- Email: [email protected]
The value of daughters of bull dams in dairy cattle progeny testing schemes
Livestock Production Science, 23 (1989) 267-274
267
Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands
The Value of Daughters of Bull Dams in Dairy Cattle Progeny Testing Schemes T.H.E. MEUWISSEN 1 and J. RUANE 2
'Research Institute for Animal Production "Schoonoord", P.O. Box 501, 3700 AM Zeist (The Netherlands) 2Institute of Animal Physiology and Genetics Research, West Mains Road, Edinburgh EH9 3JQ (Gt. Britain) (Accepted for publication 22 May 1989 )
ABSTRACT Meuwissen, T.H.E. and Ruane, J., 1989. The value of daughters of bull dams in dairy cattle progeny testing schemes. Livest. Prod. Sci., 23: 267-274. In dairy cattle progeny testing schemes, bull dams of high genetic merit are identified and bred to produce young bulls. Females will also be produced and any of them may themselves become bull dams. The impact of these (daughters of bull dams (DBD)) on genetic gain has been underestimated in the past. The predicted rate of genetic gain increased by 2%, when the DBD were included in the calculations for a conventional progeny testing scheme. When MOET (multiple ovulation and embryo transfer) was used on bull dams to produce 10 offspring dam -1 year-', this difference was 7%. When accounting for preferential treatment of bull dams by limiting the accuracy of selection or the selective intensity, including DBD increased genetic gain by 2-11%.
INTRODUCTION
In dairy cattle breeding genetic progress is made along four pathways: selection of bulls to breed bulls (BB), selection of bulls to breed cows (BC), selection of cows to breed bulls (CB) and selection of cows to breed cows (CC). About 76% of the total genetic gain is made in selecting the parents of males (i.e. on the BB and CB pathways) and 24% in selecting the parents of females (Everett, 1984). However, the matings of bull sires and bull dams will on average produce equal numbers of males and females. The daughters of bull dams (DBD) born have a higher genetic mean than the daughters of cow dams (DCD) born at the same time. With the scheme in steady-state equilibrium, the difference in means ( D B D - D C D ) is 0301-6226/89/$03.50
© 1989 Elsevier Science Publishers B.V.
268
IBB -
T.H.E. MEUWISSENAND J, RUANE
zIGLBB+_ICB - AGLcB~ (Icc - AGLcc +_IBC 2
/
\
ZGLBc
/
2
where AG, I and L represent the genetic gain year-1, the genetic selection differential and the generation interval, respectively. In a typical progeny testing scheme this difference may represent 3.5AG to 4.5riG (e.g. 3.9AGin the example of Table 1 ). The value of DBD has previously been underestimated, because the usual pooling of the DBD and the DCD population into one single population of potential bull dams leads to an underestimation of the contribution of the best population to the selected part. The existence of DBD has previously been ignored in calculations of genetic response, because their number is small. The contribution of the DBD to the CB pathway may be significant, especially when the number of selected bull dams is small owing to the use of MOET (multiple ovulation and embryo transfer). Many authors (e.g. Cunningham, 1976) have shown that by using MOET on bull dams, selection on the CB pathway can be more intense and hence response can be increased by up to 10%. A consequence of this strategy, which has not been examined, is that the DBD born are of increased genetic merit which may further increase the genetic response. The genetic merit of selected bull dams is usually lower than expected. This is thought to be due, in part at least, to preferential treatment of high yielding cows (Van Vleck, 1977). The aim of this study is to investigate the importance of the DBD in a progeny testing scheme, with or without using MOET on bull dams and to study TABLE 1 Predicted genetic gain (ziG) in a progeny testing scheme w i t h o u t t h e use of M O E T a n d daughters of bull dams ( D B D ) Path
Age class
Selected fraction (%)
Selection accuracy
Contribution 1 of age class (%)
Genetic 2 selection differential (S .D. units )
Generation interval (years) 2
BB BC
6 2 6 4 5 6 7 2-7
2 100 5 1.3 1.1 0.6 0.2 100
0.933 0.0 0.933 0.5 0.6 0.65 0.675 0.0
100 20 80 54.1 31.2 11.5 3.2 100
1.13 0.77
6 5.2
0,74
4.6
0.0
4
CB
CC
ziG = ( 1.13 + 0.77 + 0.74 ) / ( 6 + 5.2 + 4.6 + 4 ) = 0.133 phenotypic s t a n d a r d deviations year - 1. 1The contributions of the age classes sum to 100% within each path. 2These mean values are weighted by the contributions of the age classes within a path.
DAUGHTERS OF BULL DAMS IN CATTLE PROGENY TESTING
269
the effect of preferential treatment on the importance of DBD. The impact of DBD on the CC pathway is neglected here, because many cows have to be selected for this pathway and so the contribution of DBD can only be small. MATERIALSAND METHODS To predict the steady-state genetic gain of a progeny testing scheme all four selection pathways have to be included, although our interest is only in the CB pathway. Consider the following breeding plan shown in Table 1. Annually 100 young bulls are progeny tested on 100 daughter records each. Without the use of M O E T about six bull dams are needed in practice to produce each young bull (Hinks, 1978 ). The generation interval of the selected bulls is 6 years. The cow dams are assumed to be selected at random with respect to the breeding goal, giving a generation interval of about 4 years. Some 50 000 heifers enter the cow population each year. Culling of cows is taken to be independent of the breeding goal and is 30% annually of the total number. Thus for example there are 35 000 3-year-old cows. The total cow population amounts to 167 000 cows. The trait selected for is milk production, which is assumed to have a heritability of 0.25 and a repeatability between lactations of 0.4. The genetic correlations between lactations are 1, unless stated otherwise. Modern sire and cow evaluation methods make it possible to compare the estimated breeding values of animals of different ages. Thus the best cows can be selected irrespective of their age. In the present breeding plan the bull dams are selected from Age Classes 4, 5, 6 and 7 (the age class number is the age at birth of their progeny), being selected on one, two, three and four lactation records, respectively. Each age class has a distribution of estimated breeding values. The means of these distributions differ due to genetic progress (e.g. the mean of Age Class 2 is zig higher than the mean of Age Class 3) and the variances differ due to the amount of information that is available. Ducrocq and Quaas (1988) described an algorithm to predict the fractions selected from each distribution if truncation selection is applied across several distributions. Because the differences in mean between the distributions are determined by the genetic progress, the steady-state genetic gain is calculated iteratively using this algorithm. If bull dams can also be selected from the DBD, there are four more distributions to select from, i.e. the distributions of estimated breeding values of the DBD of Age Classes 4, 5, 6 and 7. Genetic progress is calculated both with and without DBD included for the following schemes. For each scheme the total number of young bulls and DBD produced is constant. (1) No MOET used on bull dams. Six hundred bull dams are selected with one offspring each. (2) One flush per bull dam. Two hundred bull dams are selected with three offspring bull dam-1 year-1. This procedure is common practice among cer-
270
T.H.E. MEUWISSEN AND J. RUANE
tain AI organizations in order to increase the probability of producing a male calf. Only the female's own records are used for evaluation and information on full sibs is ignored. (3) Repeated use of M O E T on bull dams. Sixty bull dams are selected with 10 offspring bull d a m - 1year- 1. Each DBD is thus evaluated on her own record and those of four full sibs. This scheme may be applied to sib test the young bulls (Smith and Ruane, 1987). {4) This is the same as Scheme 3 except that preferential treatment of potential bull dams is considered. The effects of preferential treatment are modelled in three ways (Schemes 4a, 4b and 4 c ) . In Scheme 4a it is assumed that all the "best" cows are treated preferentially, so that it is not possible to select the "very best" cows among the "best" cows. This situation is modelled by setting a limit on the maximum selection differential possible within an age class. The maximum standardized selection differential is set to 2.421, which corresponds to a fraction selected of 2%. For example, when the proportion of animals selected from Age Class 7 is 0.1%, the genetic merit of the selected cows in this Age Class 7 becomes ~(7) +2.421rIo(7)ao instead o f ~ ( 7 ) + 3 . 3 6 7 r I v ( 7 ) a G (where ]~(7), riG(7), a c are the mean of Age Class 7, the accuracy of selection of Age Class 7 and the genetic standard deviation of milk production). In Scheme 4b it is assumed that the accuracy of the bull dam selection is not increased by including any records other than from the first lactation, as suggested by Van Vleck (1986). The reason for this is that cows seem to be treated evenly during the first lactation but preferentially thereafter (Murphy et al., 1982 ). This situation is modelled by at first assuming that the phenotypic variances and covariances are unaffected by preferential treatment. Second, it is assumed that the evaluation method does not account for preferential treatment, i.e. the weighting factors for the estimated breeding values (EBV) are unchanged. These assumptions imply that the variances of the EBV are not affected by preferential treatment. Third, the genetic correlations between the first lactation and the second, third and fourth lactation record are assumed to be reduced to 0.67, 0.65 and 0.64, respectively. These correlations are chosen such that rig ( 4 ) = rig ( 5 ) = r~G(6) = rra ( 7 ). The breeding goal is first lactation milk yield. Adult lactations are not ignored by the breeding value estimation procedure because it is not known to the breeders how preferential treatment affects the records of the cows. Scheme 4c is the same as Scheme 4b except that it is assumed that DBD are treated also preferentially during the first lactation (based on their pedigree). It is assumed that the accuracies of the EBV of DBD are reduced by 30%. The accuracies of the EBV of DCD are the same as in Scheme 4b. It is assumed that the population is large enough to neglect the effects of reduced selection differentials due to finite population size and inbreeding. In addition, the reduction in variance due to selection is ignored. This effect is
DAUGHTERSOF BULLDAMSIN CATTLEPROGENYTESTING
271
TABLE 2 The predicted genetic gain {AG) and the fraction of bull dams that are themselves daughters of bull dams (PDBD) for Schemes 1, 2, 3, 4a, 4b and 4c Scheme
1 2 3 4a 4b 4c
DBD ignored AG (S.D. year -1)
0.133 0.138 0.143 0.130 0.138 0.138
DBD included
AG (S.D. year -1)
PDBD (%)
0.135 0.141 0.153 0.144 0.149 0.141
14 30 80 65 74 60
( + 1.5%) (+2%) (+7%) ( + 11% ) ( +8% ) ( +2% )
partly avoided by only selecting once on the same information source. For example, DBD are selected on their own performance (and that of their contemporary full sibs) and not on the performance of their dam, because their dams were already selected on own performance or DBD are not selected on half-sib records, because their sires were selected on progeny records. RESULTS
Table I gives the contribution of the four paths and of the age classes within the paths to the selection response of Scheme 1. The steady-state genetic response in this situation is 0.133 phenotypic standard deviations year-1. In Table 2 the use of DBD is included. By selecting fewer bull dams the annual genetic gain can be increased from 0.133 up to 0.143 S.D. units. This increase of 8% is in agreement with previous results (e.g. Cunningham, 1976). The response is further increased by including DBD in our calculations of genetic gain. Without MOET, this increase is 1.5%; with intense MOET (10 offspring) it is 7%. The influence of the DBD increases as the number of bull dams selected decreases. Thus combining these two features for Scheme 3, response is increased by 13% (0.155/0.135) by selecting 60 bull dams to produce the young bulls needed instead of 600. Table 2 also shows the impact that including DBD has when preferential treatment is present. In Schemes 4a, 4b and 4c the genetic response is increased by 11, 8 and 2%, respectively. DISCUSSION
The genetic response predicted when including DBD in calculations on the CB pathway in a progeny testing scheme where M O E T is not used is only slightly higher than expected if they are excluded (Scheme 1, Table 2). How-
272
T.H.E. MEUWISSEN AND J. RUANE
ever, through using MOET a smaller group of highly superior bull dams can be selected, who produce daughters of higher average genetic merit. In addition the breeding values of these daughters can be estimated with a higher accuracy because of the full-sib information available. As a consequence, a greater number of bull dams selected are themselves DBD and so they have a greater impact on genetic progress. For example, in Scheme 3 80% of the bull dams were DBD. Apart from the generation interval of the bull dams, the progeny testing scheme approaches the MOET-Hybrid schemes described by Colleau (1985), where all bull dams are DBD. If the young age classes are also included for selection, the generation intervals of the bull dams in the progeny testing and MOETHybrid schemes will be more alike. This also increases the genetic gain of the progeny testing scheme, since the number of selection candidates increases. Thus, using MOET on bull dams in a progeny testing scheme has a greater effect on genetic progress than previously considered. This should also lead to increased inbreeding rates owing to the reduction of effective population size, since the BB and DBD are closely related. The genetic merit of bull dams selected is equal to the genetic mean of the group from which they are selected plus the genetic selection differential (/~(DBD) ÷ I (DBD) or/~ (DCD) ÷ I (DCD) for DBD and DCD, respectively). The genetic selection differentials I(DBD) and I{DCD) are both reduced by preferential treatment. But I(DBD) is smaller than I(DCD), because the mean genetic level of the DBD (p(DBD) ) is higher than that of the DCD (p(DCD) ) and the truncation point is the same across these two distributions. Thus the absolute reduction owing to preferential treatment of I(DBD) is smaller than that of I(DCD ), if preferential treatment is applied equally to DCD and DBD (Schemes 4a and 4b). This implies that/z(DBD) ÷ I ( D B D ) is less reduced than p (DCD) ÷ I(DCD ), which leads to an increased importance of DBD, when DBD and DCD are equally treated preferentially. If greater preferential treatment is exercised among DBD due to their superior pedigrees (Scheme 4c), the value of including DBD decreases substantially (i.e. from 7% in Scheme 3 to 2% in Scheme 4c). Preferential treatment was modelled in three ways, the first limiting the selection intensity possible within an age class and the second and third limiting the accuracy of selection. When DBD are included in calculations, preferential treatment reduced response by 6, 3 and 8%, respectively {0.149/0.153, 0.144/0.153 and 0.141/0.153, respectively). These calculations suggest that preferential treatment on the CB pathway cannot account for the large difference between the theoretical rates and the responses obtained in practice by progeny testing schemes. Factors such as the reduction in variance due to selection are likely to be far more important (Meyer and Smith, 1989). This study has examined the impact of DBD on estimates of theoretical rates of response. In practice, DBD contribute to the realized response since they are evaluated and selected if their estimated breeding values are high enough.
DAUGHTERSOFBULLDAMSIN CATTLEPROGENYTESTING
273
T h e i m p a c t of i n c l u d i n g D B D in c a l c u l a t i o n s o f t h e t h e o r e t i c a l r e s p o n s e has b e e n e x a m i n e d a s s u m i n g t h a t bull d a m s h a v e a n equal c h a n c e of h a v i n g proge n y of e i t h e r sex. I f reliable e m b r y o or s e m e n sexing is used in p r o g e n y t e s t i n g s c h e m e s to o b t a i n m o r e bulls p e r bull d a m or to reduce costs, t h e n u m b e r s of D B D will be r e d u c e d a n d h e n c e t h e i r beneficial effect on t h e genetic gain m a y b e c o m e negligible. T h i s s h o u l d be t a k e n into a c c o u n t w h e n c o n v e n t i o n a l proge n y t e s t i n g s c h e m e s are c o m p a r e d to b r e e d i n g s c h e m e s w h i c h use e m b r y o or s e m e n sexing. I n c e r t a i n situations, for e x a m p l e if sib t e s t i n g y o u n g bulls ( S m i t h a n d R u a n e , 1987 ), D B D m a y still be p r o d u c e d in sufficient n u m b e r s to h a v e a c o n s i d e r a b l e i m p a c t o n genetic response.
REFERENCES Colleau, J.J., 1985. Genetic improvement by ET within selection nuclei in dairy cattle. Genet. Sel. Evol., 17: 499-538. Cunningham, E.P., 1976. The use of egg transfer techniques in genetic improvement. In: L.E.A. Rowson (Editor), Proc. EEC Seminar on Egg Transfer in Cattle, pp. 345-353. Ducrocq, V. and Quaas, R.L., 1988. Prediction of genetic response to truncation selection across generations. J. Dairy Sci., 71: 2543-2553. Everett, R.W., 1984. Impact of genetic manipulation. J. Dairy Sci., 67: 2812-2818. Hinks, C.J.M., 1978. The development of nucleus herd selection programmes in dairy cattle breeding. Z. Tierz. Zuchtungsbiol., 94: 44-54. Meyer, K. and Smith, C., 1989. Comparison of theoretical and simulated equilibrium genetic response rates with progeny testing in dairy cattle. Paper in preparation. Murphy, P.A., Everett, R.W. and Van Vleck, L.D., 1982. Comparison of first and all lactations of dams to predict sons' milk evaluations. J. Dairy Sci., 65: 1999-2005. Smith, C. and Ruane, 1987. Use of sib testing as a supplement to progeny testing to improve the genetic merit of commercial semen in dairy cattle. Can. J. Anim. Sci., 67: 985-990. Van Vleck, L.D., 1977. Theoretical and actual genetic progress in dairy cattle. In: E. Pollak, O. Kempthorne and T.B. Bailey (Editors), Proc. Int. Conf. Quantitative Genetics, pp. 543-568. Van Vleck, L.D., 1986. Evaluation of dairy cattle breeding programs: specialized milk production. Proc. 3rd World Congress on Genetics Applied to Livestock Production, Lincoln, Vol. 9, pp. 141-152. RESUME Meuwissen, T.H.E. et Ruane, J., 1989. La valeur des filles des m~res ~ taureaux dans les schemas de testage des bovins laitiers. Livest. Prod. Sci., 23:267-274 (en anglais). Dans les schemas de testage des bovins laitiers, les m~res taureaux de haute valeur g~n~tique sont identifi~es et insdmin~es pour produire de jeunes taureaux. Elles donnent aussi naissance des filles qui peuvent ~ leur tour devenir des m~res ~ taureaux. La contribution de ces filles au gain g~ndtique a ~t~ sous estimde dans le passd. Quand on les inclut dans les calculs d'un schdme de testage classique, on accrolt de 2% le progr~s g~ndtique p~evu. Ce gain s'dl~ve~ 7% quand on utilise la superovulaton et le transfert des embryons (MOET) pour obtenir 10 descendants par an par m~re ~taureau. Le gain varie de 2 ~ 11% quand
274
T.H.E. MEUWISSENANDJ. RUANE
on tient du traitement prdfdrentiel des m~res h taureaux en limitant la prdcision ou l'intensitd de la s6lection. KURZFASSUNG Meuwissen, T.H.E. und Ruane, J., 1989. Die Bedeutung der T~ichter von Bullenmtittern in Nachkommenpriifprogrammen yon Milchrindern. Livest. Prod. Sci., 23:267-274 (auf englisch). In Nachkommenprtifprogrammen von Milehrindern werden Bullenmtitter mit hoher genetischer Veranlagung ausgew~ihlt und zur Erzeugung von Jungbullen angepaart. Dabei entstehen auch weibliche Nachkommen, yon denen wiederum jedes Tier selbst Bullenmutter werden kann. Die Bedeutung dieser Gruppe (TSchter von Bullenmtittern (DBD)) ftir den Zuchtfortschritt wurde bisher untersch~itzt. Der vorausgescha~itzte Zuchtfortschritt stieg um 2%, wenn die DBD in den Berechnungen ftir ein konventionellesNachkommenprtifprogramm beriicksichtigt wurden. Mit Einsatz von MOET (Multiple Ovulation und Embryo Transfer) bei Bullenmiittern zur Erzeugung von 10 Nachkommen pro Mutter und Jahr betrug diese Differenz 7%. Bei Berticksichtigung von Vorzugsbehandlungen yon Bullenmtittern durch Beschr~inkung der Genauigkeit der Selektion oder der Selektionsintensit~it, fiihrte die Einbeziehung yon DBD zu einem Zuchtfortschritt von 2 his 11%.
267
Elsevier Science Publishers B.V., Amsterdam - - Printed in The Netherlands
The Value of Daughters of Bull Dams in Dairy Cattle Progeny Testing Schemes T.H.E. MEUWISSEN 1 and J. RUANE 2
'Research Institute for Animal Production "Schoonoord", P.O. Box 501, 3700 AM Zeist (The Netherlands) 2Institute of Animal Physiology and Genetics Research, West Mains Road, Edinburgh EH9 3JQ (Gt. Britain) (Accepted for publication 22 May 1989 )
ABSTRACT Meuwissen, T.H.E. and Ruane, J., 1989. The value of daughters of bull dams in dairy cattle progeny testing schemes. Livest. Prod. Sci., 23: 267-274. In dairy cattle progeny testing schemes, bull dams of high genetic merit are identified and bred to produce young bulls. Females will also be produced and any of them may themselves become bull dams. The impact of these (daughters of bull dams (DBD)) on genetic gain has been underestimated in the past. The predicted rate of genetic gain increased by 2%, when the DBD were included in the calculations for a conventional progeny testing scheme. When MOET (multiple ovulation and embryo transfer) was used on bull dams to produce 10 offspring dam -1 year-', this difference was 7%. When accounting for preferential treatment of bull dams by limiting the accuracy of selection or the selective intensity, including DBD increased genetic gain by 2-11%.
INTRODUCTION
In dairy cattle breeding genetic progress is made along four pathways: selection of bulls to breed bulls (BB), selection of bulls to breed cows (BC), selection of cows to breed bulls (CB) and selection of cows to breed cows (CC). About 76% of the total genetic gain is made in selecting the parents of males (i.e. on the BB and CB pathways) and 24% in selecting the parents of females (Everett, 1984). However, the matings of bull sires and bull dams will on average produce equal numbers of males and females. The daughters of bull dams (DBD) born have a higher genetic mean than the daughters of cow dams (DCD) born at the same time. With the scheme in steady-state equilibrium, the difference in means ( D B D - D C D ) is 0301-6226/89/$03.50
© 1989 Elsevier Science Publishers B.V.
268
IBB -
T.H.E. MEUWISSENAND J, RUANE
zIGLBB+_ICB - AGLcB~ (Icc - AGLcc +_IBC 2
/
\
ZGLBc
/
2
where AG, I and L represent the genetic gain year-1, the genetic selection differential and the generation interval, respectively. In a typical progeny testing scheme this difference may represent 3.5AG to 4.5riG (e.g. 3.9AGin the example of Table 1 ). The value of DBD has previously been underestimated, because the usual pooling of the DBD and the DCD population into one single population of potential bull dams leads to an underestimation of the contribution of the best population to the selected part. The existence of DBD has previously been ignored in calculations of genetic response, because their number is small. The contribution of the DBD to the CB pathway may be significant, especially when the number of selected bull dams is small owing to the use of MOET (multiple ovulation and embryo transfer). Many authors (e.g. Cunningham, 1976) have shown that by using MOET on bull dams, selection on the CB pathway can be more intense and hence response can be increased by up to 10%. A consequence of this strategy, which has not been examined, is that the DBD born are of increased genetic merit which may further increase the genetic response. The genetic merit of selected bull dams is usually lower than expected. This is thought to be due, in part at least, to preferential treatment of high yielding cows (Van Vleck, 1977). The aim of this study is to investigate the importance of the DBD in a progeny testing scheme, with or without using MOET on bull dams and to study TABLE 1 Predicted genetic gain (ziG) in a progeny testing scheme w i t h o u t t h e use of M O E T a n d daughters of bull dams ( D B D ) Path
Age class
Selected fraction (%)
Selection accuracy
Contribution 1 of age class (%)
Genetic 2 selection differential (S .D. units )
Generation interval (years) 2
BB BC
6 2 6 4 5 6 7 2-7
2 100 5 1.3 1.1 0.6 0.2 100
0.933 0.0 0.933 0.5 0.6 0.65 0.675 0.0
100 20 80 54.1 31.2 11.5 3.2 100
1.13 0.77
6 5.2
0,74
4.6
0.0
4
CB
CC
ziG = ( 1.13 + 0.77 + 0.74 ) / ( 6 + 5.2 + 4.6 + 4 ) = 0.133 phenotypic s t a n d a r d deviations year - 1. 1The contributions of the age classes sum to 100% within each path. 2These mean values are weighted by the contributions of the age classes within a path.
DAUGHTERS OF BULL DAMS IN CATTLE PROGENY TESTING
269
the effect of preferential treatment on the importance of DBD. The impact of DBD on the CC pathway is neglected here, because many cows have to be selected for this pathway and so the contribution of DBD can only be small. MATERIALSAND METHODS To predict the steady-state genetic gain of a progeny testing scheme all four selection pathways have to be included, although our interest is only in the CB pathway. Consider the following breeding plan shown in Table 1. Annually 100 young bulls are progeny tested on 100 daughter records each. Without the use of M O E T about six bull dams are needed in practice to produce each young bull (Hinks, 1978 ). The generation interval of the selected bulls is 6 years. The cow dams are assumed to be selected at random with respect to the breeding goal, giving a generation interval of about 4 years. Some 50 000 heifers enter the cow population each year. Culling of cows is taken to be independent of the breeding goal and is 30% annually of the total number. Thus for example there are 35 000 3-year-old cows. The total cow population amounts to 167 000 cows. The trait selected for is milk production, which is assumed to have a heritability of 0.25 and a repeatability between lactations of 0.4. The genetic correlations between lactations are 1, unless stated otherwise. Modern sire and cow evaluation methods make it possible to compare the estimated breeding values of animals of different ages. Thus the best cows can be selected irrespective of their age. In the present breeding plan the bull dams are selected from Age Classes 4, 5, 6 and 7 (the age class number is the age at birth of their progeny), being selected on one, two, three and four lactation records, respectively. Each age class has a distribution of estimated breeding values. The means of these distributions differ due to genetic progress (e.g. the mean of Age Class 2 is zig higher than the mean of Age Class 3) and the variances differ due to the amount of information that is available. Ducrocq and Quaas (1988) described an algorithm to predict the fractions selected from each distribution if truncation selection is applied across several distributions. Because the differences in mean between the distributions are determined by the genetic progress, the steady-state genetic gain is calculated iteratively using this algorithm. If bull dams can also be selected from the DBD, there are four more distributions to select from, i.e. the distributions of estimated breeding values of the DBD of Age Classes 4, 5, 6 and 7. Genetic progress is calculated both with and without DBD included for the following schemes. For each scheme the total number of young bulls and DBD produced is constant. (1) No MOET used on bull dams. Six hundred bull dams are selected with one offspring each. (2) One flush per bull dam. Two hundred bull dams are selected with three offspring bull dam-1 year-1. This procedure is common practice among cer-
270
T.H.E. MEUWISSEN AND J. RUANE
tain AI organizations in order to increase the probability of producing a male calf. Only the female's own records are used for evaluation and information on full sibs is ignored. (3) Repeated use of M O E T on bull dams. Sixty bull dams are selected with 10 offspring bull d a m - 1year- 1. Each DBD is thus evaluated on her own record and those of four full sibs. This scheme may be applied to sib test the young bulls (Smith and Ruane, 1987). {4) This is the same as Scheme 3 except that preferential treatment of potential bull dams is considered. The effects of preferential treatment are modelled in three ways (Schemes 4a, 4b and 4 c ) . In Scheme 4a it is assumed that all the "best" cows are treated preferentially, so that it is not possible to select the "very best" cows among the "best" cows. This situation is modelled by setting a limit on the maximum selection differential possible within an age class. The maximum standardized selection differential is set to 2.421, which corresponds to a fraction selected of 2%. For example, when the proportion of animals selected from Age Class 7 is 0.1%, the genetic merit of the selected cows in this Age Class 7 becomes ~(7) +2.421rIo(7)ao instead o f ~ ( 7 ) + 3 . 3 6 7 r I v ( 7 ) a G (where ]~(7), riG(7), a c are the mean of Age Class 7, the accuracy of selection of Age Class 7 and the genetic standard deviation of milk production). In Scheme 4b it is assumed that the accuracy of the bull dam selection is not increased by including any records other than from the first lactation, as suggested by Van Vleck (1986). The reason for this is that cows seem to be treated evenly during the first lactation but preferentially thereafter (Murphy et al., 1982 ). This situation is modelled by at first assuming that the phenotypic variances and covariances are unaffected by preferential treatment. Second, it is assumed that the evaluation method does not account for preferential treatment, i.e. the weighting factors for the estimated breeding values (EBV) are unchanged. These assumptions imply that the variances of the EBV are not affected by preferential treatment. Third, the genetic correlations between the first lactation and the second, third and fourth lactation record are assumed to be reduced to 0.67, 0.65 and 0.64, respectively. These correlations are chosen such that rig ( 4 ) = rig ( 5 ) = r~G(6) = rra ( 7 ). The breeding goal is first lactation milk yield. Adult lactations are not ignored by the breeding value estimation procedure because it is not known to the breeders how preferential treatment affects the records of the cows. Scheme 4c is the same as Scheme 4b except that it is assumed that DBD are treated also preferentially during the first lactation (based on their pedigree). It is assumed that the accuracies of the EBV of DBD are reduced by 30%. The accuracies of the EBV of DCD are the same as in Scheme 4b. It is assumed that the population is large enough to neglect the effects of reduced selection differentials due to finite population size and inbreeding. In addition, the reduction in variance due to selection is ignored. This effect is
DAUGHTERSOF BULLDAMSIN CATTLEPROGENYTESTING
271
TABLE 2 The predicted genetic gain {AG) and the fraction of bull dams that are themselves daughters of bull dams (PDBD) for Schemes 1, 2, 3, 4a, 4b and 4c Scheme
1 2 3 4a 4b 4c
DBD ignored AG (S.D. year -1)
0.133 0.138 0.143 0.130 0.138 0.138
DBD included
AG (S.D. year -1)
PDBD (%)
0.135 0.141 0.153 0.144 0.149 0.141
14 30 80 65 74 60
( + 1.5%) (+2%) (+7%) ( + 11% ) ( +8% ) ( +2% )
partly avoided by only selecting once on the same information source. For example, DBD are selected on their own performance (and that of their contemporary full sibs) and not on the performance of their dam, because their dams were already selected on own performance or DBD are not selected on half-sib records, because their sires were selected on progeny records. RESULTS
Table I gives the contribution of the four paths and of the age classes within the paths to the selection response of Scheme 1. The steady-state genetic response in this situation is 0.133 phenotypic standard deviations year-1. In Table 2 the use of DBD is included. By selecting fewer bull dams the annual genetic gain can be increased from 0.133 up to 0.143 S.D. units. This increase of 8% is in agreement with previous results (e.g. Cunningham, 1976). The response is further increased by including DBD in our calculations of genetic gain. Without MOET, this increase is 1.5%; with intense MOET (10 offspring) it is 7%. The influence of the DBD increases as the number of bull dams selected decreases. Thus combining these two features for Scheme 3, response is increased by 13% (0.155/0.135) by selecting 60 bull dams to produce the young bulls needed instead of 600. Table 2 also shows the impact that including DBD has when preferential treatment is present. In Schemes 4a, 4b and 4c the genetic response is increased by 11, 8 and 2%, respectively. DISCUSSION
The genetic response predicted when including DBD in calculations on the CB pathway in a progeny testing scheme where M O E T is not used is only slightly higher than expected if they are excluded (Scheme 1, Table 2). How-
272
T.H.E. MEUWISSEN AND J. RUANE
ever, through using MOET a smaller group of highly superior bull dams can be selected, who produce daughters of higher average genetic merit. In addition the breeding values of these daughters can be estimated with a higher accuracy because of the full-sib information available. As a consequence, a greater number of bull dams selected are themselves DBD and so they have a greater impact on genetic progress. For example, in Scheme 3 80% of the bull dams were DBD. Apart from the generation interval of the bull dams, the progeny testing scheme approaches the MOET-Hybrid schemes described by Colleau (1985), where all bull dams are DBD. If the young age classes are also included for selection, the generation intervals of the bull dams in the progeny testing and MOETHybrid schemes will be more alike. This also increases the genetic gain of the progeny testing scheme, since the number of selection candidates increases. Thus, using MOET on bull dams in a progeny testing scheme has a greater effect on genetic progress than previously considered. This should also lead to increased inbreeding rates owing to the reduction of effective population size, since the BB and DBD are closely related. The genetic merit of bull dams selected is equal to the genetic mean of the group from which they are selected plus the genetic selection differential (/~(DBD) ÷ I (DBD) or/~ (DCD) ÷ I (DCD) for DBD and DCD, respectively). The genetic selection differentials I(DBD) and I{DCD) are both reduced by preferential treatment. But I(DBD) is smaller than I(DCD), because the mean genetic level of the DBD (p(DBD) ) is higher than that of the DCD (p(DCD) ) and the truncation point is the same across these two distributions. Thus the absolute reduction owing to preferential treatment of I(DBD) is smaller than that of I(DCD ), if preferential treatment is applied equally to DCD and DBD (Schemes 4a and 4b). This implies that/z(DBD) ÷ I ( D B D ) is less reduced than p (DCD) ÷ I(DCD ), which leads to an increased importance of DBD, when DBD and DCD are equally treated preferentially. If greater preferential treatment is exercised among DBD due to their superior pedigrees (Scheme 4c), the value of including DBD decreases substantially (i.e. from 7% in Scheme 3 to 2% in Scheme 4c). Preferential treatment was modelled in three ways, the first limiting the selection intensity possible within an age class and the second and third limiting the accuracy of selection. When DBD are included in calculations, preferential treatment reduced response by 6, 3 and 8%, respectively {0.149/0.153, 0.144/0.153 and 0.141/0.153, respectively). These calculations suggest that preferential treatment on the CB pathway cannot account for the large difference between the theoretical rates and the responses obtained in practice by progeny testing schemes. Factors such as the reduction in variance due to selection are likely to be far more important (Meyer and Smith, 1989). This study has examined the impact of DBD on estimates of theoretical rates of response. In practice, DBD contribute to the realized response since they are evaluated and selected if their estimated breeding values are high enough.
DAUGHTERSOFBULLDAMSIN CATTLEPROGENYTESTING
273
T h e i m p a c t of i n c l u d i n g D B D in c a l c u l a t i o n s o f t h e t h e o r e t i c a l r e s p o n s e has b e e n e x a m i n e d a s s u m i n g t h a t bull d a m s h a v e a n equal c h a n c e of h a v i n g proge n y of e i t h e r sex. I f reliable e m b r y o or s e m e n sexing is used in p r o g e n y t e s t i n g s c h e m e s to o b t a i n m o r e bulls p e r bull d a m or to reduce costs, t h e n u m b e r s of D B D will be r e d u c e d a n d h e n c e t h e i r beneficial effect on t h e genetic gain m a y b e c o m e negligible. T h i s s h o u l d be t a k e n into a c c o u n t w h e n c o n v e n t i o n a l proge n y t e s t i n g s c h e m e s are c o m p a r e d to b r e e d i n g s c h e m e s w h i c h use e m b r y o or s e m e n sexing. I n c e r t a i n situations, for e x a m p l e if sib t e s t i n g y o u n g bulls ( S m i t h a n d R u a n e , 1987 ), D B D m a y still be p r o d u c e d in sufficient n u m b e r s to h a v e a c o n s i d e r a b l e i m p a c t o n genetic response.
REFERENCES Colleau, J.J., 1985. Genetic improvement by ET within selection nuclei in dairy cattle. Genet. Sel. Evol., 17: 499-538. Cunningham, E.P., 1976. The use of egg transfer techniques in genetic improvement. In: L.E.A. Rowson (Editor), Proc. EEC Seminar on Egg Transfer in Cattle, pp. 345-353. Ducrocq, V. and Quaas, R.L., 1988. Prediction of genetic response to truncation selection across generations. J. Dairy Sci., 71: 2543-2553. Everett, R.W., 1984. Impact of genetic manipulation. J. Dairy Sci., 67: 2812-2818. Hinks, C.J.M., 1978. The development of nucleus herd selection programmes in dairy cattle breeding. Z. Tierz. Zuchtungsbiol., 94: 44-54. Meyer, K. and Smith, C., 1989. Comparison of theoretical and simulated equilibrium genetic response rates with progeny testing in dairy cattle. Paper in preparation. Murphy, P.A., Everett, R.W. and Van Vleck, L.D., 1982. Comparison of first and all lactations of dams to predict sons' milk evaluations. J. Dairy Sci., 65: 1999-2005. Smith, C. and Ruane, 1987. Use of sib testing as a supplement to progeny testing to improve the genetic merit of commercial semen in dairy cattle. Can. J. Anim. Sci., 67: 985-990. Van Vleck, L.D., 1977. Theoretical and actual genetic progress in dairy cattle. In: E. Pollak, O. Kempthorne and T.B. Bailey (Editors), Proc. Int. Conf. Quantitative Genetics, pp. 543-568. Van Vleck, L.D., 1986. Evaluation of dairy cattle breeding programs: specialized milk production. Proc. 3rd World Congress on Genetics Applied to Livestock Production, Lincoln, Vol. 9, pp. 141-152. RESUME Meuwissen, T.H.E. et Ruane, J., 1989. La valeur des filles des m~res ~ taureaux dans les schemas de testage des bovins laitiers. Livest. Prod. Sci., 23:267-274 (en anglais). Dans les schemas de testage des bovins laitiers, les m~res taureaux de haute valeur g~n~tique sont identifi~es et insdmin~es pour produire de jeunes taureaux. Elles donnent aussi naissance des filles qui peuvent ~ leur tour devenir des m~res ~ taureaux. La contribution de ces filles au gain g~ndtique a ~t~ sous estimde dans le passd. Quand on les inclut dans les calculs d'un schdme de testage classique, on accrolt de 2% le progr~s g~ndtique p~evu. Ce gain s'dl~ve~ 7% quand on utilise la superovulaton et le transfert des embryons (MOET) pour obtenir 10 descendants par an par m~re ~taureau. Le gain varie de 2 ~ 11% quand
274
T.H.E. MEUWISSENANDJ. RUANE
on tient du traitement prdfdrentiel des m~res h taureaux en limitant la prdcision ou l'intensitd de la s6lection. KURZFASSUNG Meuwissen, T.H.E. und Ruane, J., 1989. Die Bedeutung der T~ichter von Bullenmtittern in Nachkommenpriifprogrammen yon Milchrindern. Livest. Prod. Sci., 23:267-274 (auf englisch). In Nachkommenprtifprogrammen von Milehrindern werden Bullenmtitter mit hoher genetischer Veranlagung ausgew~ihlt und zur Erzeugung von Jungbullen angepaart. Dabei entstehen auch weibliche Nachkommen, yon denen wiederum jedes Tier selbst Bullenmutter werden kann. Die Bedeutung dieser Gruppe (TSchter von Bullenmtittern (DBD)) ftir den Zuchtfortschritt wurde bisher untersch~itzt. Der vorausgescha~itzte Zuchtfortschritt stieg um 2%, wenn die DBD in den Berechnungen ftir ein konventionellesNachkommenprtifprogramm beriicksichtigt wurden. Mit Einsatz von MOET (Multiple Ovulation und Embryo Transfer) bei Bullenmiittern zur Erzeugung von 10 Nachkommen pro Mutter und Jahr betrug diese Differenz 7%. Bei Berticksichtigung von Vorzugsbehandlungen yon Bullenmtittern durch Beschr~inkung der Genauigkeit der Selektion oder der Selektionsintensit~it, fiihrte die Einbeziehung yon DBD zu einem Zuchtfortschritt von 2 his 11%.