_\._\ —‘O #0) “I, u} DIFFEREEW'EAL REMGVAL C3? {Exsz‘GH‘fERS AMO .THS_ -- mum: w “5".” unarlruuwu 3:"..s. ! ' "r1&‘...- :1. “011:4 5113:“-Y‘? 01.1.1 5 t 4'"! ; : "13‘5u‘z: ‘ ’1’ “$3M ’ v a"; ' "6 ~51 f? I ~- 3’"); 3. 3 mi; ’:1 . 3m :1. 3 ‘3'. {e (“a :3 ‘ A ' I at} 2”,"5: M “ _ LIBRARY T! w: ‘1'} MiChigan Stan University ABSTRACT DIFFERENTIAL REMOVAL OF DAUGHTERS AMONG AI SIRES by Arthur D. Dayton Accurately comparing sires in AI may be difficult if the number of daughters eliminated from test is disprOportion- ate among bulls and related to performance. The purposes of this study were to determine if rates of removal at young ages among the daughters of AI sires are equal or if some sires have a larger per cent of their daughters leaving the herd than others, to determine the dis- tribution of removals among various voluntary and involuntary reasons, and to measure the extent individual differences in reasons for disposal are heritable. The data were taken from Michigan DHIA records from 1957 through 1962. Only daughters resulting from artificial insemination by sires in the Michigan Animal Breeders Co—op., American Breeders Service, and Curtiss Breeding Service were included. Over 1,300 Guernsey cows out of 42 sires and 7,800 Holstein cows out of 266 sires were used. To compare fairly bulls which were not contemporary, the analysis was within lactations and by numbers of tested daughters per sire. Clearly, cows in first lactation were removed at a dis- prOportionate rate among sires regardless of the number of tested daughters per sire. Arthur D. Dayton Heritability of reasons for diSposal during the first lactation only ranged from 0 to .18 for Specific reasons and was .24 for total removals in the Holsteins. When the heritabilities were adjusted for average incidences by transformation to the probit scale, heritability ranged from 0 to .67 for Specific reasons and was .62 for total removals. The Guernsey data were similar: however, herit- ability for total removals was unreasonably large, perhaps due to the small number available. DIFFERENTIAL REMOVAL OF DAUGHTERS AMONG AI SIRES BY Arthur D. Dayton A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER.OF SCIENCE Department of Dairy 1966 ACKNOWLEDGMENTS The author is grateful to Dr. Lon D. McGilliard for his interest and advice in handling the problem and for his critical reading of this manuscript. The financial assistance of Michigan State University in the form of a Graduate Research Assistantship is gratefully acknowledged. To his wife, LaVonia, the author is eSpecially grate- ful for her encouragement, understanding, patience, and sacrifice. ii TABLE OF CONTENTS Page ACKDIO'NIJEDGEI‘IEquS . o o o o o o o o o o o o o o o o o 0 ii LIST OF TABLES o o o o o o o o o o o o o o o o o o o 0 iv IMRODUCTION o o o o o o o o o o o o o o o o o o o o o l REVIW OF LITERJKTURE o o o o o o o o o o o o o o o o o 4 Average age of cows . . . . . . . . . . . . . 4 Reruoval rates 0 o o o o o o o o o o o o o o o 7 Heritabilities o o o o o o o o o o o o o o o o 13 SO URCE OF DATA 0 O O O O O O O O O O O O O O O O O O O l 5 Di sposals O O O O O C O O O O O O O O O O O I 15 I‘Ion-di Spos a]- S O O O O O O O O O O O C O O O O l 8 METHODS AND RESULTS . . . . . . . . . . . . . . . . . 19 Heritabilities of reasons for diSposal . . . . 29 COEECLUSIOEES o o o o o o o o o o o o o o o o o o o o o 36 Reasons for diSposal . . . . . . . . . . . . . 36 Removal rates between sires . . . . . . . . . 38 Heritabilities of reasons for diSposal . . . . 39 s m'fi/IA RY O O O C O O O O O O O O O O O O O O O C O O O 4 2 LITERATURE CITED 0 O O O O O O O O O O O O O O O O O O 43 iii LIST OF TABLES Table l. 2. 10. ll. 12. 13, Annual Rates of Removal Reported in the Literature . . . . . . . . . . . . . . . . . . . . Causes and Frequencies of Removals (Per cent of Total Removals) Reported in the Literature . . . . Distribution of Holstein and Guernsey AI Daughters by Reason for Removal . . . . . . . . . Distribution of Michigan DHIA Cows by Reason for Ranoval O O O O O O O O O O O O O O O O O O I O 0 Number of Holstein DiSpOSalS and Non-diSposals by Reasons Within Age-Lactation . . . . . . . . . Number of Guernsey DiSposals and Non-diSposals by Reasons Within Age-Lactation . . . . . . . . . Holstein DiSposals and Non-disposals for Number of Tested Daughters per Sire Within Age-Lactation. Guernsey DiSposals and Non-diSposals for Number of Tested Daughters per Sire Within Age- LaCtatiOn o o o o o o o o o o o o o o o o o o o 0 .Average Production, Age at Calving in Months, and Days in Milk During Terminal Records of HOlStej-ns O O O O O 0 O O O O O O O O O O O O O 0 Average Production, Age at Calving in Months, and Days in Milk During Terminal Records of Guernseys . . . . . . . . . . . . . . . . . . . . Average Production of Holstein DiSposals and Non-disposals Within.Age-Lactation . . . . . . . . Average Production of Guernsey DiSpOSalS and Non-diSposals Within.Age-Lactation . . . . . . . . Analysis of Variance of DiSpOSalS in First Lactation (DiSposals for All Reasons) Among AI Daughters . O O O O O O O O I O O O O O O O O O 0 iv Page 10 16 17 21 21 24 25 27 28 31 32 33 LIST OF TABLES (Continued .......) Table A L Page 14. Heritability Estimates of Reasons for DiSposal Before and After Transformation to the Probit scale 0 O O O O O O C O O O O O O O C O O O O O O O 35 INTRODUCTION Many dairymen today depend for their breeding programs either wholly or in part on sires from the different Artifi- cial Insemination (AI) units to sire the next generation of milking cows. Fifty-five per cent of the milk cows in Michigan, for example, were bred to AI dairy bulls in 1964 (Dairy Herd Improvement Letter 1965a). The dairymen select the bulls to be used in their herds by such criteria as the sire's daughters' production and type or his rate of conception. The initial summary of a sire used in AI service is usually made from the first records of 305-day lactations of his daughters. Biased evaluations can result when sires have diSproportionate numbers of daughters removed from lactation prior to completing a record and when these incomplete terminal records are not included in their summaries or are included in the summary but are extended to 305 days with ratios or regression coefficients for normal records in progress rather than with values developed for incomplete terminal records, Aulerich (1965). The rank of a bull having a large prOportion of his daughters removed as compared with a bull with few daughters discarded might not then be accurate if any differ- ence in production is associated with the difference in removals. 2 The current policy of the United States Department of Agriculture (USDA) for sire evaluation is to project or ex- tend incomplete records to a 305-day basis by ratios apprOpri- ate for normal records in progress. Incomplete records are those reported to the USDA with conditions affecting records (CAR) codes 2, 3, and 8 with days in milk less than 305 and more than 14. Records reported with CAR codes of 0, 4, 5, 6, and 7 are complete, are not projected, and are used in sire evaluations if the days in milk range from 180 to 305 days. They are not used if days in milk are less than 180 days. Records reported with CAR codes of l and 9 are not used for sire evaluations (Dairy Herd Improvement Letter 1965b). CAR codes used by USDA. Code Interpretation 0 Dry or 305—day record with no other conditions affecting it. 1 Estimated (Incomplete or missing first part of lactation). 2 Sold presumably for dairy purposes. 3 Died or sold for beef. 4 Injury 5 Mastitis. 6 Ketosis. 7 Other sickness. 8 Record terminated by abortion. 9 Nurse cow. From every dairy herd a number of milking cows are removed each year for various reasons. Little information is available in the literature to indicate whether these causes are genetically influenced or are caused mainly by environ- mental and managerial factors. 3 The purposes of this study were to determine if rates of removal of daughters at young ages among AI sires are equal -— do some sires have a larger per cent of their daughters leaving the herd than others, to determine the distribution of removals among various reasons, and to measure the extent individual differences in reasons for diSposal are heritable. REVIEW OF LITERATURE Literature concerning removals of cows contains a con- siderable amount of information on age, production, per cent removals, and causes for diSposal. This information has been collected in large part by survey. Average age of cows The average age of cows and the average length of time cows stay in the herd indicate what prOportion of the cows leave the herd each year. Becker and Arnold (1954) found cows raised at home remained in 14 Florida herds 4.7 years after attaining producing age; whereas, in 101 herds that purchased their replacements, the average was 3.9 years. In a similar study of 138 Jersey and 174 Holstein cows, Seath 22 El. (1943) observed 3 years 8.3 months and 3 years 9.3 months, reSpectively for Jersey and Holstein cows, as the actual time Spent in the Louisiana milking herd. O'Connor and Hodges (1963) estimated the average herd life in Private Milk Record herds from first calving to diSposal was between 3.5 and 4.5 years, and the average life Span from first calving to death was between 5 and 6 years. The average milk producing life of Michigan DHIA cattle from 1931—39 was 3.7 to 4.0 years (Baltzer, 1940). Rendel and Robertson (1950) put four lactations as the average productive life and stated further that an increase in 4 5 productive life by one lactation would raise the mean produc- tion of the herd by less than one per cent. Horton g3 a1. (1960), reporting on 894 animals of four dairy breeds from the Arkansas Agricultural Experiment Station herd, found an average number of lactations per cow of 3.4 for Holsteins, 3.3 for Jerseys, 3.0 for Guernseys, and 2.2 for Brown Swiss. Seath 23 a1. (1943) found the average number of freshenings for Jersey and Holstein cows was 3.78. One concept of average age, as pointed out by Slobodkin (1962), is that of the age attained by a median indi- vidual. He eXplains the median age by imagining a group of 100 animals of a particular Species born at the time instant t . There will be a time when the 50th individual has just 0 died. If this time is referred to as t the median age o+md' of the animals or the median life eXpectancy at birth is equal to the interval (t -to). This method has a tendency to o+md make the median life expectancy short in Species with high mortality rates in the young. Slobodkin eXplains another concept of average life called the mean life eXpectancy. This is the number of years of life that will be lived by an average animal in a group. Defining the number of individuals born alive at the time in- stant to as 10 and defining the number of these individuals alive at a subsequent instant tX as 1x' the graph of 1x against time can be constructed, Figure l. The graph shows four basic types of survivorships. Mean life expectancy is 6 calculated for any given age by computing the area under the survivorship curve for all subsequent ages and dividing by 1x' 100 1 Survivors Iime Figure 1. The four simple types of survivorship curves. In type I, mortality is concentrated in the old animals. Type II is characterized by a constant number of deaths per unit of time. Type III is found when the risk of death is constant with age. Type IV has mortality concentrated at the young stages. Cannon and Hansen (1939) using information from the Iowa Cow Testing Association during 1927-28 and 1930-36 cal- culated the life expectancies for 147,596 dairy cows. These life expectancies were compared with the life histories of cows that had passed through the Iowa State College herd. In most cases the life expectancies for the two groups were similar. In the association herds a cow freshening at an age less than 2 years was considered 2-3 years old; whereas, in 7 the college herd uncalved two-year-olds were included, but yearlings that had freshened were not used in the calculations. For every 100 cows in the 2-3 year-old group, 72.3 reached the 3-4 year-old group in the cow testing association herds, but in the college herd 77.1 of the 2-3 year-olds reached the 3-4 year-old group. Cannon and Hansen attributed this differ- ence to more removals for low production in the association lierds than in the college herd. The average ages of the cows 111 the association and college herds were about 4.7 and 4.2 ynears, reSpectively. Even though drought conditions and the fussiness depression were confounded with years, the yearly differences for rate and age of removals were small with the exception for 1927-28. Using herd book records of four dairy breeds in Great Eiritain (Shorthorn, Ayrshire, Jersey, and British Friesian), Smith and Robison (1950) gave the average age of cows at mNma . OZ aoma v-i (DMOOV'MN \Or-i VOMNOKDF ehqcv OO mbammom r-l v-Ifi' & momma mm mm mm mmm Nma mma Nv Nma mmv hmhm NNm vmm aNN moma amma maoo hmha oOZ owma mumma mm 00 mom mva Oha Nv mma mmv mmmm omN mmm mam mwaa hmma mhwm mmma OOZ mmma use» vawmmfi‘ o O omwmmv - a H ms oaNmom O 0 HM raw 3Q hamm vmm mm mm mg mp ha ow mmwa hma mmm mad boo Nmm mmme, mvaa OOZ mmma .mamSOEmH amuou MO “Emu Hmm X momh mwa ON mm mm aw oa vv toaoommm 0 Va mmoa a oma .a Noa m mwm 0 50% m mow mmvm hmm X .oz bmma we me mm mm N¢ aw 0v mm mm mm mm vm mm mm am om moon am>oemm now cOmmmm m9 msoo «Hmn comanoaz mo COHDSQanumaQ e magma amuoe czocxco mcac0maom was vac ucmcaoua HMOam mumzcnmm maemcoumod Hm>wm Maaz Uman was 6H0 mueeeumum umxaes eumm .mC—H mmcmm mapaummz husflcH Hmoemsnm c0auoscoum 30A mmmomusm whamn Gaom c0mmmm 18 gives the number and per cent removals for each reason for all Michigan DHIA cows, all breeds, for 1957 through 1963. Non-disEosals Records for non-diSposals were by all daughters of the AI sires of diSposals that completed one or more lactations through 1962 and included any complete records for cows sub- sequently removed. Complete records of milk and fat were corrected for age and adjusted to twice-a-day milking, but complete records less than 305 days were not adjusted for ““‘ ""‘"““- “ “’“~“ " ’“T'izw- M length regardless of their duration. The Holstein sires had 1* 30,308 complete records and the Guernsey sires had 2,763 com- pleted records through 1962. METHODS AND RE SULT S The first problem in analyzing removal rates of daughters of AI sires was to find some method of comparing non-contemporary bulls since daughters of varying ages dur- ing the 5 years represented bulls from a much larger Span of time. Since the literature indicates a larger rate of removal during the first lactation than during later lacta— tions, the data were grouped by parity into three categories. Group I was first lactation: group II, second and third lactation: and group IIIwas fourth or later lactation. Sires were compared on the prOportion of their daughters that left the tested herds during certain lactations of their daughters regardless of when the bulls were in service. The cows were classified according to their lactation number when it was available or by age if no number of lactation was indicated. Fritz (1960) found less than five per cent of the first lactations of Michigan DHIA cows started after 36 months of age at calving and less than three per cent of the third lactations began after 62 months of age at calving. Therefore, all records without lactation number were designated first lactations if the cow was less than 36 months of age at calving, second or third lactations if the cow was 36-62 months of age at calving, and all older 19 I lllllll‘ll Hz]! iii if lit illl'll‘ l l‘l ll.“ I'll, Ill I'll I!!! I‘ll-III- 20 cows were considered to have four or more lactations. Tables 5 and 6 give the number and per cent of total removals by reason within age-lactation for Holsteins and Guernseys. A cow with a terminated record in age-lactation II (second and third lactation) could have a completed record in age-lactation I. She was a non-diSposal in the age- lactation I but a diSposal in age-lactation II. The analysis of disprOportionate removal rates of daughters of sires was done by Chi Square goodness of fit. The question was whether the distribution of terminated and complete records was the same for all sires or whether selec- tion was much stronger and removals more frequent for Specific reasons among daughters of some sires. Since the test is of independence or agreement between eXpected and observed frequencies and not total numbers in each cell, the hypothetical frequency or the eXpected value for each cell was computed from the marginal totals in the correSponding rows and columns: Expected cell value = (frow total) .(leumn total) N N = total number of observations. In Chi Square goodness of fit the number eXpected in any cell should be not less than five for the probability dis- tribution to be reasonably accurate. In cases where sires had only a few tested daughters, the frequencies eXpected in many of the removal classes were quite likely to be very low 21 man mama awn maMmommanIcoz scaumuomqlmmd casuaz c0mmmm >9 ovvh momma moooa mammommaolcoz GOHDMDUMAImm4 casuaz mCOmmmm >.N oa m.v mm N.® em amen h.m wma o.m woa h.m mma amen ram .4 o N OCD Hf“ h.m oma H. m N. v mm ©.ma gm o.aa mm m.m mm mm ¢.Na mmN 5.0a mmm m.h ona hm O‘l‘ 030 m.m m mm o m.N O m w. o.a N m O.a m.a v N mm mm o.m mm ®.v cm m.N oa mm UaOm .mam>OEmn amuou m0 pcmo Hmm X m.m hm v.m we m.m mm mm m.mm 0am m.oo mam m.hm omm am m.© hm N.m mm a.¢a mm Om X HHH R HH & H c0aumuoqumm4 mamwommaplcoz cam mammomman mmmcumso Mo Hmnfisz m.m ow ®.N mm m.m em on o magma ©.a m.a mm mm o.m m.a ooa mv N.N ®.a mv mm mm «m b.0a aow mom mmm N.¢ am mm paom h.ma mom o.aa aom Nooa mam mm a.Nm Nhh a.hv omma 0.0v 5mm am Moo ama .mam>oewu amuou no name new R a HHH m.oa X owm a.ma mmm om H H HBQ ceaumuomqlmmd an mammommanlcoz cam madmomman camumaom m0 nonfidz m manme 22 if all sires were analyzed in one group without regarding the number of daughters per sire. In order to prevent expected values of less than five in many of the removal classes, sires were classified within age-lactation group according to the number of tested daughters. The first group was sires with 11-20 tested daughters, group two was 21-49 tested daughters per sire, and the last group was sires with 50 or more tested daughters. In each group there were as many different classes of removal as possible with at least five eXpected in each cell. In cases where eXpected values were less than five, observations in two or more classes of removal were pooled into one class. If, for example, the eXpected frequencies of removal for physical injury and mastitis each were less than five, the observations were pooled into one class. Tables 7 and 8 give the number of tested daughters per sire within age-lactation and the Chi Square values for diSprOportionate removals between the sires. Table 7 indi- cates clearly in every case for Holsteins except for sires with 11—20 tested daughters in age-lactation II and sires with 11-20 tested daughters in age-lactation III, a diSpro- portionate rate of removal between sires. The probability was less than .05 for the former group: the latter group showed no statistical difference in removal rates among the 37 sires represented. Table 8 reveals similar results for Guernseys with significant Chi Square values at the probability of .01 with 23 the exception of three groups where no statistical differ- ences in removal rate were observed. The three groups were sires with 11-20 tested daughters in age-lactation I and III and sires with 50 or more tested daughters in age-lactation III. The trend in AI, especially in young sire programs, is to put a great deal of emphasis in selection on the sire summary based on first lactations of the sire's daughters. An inaccurate evaluation can result when sires have diSpro- portionate numbers of daughters removed from lactation prior to completing a record, the incomplete records are not in- cluded in their summaries, and the difference in removals is associated with any difference in production. The diSprOpor- tionate removals among sires is shown clearly for Holsteins by the highly significant Chi Square values for every group within age-lactation I regardless of how many tested daughters a sire had. The Guernsey sires with 21-49 tested daughters and more than 49 tested daughters differed in removal rates during the first lactation. In Spite of the different rates of removal for various sires, bulls could be compared fairly if the removals were not related to production or if all incomplete records were extended to a 305-day basis with the proper extension fac- tors and were included in the sire summaries. The average milk and fat production, age in months at calving, and days in milk during terminal records are shown for the two breeds in Tables 9 and 10. The difference 24 .maasn assoa>aoca How om>némn COHDHOQOHQ ca mmcmm a mo. m a a0. m st omm ssmmm mw.lmo. mmmh mmmm mmha an m? A HHH awa stomm mm.lmo. mhva mmaa Nmm me mwlam HHH Nb mmm mm.loa. omm mam wva hm omlaa HHH mvm ssamaa mo.l¢o. omnma wmaaa wmmm on va HH mma «£50m mm.lma. mmma mmaa wmm ow mvlam HH mu smoa wo.INN. hmm mow wma ow omlaa HH Nha stvmo Nefuaa. onom omom vmma 3 may A H om stmma vm.lma. Nmma mmaa mha aw mVIaN H om stmma mmolmo. mam mny mva aw ONIaa H muam mamm nmm.smo Iommao matmom cmpmmp Icoc twat mo mmuam mo .u.o mx ammcmm amuoe mo .02 Honesz mo .02 amnesz cOHDMDumalmm4 aceumuomaumms senses muam mom mumuflmsmn pounce no quEsz How maMmomwaolcoz cam mammOQman samumaom h maQMB I lull. I'll: IIIIII {I If I I ll]!!! .1"! [fl [II-ll" [I )|I i " l‘l {III I." I'll llllfll fl '7 Ill’lll l I‘ll (I'll ‘ :1 25 mm mm ma mm wm ma mm ma .maadQ assoa>aoca Mom om>oewu ceapuomoum Ca mmcmm m.m Nm.lmN. fiON ava tsflomh amolba. 00v th womN mm.lmo. mam mva «taomfl ¢MolNa. mmo vvm ttbomh mh.lhoo 0am Nmm kthowh moolmo. mom hma «SN.Nm mo.loa. avm hem ssmomoa abolmo. ®N¢ mNN woma awolhoo mva vaa mm mva ah vwa mNa mm mm hma mm mammommao mammom Icon Imao no N mmcmm amuoa mo .02 HmnEdZ ceapmuomqlmmd Casuaz va ma ha ma ma oa m0ham. no .02 EVA mvbam omlaa me A mwlam omlaa mv.A mwlam omlaa mnam umm .msmo omummu ao. HHH HHH HHH HH HH HH a m is HmQESZ ceaumpomqlmm4 sham Hod mucusmsmn pounce mo Honesz How maMmommaclcoz one maMmommaQ wmwcumsm m maQMB 26 between production of the non-diSposals and disposals is shown for each reason for diSposal. The differences in production support the accuracy of the reason given for dis— posal in cases of low production. Some of the other reasons may indicate cows are removed because of mastitis, sterility, or physical injury only when production has been affected. For example, if a cow had severe mastitis and _' sq.- u)‘rr.04 v FA was a good producer, the dairyman could justify drying up one quarter and keeping the cow in the herd; whereas, an average or below average producer would be removed. r. The tenninal records in the table were extended to g. 305 days with factors developed by Lamb and McGilliard (1960). The differences in production between non-diSposals and diSposals would be larger had the factors develOped by Aulerich (1965) been used to extend the terminal records to 305 days. Aulerich found the lactation curves of cows re- moved involuntarily from the herd--cows which could not have been retained even if the owner wanted-~were similar to those of cows completing their records. However, cows removed involuntarily from the herd generally initiated their lactations with a slightly lower yield than the cows completing their records. The projection factors used for extending an incomplete non-terminal lactation can be em- ployed to estimate 305-day milk yields of cows involuntar- ily removed from the herd during lactation: however, they may underestimate the 305-day yield due to the slightly faster 27 Na am mm vv mm mm 00a “mm amv amo th mom omal mm dam mom Nmmm oval xaaz AmacmommaolmaMmommaolsozv .l- | O .I- O In mw o+mm¢ m®a+oomma m v+mma m a+mm mmv DD 0 mamas phenomome H.6uema m Named mea mam 6H0 suave amummoma m Hammm m.uem mmm sueaenmum eueme mmeuyomaa m.musea e.awoe mam umxees 666m enemy mmauaemma e mumee 6 Hubs ewe .m.e Haueme vemumeema «.mumee H.~Hme mad mmemm vummv mmauemmea e mumme m “me mes mapepmms eumev nonmeoma e.musme m.Hem emm menace Hmuemmcm muckm Neummmoa ~.Hueme e.uee momm eoHuUSUOMe Sou ensue emusmmme m.mumee s.ume mew mmmoeesm mused maMmommao MaaE no acmommao use xaaz ca mmma mmd Honesz you c0mmmm namumaom m0 monoomm amcHEHmB moanso xaaz ea mama ocm .msucoz ca mca>amo mo mmdt.coHuoso0Hm mmmum>4 m manta mm mm mm 0% mm aoa NI Dom AmammommaolmammommaolcozV mmm aom me aVI wvm mow hvm mmma NNI Xaaz oaHoav mmHaov mmuvow haummm wwuvmw maumaw aaHmmm aauvow m Hmmm oaHavw Q mm 2 Dom WE 1;")... «(II--1llilll: 11V... \II-Ihh hmmflomom mmouhmwm voawmbam mwmumhom mmonmom ommuwhbm ommHmamm aomHahaw mo “moon aomHowom cmmE Xaaeg mwmmcumso mo oncomm emceeume madman sees ca mate 6:6 v.0aHhaa mcaNHNON doomHvNN a.oaH>ma m.®mHmaN a.mmHooa m.mHNoa m.eueea b.NHmba N mHNwa cmmz XaaE ca whom oa maQMB m mfihw a mHaoa o.mHoo o.mHom h.mHmN m.mHom a.mHmm m.mHom m. Hmv m.NHmv cmmz wmd mammommap mo HmQESZ hm Spawn ma mmm cao mme sueeeumum om umxaaE whom o .m.B am mmcmm we mapepmms maa musth amuammnm mmh COHDUSGOHQ 30A mma mmmomusm muamm ammommao How Cowmmm .m£DCOE Ca mca>amo um mm¢ .c0auosooum mmmum>¢ 29 decline of the lactations terminated involuntarily. The shapes of the curves of voluntarily incomplete terminal lactations--lactations terminated by the owner willfully removing the cow-~were unlike those of the cows completing records. The cows voluntarily removed from the herd initi- ated their lactations at a lower milk yield and declined at a faster rate during lactation than cows completing 305 days. Aulerich develOped separate factors for extending voluntarily tenminal records to 305 days. The two breeds differed little in average days in milk during terminal records or in mean age at calving for the various reasons for removal. The same information for milk and fat production in the two breeds but within age—lactation is given in Tables 11 and 12. The differences in production between the non- diSposals and diSposals as well as the differences in produc- tion between non-diSposals and all cows are indicated. The difference in production between the selected cows (non- diSposals) and all cows was nearly twice as large for Guernseys as for Holsteins. Heritabilities of reasons for diSposal Heritabilities of reasons for disposal during the first lactation only were calculated by analysis of variance within and between sires. Estimates of heritability from cows in first lactation should be more meaningful than from any other age-lactation group or combination of all the data because of the lack of previous selection for production. W. .nr in? 30 Only sires with eleven or more tested daughters were used. The data are an all-or-none kind in that each animal was either removed or kept in the herd during the 305 day lactation period. The results for total disposals for Holsteins and Guernseys are presented in Table 13. The components of variation expected in the mean squares are designated as E and S. E is the variance between paternal half sisters and A.XleL.fl*—1 : S is the extra variance between means of groups of paternal sap .‘ “'“Nl sisters. The heritabilities of differences in fates of indi- 11'1: ._ viduals in Table 14 were calculated by 4S/(S+E). For ex- ample, the heritability of total removals in Holsteins was .276 where S = .0085 and E = .1289. Most of the applications of heritability have been to characters continuously distributed on the phenotypic scale. However, the concept of heritability extends equally to traits which may be eXpressed phenotypically on an all-or- none or discontinuous basis. To treat a character with an all-or-none phenotypic expression as dependent on an underlying continuous variable seems reasonable, Lush (1948). The value of each underlying variate in a particular individual would depend on both genetic and non-genetic factors, and the phenotypic ex- pression of the character would require that some threshold be exceeded. Heritabilities measured from the all-or-none data could provide a basis for computation of genetic gains 31 aa mom mm ohma Na mvm we mom ma wvm mo omoa m aNN om amma Dom MaaE Dom MaaE Amsou aam Imammommaolcozv Awammommaolmammommaolcozv mow Ohw mav owv mmv oav mow ohm mav mow th NNV Dom mmwma amhma movaa haama momma oomaa wmoma Nooma Namaa mmoNa gamma momaa Xaaz hvamm mOMOm mmmh fivmm O¢Vh wowm mmaoa momma mmmm omaNa moooa beam quESZ mzoo aad mammOQmaoIcoz mammommao msou aa¢ mammommaolcoz mammomman msou aad mammommaolsoz mammomman mzoo aa4 oatmommaolcoz mammommao Hmuoe HHH HH COHDMDUMAIomd coHDMDUMAImmé casuaz maMmOdmaoIcoz pom mammommaa saoumaom mo coHuusoonm ommum>4 aa magma mm mm? 2 3 mm one am mmv NN ovv pom xaaz Anson aam Imammommaolcozv Amammommaoumammommaoncozv scaumuUMQImmd casuaz maMmommaoIcoz com mammomman mmmcumso mo mo Nb mo Dom vmma omma omva o0ma Maaz oav mmv Ohm how Nmm mom mav omv vom hmw mom mom Dom ooom maom vooh ommm ooom amwh mndm mmmm mosh mmho moam momh Xaaz Na maQMB mwov mohN mmma haaa man omm omha moma hmm whaa amh hmm umpesz msoo aad mammommaolcoz maMmommaQ mzoo aav. mammommaolcoz maMmommaD mzoo aad madmommaolcoz maMmommaQ msoo aaq maMmommaolsoz mammomman sauce HHH HH COHDMDUmaIwmd Mo c0auosooum mmmum>¢ 33 Table 13 Analysis of Variance of Disposals in First Lactation (Disposals for all Reasons) Among AI Daughters Holsteins Source of d f Sum of Mean Components Variation ' ' Squares Squares of Mean Squares Total 11617 1593.31 f Between 125 111.68 .8934 E+90.04S i Sires 5 Within 11492 1481.63 .1289 E g Sires ) Components: B = .1289 S = .0085 E Guernseys Total 914 183.93 Between 27 82.61 3.0597 E+32.03S Sires Within 887 101.32 .1142 E Sires Components: E = .1142 S = .0920 expected from selection. However, in some instances in- accuracies could result in the estimates since the genetic variance which may be additive for the underlying variate could lose this prOperty on the phenotypic scale. Since the limiting genotypic values for the all-or-none trait are 0 and 1, a given gene substitution is unlikely to have the same effect near these limits as in the middle of the range. 34 Lush egpgl. (1948) held that probit transformation avoids some of the objections of the coarse phenotypic scale. The transformation is based on the concept of an underlying variate with a normal environmental distribution whose vari- ance is independent of the genotypic level. The heritability on the probit scale is independent of the threshold value; whereas, on the phenotypic scale, heritability varies approx- imately in proportion to zZ/pq where g is the ordinate of a unit standard normal curve cutting off an area equal to p. p equals the fraction of the pOpulation removed from the herd 15”" and q equals (l-p). Heritability on the phenotypic scale would be low for values of 2 near zero or unity and relative- ly high for intermediate values. The heritabilities estimated from the data were trans- formed to correct for average incidences of the removals. The transformations were done by multiplying the heritabil- ities actually observed by pq/zz. Table 14 gives the heritabilities of reasons for dis- posal for each breed in original units of removal and re- tained and in the transformed units of the probit scales 35 .wmumEaumm muaaanmuaumfl Ca ooUSauca msou aspou Bonn om>oemn coapomum m bem.m vmh.a owm. mmo. ohm. O O maO. wma. NaO. O O maO. O O aOm. maO. #00. vom. ONO. mah.m moo. mOO. woo. woo. mvo. NOO. mOO. mod. aaO. bmm. ONO. mOO. mom. maO. own. No0. bNO. mmo. ONO. Nov. mna. ama. mmv. mwa. aba.a mom. ovO. on. amO. ooEuoumcmHB mumEaumm mmmcnoso am mamom panoum ms» cmeuommcmue mDMEapmm samumaom a mammommao amuoe mnpmma xueaauoum umxaaE oumm .m.B mmcmm mauaummz hushCa amoammcm GOHDUDUOHQ Boa mmmomHSQ whamn How c0mmom 0p COHDMDMOQQOHB umuw¢ one oncomm acmomman How wQOmmom no mmDMEaumm muaaanmuaumm va maQMB CONCLUSIONS Reasons for disposal The literature for the most part indicates that dairy cows are removed from herds primarily for low produc- tion and reproductive difficulties. This study supports low production and other voluntary diSposals as the predominant reason cows sired by AI bulls leave herds and indicates in- voluntary diSposals are considerably less frequent. However, Meek (1962) found in the Iowa institution herds involuntary reasons were a larger percentage of total removals when calves and heifers were included because calf and heifer losses are usually forced diSposals. Death of animals from first calving onward was 8% of the total but death as a reason for diSposal when calves and heifers were included was 13.5% in the Iowa data (Meek, 1962). This study indi- cates about 5%.of the total diSposals are caused by death. The other reasons for involuntary losses in this study were sterility, bangs, and tuberculosis. These three reasons accounted for 14-17% of all disposals, with steril- ity accounting for 11%. Sterility or reduced fertility is effectively natural selection. Therefore, little progress could be made to reduce involuntary losses by non-managerial practices. 37 Physical injuries, mastitis, and hard milker are in- voluntary reasons for diSposal. Severe injuries, acute mastitis or a chronic hard milker may force the dairyman to remove the cow in order to conserve labor even though the cow's production would warrant her remaining in the herd. This study shows 8.7%.of the Guernseys and 12.2%.of the Holsteins leaving the herds were removed for physical in- juries—~more than reported in the literature. However, this category could include animals that are sold because of bad feet and legs or undesirable conformation since no other category is available for such cows. Removal for mastitis was the fourth most frequent involuntary reason for diSposal accounting for nearly 10% of the Holsteins and 5% of the Guernseys. This category has much economic importance to the dairyman because mastitis causes a loss of milk production for part of the lactation period and possibly permanent damage to udder tissue. Since only cows with severe cases of mastitis would be removed from the herd and recorded as removed for this reason, that some cows have mastitis and are not removed and others are removed for low production-—the direct result of mastitis-- suggest mastitis probably occurs more frequently in the herd than the data on diSposal indicate. "Sold for dairy purposes" includes cows that leave one herd but join another herd: that is, they still remain in the dairy pOpulation. Animals sold in consignment sales or .‘7. 1.! ;' Giza- magnum " i {IT—Vi”..- ~.; 38 breed promotion sales would be included in this group. This study shows 9-11% of the diSposals were sold for dairy pur- poses. "Sold for low production" is by far the largest single classification given to cow removals in this study and also the most frequent reason reported in the literature. However, this category may be used as a catch-all for animals whose production has been reduced by some other condition such as mastitis, milk fever, or hard milking. Removal rates between sires The initial proof of an AI sire is based primarily on the production of his daughters' first 305-day records and can be biased if a large proportion of his daughters—- those producing less milk--are removed prior to completing a record. If AI sire summaries are determined from completed records only or if incomplete records are extended with nor- mal factors and included, the ranking of bulls could be in- accurate when diSproportionate numbers of daughters are removed. Aulerich (1965) found involuntarily tenninated records are probably slightly but not seriously underestimated when extended to 305 days with normal factors. However, cows voluntarily removed should have their 305—day estimate deter- mined with special factors. Van Vleck (1962) concluded from a study of New York DHIA first lactation records of AI Holstein cows that no bias in sire evaluation results when incomplete records are ex- cluded from sire proofs. The New York data included: 39 (l) A relatively small fraction (5-7%) of first lactation records were incomplete and (2) the magnitude of the sire summary and the fraction of incomplete records were independ- ent. In a later study, Van Vleck and Henderson (1963) used records of New York AI Holstein daughters to measure the effect of removal after the first lactation on sire evalu- ation based on both first and second lactation records. Van Vleck and Henderson concluded that a differential removal rate after the first lactation did not bias the sire evalu- ation based on first and second lactation records only. On the other hand this study indicated 18% of the AI Holstein cows and 34% of the AI Guernsey cows were removed during the first lactation and Aulerich (1965) estimated that 16% of the Holstein cows in first lactation in Michigan were removed before completing a lactation. These larger rates of removal would increase the opportunity for dis- prOportionate removals among sires to affect their test. This study indicated, eSpecially clearly in first lactation cows, a disprOportionate rate of removal between sires for Holsteins and Guernseys Tables 7 and 8. Also un- equal rates of diSposal were between sires with daughters in their second and third lactations and in their fourth and later lactations. Heritabilities of reasons for diSposal Estimates of heritability indicate the extent to which individual differences in causes for diSposal are 40 transmitted. Heritabilities provide a basis for computation of genetic gains expected from selection. Meek (1962), using Iowa data, measured heritabilities by four different paternal sister analyses. In each of the models slightly different assumptions were made. In the first case all the diSposal records were treated as a single large population and the analysis was within and between sires. In the second analy- sis the population was first divided into herds and then in- to groups of paternal sisters within herds. The third anal— ysis was also hierarchical but the intraherd differences between years of diSposal of the cows were removed before computing the intrayear differences between groups of pater- nal sibs. This would be importantly different from the pre- ceding if some factors, other than sires, which varied from year to year had a marked effect on the reasons for diSposal. Another analysis was run using the same model but classifying the cows on year of birth rather than by year of disposal. In the last analysis the subclass numbers in the sire com- ponent were almost three times as large as in the previous one because the daughters of a sire were usually born within one or two years but their diSposal usually extended over many years. Meek reported a range in heritability from .03 for death to .35 for low production and concluded that analy- sis should be within herds but not within year of diSposal because the sire component is automatically inflated whenever the probability of diSposal varies according to age. 41 The heritabilities in Michigan ranged from 0 to .18 for Specific reasons for diSposal and the estimate for total removals was .24 in Holsteins, Table 14. When the heritabil- ities were adjusted for average incidences by transform- ation to the probit scale, the range was 0 to .67. Adjust- ment to probit scale was necessary to allow comparison be- tween the different reasons for diSposal since the per cent removals for each reason varied considerably. The Guernsey data were similar: however, the heritability estimate for total removals was 1.86 indicating a large sampling error in estimation, perhaps due to the small number available. Using the first lactation rather than all lactation records should have eliminated the effects of factors such as previous selection for production and different prob- abilities of disposal for different ages, which could bias the estimates of heritability. SUMMARY The most frequent reasons given for disposals of 7,839 Holstein cows out of 266 AI sires and 1,322 Guernsey cows out of 42 AI sires were low production, dairy purposes, physical injury, and sterility. Removals for low production and dairy purposes decreased with parity while losses from injury, sterility, and mastitis were more frequent in later lactations. DiSprOportionate removals among Sires were shown clearly for first lactation daughters and suggest a possible source of bias in sire evaluation if incomplete records are not included in the data used for the evaluation. Aulerich's (1965) results indicate bias caused by differences in pro— duction being associated with differences in removal can be eliminated if incomplete records are extended to 305 days with ratio factors develOped for incomplete terminal records and are included in the evaluation. Heritability estimates ranged from 0 to .18 for Specific reasons for diSposal and indicate that some of the variation in rates of total removals and in rates of removals given for specific reasons is genetically caused. 42 LITERATURE CITED Arnold, P. T. D., Becker, R. B. and Spurlock, A. H. 1949. Management of Dairy Cattle in Florida. Florida Agr. Expt. Sta. Bull. 464. Asdell, S. A. 1951. Variations in Amount of Culling From DHIA Herds. J. Dairy Sci., 34:529. .Aulerich, C. C. 1965. Differences in Extending Terminal and Non-terminal Incomplete Lactations to 305 'Days. Unpublished M.S. Thesis. Michigan State University Library, East Lansing. Baltzer, A. C. 1940. Causes for Cow Removals in Michigan Herds Under Test in Dairy Herd Improvement Associ- ations. Michigan Agr. Expt. Sta. Quart. Bull. 22: 147. Becker, R. 8., Arnold, P. T. D.'and Spurlock, A. H. 1954. Productive.Life Span of Dairy Cattle. Florida Agr. EXpt. Sta. Bull. 540. Cannon, C. Y. and Hansen, E. N. 1939. EXpectation of Life in Dairy Cows. J. Dairy Sci., 22:1025. Clark, C. H. 1952. A Survey of Dairy Herds Wastage in Queensland for the Years 1948-51. Dairy Sci. Abst., 14:498. Clark, C. H. and Paul, R. A. 1954. A Survey of Dairy Herds Wastage in Queensland for 1951—52. Dairy Sci. Abst., 16:185. Clark, C. H. 1958. A Survey of Herd Wastage in Queensland. Dairy Sci. Abst., 20:497. Dairy Herd Improvement Letter. 1965a. ARS—44-158, 41:no. 3. Dairy Herd Improvement Letter. 1965b. ARS-44—164, 41:no.6. Fritz, G. R. 1960. Environmental Influences on Regression Factors for Estimating 305—day Production from Part Lactations. J. Dairy Sci., 43:1108. 43 44 Guba, S. and Illes, A. 1959. Herd Book Cows Using Herd Book Data. Dairy Sci. Abst., 21:2707. Hoekstra, P. 1960. The Productive Life of Dutch Cows. I, II, III, IV. Dairy Sci. Abst., 22:1785. Horton, O. H., Stallcup, O. T. and Rakes, J. M. 1960. Dairy Cattle Losses and DiSposals. Arkansas Agr. EXpt. Sta. Bull. 623. Jeske, A. 1958. The Results of Milk Recording From 1946 to 1955 in German Democratic Republic. Dairy Sci. Abst., 20:767. Johansson, I. 1961. Genetic ASpects of Dairy Cattle Breed- ing. University of Illinois Press, Urbana. Johnson, L. A. and Thelen, A. J. 1959, 1960, 1961, 1962, 1963. Annual Summary. Michigan Dairy Herd Improve- ment Records. Michigan State University, East Lansing. Lamb, R. C. and McGilliard, L. D. 1960. Variables Affect- ing Ratio Factors for Estimating 305-day Production from Part Lactations. J. Dairy Sci., 43:519. Leali, L. 1956. Fertility of Cattle Under Veterinary Con- trol in the Milan Province. Dairy Sci. Abst., 18: 776. Lush, J. L. 1948. The Genetics of POpulations. Mimeo- graphed Notes. Iowa State University, Ames. Lush, J. L., Lamoreux,‘W. F. and Hazel, L. N. 1948. The Heritability of Resistance to Death in the Fowl. Poultry Sci., 27:375. Meek, A. M. 1962. Survivorship in Dairy Cattle. Unpub— lished Ph.D. Thesis. Iowa State University Library, Arnes. Mutovin, V. I. 1961. Control of Mastitis in Cows. Dairy Sci. Abst., 23:1724. Nai, D. D. 1958. Wastage and Culling in Dairy Herds in Lombardy Plains. Dairy Sci. Abst., 20:410. New Zealand Dairy Board. 1958. Herd Wastage. Dairy Sci. Abst., 20:1499. New Zealand Dairy Board. 1963. Farm Production Report and Summary of Board's Work. No. 39:70. 45 O'Bleness, G. V. and Van Vleck, L. D. 1962. Reasons for DiSposals of Dairy Cows from New Ybrk Herds. J. Dairy Sci., 45:1087. O'Connor, L. K. and Hodges, J. 1963. Wastage and Culling in Dairy Herds. Animal Prod., 5:165. Parker, J. B., Bayley, N. D., Fohram, M. H. and Plowman, R. D. 1960. Factors Influencing Dairy Cattle Longevity. J. Dairy Sci., 43:401. Rabold, K. 1958. Investigations on the Use, Length of Life and Causes of Wastage in Spotted Mountain Cows on a Small Farm in the Kraich District of North Baden. Dairy Sci. Abst., 20:996. Rendel, J. M. and Robertson, A. 1950. Some ASpects of Longevity in Dairy Cows. Empire J. EXpt. Agr., 18: 49. Seath, D. M. 1940. The Intensity and Kinds of Selection Actually Practiced in Dairy Herds. J. Dairy Sci., 23:931. Seath, D. M., Staples, C. H. and Nearham, E. W. 1943. A Study of Breeding Records in Dairy Herds. Louisiana Agr. Expt. Sta. Bull. 370. Slobodkin, L. B. 1962. Growth and Regulation of Animal POpulations. Holt, Rinehart and Winston, New York. Smith, A. D. B. and Robison, O. J. 1950. The Averages of Cows and Bulls in Six Breeds of Cattle. J. Agr. Sci., 21:136. Snick, G. and Kinit, G. 1964. Causes of Culling of Dairy Cows in Belgium. Dairy Sci. Abst., 26:611. Specht, L. W. and McGilliard, L. D. 1960. Rates of Improve- ment by Progeny Testing in Dairy Herds of Various Sizes. J. Dairy Sci., 43:63. Van Vleck, L. D. 1962. Effect of Incomplete Records on Sire Evaluation. J. Dairy Sci., 45:1511. Van Vleck, L. D. and Henderson, C. R. 1963. Bias in Sire Evaluation Due to Selection. J. Dairy Sci., 46:976. Withers, F. W. 1955. Wastage and Diseases in Dairy Herds. The Veterinary Record, 67:605. 46 Withers, F. W. 1957. Wastage and Disease Incidence in Dairy Herds. The Veterinary Record, 69:446. Withers, F. W. 1959. Surveys of Wastage and Diseases in Dairy Herds. The Veterinary Record, 71:864. Ziegenhagen, G. 1952. The Problem of Duration of Usefulness of Cattle in the Angeln Breed. Dairy Sci. Abst., 14:577. "TIT 11111113111)?“fiTjflTflifiiljljflifllflr