: University of Illinois Press; :-. Sponenberg DP and Beaver BV, 98. Horse color. College Station: Texas A&M University Press; p. Sponenberg DP, Harper HT, and Harper AL, 98. Direct evidence for linkage of roan and extension loci in Belgian horses. J Hered 7:-. Walther AR, 9. Die Vererbung unpigmentierter Haare (Schimmelung) und Hautstellen ("Abzeichen") bei Rind und Pferd als Beispiele transgressiv fluktuierender Factoren. Z Indukt Abst :-8. Woolf CM, 988. Evidence of eumelanin and pheomelanin producing genotypes in the Arabian horse. J Hered Woolf CM, 989. Multifactorial inheritance of white facial markings in the Arabian horse. J Hered 8:7-78. Downloaded from at Penn State University (Paterno Lib) on March, 6 Table 9. Percentage distribution of scores by sex in,77 foals Male foals (n=,787) Female foals (n=,986) Table. Percentages of white facial markings in, bay and chestnut foals classified by sex in seven sire families Male (n = 88) Female (n = 98) Male (n = ) Female (n = 68) The Journal of Heredity 99:8()