i _ , i : I : . ,_ “5;, . ‘,.:C. \’£L“ . 5 %~:’ 5 - 4.“ t“ .i \ v s : O . _ 3 —--\.‘ .m- -. 'onl ‘ c‘ 2“ B' \ "f‘ Ix: 5“: '- u 9 "J . r ‘t. I“... “.5. . - "‘Q‘ a. b We. 81" t O ‘IIV' cut-5.; Mm: E ,v 9.? .O..-o PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE L19! my. '7 I s » ,7 n ' A x ‘\ . I new to a“! - - c I , . . a. , . : e __ JLJ IV 7- ? C: % ll MSU Is An Affirmetlve Action/Equal Opportunity lmtltulon INTERACTIONS BETWEEN AI SIRES AND MANAGERIAL PRACTICES IN DAIRI HMS JOHN MILTON BURDICK II An Abstract Su‘mitted to the College of Agriculture Michigan State University of Agriculture en! Applied Science in parthl fulfilllmt of the requirenents for the degree or MASTER OF SCIENCE Department of Dairy 1959 APPI‘OVed 1071) b: 7n Qtflfiwwt ABSTRACT JOHN HILTQI BUEDICX Five environnental factcrs were studied to determine the existence and amount of interaction between these factors and milk. fat. and test of sires' daughters. The factors were based on lard intonation and included level of production. location. days dry. calving interval. and type of housing. Data were first records reported in Michigan Dairy Herd Improvemnt Association for AI daughters in tested herds with at least one yearly herd average reported. The analysis of the five herd classifications involved the followim data: (1) 8638 Holstein AI daughters of 192 AI sires in 1211 herds an! 1872 Guernsey AI daughters of 138 AI Gmrnsey bulls in 355 herds were divided into three level'sof graduation on the average annual production of their respective herds. (2) The 1211 Holstein herds were divided into six geographical locations of the state. (3) and (’4) 39’4 301315911! herds were divided into three levels based on the average number of days cows were dry and the average number of months per cow between calvims. These data included #081 A1 daughters from 172 sires. (5) 627 Holstein herds were divided into 3 groups on the type of housing for dairy cows. This classification included 5240 AI daughters of 186 AI bulls. ‘ The general component of interaction between herds and sires accounted for between zero and nine per cent of the variation in nilk ard fat production and test. When herds and herd-by—sire components were split into one of the five constitutive environments. there was no apparent interaction with sires. Therefore. the ranking of AI sires in herds on the basis of their daughters' performance will be indeperdent of am of the five envirorlental segments studied. Holstein AI sires are not equally represemed in all herds. INTERACTIONS BETWEEN AI SITES AND MANAGERIAL PRACTICES IN DAIRY HERDS JOHN MILTON BURDICK II A Thesis 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 (F SCIENCE Department of Dairy 1999 ACWGEMEN TS The writer wishes to express his sincere gratitude to Dr. L. D. McGilliard, Associate Professor of Dairy, for his constant guidance and expert advice so generously given at all times. to Dr. N. P. Ralston. Professor and former Head of Dairy, for making available the Research Assistantship that made this study possible. and to his wife Madlyn Lou Burdick for her patient and considerate support during all stages of the investigation. Grateful aclmavaedgement is also made of the help so freely given by R. P. Witte, D.H.I.A. Analyst. and A. J. Thelen. Tabulating Supervisor. with the D.H.I.A. records ard I.B.M. equipment used in the cornpiling arr! analysis of the data. iii TABLEOFCONTENTS PAGE INTRODIDTION.........................1 REVIEJOFIITmATURE.....................3 MATERIAISANDMETHCDS....................10 Levels of Product ion . . . . . 10 location e e e e e e e e e e e e e e e e e e e 11 Days Dry am Calving Inteflh; e e e e e e e e e e e h e 1.1 TYPB of Housing. 0 e e e e e e e e e e e e e e e e e e 13 lbthd of Awais e e e e e e e e e e e e e e e e e e 1"" RESULTSANDDISCUSSION....................16 Lovels of Production (Holstein) . l6 Levels of Production (Guernsey). . . . . . . . . . . . l9 Areas.........................23 DaysDryaniCalvingIntervals............ 25 TypeofHousing....................27 mNCIJIBIONS O O O O O O O O O O O O O O O O 0 O O O O O O O O 30 W O O O O O O 0 O O O O O O O I O O O O O O O O O O O O O 3 2 mm CITATION O O O 0 O O O O O O O O O O O O O O O O O 3 3 LIST OF TABLES TABLE PAGE 1. Size. Per Cent AI. an Average Production for Herds at Three Levels of PrOduCtione e e e e e e e e e e e e e e 11 2. Number of Herds and Range of Three Managerial Lavels for Days Dry and for C8171ng Interval e e e e e e e e e e e 13 3. Components of Variance for Levels of Production. Herds and Sires for HOlBteins e e e a e e e e e e e e e e e e 17 1+. Production by Holstein AI Daughters at Various Lavels of Praduction.0flfierds e e e e e e e e e e e e e e e e e e 19 5. Cmponents of Variance for Levels of Production , Herds. and Sires for Guarnseys e e e e e e e e e e e e e e e e 20 6. Praduction by Guernsey AI Daughters at Various Levels of PrOduction 0f Herds e e e e e e e e e e e e e e e e e e 21 7. Componmts of Variance for Areas, Herds. ard Sires. . . 22 8. Average by Areas of Records First Reported for AI Daughters e e e e a e e e e e e e e e e e e e e e e e e 23 9. Cmponents of Variance for Days Dry, Herds and Sizes. . 24 10. Average by Days Dry of Records First Repcrted for AI Daughters e e e e e e e e e o e e e e e e e e e e e e e 25 11. Components of Variance for Calving Intervals. Herds and Sires a e e e e e e e e e e e e e e e e e e e e e e e e 26 12. Average by Calving Intervals of Records First Reported forAIDaughterSeeeeeeeeeeeeoeeeeeoo27 13. Conponents of Variance for Types of Housing, Herds an! Sires e o e e e e e e e e e e e e e a e e e e e e e e e 28 14. Average by Type of Housing of Records First Reported for‘AI Daughters. e e e e e e e e e e e e e e e e e e e 29 INTRODUCTION Interaction between sires and herds may be principally a differ— ential response of a sires' daughters to different envirornents. In dairy cattle the interaction between sires and herds constitutes only a stall part, if am. of the total variation in milk and fat yield and test. The literature generally ascribes from none to seven per cent of the total variation to the interaction of sires with herds. Absence of interaction indicates that differences between groups of daughters by different sires remain the seine regardless of the envirorment in which they are compared. To the individual breeder of dairy cattle and to the committees for selection of sires for use in AI (artificial insemination) the negligible interaction has neant that different bulls need not be selected for specific envirornents. A snall general interaction between sires arr! herds does not mean. per se. that interactions between specific enviroruents and the gene- types of individuals living unier these environments may not be of some importance. Environments of herds are a canposite of many envirornents with the magerial practices of the dairynan being an integral part of these environments. The constitutive enviroments may be the primary enviroments with which sires' daughters interact. In the gemral sire by herd interaction. individually important interact ions between sires' daughters and the constitutive enviromnents of the horde may be swanped and cancelled out. This cancelling effect nay reduce the sire by herd interaction to mar zero even when interactions involving specific enviroments are important. The purpose of this study is to ascertain the existence and I size of interactions between sires and a few measurable constitutive environments . KEVIN OF IITRATURE Recent estimates by Pirchner an! Lush (1959) , Specht (1957) an! Mason an! Robertson (1956) ascribe to differences between herds 30 to #0 per cent of the total variation in milk and fat production. of this alount only a small fract ion is heritable. Pirchner and Lush (1959) indicate that 6.5 per cent of the differences in milk and fat production between herds within years are heritable. Other estimates from the literature place genetic differences between herds between zero and 12 per cent. Little work has been reported concerning the fraction of the total variation for milk and fat production dimctly due to sons seg- ment of the total herd environment. mm (1935) attributed 114+ per cent of the total variation in fat production in Iowa DHIA records to differences between herds in feeding mthods. More recent strxlies of feedirg methods do not indicate how much of the total variation in production may be accounted for by specific feeding and runagerial practices but do estimate the relative impor- tance of specific feeding and managerial practices. Bayley and Heizer (1952) conducted for 20 months a study of 96? grade and registered Holsteins in 1+? Wisconsin herds to determine the relationships bettteen five measures of environmnt or management and production of milk an! fat. They used the last capleted lactation record for all cows bred. born. raised. and still living in the herd. Records were standardized to 305 days. included only 21 milking and were uncorrected for age. Specific factors reported were condition before freshenirg, selection rating, TDN fed per 1,000 lb. of body weight. nutritive ratio, and size of herd. Ratings and systems of scoring were devised for influences not readily evaluated numerically. The selection rating had a range of from 1 to #5 points for the methods of selection of breeding stock in the herds. The four grades for condition at the time of calving ranged from excellent to poor with 85 per cent of the cows in the excellent and good grades and less than 1 per cent in the poor grade. TDN fed per 1,000 lb. body weight arr! nutritive ratios were based on rations fed during the winter months in the barn ard ranged from 12.4 to 27.0 and 5.2 to 9.2. respectively. The number of cows of milking age in the herds determined the size of herd which had a range of 12 to 165 cows. Multiple regressionwas used to estimate the importance of the specific factors. The regression of milk yield on the selection rating indicated that with each increase of five points in the selection rating, there was an average increase of 305 1b. in milk yield. No indication of curvilinearity was reported for regression of milk produc- tion on selection ratings. The effects of condition at time of fresh- ening were reported as average differences between the observed milk and fat production records ani the predicted milk and fat production as determined by the multiple regression of the quantitatively measured factors. This method showed an increase of 146 lb. of milk from "fair" to I'good" grades and an increase of 850 lb. from ”good" to I'excellent". For a daily increase of 1 lb. of TDN fed per 1,000 lb. body weight, there was an increase of 551 lb. of milk produced. A change in the nutritive ratio from 9.2 to 5.2 resulted in an increase in milk yield of 2,952 lb. Herd size had a negative effect on milk production with a decrease of 775 1b. milk for an increase in herd size from 20 cows to 49. The decline was 521 lb. for an increase in herd size from 50 to 79 cows with only slight decreases in milk production for herds containing more than 80 cows. In a continuation.of the previous study. Starkey. Carley. and Kaiser (1958) collected 1.168 records of production.and environmental information fromLNB DHIA Holstein.herds infiflisconsin.during a three year period.. Results, which were comparable with the previous study, are in fair agreement. Of particular interest. however. to the present stmiywas the highly significant partial regression coefficient for milk.yie1d on calving interval and previous dry period. A lack of linearity in the data was noted which would make the regression coeffic- ients given.reliable for only certain parts of the ranges listed for the environmental factors considered. while the studies cited give indications of the relative impor- tance of some of the constitutive environments that go into the overall herd environment. they by no means cover all possibilities. The studies of Bayley and Kaiser and of Starkey at, 31. are plagued with problems ccmmon.to this type of study such as the paucity of exact measurements. lack of linearity within the ranges of the variables being comidered. and all variables considered not being completely indepenient of each other. The motivation.fcr these studies has been to develop correction.factors to standardize environment and. thereby; to reduce environmental.differences between herds. Where sufficient numbers of animals are inrdrved. correction factors are an economical and reason- ably accurate method of staniardizing environment for sire analysis work. In order to escape the vagaries of correction factors as much as possible and at the same time to standardize environment, special testing stations have been established in Denmark for evaluating daughters of potential AI sires. Robertson and Mason (1956) found that the variation beWeen sires' daughters in milk yield was much larger at the stations than in farmer herds of the same production level. The authors point out that there is extra variation between progeny groups at the teeth: stations that is not repeated in the field. A possible environmental sauce of this extra variation is suggested to be the manner in which grogem‘ groups are segregated in the testing stations. The authors smgest desegregating these groups in order to reduce variation between sires. They state: ”It is doubted whether the test stations can give as much useful information on the numerical aspects of performnce as the field records usually available. The principle value of the test stat ions is one of demonstration of managerial methods“. Touchberry and Rottensten (1958) studied the Danish testing station ani field records covering an eight year period. The station reccn‘ds included 5H5“ daughters of 305 Red Danish Hillu‘ace sires. Farmer herd records were from 110 of these sires with 3270 daughters. The testitg station data showed couponents of variance betwaen sires to be 223 and intra sire to be 11140 after adjustment for age at calving and days in milk. Data of sires' daughters from farmer herds were assessed as deviations from the mean annual production of the herd. The components between ard within sires were 47 and 887. respectively. The authors observed by comparing the between an! within components of variance for sires. that tie test stations were injecting environmental variation that was not found in the farmer herds. The authors found that the correlation between station and farmer-herd tests was .16. A negative component of interaction between sires and herds within AI centers was reported. The studies reviewed indicate mam constitutive environments make up the herd enviroment. If interactions are taking place they are between sires an! these constitutive enviroments and not between sires and herds. m n. Wadell (1957) noted no interaction between sires and herds for 1.496 Holsteins sired by 199 sires in 282 Michigan herds. This study analyzed each AI record as a deviation from the average of all the natural records in the herd. am! for cows with more than one record all but one record were eliminated in a random manner. Specht (1957) observed a positive herd-sire interaction that accounted for 7 per cent of the total variation in milk yield for 5.098 Holstein AI daughters sired by 30 bulls in le Michigan herds. The interaction components from the records of 2,631 AI progem‘ in the three years 1953. 1951+. ani 1955 accounted for three. two. and zero per cent respectively. of the total variation in milk production. When the data for these three years were recombined and reanalyzed. an interaction component accounting for 9 per cent of the total variation in milk production was noted. Pirchner and Lush (1959) analyzed 2,903 Holstein AI heifers in 1.177 Iowa herds on an intra-year stuch involving I+81 hires. Sire-herd interactions accounted for it an! 3 per cent of the total variation in milk and fat [reduction of AI heifers. Genet ic differernes between herds accounted for 6. 5 per cent of the differences between herds. On an intra-year-season basis. interaction components amounted to 7 per cent of the total variation for both milk and fat. This basis. however. led to estimates of 28 and 24 per cent for genetic differences be- tween herds for fat and milk production, respectively. The authors suggest sampling errors as a possible cause for the larger proportion of genetic differences. Korkman (1953) studied 35 AI sires with 10 daughters in an of the 252 herds analyzed. His results suggest that herd differences are due to Specific environmmtal influences. By dividing herds into three planes of nutrition. he found that within a given plane of nutrition there was a significant difference between the breeding values of sires. but there is no difference between the adaptability of the daughter groups of different sires to different planes of nutrition. Mason and Robertson (1956) divided herds into three groups on the annual average production for the herds. Analyzing first annual records from 13,000 AI daughters of 152 sires in Denmark, they founi no evidence of sire-herd interaction either between or within the three herd levels. They concluded that the ranking of sires within levels Will remain the same from herd to herd. They also noted a decrease in the coefficient of variation as production of herds increased. Legates, Verlinden, and Kendrick (1956) worked with 2n.75u AI daughters of Guernsey, Holstein, and Jersey sires throughout the United States for the years l9lt6 through 1950 to estimate herd-sire interaction. They founi that interaction for test was not important except in Jerseys where it approached the variation of sires in magnitude. Sire by herd interaction for milk yield was indicated to be at or near zero. They concluded that specific sire by herd differences are not of a major importance. In other words. the ranking of sires should remain relatively the same from herd to hard regardless of environmental differences between herds. Hancock (1953) estimated interaction between environment ard genotypes by a method different frat those previously discussed. Fifteen sets of monzygous twins were divided into three group of ten each and each group was divided over three different levels of enviroment. In design. this experiment was a balanced incomplete block. The study covered a three year period with milk. fat, and casein content recorded daily. All records were terminated (1) after fat production fell below 8 lb. for 28 days; (2) six weeks prior to freshening; or (3) 309 days after calving. Hill: yield was corrected for age (ME) . The analysis of variance was by years with the range of components of interaction between enviroment and genotypes for the three years accounting for between 5 and 9 per cent of the total variation in milk production am! between 6 an! 11 per cent of the total variation in fat yield. The components of interaction. as noted by the author, are actually residual components only partially made up of interaction and contain errors of smasureunt and errors due to uncontrollable factors specific to individuals . MATERIALS AND METHODS First records reported in the Michigan Dairy Herd Improvement Association fcr AI daughters in tested herds with at least one yearly herd average reported were used. These records were from stufiard DHIA and REA-IBM an! included records cmpleted through Septanber 1958. All records were BOS-K-ME and were canpleted lactations of 180 to 305 days in length. Lactations longer than 305 days were terminated at 335 days. The pounds of milk and fat produced and per cent fat were recorded for each cow together with her sire and hard. Herds were classified by level of production, location. average days dry. calving interval. and type of housing. W. In a first attempt to identify a potential source of sire-environmental interaction. it was decided that yearly herd averages might provide a guide to reduce non-genetic differences between herds. Three levels of production were established ‘cy dividing the array of herds (as based on an average of all. available yearly herd averages) into essentially equal numbers of herds within each level. Table 1 presents for each level the lumber of herds in that level. the average nuber of cows per herd. and the percent of cows that were produced by £1. The average milk production per level is also iniicated an! is calculated from the average of the annual herd averages. Some automatic relationships exist between the level of production of a herd and the AI daughters in the herd since it was possible for AI daughters to be included in the calculation of the annual herd average. Complete indeperdence between annual herd average an! AI daughters would 10 be desirable since there should be no [redetermined connection between levels based on herd averages and the AI daughters in them. The amount and effect of the automaticity is considered small. TAELEl SIZE, NUMBER AI . AND AVERAGE PRODUCTION FOR HERDS AT THREE LEVELS OF PRODUCTION Level Holstein Guernsey No. Herds No. Herd Ave. Milk No. Herds No. Herd Ave. Milk AI S ize AI Size 1 401+ 2325 21+ 9.1tro 117 601 21» 6.762 2 #03 3058 25 10.983 116 832 22 8.307 3 #04 3255 23 12.775 122 1+39 26 10,2144 in 1211 2a 10,966 355 2a 8.46; Lagging. Since the nine crop reporting areas of Michigan contain within each area a somewhat uniform type of farming with a different type of farming between areas. it was decided that some uniformity of environment might exist on dairy farms within each reporting area. Areas 1 through 5 (Figure l) were combined into two composite areas because of insufficient numbers of cows on test. Crop reporting areas 1. 2. and 3 were combined into Group 1. Crop reporting areas ’4 and 5 were combined into Group ’4. This study included all of the Holstein herds included in the previous division into groups by level of production of the herd. Numbers of Guernseys were too mall to use here or in am of the subsequent groupings. WW1. calving interval is considered 07° DICK/II D h‘ SON MAC/(WAC C ”AR ANTflM MLKASM RAWF'D T5160 CROP REPORTING COMBINED ‘ . 0““ an) my AREAS AREAS MAINS. IV a Miami scan. 1 -------—------ 1 “4'" 2 __ 1 A “a uR°" 3 --- __ 1 r ocean no (60514 menu MIDLAND l4 _ _ q, ‘ umuc MUSKE. amaw par/or INAIV 5 -.- "‘— =- u l 43' ‘7 6 6 0mm 0 IA 7 -- — 7 mum: ALLEGAN r EATON menu I I :77! 9 9 VA]! sum 1m ALHOU JACKSON asilmwv m m: .- “‘ % FIGURE 1 - CROP mm AREAS' év as: most»! ”AMI! mum mm: MONROE 42. OF MICHIGAN l 1 J ‘3. ° ° e e as or u ,5 54' as 0-1219 13 one of the best m criteria for rating good overall management on a dairy farm (Benne, 1958). It implies in one readily available value the level of such managerial factors as the reproductive efficiency of the herd which is in turn directly affected by the health and condition of the herd and by the ability of the dairyman to cope with reproductive disorders. Days dry seemed also to present a good irdication of management in general. Four hundred an! ten herds with information of days dry and calving interval previously studied by Benne (1958) were used to code latching herds in the present study. Table 2 irdicates the division of the herds by the range of days dry and calvirg intervals. TABLEZ NUMBEROFHEDSANDRALBECF THREEMANAGERIALIEVEISFOR DAYS DRY AND FOR CALVING INTERVAL Days Dry Calving Interval No. Herds Range (days) No. Herds Range (months) 36 0&5 62 10-11 335 #6-75 222 12 “9 76.111 126 13-16 W- Types of housing as analyzed by Knisely (1959) involved 627 herds with qualifications for inclusion in the present study. The types of housing included 471 herds with stanchion type, 199 with loose type. and 37 with switch type. Switch was the case where the herd was too large for the milking facilities of the stanchion barn. and as a result. cove were held as in loose housing for certain periods of the day but were milked in the stanchion barn. 14 W. All sires were not represented equally in all herds. The majority of the sires were represented by daughters in few herds with the number of daughters being from 0 to 6 in a herd. Henderson's (1953) method 1 for coulponents of variance of non-orthogonal data was used for the following two arrangements. First, an over-all analysis was done to estimate the components of variance for herds (H), sires (S), interaction between herds and sires (HS), and residual (E). The second method involved sub—dividing the components of variance H and HS on each of the classifications. levels of production (LP). locations (A), days dry (DD), calving intervals (CI), types of housing (TH). For example. H was divided into variance between levels of production LP and H within LP; HS was divided into interaction between sires and levels of production 3 by LP and into HS within levels. This procedure estimated the interaction between sires and various characteristics of herds related to their environment and mamgement. In the model for the over-all analysis yijk denotes the record made by the km daughter of the jib sire in the ith herd: Yauehiese-hs as. 131: 8 J 31:) 13k T571713 1’31 v h"i31 u is cannon to all observations. hi is the deviation from the mean caused by the m herd and s3 is the deviation of the daughters of the jib sire from the mean. he” is peculiar to records of the daughters of the jib sire in the 1th herd. °ijk is a random element in each daughter's record. The notations below hi ard he” of the model represent graph- ioally the division of the first model for the second method of analysis .15 as given in the previous paragraph. Henderson (1953) points out in the description of his method 1 that it is assmed that u is a constant and other elements of the first model are uncorrelated variables with means zero and variances H. 8. HS. E. It is not known to what extent. if at all. this condition is met in this study. Conceivably a nominated rating system could impart some degree of correlation between h: and s 3. As Pirchner and Lush (1959) point out. ”The infomtion on a bull's breeding value at the time he is selected is generally meager and would seem to warrant assumirg that the correlation between the genetic merit and the enviromental level of DHIA or HIR herds is negligible." The process of computation consisted of calculating sums of squares ani equatirg them to their expectations. Estimates of the components of variance for the elements of the Innar model were obtained by solvim the equations of the competent matrix. RESULTS AND DISCUSSION The results and discussion of each constituent environment will be deferred to its separate heading together with its respective tables. WW. When AI daughters were grouped into three environmental levels based on the average annual prodmtion of the herds in which they were. the component of interact ion between sires an! levels of production accounted for none of the total variation in milk. fat, or test (Table 3). The ranki’tg of sires would be the same in am of the three enviromental levels measured by annual herd averages. Interactions within levels between sires ard herds were still taking place. however. and‘accounted for 3 per cent of the total varia- tion in milk an! fat yield. This is only slightly less than the it per cent calculated for all herds by sire interactions for milk and fat yield. Mason and Robertson (1956) . and Legates gt, 5].. (1956) indicated that interaction between sires an! herds was near zero. but recent esti- mates cf interaction place the value at 6 per cent of the total variation within year in milk and fat production (Pirchner and Lush, 1959). Specht (1957) estimted that interact ion between sires and herd accounted for 9 per cent of the total variation in milk and fat production. Interaction between herds and sires of 1+ per cent of the total variation falls within the range cited by others. Grouping herds according to levels of production effectively placed most of the variation between herds in the component of variance for levels. About half of the differences between herds for milk and fat yield were removed by grouping AI daughters in this manner. If 17 TABLE 3 COMPONENTS OF VARIANCE FOR IEVEIS G‘ TRODUCTION, HERDS AN) SIRES Fm HOISTEINS Source of Mean Component Percent of Wu df 39mm: of Vgnjgmg Totg11 an? Sire 192 16.866 231 3 Herd 1.210 19.041 2.066 33 SireIHerd 5,385 3.923 2u1 u Residual 1.850 3.760 3.760 60 Total 8,637 6.298 Sire 192 16,866 222 3 Level 2 14,856,626 1,695 25 Herd/Level 1.208 11.032 941 1a Sire/Level 310 4,951 31 0 Sire-Herd/Level 5.075 3.860 227 3 Residual 1.850 3.760 3.760 55 Total 8,637 6.876 nu Sire 192 - 16.718 178 2 Herd 1.210 26.212 2.925 36 SireIHerd 5, 385 5,016 288 1+ Residual 1.850 4.813 1+,Bl3 58 Total 8,637 8,204 Sire 192 16.718 186 2 Level 2 5,878,794 2,051 23 Herd/Level 1.208 16,522 1. 563 18 Sire /Level 310 5.826 .. 19 0 Sire—Herd [Level 5 .075 9,966 312 3 Residual 1,850 15813 £5813 5h Total 8.637 8,906 last. Sire 192 0.62 0.01 6 Herd 1.210 0.23 0.02 11 SirexHerd 5.385 0.09 .o.ou 0 Residual 1.850 0.15 0.15 83 Total 8.637 0.14 Level ‘ 2 0.80 0.00 o Herd/Lavel 1.208 0.23 0.02 11 Sire/Level 310 0.10 0.00 0 Sire-Herd/Level 5.075 0.09 —0.0u 0 Residual 1,850 0.15 0.15 78 8.632 0.15 INegative variance components were considered as essentially zero in the calculations of per cent of total. zMean squares and components of variance for milk multiplied by 10'3. 18 herd differences are between 6 and 10 per cent heritable as estimated by Lush ard Straus (1942) . Pirchner and Lush (1959) . Robertson and Rendel (19514). and others. then at least 90 per cent of the differences between levels is non-genet ic if genetic herd differences are the same for all levels of environment. It has generally been thought that since dairymen have had equal Opportunity to select between bulls in AI. genetic differences betvmeen AI herds are small. and are being reduced by the same bulls being used in may herds. To determine if dairymen with herds in one of the three levels of environment were selecting bulls more or less than dairymen with herds in some other level. a chi-square test of the preportionate use of sires in the three levels was made. Seventy-five bulls having five or more AI daughters per sire-level cell were used. The results of the test (12 = 178.8. d.f. . 11+8, P< .05) indicated that selection has been taking place. Various AI bulls were not unifomly represented in all environmental levels. This might indicate that tin genetic difference between levels could be something more than the 10 per cent between herds indicated by the literature. The fallacy of this assump- tion lies in the observations of Pirchner ani Lush (1959) that no dairyman is going to select bulls for low production and that the generally meager information available at the time the dairy-man selects an AI sire for use in his herd does not allow intelligent selection to be made based on the true genetic merit of the bull. Table 1+ presents the average production of the AI daughters by the level of production of the herd in which they were. In comparing these results with those in Table 1, while each level consisted of an 19 TABLE“ PRODUCTION BY HOLSTEIN AI DAUGHTERS AT VARIOUS IEVEIS (F PRODUCTION OF I'IERDS Level No . of Records Milk Fat Test 1 2325 11.018 1402 3.67 2 3053 12 e103 it39 3.65 3 3255 13 .640 “'93 3.63 Total 8638 Average 12 , 390 1+ 50 3 . 65 equal number of herds essentially of the same size. there is a definite increase in the preportion of AI daughters in herds of the higher level. If there is no correlation betwaen environment level and genetic merit of herd then these herds in the high level have high annual production more because of their environments (which includes management), than because the genetic quality of their AI daughters is superior to the genetic quality of AI daughters in the lower herds. W. The results obtained from the analysis of levels of production for Guernseys were essentially the same as for Holsteins. Table 5 summarizes these results. The component of variance betwaen Guernsey sires was relatively large as compared with Holsteins. If differences between progerw groups are mainly due to genetic differences, then it would appear that Guernsey sires have larger genetic differences than Holstein sires and that close attention to sires selected would be a profitable consideration for Guernsey dairymen. TABLE5 20 COMPONENTS OF VARIANCE FOR IEV'EIS OF TRCDUCTION. I-IERDS AND SIRES FOR GUERNSEYS Source of Mean Component Percent of Isaiaiien 41 W5 V T 1 am? Sire 138 6.886 259 8 Herd 355 8.425 1.181 35 Siremerd 865 1.549 80 2 Residual 513 1.819 1.819 55 Total ' 1.871 30339 Sire 138 6.886 274 8 level 2 436.531 665 18 Herd/Level 353 6.000 721 20 Sire/level 159 1.388 .. 115 0 Sire-Herd/Level 706 1. 586 180 5 Residual 513 1.819 1.819 49 Total 1.871 3.544 fat. Sire 138 16.635 663 8 Hard 355 21 .5143 3 o 265 38 Siremerd 865 2.644 - 682 0 Residual 51’) 4.602 4.602 54 Total 1.871 7.848 Sire 138 16.635 657 7 16761 2 973 9507 1 .467 16 Herd/Level 353 16.150 2.250 25 Sire/Level 159 3.939 - 140 0 Sire.Herd/Love1 706 2.352 - 539 0 Residual 513 4.602 4.602 52 Total 1.871 8.297 Teal Sire 138 0.51 0.02 11 Herd 355 0.32 0.02 10 SireXHerd 865 0 .16 0.02 9 Residual 513 0.15 0.15 70 Total 1.871 0.21 Sire 138 0. 51 0.03 13 Level 2 0.21 0.00 0 Herd/Level 353 0.32 0.02 9 Sire/Level 159 0.14 .0.01 0 Sire-Herd/Level 706 0.17 0.03 13 Residual 513 0.15 0.15 65 1.871 0.22 TNegative variance components were considered as essentially zero in tin calculations of per cent of total. 2Mean squares and components of variance for milk multiplied by 10‘3. 21 The total variance in fat ani milk production for Guernseys was smaller than for Holsteins. with the two min sources of variation being herds and residual. Differences between Guernsey herds were less than those of Holsteins and residual components of variance for Guernseys were about one-half those of Holsteins. 'wadell (1957) also noted this difference between the two breeds. While these observations hold for milk and fat yield. they do not hold for test where Guernseys exhibit more total variability than Holsteins. This criterion is also the location of the largest component of interaction between sires ard herds found in this study. contributing 9 per cent of the total variation. The sunnnaries of AI daughters as grouped by herd levels for Guernseys in Table 6 indicate a marked decrease in the number of AI TABLE. 6 PRODUCTION BY GUERNSEY AI DAUGHTERS AT VARIOUS LEVELS OF PRODUCTION OF HERDS Leul No. of Records Milk Fat Test 1 601 7.465 363 4.89 2 832 8,662 418 4.86 3 439 9 s 213 447 4.87 Total 1872 Average 8 .407 407 4.87 daughters in herds with higher environmental levels. In both Holsteins and Guernseys. averages of the herd levels in Table l have been below those of the AI daughter averages for the same levels with one exception. This was in the high environmmtal level of Guernsey where the herd averages exceed the AI daughter averages. As pointed out byWadell (1957) . I T I . f . Y T T T \ '. 22 TABLE 7 COMPONENTS W VARIANCE FOR AREAS. HERBS. AND SIRFS Source of Mean Component Percent of Vania: ion df 59952;: or vegan“ Tgtaa} um? Sire 192 16 . 866 231 3 Herd 1 .210 19.041 2 .066 33 Siremerd 5 . 385 3 .923 241 4 Residual 1.850 3 .760 3 .760 60 Total 8 .637 6 .298 Sire 192 16 .866 318 5 Locat ion 5 198 .515 119 2 Herd/Location 1.20 5 18.296 1.963 29 SireILocation 544 2 .314 - 375 0 Sire-Herd/Locat ion 4 .841 4.104 537 8 Residual 1.850 3 0760 3 9760 56 Total 8 .367 6 9322 Eat s ire 192 16 .718 178 2 Herd 1.210 26.212 2.925 36 Siremerd 5 .385 5 .016 288 4 Residual 1 .850 4 .813 4 .813 58 Total 8 . 637 8 .204 s ire 192 16 .718 305 3 Location 5 356 .193 236 3 Herd/Location 1.205 24.843 2.731 31 SireILo cat ion 544 2 .196 - 570 0 Sire-Herd [Location 4 .841 5 .333 739 8 Residual 1.850 4.813 4.813 55 Total 8 .637 8 .257 Inst Sire 192 0.62 0 .01 6 Herd 1.210 0 .23 0.02 ll Siremerd 5 .385 0 .09 .0 .04 0 Residual 1.850 0.15 0.15 83 Total 8 .637 Sire 192 0 .62 0 .00 0 locat ion 5 4 . 81 0 . 01 5 Herd/Location 1.205 .21 0.02 10 SireILocation 544 - .30 -0 .04 0 Sire-Herd/Location 4.841 .13 0.02 10 Residual @250 .15 O .15 75 . 37 jNegative variance components were considered as essentially zero in the calculations of per cent of total. 2Mean squares an! cornponents of variance for milk multiplied by 10'3. 23 bull studs should provide bulls of sufficient genetic merit as to be at least as good as the sires being used naturally now by the better herds. It would seem from the results of Tables 1 and 6 that this is not being done in the case of Guernseys. While Guernseys exhibit a large component of interaction between sires and herds for test. when considered on the basis of environmental groups. they failed to give any indication of interaction. 5:93;. This section contains the same herd and sire information as does the first section for Holstein levels. However. as Table 7 shows. very little was removed from the component of variance between herds as a result of grouping herds by areas. While the areas were relatively close together geographically as compared to the areas covered by Legates gt, :1. (1956). the conclusions are the same; there is no need to designate certain sires for use in certain areas. at least not in Michigan. Table 8 gives a tabulation by area of the AI daughter averages. TABLE 8 AVERAGE BY AREAS OF RECDRDS FIRST REPCRTED FOR AI DAIGHTERS Area No. Milk Fat Test 1 722 11.467 417 3.66 4 648 12.493 456 3.67 6 2196 12.353 443 3.60 7 734 12.587 462 3.70 8 2593 12.695 464 3.68 9 1745 12.250 441 3.62 Total 8638 Average 12 .390 450 3 .65 24 TABLE 9 COMPONENTS OF VARIANCE FOR DAI S DRY. HERBS. AND SIRES Source of Mean Component Percent of mien df Square: 9; Vgnjangg Tgml um? Sire 172 11.769 222 3 Herd 394 214.480 1.955 30 SireXHerd 2.1496 3.871 - 169 0 Residual 1.018 4.313 4.313 67 Total 4.080 6.321 Sire 172 11,769 294 4 Days Dry 2 134.959 229 3 Herd/Days Dry 392 24.916 1.906 28 SireXDaya Dry 157 5.679 - 91 0 Sire-Herd/Days Dry 20339 30750 " 152 0 Residual 1.018 4.313 24.313 64 Total 4.080 6.1499 Eat Sire 172 11.439 110 1 Hard 39+ 33.816 2.755 33 Sirsmerd 2.496 5.049 - 188 0 Residual 1.018 5.566 5.566 66 Sire 172 11.439 144 2 Days Dry 2 1914.709 6334 1* Herd/Days Dry 392 32.995 2.683 31 SireXDays Dry 157 7.987 - 42 O Sire-flerd/Days Dry 2.339 4,852 - 181 0 Residual 1.018 5.566 5.566 64 Total 4.080 8.504 19.21 Sire 172 0.38 0.01 5 Herd 394 0.23 0.01 5 Sirsmsrd 2.496 0.07 -0.06 0 Residual 1.018 0.19 0.19 90 Total 4.080 0.15 Days Dry 2 8.01 0.00 0 Herd/Days Dry 392 0.23 0.01 5 SireIDays Dry 157 0.11 0.00 0 Sire-Herd/Days Dry 2.339 0.07 -o.09 0 Residual 1.018 0.19 0.19 90 4 .080 0.12 megative variance components were considered as essentially zero in the calculations of per cent of total. 2Mean squares and components of variance for milk multiplied by 10'3. 25 W. Days dry and calving intervals are cmsidered to be good indicators of the overall namgerial ability of the dairyalan (Benne. 1958). However. very little of the caaponent of variance between herds was accounted for by these criteria. 2 or 3 per cent as iniicated in Tables 9 ani 11. This is much snallnr than expected. As previously nentioned. it is possible to have sire by herd interactions overwhelmed by herd differences an! by cancelling effects within herds. In the sire by herd analysis for milk an! fat production in Table 11 the component of interaction between sires a!!! bonds is negative or essentially zero. Yet. when this component of interaction between sires and herds is divided into herd groups based on calving intervals in the herd. the component of interaction between sires and calving interval groups become a positive value. While this is less than one per cent of the variance. it could indicate that some interaction was being identified - albeit very little. Sanpling error light be a more plausible reason. The criterion of due dry as shown in Table 10 would seem to favor shorter dry periods. however. no conclusions should be drawn. TABLE 10 AWGE BI DAYS DRY OF RECORDS FIRST REHETED FOR AI DAUGHTERS Dqs Dry No . Milk Fat Test 1 114 13.540 493 3.65 2 3592 12 . 620 #58 3 .65 3 375 11 .929 #32 3 . 64 Total 4081 Average 12. 582 457 3.65 26 TABLE 11 00st CF VARIANCE F02 CALVING INTERVALS. HFRDS. AND SIRES Source of Mean Component Percent of Damion d: W Total];_ 81.1112 Sire 172 11.769 222 3 Herd 39+ 24.480 1.955 30 Siremerd 2.496 3.871 - 169 0 Residual 1.018 4,313 4,313 67 Total 4.080 6.321 Sire 172 11.769 210 3 Calving Interval 2 191.928 124 2 Herd/Calving Interval 392 23.626 1.883 29 Sire/Calving Interval 237 6. 543 32 0 Sire-Herd/Calving Interval 2.259 3.591 - 196 0 Residual 1.018 4,313 4.313 66 Total 6.366 Eat. Sire 172 11.439 110 1 Herd 391+ 33.816 2.755 33 SirexHerd 2.496 5.049 - 188 0' Residual 1.018 5.566 5.566 66 Total 4.080 8.243 Sire 172 11.439 104 l Calving Interval 2 237.798 149 .2 Herd/Calving Interval 392 32.775 2.669 31 Sire-calving Interval 237 8 .586 19 0 Sire-Herd/Calving Interval 2.259 4.678 - 210 0 Residual 1.018 5.566 5, 566 65 Total 8.297 Teal . Sire 172 0.38 0.01 5 Herd 394 0.23 0.01 5 Siremerd 2.496 0.07 -0.06 0 Residual l .018 O .19 0 . 19 90 Total 4.080 Sire . 172 0.38 0.01 5 Calving Interval 2 2.61 0.00 0 Herd/Calving Interval 392 .22 0.01 5 Sire/Calving Interval 237 .11 0.00 0 Sire-Herd/Calving Interval 2.259 .07 .0.09 0 Residual 1.018 .19 0.19 90 1.21.11 0412 INegative variance components were considered as essentially zero in the calculations of per cent of total. 2Mean squares and components of variance for milk multiplied by 10'3. 2? Table 12 indicates small differences between the averages of AI daughters in herds where. as Table 2 suggests. the recommended managerial practices are not being carried on under calving intervals 1 an! 3. TABLEJZ AVERAGE BI CALVING DITEWAL 0F REWRDS FIRST REHETED FOR AI DAUGHTERS Calving Interval No. Milk Fat Test 1 550 12.893 “77 3.72 2 2244 12.305 “-48 3.65 3 1287 12.936 465 3.61 Total 4081 Average 12. 582 457 3.65 W. Here again. an attempt has been made to separate differences betwaen herds by grouping AI daughters by the type of housing under which they are confined. As Table 13 indicates. type of housing has very little effect on dividing differences betvmen brds. ~ Table 14 presents averages of AI daughters for the three types of housing. Milk an! fat yield are not affected by the type of housing. but test seems to vary more than would be apected. If the sire by herd interaction for test is valid. it might explain this difference in test between the different types of housing. 28 TABLE 13 (DMPONENTS CF VARIANCE FOR TYPES OF HOUSING. HERBS. AND SIRES Scurce of Mean Component Percent of Mien df satires of Variance Totanl 14.1122 Sire 185 12.565 209 3 Herd 626 20 .967 1.979 31 SireXHerd 3 .214 4.02? 208 3 Residual 1.214 3 .943 3 .943 63 Total 5.239 6.339 Sire 185 12. 565 22 0 Housing 2 74. 502 23 0 Herd/Housing 621+ 20 .796 2 .000 32 Sire/Housing 229 6.252 - 23 0 Sire-Herd/Housing 2 .985 3 .856 194 3 Residual 1.214 3.943 3.943 64 Total 5.239 6.159 {at Sire 185 14.198 193 2 Herd 626 29.047 2.826 34 Sir eXHerd 3 .214 5 .010 47 1 Residual 1.214 5.191 5.191 63 Total 5.239 8.257 Sire 185 14.198 19 0 Housing 2 50 .501 - 12 O Herd/Housing 624 28.979 2.859 35 Sire /Housing 229 8 .519 - 1 0 Sire-Herd/Housing 2.985 4.741 29 0 Residual 1.214 5.191 5.191 65 Total 5.239 8.085 last Herd 626 0.25 0.02 10 SirexHerd 3.214 0.08 -o.07 0 Residual 1.214 0.18 0.18 85 Total 5 .239 0.14 Sire 185 0.41 0.00 0 Housing 2 0.83 0.00 0 Herd/Housing 624 0.24 0.02 10 Sire/Housing 229 0 .13 0.00 0 Sire-Herd/Housing 2.985 0.08 .0.07 0 Residual 1.214 0.18 0.18 90 5.239 0.13 Negative variance components were considered as essentially zero in the calculations of per cent of total. 2Mean squares an'l components of variance for milk multiplied by 10‘3. 29 TABLE 115 AWAGE BI TYPE OF HOUSING OF RECORDS FIRST REPCRTED FOR AI DAUGHTES Type of Housing No. . Hill: Fat Test 1 3570 12.622 458 3.64 2 1309 12.235 448 3.67 3 361 12.638 457 3.61 Total 5240 Average 12.526 455 3.65 CONCLUSION The general component of interaction between herds an! sires accounted for between zero and nine per cent of the variation in milk and fat production and test. When components for herds and herd by sire interaction were split into one of the five constitutive environments. there was no apparent interaction with sires. Therefore. the rankiig of AI sires in herds on the basis of their daughter's performance Will. be indepenient of any one of the five environnsntal segments studied. From a practical standpoint this means that the committees for selection of AI sires need not be concerned with having to provide different bulls for the different types of environment. That no interaction was found between sires and am of the five environmental factors studied does not mean that interaction is not present. This is evident from the general sire by herd interaction observed and from the sire by herd interactions within three of the envirommtal segments. production levels. types of housim. and loca- tions. How much effect cancelling within herds of opposing interaction may have on the Ingnitaie of the component of interaction is not known. It seens reasonable to assume that the cancelling effect does exist. The literature suggests that year and season effects may be important in an analysis of this type. As more records become tvailable through the expanded DHIA-IBM program these two parameters should be included in am future study. The chi-square test indicated that Holstein AI sires were not equally represented in all herds. While probably not affecting the genetic difference between herds. it does indicate that dairymen are 30 31 aware of some kind of difference between the bulls offered by the AI studs. On what basis dairymen select bulls for use in their own herds might be of interest. SUMMARY Five envirorauental factors were studied to determine the existence and magnitude of interaction between sires' daughters and these factors. The factors were based on herd information and included level of produc- tion. location. days dry. calving interval. an! type of housirg. In no instance was a canponent of intonation between sires' daughters and an enviror-ental factor detected. The rankirg of sires by the performance of their daughters would be generally the same in am of the environ- ments examined . 32 2. 10. LITERATURE CITATION Bayley. N.D.. and Heizer. E.E. 1952. Herd data measures of the effects of certain environmental influences on dairy cattle. Jour. Dairy Sci. 35,: 540-5149. Benne. M.E. 1958. Relation of calving interval. days dry. and period of calving to milk production. M.S. thesis. Michigan State University. East Lansing. Hancock. J. 1953. The relative importance of inheritance and environment in the production of dairy cattle. New Zeal. Jour. Sci. and Technol. 35,: 67-92. Herrlerson. C.R. 1953. Estimation of variance and covariance components. Bio- metrics 9: 226-252. Knisely. R.C. 1959. Unpublished data. KorleIan, Ne 1953. Versuch einer vergleichenden Nachkommenschafts-untersuchurg von Bullen die in Herden mit verschieden starker Futterung wirken. (original not seen) (Eng. abst.) Z. Tierzucht. Zuchtungsbiol. 6],: 376-390. Legates. J.E.. Verlinden. F.J.. and Kerdrick. JJ“. 1956. Sire by herd interaction in production traits in dairy cattle. Jour. Dairy Sci. 19: 1055-1063. Lush. J.L.. and Straus. F.S. 1942. The heritability of butterfat production in dairy cattle. Jour. Dairy Sci. 25,: 975-982. Mason. LL. and Robertson. A. 1956. The progerv testing of dairy bulls at different levels of production. Jour. Agr. Sci. 5L2: 367-375. Firehmr. Fe, and LuSh. JOL. 1959. Genetic and environmental pcrtions of the variation among herds in butterfat production. Jour. Dairy Sci. £2: 115-122. Plum. M. 1935. Causes of differences in butterfat [reduction of cows in Iowa cow testing associations. Join. Dairy Sci. 18: 811-825. 33 13. 1’4. 15. 16. 17. Robertson. A.. and Rendel. J.M. 1954. The performnce of heifers got by artificial insemination. Jour. Agr. Sci. 51: 1814-192. Robertson. A.. and Mason. LL. 1956. The progeny testing of dairy bulls: A comprison of special station anzl field results. Jour. Agr. Sci. 52: 376—381. swCht, Lowe 1957. Rates of improvement by progery testing in dairy herds of various sizes. Ph.D. thesis. Michigan State University. East Lansing. Starkey. E.E.. Corley. E.L.. and Heizer. E.E. 1958. The effects of certain measured environmental influences on tin p'oduction of Holstein-Friesian cattle. (Abst. ) Jour. Dairy Sci. EL: 722. Touchberry. R.W.. and Rottensten. K. 1958. A comparison of sire tests made at special Danish oge testing stations with tests made in farmer herds. grbst Jour. Anim. Sci. 12: Ill-#6. Wadell. L.H.. and McGilliard. L. D. 1959. Influence of artificial breeding on Iroduction in Michigan dairy herds. JO‘n'e Dairy seie L3: 1079‘1085' ROOM USE (Xi-ELY HICHIGON STRTE UNIV. LIBRQRIES 31293010725335