H ‘\ NI l W || WWW‘ COMPNHSQNS OF MATES OF $§RES USED ARTIFICIALLY AND MWRALEX W‘ THE HEMTABBUTY OF MIG-MEAN B‘HEA HQfiLS-‘FEIN FROWEQN RECORDS L W Kl; l Thesis go? $5M Degree of M. S. MECEEGAN STATE UNIVERSE“ Victor Clement Bea}, Jr. 3957 WW! HWHNIWIIWWWW f _3 293010725327 } . LIBRARY Michigan State University PLACE IN RETURN BOX to roman this chockout from your ncord. To AVOID FINES Mum on or befor- ddo duo. DATE DUE DATE DUE DATE DUE sass-em __ N d k ‘— ‘3) I if“ 3- ‘nnn ‘ 4— * l ' , \ Ej —— L. — I— E — “1% 7—7:: MSU IoAn AfflrmEtlvo Action/Emil Oppomnny Imam W nus-o. COMPARISONS OF MATLS OF SIRES USED ARTIFICIALLY AnD NATURALLY AND THE HERITABILITY OF MICHIGAN DHIA HOLSTEIN PRODUCTION RECORDS By Victor Clement Beal, Jr. AN ABSTRACT Submitted to the College of Agriculture Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Dairy 1957 Approved [D m E ' Wadflm’ ABSTRACT Victor Clement Beal, Jr. Genetic improvement of dairy cattle is an important factor in successful dairying and is a function (a) of the genetic level of the individuals (sires and their mates) chosen to reproduce and (b) of the accuracy of selection. A knowledge of the level of mates of sires used arti- ficially (AB) and naturally (non-AB) is pertinent to the evaluation of daughters of these sires. When daughter aver— ages are compared, is it valid to assume that when these daughters are equal the sires have equal transmitting abili- ties or are there differences in levels of their mates. Heritability is a measure of the accuracy of selection and indicates the methods of most effective selection in choosing individuals to reproduce. Heritability values for Michigan DHIA Holstein cattle have not been reported pre- viously. Approximately 45,000 Dairy Herd Improvement Association records made by 20,000 Michigan Holstein cows for the period 1944 to 1956 were analyzed (a) to compare production levels of mates of AB and non-AB sires and (b) to estimate herita- bility of milk and fat production. Mates were compared in herds with both AB and non-AB daughters. Purebred and grade mates were analyzed separately since the former had higher levels of production and higher ABSTRACT Victor Clement Beal, Jr. average number of records than the latter. Purebred mates of AB sires did not produce significantly more milk or fat but did have significantly more records than non-AB pure- bred mates (12,794 pounds milk, 452 pounds fat, and 2.79 records for AB mates and 12,711 pounds milk, 449 pounds fat, and 2.60 records for non-AB mates). Grade mates of AB sires did not produce significantly less milk or fat or have significantly more records than non-AB grade mates (11,537 pounds milk, 421 pounds fat, and 2.45 records for AB mates and 11,659 pounds milk, 422 pounds fat, and 2.54 records for non-AB mates). Although some dairymen intend to mate certain levels of their cows in a given manner, evi- dence from this study indicates that when producing ability is represented by the average of a cow's records, these dairymen actually have not recognized their better cows. Heritability was computed by doubling the intra-sire (for non-AB situations) or intra-herd intra-sire (for AB situations) regression of daughter on dam for the various groups (purebred and grade, AB and non-AB). Heritabilities of single records ranged from .17 to .44 for milk and from .05 to .51 for fat. Regressions for these groups did not differ significantly and were combined to provide a pooled heritability value of .26 for milk and fat. COMPARISONS OF RATES OF SIRLS USED ARTIFICIALLY AND NATURALLY AND THE HERITABILITY OF MICHIGAN DHIA HOLSTLIN PRODUCTION RECORDS By Victor Clement Beal, Jr. Submitted to the College of Agriculture Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Dairy 1957 /-7~5£ I»; 2! 73 TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . INTRODUCTIOI‘I o o o o o o o o o o o o o o o 0 REVIEW OF LITERATURE . . . . . . . . . . . . Comparison of Mates of AB and Non-AB Sires Heritability of Milk and Fat Production SOURCE AND ADJUSTMENT OF DATA . . . . . . . METHODS OF ANALYSIS AND RESULTS . . . . . . Comparison of Mates of AB and Non-AB Sires Heritability of Milk and Fat Production DISCUSSION 0 O O O O O O O O O O O O O O O 0 Comparison of Mates of AB and Non-AB Sires Heritability of Milk and Fat Production SUMRY O O O O O O O O O O O O O O 0 O O 0 LITERATURE CITED 0 O O O O O O O O O O O 0 0 AC m 0 LUVLE D GEIULE‘: T S O O O O O O C O O O O O O O \OKOOW-PWWH 13 20 20 23 26 27 29 Table LIST OF TABLES Production levels of Michigan daughters of Holstein sires. Average milk and fat production for purebred and grade mates of AB and non—AB sires. Adjusted purebred Analysis per mate Analysis average milk and fat production for and grade mates of AB and non-AB sires. of differences in number of records for mates of AB and non-AB sires. of differences in milk and fat pro- duction for purebred and grade mates of AB and non—AB sires. Production levels of daughters and dams used in the heritability analyses. Variances and covariances for estimating her- itability of milk and fat. Regressions of daughter on dam and heritabili- ties of milk and fat production. Page 12 15 15 l7 l9 INTRODUCTION Genetic improvement of dairy cattle is an important . factor in successful dairying and is a function (a) of the genetic level of the individuals (sires and dams) chosen to reproduce and (b) of the accuracy of selection. Sires may be used artificially (AB) and/or naturally (non-AB). The production levels of daughters of such sires have been com- pared to ascertain differences in the transmitting ability of Michigan Holstein sires (Wadell, 1957 and Specht, 1957). No evaluations have been made of the production levels of mates of these AB and non-AB sires nor have heritabilities been estimated by daughter—dam regressions for Michigan Dairy Herd Improvement Association (DHIA) Holstein produc- tion records. A knowledge of the level of mates of AB and non-AB sires is pertinent to the evaluation of daughters of these sires. When daughter averages of AB and non-AB sires are compared, is it valid to assume that when these daughters are equal the sires have equal transmitting abilities or are there differences in the levels of their mates. Heritability represents the accuracy of selection and indicates the methods of most effective selection in choos— ing individuals to reproduce. Heritability values computed by the regression of daughter on dam have not been reported for Michigan DHIA Holstein cattle. The purpose of this study is (a) to compare production levels of mates of AB and non-AB sires and (b) to estimate the heritability or accuracy of selection of milk and fat production for Michigan DHIA Holstein cattle. General dairy statistics for Michigan indicate that of the 817,000 Michigan dairy cows, the approximately 60,000 cows enrolled in DHIA programs averaged 582 pounds fat and 9,725 pounds milk (Murray, 1957). Over 30 percent of Michi- gan dairy cows are serviced artificially and approximately 70 percent of the cattle are Holstein. REVIEW OF LITERATURE Comparison of Mates of AB and non-AB Sires The evaluation of a sire's transmitting ability has been considered by several authors (Gaunt and Legates, 1955; Lush and McGilliard, 1955). Various methods of sire evalu— ation are the daughter average, the daughter—dam compari— son, the equal parent index, the daughter-contemporary com- parison in the same herd, and the daughter-contemporary herd index. No one measure was found to be superior although the daughter-contemporary comparison and index were useful for correcting for large environmental differences between herds (Gaunt and Legates, 1955). A sire's transmitting ability can be defined as the amount by which he makes his daughters deviate from breed average. Biases in evaluation can be introduced in several ways, namely, through select- ing the best daughters, selecting the mates, and/or select- ing the records used for each daughter (Lush and McGilliard, 1955)- James and Southcombe (1948) compared AB heifers with mature non-AB cows in the same herd and concluded that New Zealand dairy production was being increased more‘by AB than non-AB. Robertson and Rendel (1954) and Wadell (1957) found no difference between contemporary AB and non-AB -3... daughters. These latter workers did not report on differu- ences between the dams of daughters used in their studies. No reports have been observed of comparisons of AB and non— AB mates to determine these differences. Heritability of Milk and Fat Production Heritability is defined as the fraction of the observed or phenotypic variance which is caused by differences be- tween the genotypes of the individuals (Lush, 1949). Her- itability in the narrow sense refers to that portion of the phenotypic variance which is due to the average effects of the genes, referred to as the additively genetic variance. Heritability in the broad sense also includes deviations resulting from interactions between alleles and between non- alleles. Choicmsof methods of selection depend partially on the magnitude of heritability (Lush, 1940). If heritability is high, individual selection will be effective. If herit- ability is low, but with little epistatic variance, consid- erable use of pedigrees, progeny tests, and family selection without inbreeding would be a better plan. If there is little additively genetic variance and much epistatic vari- ance, inbreeding is useful. Heritability is computed from associations between relatives. For dairy production traits, the intra-sire regression of daughter on dam is preferred to the daughter- dam correlation, which is biased by selection of the dams (Lush, 1940). Environmental differences and peculiarities. of the mating system are largely discounted by analyzing groups of females bred to one sire. When the daughters and mates of a sire are in various herds, as in AB data, an intra-herd intra—sire regression will accomplish essen- tially the same results. Reported estimates of heritability are listed below: Heritability Breed iilk Fat Source Holsteins .22 Legates E1957) Holsteins .25 Gifford 1950) Holsteins .25 - .50 Lush, et a1. (1941) Holsteins .19 — .24 .17 - .21 Mitchell, et a1. (1957) Several Breeds .27 Beardsley, et a1. (1950) Several Breeds .25 Lush and Schultz (1956) Several Breeds .14 - .17 Lush and Straus (1942) Ayrshire .25 - .50 Hahadevan (1951) Ayrshire .51 .28 Tyler and Hyatt (1947) Heritability, as a biological parameter, is not a con- stant value to be applied to all groups of dairy cattle indiscriminantly but may be expected to vary from situation to situation. The interesting point is that agreement is as general as reported above, since heritability values from .2 to .5 encompass most of the reported values. Removal of recognized sources of variation usually result in values comparable to those generally reported. SOURCE AND ADJUSTMENT OF DATA The data used in this study were DHIA milk and fat production records of Michigan Holstein cows since 1944. These records were available from two sources: records reported to the United States Department of Agriculture by the DHIA Supervisors and returned to the Michigan State University Dairy Department, and records available from the Michigan DHIA-IBM program. The former records were lactations initiated between January 1944 and September 1955 while the latter were lactations initiated between the start of the Michigan DHIA—IBM program in 1955 and January 1, 1956. The difference in the final dates results from the extended period required to process the DHIA rec- ords. Completed lactations of 180 to 505 days duration or the first 505 days of longer lactations were included. Only lactations with twice—a—day milking were used. These records were adjusted for age differences using the fac- tors presented by Kendrick (1955) thereby providing 2x-505 day-M.E. milk and fat records. After eliminating the rec- ords not fulfilling the above conditions, there were about 45,000 purebred and 4,000 grade lactations from the first source and 8,000 purebred and 10,000 grade lactations from _ g _ the second source. Approximately one-third of the grade lactations from the second source were discarded because of missing sire and/or dam identification. Specht (1957) is presented in Table 1 to indicate the gen- A summary by eral levels of production for various breeding methods, and the overall level for Michigan DHIA Holsteins. data are grouped as purebred and grade daughters of AB and non-AB sires. These Table 1--Producgion levels of Michigan daughters of Holstein sires. Average Source of data Number of pounds of Sires Daughters Sires Daughters Records Milk Fat AB Purebred AB 159 2,678 4,958 12,557 451 AB Purebred non-AB 141 1,985 4,515 15,296 484 AB Grade AB 117 2,420 5,755 12,049 457 AB Grade non-AB 75 292 525 12,526 444 Ron—AB Purebred non-AB 6,716 22,274 44,165 12,215 451 Non-AB Grade non-AB 142§2 4,426 2,500 11,285 421 Total 8,475 54,075 65,592 12,257 455 aSpecht (1957). From the data presented in Table 1, dams were selected which had records in the same herds as their daughters. Only mates of registered sires were used. Daughters were INTRODUCTION Genetic improvement of dairy cattle is an important .factor in successful dairying and is a function (a) of the genetic level of the individuals (sires and dams) chosen to reproduce and (b) of the accuracy of selection. Sires may be used artificially (AB) and/or naturally (non—AB). The production levels of daughters of such sires have been com- pared to ascertain differences in the transmitting ability of Michigan Holstein sires (Wadell, 1957 and Specht, 1957). No evaluations have been made of the production levels of mates of these AB and non-AB sires nor have heritabilities been estimated by daughter-dam regressions for Michigan Dairy Herd Improvement Association (DHIA) Holstein produc- tion records. A knowledge of the level of mates of AB and non—AB sires is pertinent to the evaluation of daughters of these sires. Vhen daughter averages of AB and non-AB sires are compared, is it valid to assume that when these daughters are equal the sires have equal transmitting abilities or are there differences in the levels of their mates. Heritability represents the accuracy of selection and indicates the methods of most effective selection in choos- ing individuals to reproduce. Heritability values computed by the regression of daughter on dam have not been reported for Michigan DHIA Holstein cattle. The purpose of this study is (a) to compare production levels of mates of AB and non-AB sires and (b) to estimate the heritability or accuracy of selection of milk and fat production for Michigan DHIA Holstein cattle. General dairy statistics for Michigan indicate that of the 817,000 Michigan dairy cows, the approximately 60,000 cows enrolled in DHIA programs averaged 582 pounds fat and 9,725 pounds milk (Murray, 1957). Over 50 percent of Michi- gan dairy cows are serviced artificially and approximately 70 percent of the cattle are Holstein. REVIEW OF LITERATURE Comparison of Mates of AB and non-AB Sires The evaluation of a sire's transmitting ability has been considered by several authors (Gaunt and Legates, 1955; Lush and McGilliard, 1955). Various methods of sire evalu- ation are the daughter average, the daughter-dam compari- son, the equal parent index, the daughter-contemporary com- parison in the same herd, and the daughter-contemporary herd index. No one measure was found to be superior although the daughter-contemporary comparison and index were useful for correcting for large environmental differences between herds (Gaunt and Legates, 1955). A sire's transmitting ability can be defined as the amount by which he makes his daughters deviate from breed average. Biases in evaluation can be introduced in several ways, namely, through select— ing the best daughters, selecting the mates, and/or select- ing the records used for each daughter (Lush and McGilliard, 1955). James and Southcombe (1948) compared AB heifers with mature non-AB cows in the same herd and concluded that New Zealand dairy production was being increased more‘by AB than non-AB. Robertson and Rendel (1954) and Wadell (1957) found no difference between contemporary AB and non-AB _3- daughters. These latter workers did not report on differ- ences between the dams of daughters used in their studies. No reports have been observed of comparisons of AB and non— AB mates to determine these differences. Heritability of Milk and Fat Production Heritability is defined as the fraction of the observed or phenotypic variance which is caused by differences be- tween the genotypes of the individuals (Lush, 1949). Her- itability in the narrow sense refers to that portion of the phenotypic variance which is due to the average effects of the genes, referred to as the additively genetic variance. Heritability in the broad sense also includes deviations resulting from interactions between alleles and between non- alleles. Choicasof methods of selection depend partially on the magnitude of heritability (Lush, 1940). If heritability is high, individual selection will be effective. If herit- ability is 1ow, but with little epistatic variance, consid- erable use of pedigrees, progeny tests, and family selection without inbreeding would be a better plan. If there is little additively genetic variance and much epistatic vari— ance, inbreeding is useful. Heritability is computed from associations between relatives. For dairy production traits, the intra—sire regression of daughter on dam is preferred to the daughter— dam correlation, which is biased by selection of the dams (Lush, 1940). Environmental differences and peculiarities. of the mating system are largely discounted by analyzing groups of females bred to one sire. When the daughters and mates of a sire are in various herds, as in AB data, an intra—herd intra—sire regression will accomplish essen- tially the same results. Reported estimates of heritability are listed below: Heritability Breed Milk Fat Source Holsteins .22 Legates E1957) Holsteins .23 Gifford 1930) Holsteins .25 - .30 Lush, et al. (1941) Holsteins .19 — .24 .17 - .21 Mitchell, et al. (1957) Several Breeds .27 Beardsley, et al. (1950) Several Breeds .25 Lush and Schultz (1936) Several Breeds .14 - .17 Lush and Straus (1942) Ayrshire .25 - .30 Mahadevan (1951) Ayrshire .31 .28 Tyler and Hyatt (1947) Heritability, as a biological parameter, is not a con- stant value to be applied to all groups of dairy cattle indiscriminantly but may be expected to vary from situation to situation. The interesting point is that agreement is as general as reported above, since heritability values from .2 to .3 encompass most of the reported values. Removal of recognized sources of variation usually result in values comparable to those generally reported. SOURCE AND ADJUSTMLNT OF DATA The data used in this study were DHIA milk and fat production records of Michigan Holstein cows since 1944. These records were available from two sources: records reported to the United States Department of Agriculture by the DHIA Supervisors and returned to the Michigan State University Dairy Department, and records available from the Michigan DHIA-IBM program. The former records were lactations initiated between January 1944 and September 1955 while the latter were lactations initiated between the start of the Michigan DHIA-IBM program in 1953 and January 1, 1956. The difference in the final dates results from the extended period required to process the DHIA rec- ords. Completed lactations of 180 to 305 days duration or the first 305 days of longer lactations were included. Only lactations with twice—a-day milking were used. These records were adjusted for age differences using the fac- tors presented by Kendrick (1953) thereby providing 2x-305 day-H.E. milk and fat records. After eliminating the rec- ords not fulfilling the above conditions, there were about 45,000 purebred and 4,000 grade lactations from the first source and 8,000 purebred and 10,000 grade lactations from _ 6 _ the second source. Approximately one—third of the grade lactations from the second source were discarded because of missing sire and/or dam identification. A summary by Specht (1957) is presented in Table l to indicate the gen- eral levels of production for various breeding methods, and the overall level for Michigan DHIA Holsteins. data are grouped as purebred and grade daughters of AB and non-AB sires. These Table 1—-Producgion levels of Michigan daughters of Holstein sires. Average Source of data Number of pounds of Sires Daughters Sires Daughters Records Milk Fat AB Purebred AB 139 2,678 4,938 12,537 451 AB Purebred non-AB 141 1,983 4,513 13,296 484 AB Grade AB 117 2,420 3,753 12,049 437 AB Grade non-AB 75 292 523 12,326 444 Non-AB Purebred non-AB 6,716 22,274 44,165 12,215 431 Non-AB Grade non—AB ;,g§2 4,426 2,500 11,283 421 Total 8,475 34,073 65,392 12,237 435 aSpecht (1957). which had records in the same herds as their daughters. Only mates of registered sires were used. Daughters were From the data presented in Table 1, dams were selected classified as purebred or grade depending on the presence or absence of a registration number and their dams were classified accordingly. Therefore, if an unregistered daugh— ter had a registered dam, both were classified as grades. Daughters were classified as daughters of AB and non-AB sires and their dams accordingly were classified as mates of AB and non-AB sires. The dam was included with each daughter and therefore might appear both as an AB and a non-AB dam. Average records were used as the measure of producing ability. In comparing mates of AB sires and non-AB sires, data were included only for herds which had both AB and non-AB mates. In estimating heritability, data were in- cluded for all herds with daughter-dam comparisons. Hat- ural daughters of AB sires are listed in Table l as non-AB daughters of AB sires. These were not used in the compari- son of mates but were used in the heritability analyses. MLTHODS 0F ANALYSIS AKD RESULTS Comparison of Mates of AB and non-AB sires In the comparison of mates of AB and non-AB sires, mates were grouped also as purebred or grade as noted above. Sta— tistics for these four groups of dams are presented in Table 2, where average production levels were obtained from the sum of all average records divided by the number of cows. Table 2--Average milk and fat production for purebred and grade mates of AB and non-AB sires. Source of data NEE Records Milk (lbs.) Fat (1bs.) Sires Mates mates Avg. Std. Avg. Std. Avg. Std. No. dev. dev. dev. AB Purebred 1,055 2.89 1.70 12,623 2,322 448 84 Non-AB Purebred 2,286 2.77 1.53 12,723 2,415 451 90 AB Grade 126 2.52 1.59 11,756 2,145 422 71 Non-AB Grade 182. 2.39 1.55 11 66 2 o 6 Egg zg Total 3,654 2.79 1.67 12,621 2,575 448 88 Purebred mates of AB and non-AB sires differ little in aver- age number of records and production levels. Grade mates of A3 and non-AB sires differ little in average number of rec— ords and production levels but are distinctly lower than purebred mates. 10 Because of wide variation in numbers of comparisons within herds, an analysis of variance for two classifica— tions and disproportionate subclass numbers using unweighted means (Snedecor, 1946) was chosen to test for differences between AB and non-AB mates. Herds with only one mate in a Herd x Method of Breeding subclass were not included when comparing mates since these provided no information for es- timating the error or within subclass variance. Average num— bers of records and production levels were obtained for AB and non—AB mates in the same herd. These intra—herd AB and non—AB averages were then averaged over all herds and are presented in Table 3 as adjusted averages. Table 3--Adjusted average milk and fat production for purebred and grade mates of AB and non-AB sires. Source of data No. Records Milk filbs.) Fat (lbs.) Sires Mates h 2: Avg. Std. Avg. Std. Avg. Std. e S No. dev. dev. dev. AB Purebred 183 2.79 1.02 12,794 1,876 452 69 Non-AB Purebred 183 2.60 .82 12,711 1,713 449 63 AB Grade 31 2.43 .95 11,557 1,678 421 59 Non-AB Grade 31 2.34 1.08 11,659 1,628 422 69 Total 214 2.65 .94 12,585 1,819 446 66 The same relative situation is evident as in Table 2 Since purebred mates did not appear different, grade mates did not 11 appear different, but purebred mates had greater numbers of records and higher production than grade mates. The sources of variation considered in analyzing these data to compare mates were Herds, Methods of Breeding, Herds by Methods Interaction, and Error. The first three of these were found directly by obtaining sums of squares on the ad- justed averages, i.e., using subclass means as the individual observation. A within subclass error term was computed as the difference between total and subclass sums of squares with the individual cow average production as the variable. This error term was then adjusted for use with the other three sources of variation by multiplying by the product of the reciprocal of the number of subclasses and the sum of the reciprocals of the number (ni) of mates in each sub— class, so that: 1 l 1 1, Within subclass — -+000 2 07‘ ’ Number of subclasses n1. 112 n mean square Since average records were used in this study, it was impossible to remove the effects of years and of the number of records per mate. In order to ascertain whether differ- ences existed in the number of records per mate, an analysis (as described above) was made with number of records per mate as the variable. These analyses are presented in Table 4. 12 Table 4—-Analysis of differences in number of records per mate for mates of AB and non-AB sires. Source of variation Purebred mates Grade mates d.f. Mean Square d.f. Mean Square Herds 182 1.383** 30 l.586** Methods of Breeding 1 3.342‘* 1 .101 Herds X Hethods 182 .335 30 .482 Error 2,975 .429 251 .476 ‘TSignificant at the 1% level. Highly significant differences between herds in numbers of records were found for both purebred and grade mates. The purebred AB mates had significantly more records per mate than did their non-AB counterparts. No difference in num- bers of records per mate between grade AB and non-AB mates wa 8 found. Differences in levels of milk and fat production for AB and non-AB mates were determined using the analysis noted above and are presented in Table 5. 13 Table 5—-Ana1ysis of differences in milk and fat production for purebred and grade mates of AB and non-AB sires. - r Source of variation Purebred mates Grade mates d.f. Mean Square d.f. Mean Sguare Milka Fat Kilka Fat Herds 182 5,651** 7,612" 50 4,855** 6,579** Methods of Breeding 1 628 959 l 252 7 Herds X Methods 182 840 1,109 50 611 596 Error 2,975 779 1,026 251 765 950 aMilk multiplied by 10‘5. ‘*Significant at the 1% level. Significant differences between herds for levels of milk and fat production for A3 and non-AB mates were found for both purebreds and grades. Mates did not differ in production levels between the Methods of Breeding for either purebred or grade mates nor were significant Herd x Kethods interac— tions present. Heritability of Milk and Fat Production Heritability is measured by associations between rela- tives. Daughter—dam association is the usual method of com— parison for dairy production purposes. Comparisons used in this study have been grouped as purebred or trade dauthters of AB sires and include daughters conceived prior to and dur- in; AB use, and as purebred or grade daughters of non-AB sires. 14 The relative levels of production of daughters and dams in- cluded in this portion of the study are presented in Table 6. Daughters exceed dams in fat production, are about equal in milk production, and have fewer records per individual than their dams. In order to avoid the bias which selection of dams would contribute to a daughter-dam correlation, the regression of daughter's production on dam‘s production is preferred (Lush, 1940). An intra-sire regression provides information on differences which exist only between females mated to the same sire. This eliminates differences between sires and analyzes only differences or variations within sires. The number of herds, the number of sires, the number of sires in more than one herd, and the number of sires within herds involved in each comparison are presented in Table 6. In the two non-AB sire groups and the group of grade non-AB daughters of AB sires, relatively few sires were used in more than one herd. Assuming natural sires were used in only one herd, an intra—sire analysis would be appropriate. For the group of purebred non-AB daughters of AB sires, a higher percentage of sires were used in several herds than in the other non-AB situations but a lower per- centage than in AB situations. A high percentage of AB sires were used in more than one herd. An intra—herd intra—sire 15 ass mem.ma mm.m He.m mme emm.ma ew.H 0mm mmm.m mmo.m mmo.m oum.ma fleece Mme rm».aa Ho.m mm.m mm: muwwaa oe.H ea mm: ome Ham 6mm {I wedge menace menace mes www.ma om.m ee.m wee mme.ma mw.a mom omn.m mom.m mow mma.w eonnmnsm menace malnoz eme mon.mfi mw.H o¢.m ewe emm.ma mm.a m an em me mad memnm Mdlfloz Md ems oefl.ma mm.a mo.m es: wam.ma om.a me sum om mam ewe meegm me me mus mmfl.ma me.m ou.m mom www.ma ¢H.m mm mmm moa mea Ham eenpmnsm _ menace m4 wee mee.mH mm.m on.m mm: www.ma ma.H mm mem.a mmfl mod emm.a emmpmnsm me me anon Amv ono mmoma mescoen menooen damp IBREUO Ho Mo 0908 menon awn meme A.mpav A.mpav .oz A.mpav A.mpav .02 ea queues lamp and pmm Mafia NEKV .w>¢ pmm Mafia .m>< monflm mohflm moHHm mvhem lawman mumpnwzmm meHHm mama mumpmmsmm Ho nepapz awed Ho oohfiom .memmamqw hpflaapwpfinen map ma vows meme was whepnmsme mo mHm>mH moaposdoHMIlm manna l6 analysis would be appropriate for such situations. Some were not used in more than one herd because of a short stay in the AB organization, not maintained in a Michigan AB organi— zation, or erroneous calving dates. The usefulness of the intra-sire and intra—herd intra- sire covariance analyses for estimating the regression of daughter on dam depends on the validity of the assumptions listed above. The intra-sire analysis is probably less va- lid for the non—AB situation where half the sires have been used in more than one herd than for the non-AB situations where few sires have been used in more than one herd. The intra-sire analyses were computed for purebred and grade daughters of non-AB sires and for grade non-AB daughters of AB sires. Intra-herd intra—sire analyses were obtained for purebred and grade AB daughters and for purebred non—AB daughters of AB sires, the latter group being included in this type of analysis because about half of these sires were used in more than one herd. ApprOpriate variances and co- variances for the different groups are listed in Table 7. The first part contains the total variances and covariances while the second and third parts contain the intra—sire or intra—herd intra—sire variances and covariances. The ap- propriate regression values (b) presented in Table 8 are ob- tained by dividing the covariance by the variance of dams in the lower part of Table 7. Table 7--Variances and covariances of milk and fat. 1'7 for estimating heritability Source of variation d.f. Milka Fat Sires Dams and Var. Cov. Var. Var. Cov. Var. daughters Dams Daus. Dams Daus. Total AB AB purebred 1,585 5,652 2,501 5,859 7,552 5,421 7,652 AB Non-AB purebred 990 6,525 2,247 6,868 9,011 5,457 9,045 AB AB grade 626 5,686 2,468 5,962 7,580 5,280 7,572 A3 Non-AB grade 112 5,524 2,722 6,579 6,685 4,075 8,022 Non-A3 Non-AB purebred 8,126 6,460 2,648 6,451 8,654 5,774 8,508 Non-AB Non-AB grade 927 5,095 2,505 4,774 6,916 5,545 8,217 Intra-sire AB Non-AB grade 77 5,552 1,025 5,158 4,105 1,424 5,660 Non-AB Non-AB purebred 5,821 4,094 781 5,865 5,159 1,007 4,908 Non-AB Non-AB grade 508 2,750 557 5,009 5,611 612 4,174 Intra-herd intra—sire AB AB purebred 241 5,674 427 5,086 5,986 487 5,922 AB Non—AB purebred 765 5,820 458 4,090 4,897 601 5,457 AB AB grade 50 2,081 519 2,045 2,055 61 2,750 aMilk multiplied by 10‘3. 18 To correct the regressions for the reduced variation resulting from the use of average records of the dams, the factor of Lush and Straus (1942), 1+5ffi-1)r+a'2m(l-r) a m5 where a is the mean of the number of dams' records, r is the repeatability of milk or fat records,and (7mg is the variance of the dams' records, was multiplied by the computed re- gressions (b) to obtain regressions corrected to a single record basis. ApprOpriate m and.(72m values for dams are listed in Table 6. Repeatability values (r) of .46 for milk and .40 for fat were obtained by Specht (1957). Her- itability values, obtained by doubling the corrected regres- sions, and their standard errors are listed in Table 8. The original regressions were compared (Goulden, 1952) to determine if they were significantly different. Since these regressions did not differ significantly, the original sums of squares, crossproducts, and degrees of freedom for intra- sire or intra-herd intra—sire groups were pooled. The re- gression obtained was corrected to a single record basis and doubled to provide an overall heritability of .26 i .02 for both milk and fat as noted in Table 8. 19 Table 8--Regressions of daughter on gam and heritabilities of milk and fat production. Source of variation Milk Fat Sires Dams and b Heritability b Heritability Daughters AB AB purebred .116 .168 i .084 .122 .168 i .088 AB Non-AB purebred .115 .166 i .054 .125 .170 i .052 AB AB grade .155 .252 i .250 .050 .048 i .264 AB Non-AB grade .290 .458 i .204 .547 .506 t .148 Non-AB Non—AB purebred .191 .278 i .018 .195 .274 i .018 Non-AB Non-AB grade :151 .202 i .072 1182 .250 i .070 Pooled .179 .264 i .016 .186 .262 i .018 aRegressions computed for average corrected to a single-record basis. records; heritabilities DISCUBSIOK Comparison of Mates of AB and non-AB sires The main purpose of this study was to determine if mates of AB sires had milk and fat production levels differ- ent from mates of non—AB sires. Averages and adjusted av- erages in Tables 2 and 5 indicate that AB and non-AB mates do not differ in levels of production. Grade mates are lower than purebred mates and were analyzed separately. Further bases for concluding that significant differences in production do not exist between AB and non-AB mates are found in Table 5 where differences between Methods of Breed- ing are not significant for milk and fat production of pure— bred and grade mates. The conclusion that significant differ— ences do not exist between AB and non-AB mates has practical significance in two respects, namely, reports that AB and non-AB daughters do not differ and reports that certain dairymen intend to breed their best cows in a given manner. When daughters of AB and non-AB sires were compared, Robertson and Rendel (1954), working with data from several British dairy cattle breeds, and Wadell (1957), working with several breeds of Michigan dairy cattle, found no signifi- cant differences. These authors apparently considered mates of AB and non—AB sires about equal. Since daughters did not - 20 - 21 differ, they concluded that AB sires transmitted the same level of production as did non-AB sires. Since no signifi- cant differences between mates were found, the present study, using some of the same data, substantiates Wadell's (1957) conclusion of no differences between Holstein sires. That is to say, if neither daughters nor mates differ, sires do not differ. Various opinions have existed concerning which cows are mated artificially in herds which use both AB and non- AB sires. In reszonse to a questionnaire (Wadell, 1957), Michigan dairymen indicated that the grade herds are breed- ing their better cows artificially more frequently than the registered herds, that some dairymen nith both registered and grade cows may breed the formerartificially and the latter naturally, and that the grade herds are breeding their heifers naturally more frequently than the registered herds. The present study is able to offer only general in- formation on these subjects. If Michigan dairymen are at- tempting to mate certain leVels of their cows in a given manner, they have not been successful. Although their in- tentions may be to mate their best cows artificially, or vice versa, insofar as the average record is a good indica— tion of the cow's ability, the dairymen have not actually recognized their better cows. This is substantiated by the work of Dunbar and Henderson (1954) who found that when 22 comparing several approximate indexes with a "best" index for ranking cows, the owner's ranking of his cows was the least correlated with the "best" index of any of the indexes considered. lhe present study also indicates that the grade breeder is not mating his better cows artificially any oftener than is the purebred breeder. As mentioned previously, a dam may appear both as an AB and non—AB mate. The occurrence of such dams would re- duce differences between A3 and non-AB levels and supports the conclusion that dairymen actually are not picking out their better cows to mate in a given manner even if they desire to do so. Since various views have been expressed concerning length of stay of mates in herds, the number of records per cow was tested to determine if AB mates remained in the herds longer than did non-A3 mates. As illustrated in Table 4, herds differ significantly in the number of records per mate, but only purebred mates differ significantly between hethods of Breeding as to the number of records per cow. These small differences (2.79 for AB and 2.60 for non—AB mates) probably have little practical significance but statistically are significantly different, indicating that purebred AB mates remain in the herd slightly longer than purebred non-AB mates. Differences in numbers of records per cow for grade AB and grade non—AB mates were not significant. 25 Pertinent questions not considered here were: year to year effects and possible differences in variation of individual production records by the same cow. Specht (1957) found no significant differences between levels of produc- tion for different years. The use of average records would have permitted random year to year variations partially to cancel out, but permitted no measurement of variability of individual production records. Heritability of Milk and Fat Production The choice of breeding plans is dependent on the accu- racy of selection as measured by heritability. The purpose of this section of the study was to measure heritability of milk and fat production in various daughter-dam groups. The heritability values presented in Table 8 are in general agreement with those found in the literature and presented above in the Review of Literature. Several of the herita- bilities found here seem extreme and have large sampling errors. These values may be the result of limited sample sizes and/or errors in field recording, such as recording identification and production information. The values from .4 to .5 for milk and fat in non—AB grade daughters of AB sires probably result from limited numbers, since only 115 comparisons on 36 sires were available. However, S of these 56 sires, or 22 percent, were used in more than one herd. 24 An intra-sire regression method was utilized and failure to remove herd differences may have increased the likeness between daughter and dam in the same herd, producing the large heritability values. The other extreme heritability value is .05 for fat for grade AB daughters. The correspond- ing value for milk of .25 may indicate an error in one or the other (milk or fat) since milk and fat are known to be highly correlated (Tabler and Touchberry, 1955). The data ' were recomputed to determine errors in calculations but none were found. Examination of the individual daughter-dam comparisons indicated that there may be errors in identify- ing the daughter with her proper sire or dam. Some large extremes were found in milk and fat between certain daugh- ters and dams with very little difference in fat percent. With limited numbers, a few cases of a high level dam with a low level daughter, or vice versa, could have lowered the regression drastically. When certain extreme records were eliminated, the regression for milk was increased slightly but the regression for fat was increased sufficiently to make it one half instead of one fifth the magnitude of the regression for milk. This leads one to believe that there were errors in recording the data but since there is no logical basis for removing these extreme records, the origi- nal records were used. 25 The heritability values computed on an intra-sire basis are higher than those computed on an intra-herd intra—sire basis, and may have overestimated heritability. If sires which were used in more than one herd had not been included, a lower heritability value would have been obtained. Table 6 indicates the number of sires used in more than one herd. As noted previously, the regressions found in Table 8 did not differ significantly and were combined to give an overall heritability of 0.26 for milk and fat. summm Michigan DHIA Holstein records from 1944 to 1956 were analyzed (a) to compare production levels of mates of sires used artificially (AB) and naturally (non-AB) and (b) to obtain heritability of milk and fat production. Purebred and grade mates were analyzed separately since the former had higher levels of production and higher aver- age number of records than the latter. Purebred mates of AB sires did not produce significantly more milk or fat but did have significantly more records than non—AB purebred mates. Grade mates of AB sires did not produce significantly less milk or fat or have significantly fewer numbers of rec- ords than non-AB grade mates. Heritabilities, computed by doubling the intra—sire or intra-herd intra-sire regression of daughter on dam, for the various groups (purebred and grade, AB and non—AB) ranged from .17 to .44 for milk and from .05 to .51 for fat. Re- ‘gressions for these groups did not differ signifiCantly and were combined to provide a pooled heritability value of .26 for milk and fat for flichigan DEIA Holsteins. - 26 - LITERATURE CITED Beardsley, J. r., Bratton, R. w., and Salisbury, G. w. 1950. The curvilinearity of heritability of butterfat produc- tion. Jour. Dairy Sci. 33:93-97. Dunbar, R. S. and Henderson, C. R. 1954. A comparison of approximate indexes for butterfat production. (Abs.) Gaunt, S. N. and Legates, J. E. 1955. The relative merits of five measures of a dairy sire's transmitting ability. (Abs.) Jour. Dairy Sci. 38:1411. Gifford, W. 1930. The mode of inheritance of yearly butter- fat production. Mo. Agr. Expt. Sta. Res. Bul. 144. Goulden, C. H. 1952. Methods of statistical analysis. 2nd ed. New York, John Wiley and Sons, Inc. ‘ James, J. P. and Southcombe, S. A. 1948. Production rec- ords of artificially-bred heifers. New Zealand Jour. Agr. 77:363. 565-566- Kendrick, J. F. 1953. Standardizing dairy herd improve— ment association records in proving sires. U.S.D.A.‘ Bureau of Dairy Ind. Inf. 162. Legates, J. E. 1957. Heritability of fat yields in herds with different production levels. (Abs.) Jour. Dairy Sci. 40:631. Lush, J. L. 1940. Intra-sire correlations or regressions of offspring on dams as a method of estimating herita- bility of characteristics. Amer. Soc. Animal Prod. Proc. 1940:293—301. . 1949. Heritability of quantitative characters in farm animals. Eighth Internat. Congress Genetics and McGilliard, L. D. 1955. Proving dairy sires and dams. Jour. Dairy Sci. 38:163-180. g_ , Norton, H. W., III, and Arnold, Floyd. 1941. Effects which selection of dams may have on sire indexes. Jour. Dairy Sci. 24:695—721. _ 27 - 28 and Schultz, E. N. 1936. Heritability of butter- fat percentage and butterfat production in the data with which sires have been proved in Iowa. (Abs.) Jour. Dairy Sci. 19:429-430. and Straus, F. S. 1942. The heritability of butterfat production in dairy cattle. Jour. Dairy Sci. 25:975-982. Mahadevan, P. 1951. The effect of environment and heredity on lactation. I. Milk yield. Jour. Agr. Sci. 41:80— 88. Mitchell, R. G., Corley, E. L., Heizer, E. E., and Tyler, W. J. 1957. Heritability and phenotypic and genetic correlations between type ratings and milk and butter- fat production in Holstein-Friesian cattle. (Abs.) Jour. Dairy Sci. 40:632. Murray, D. L. 1957. Dairy notes. Michigan State Univer- sity Dairy Extension Publication, July 15, 1957. Robertson, A. and Rendel, J. H. 1954. The performance of heifers got by artificial insemination. Jour. Agr. Sci. 44:184-192. Snedecor, G. W. 1946. Statistical methods applied to ex- periments in agriculture and biology. 4th ed. Ames, Iowa, Iowa State College Press. Specht, L. W. 1957. Unpublished Michigan DHIA Holstein data. Tabler, K. A. and Touchberry, R. W. 1955. Selection indices based on milk and fat yield, fat percent, and type classi- fication. Jour. Dairy Sci. 38:1155-1163. Tyler, W. J. and Hyatt, G., Jr. 1947. The heritability of milk and butterfat production and percentage of butterfat in Ayrshire cattle. (Abs.) Jour, Animal Sci. 6:479-480. Wadell, L. H. 1957. Influence of artificial breeding on production in selected Michigan dairy herds. Unpublished H. S. Thesis. East Lansing, Michigan, Michigan State University Library. ACKNChLEDGEmEbTS I am greatly indebted to Dr. D. E. Madden for his as- sistance and guidance in the gathering, preparing, inter- preting, and writing of this material. I am also indebted to Dr. W. D. Baten for his assistance and guidance in a por- tion of the analyses. I would like to extend many thanks to: Dr. L. D. McGilliard for his advice and helpful sug- gestions; Dr. N. P. Ralston, Head of the Department of Dairy, for his encouraging me to proceed with graduate studies; Mr. L. W. Specht for his assistance and permission to use some results of his efforts; and Mr. R. P. Witte for his assistance and technical advice in the use of I.B.H. ma- chines for analyzing these data. Finally, my sincere thanks go to my parents who have made it possible for me to pursue these graduate studies. - 29 - ROOM use 0va Date Due HICHIGRN STQTE UN V. I L | HIIWIHIII III“ \I \llllll £13 312 010725 IBRRRIES WHIHIIWI :31?7