- Email: [email protected]
Cow-Indexing Procedure and Culling Levels for Potential Dams of Dairy Bulls
Cow-Indexing Procedure and Culling Levels for Potential Dams of Dairy Bulls E. B. BURNSIDE
Department of Animal Science, University of Guelph, Guelph, Ontario, Canada Abstract
A procedure to index dairy cows on milk production is described. First records of cows were compared to first records of contemporaries, and second and later records of cows to second and later contemporary records. The index combined this information with the proof of the cow's sire. Of 5,704 Guernsey cows, 144 had indices two or more standard deviations above the average index. Restrictions on the milk fat percentage of these "potential" dams of bulls caused the numbers to drop rapidly. Restrictions on type classification decreased numbers to a lesser degree. Mean indices suggested that cows classified "Excellent" received preferential treatment.
Introduction
Methods of indexing dairy cows for milk production have been i n use in the United States, New Zealand, Canada, and possibly other countries for several years. Canadian procedures involve the computation of an index for milk yield based on the cow's own records, each expressed as a difference from a contemporary average and the first-lactation progeny proof of the cow's sire. Breeders and artificial insemination sire-purchasing personnel consider each cow's type classification and milk fat per cent as well as milk yield in searching for potential dams of young bulls. The purpose of this paper is to describe a method used to index cows for milk yield in Canada and to show the number of cows with high milk indices which survive culling when restrictions are set on other traits. Data and Methods
Breed (Breed Class Average, or BCA). To be included in this study, a cow must have completed one or more records before IV[arch 2, 1967, and, in addition, have calved as recently as September 1, 1965. b) Sire progeny proofs for milk. These were used in the calculations when based on five or more "effective daughters." Sire proofs for milk production were taken directly from the latest official ROP milk ratings (3) which were based on first-lactation contemporary comparisons. Although a cow's first record with contemporaries contributed to her site's proof, for simplicity index weights for the cow's records and the site's proof were computed on the assumption that her first record was excluded from the proof. Sires were required to have at least five effective daughters in their proofs, thus variation in weights nnder the alternative assumptions would be small in any case. Type classification information, if available for the cow, was taken from the latest official type classification of the cow in the period January, 1953, to June, :1967. Type classification information was not considered in the calculation of the milk index. Computation of the index. The difference between the cow's records and the average of contemporaries (i.e., daughters of other bulls calving in the same herd-year-season) was computed. Each cow's first-lactation record was compared to the average of her first-lactation contemporaries, and second and later lactations were compared to the average of second and later lactation contemporaries. The difference between the cow's record and the first or second and later lactation contemporary average was adjusted for varying nmnbers of contemporaries by the formula: Adjusted contemporary difference ~-
Data. A total of 5,704 Guernsey cows was indexed using : a) Official Record of Performance (I~OP) production records. These production records were expressed on a twice-daily 305-day basis and then were age-adjusted by expressing each record as a percentage of the appropriate ageof-calving average production for the Guernsey Received for publication August 11, 1969.
Number of cont [Cow's BCA for milk - - ~ Number of c o n t + 1 \ C o n t avg BCA for m i l k ] The basis for this adjustment is the weighting of the actual difference for the inverse of its variance. This adjustment regresses differences based on small numbers of contemporaries towards zero to a more pronounced degree than the adjustment for number of herdmates used in the USDA Cow Index procedure (1).
195
196
BURNSIDE
720
600 -0
480 N ~ 360
240 --
120--
-20
,
-15
I
-10
-5
0 5 10 MILK INDEX Fro. 1. D i s t r i b u t i o n o f m i l k index of 5,704 Guernsey cows.
This difference was further adjusted for level of production in the herd in which the cow made each record as follows: F i n a l contemporary difference ---- adjusted contemporary difference 1~ Cont avg BCA for milk - • ~Breed season avg BCA for m i l k / This adjustment is more important in indexing cows than in dealing with sire proofs, since there is much greater variation in the contemporary average between cows in different herds than between sire proofs, even in different areas. No estimate of the regression of contemporary average on breed season average was available from Canadian data, and a value of 0.1 was assumed. The milk index was computed by the formula : Index ---- W I Avg of cow's final cont differences + W2 Proof of the cow's sire: where weights W1 and W2 are functions of a) the number of records the cow has completed b) the number of first lactation daughters included in the sire's progeny average c) the heritability of milk records (assumed = .25) d) the repeatability of milk records (assumed = .45) JOURNAL OF DAIRY SCIENCE VOL, 53, NO. 2
15
20
The formulae for W1 and W2 were derived by the method of path coefficients and are similar to those derived by Miller (4). I f important environmental correlations exist among paternal half-sibs (2, 5, 6), the weights may be in error, particularly in cases where the sire of the cow had a nonartificial insemination or single herd proof. Results and DiscussionM Examination of the Index
The distribution of the index, expressed in units of BCA for milk, is shown (Fig. 1). TABLE 1. Table of means and standard deviations. All cows calving since September 1, 1965 (5,704 cows) Mean SD Milk index (BCA) .95 5.78 Per cent fat 4.86 0.41 Type-classified cows calving since September 1, 1965 (3,677 cows) Mean Milk index (BCA) 1.27 Per cent fat 4.88 Type class score 81.1
SD 5.95 0.42 4.2
197
COW I N D E X I N G A N D C U L L I N G
TABLE 2. Milk index and type score. Final type class Poor Fair Good Good plus Very good Excellent Not classified
Number of cows 3 164 1,073 1,519 827 91 %027
Mean milk index (BCA) --1.67 0.43 0.32 1.26 2.26 5.06 0.38
Means for the index and milk fat per cent and standard deviations were computed for the 5,704 Guernsey cows indexed, as well as for the 3,677 of these cows that had also been type classified (Table 1). The mean index was slightly positive for both groups of cows, indicating that breeders were putting some emphasis on milk yield in culling cows or in selecting sires. The mean milk indices and milk fat per cent of classified cows were slightly but not significantly higher than those of all cows. This could easily be attributed to chance or to herd differences. Mean milk indices were computed by classification rating for the 3,677 cows classified and for the 2,027 cows not classified (Table 2). The number of "Poor" cows was insufficient to set a mean index. The index which, at least in part, is dependent on a within-herd comparison of
the cow's records with contemporaries, did rise slightly (average of +.61 BCA/interval) as classification increased from "Fair" through "Very Good." The mean change in the index was 2.80 BCA points in going from "Very Good" to "Excellent" cows. This increase may be due in part to special treatment of "Excellent" cows, an area that requires more study by researchers. Adjustment of indices of "Excellent" cows may be justified, if part or all of the difference observed can be shown to be due to preferential treatment. Genetic progress for milk yield is quite dependent on the genetic superiority of cows selected to be the dams of bulls. The effects of various restrictions for type classification and fat per cent on very high producing groups of cows were studied (Table 3). I t was assumed that the cows should have an index two standard deviations above the average index for milk, if they were to be considered as dams of bulls. Accordingly, a minimum index of +12.3 BCA points was required of all 5,704 cows indexed and 144 cows survived this initial restriction as prospective dams. Few artificial insemination studs in the United States or Canada would consider entering the son of an unclassified cow. Requiring the cow to be classified caused a drop from 144 to 119 prospects. When the final grade of type classification was required to be "Good Plus" or higher, 105 cows survived, as compared to 53 surviving prospects when the cow was required to grade "Very Good" or
TABLE 3. Effect of restrictions on type and test on cows indexing two or more standard deviations above breed average for milk. Average Restrictions
Number of cows
Index (B CA)
Type score
All cows All type-classified cows Type GP or better % F a t 4.10 % F a t 4.50 % F a t 4.90
144 119 105 96 59 21
15.9 16.0 16.0 16.0 15.4 t5.4
4.57 4.58 4.57 4.63 4.83 5.13
...... 83.6 84.5 84.6 84.5 85.1
Type YG or better % F a t 4.10 % F a t 4.50 % F a t 4.90
53 48 27 12
16.9 17.0 16.3 16.2
4.53 4.59 4.82 5.07
87.1 87.2 87.4 87.1
Type E X % F a t 4.10 % F a t 4.50 % F a t 4.90
10 9 6 2
17.7 18.0 17.5 17.7
4.53 4.60 4.82 5.20
90.6 90.7 90.6 90.8
Fat
(%)
JOURNAL OF D ~ Y
S C ~ N 0 ~ ~trOn. 53, NO, 2
] 98
BURNSIDE
TABLE 4. Effect of restrictions on type and test on cows indexing one or more standard deviations above breed average for milk. Average Restrictions
Number of cows
Index
All cows All type-classified cows Type GP or better % F a t 4.10 % F a t 4.50 % F a t 4.90 Type VG or better % F at 4.10 % F a t 4.50 % F a t 4.90 Type E X % F a t 4.10 % F a t 4.50 % F a t 4.90
818 587 454 433 329 164 215 205 160 89 36 35 30 14
10.5 10.7 10.9 10.8 10.3 9.8 11.3 11.2 10.6 10.2 12.1 12.0 11.5 10.9
Fat
Type score
(%)
higher and only ten prospects when the cow was required to grade "Excellent." Restrictions on average milk fat per cent of the cow were set at the breed average (4.9%), at one (4.5%), and at two (4.1%) standard deviations below the breed average. I f one decided that the dam of a bull must grade "Good Plus" or "Better" and, of course, be two standard deviations above the breed average in milk index, 105 cows were available. This number dropped to 96 if an average milk fat per cent of 4.1% was desired, to 59 i£ an average of 4.5% was desired, and to 21 if the cow was required to have an average milk fat test equal to the breed average. The limits set on milk fat per cent were relatively low, yet they removed large numbers of cows. A negative genetic correlation between milk fat per cent and milk index could be a major cause of these results. Similar results were obtained when the milk index was set at -b6.6 B C A points, one standard deviation above the breed average, and the other traits were studied as before (Table 4). A lower limit of one standard deviation above breed average for the milk index of dams of bulls would not lead to rapid genetic change, but it may be higher than that currently set by many artificial insemination studs. Studs tend to set fairly high limits of milk fat per cent on the dams of bulls, as well as reasonably high limits on milk, milk fat per cent, and type of the dam of the cow and other female relatives in the pedigree. In view of the recent results which suggest that relatively low emphasis should be JOURNAL OF I)AIlC¥ SCIEI~CE VOL. 53, NO. 2
4.68 4.69 4.70 4.74 4.88 5.12 4.72 4.76 4.89 5.10 4.73 4.75 4.84 5.09
.
.
.
.
.
.
82.5 84.3 84.4 84.5 84.7 87.0 87.0 87.1 86.9 90.0 91.0 91.0 91.3
given to milk fat per cent in selecting cows and sires (7, 8), it is essential that dairy cattle breeders use their influence to convince sire analysts and sire-purchasing committees: a) to avoid putting undue emphasis on milk fat per cent in selecting dams of bulls; b) to start with lists of potential dams that index high f o r milk (preferably two or more standard deviations above breed averages) and select cows only from these lists; c) to carefully explore unusual herd management practices which pertain to cows grading "Excellent" in final classification. Preferential treatment may have biased the average index values of "Excellent" cows in this study upward by as much as 0.4 of a standard deviation. Acknowledg ments
The author acknowledges the critical comments of Dr. J. W. Wilton, the programming assistance of Miss J. Longmuir, the technical assistance of Mr. John Creeden, and the financial assistance of the Ontario Department of Agriculture and Food. References
(1) Agricultural Research Service, USDA. 1969. D I t I A Cow Performance Index List. ARS 44-209. (2) Bereskin, B., and J. L. Lush. 1965. Genetic and environmental factors in dairy sire evaluation. III. Influence of environmental and other extraneous correlations among the daughters. J. Dairy Sci., 48: 356. (3) Canada Department of Agriculture. 1967.
COW I N D E X I N G The seventeenth report on sire appraisal. Production and Marketing Branch, Ottawa. (4) Miller, R. H. 1964. Indexing Progeny of A.L Sires. Unpublished mimeograph. (5) Thomson, G. M. 1969. Environmental correlations and selection of sires for use in artificial breeding. J. Dairy Sol., 5 2 : 9 2 7 . (6) Van Vleck, L. D. 1966. Environmental cor-
AND
CULLING
199
relations in sire evaluation. J. Dairy Sei., 49 : 86. (7) Wilton, J. W., a n d L. D. V a n ¥1eck. 1968. Selection of dairy cows for economic merit. J. Dairy Sci., 51: 1680. (8) Wilton, J. W., and L. D. Y a n ¥1eck. 1969. Sire evaluation for economic merit. J. Dairy Sci., 52 : 235.
JOURNAL 0¥ DAIRY SCIENOE rOD. 5~, NO.
Department of Animal Science, University of Guelph, Guelph, Ontario, Canada Abstract
A procedure to index dairy cows on milk production is described. First records of cows were compared to first records of contemporaries, and second and later records of cows to second and later contemporary records. The index combined this information with the proof of the cow's sire. Of 5,704 Guernsey cows, 144 had indices two or more standard deviations above the average index. Restrictions on the milk fat percentage of these "potential" dams of bulls caused the numbers to drop rapidly. Restrictions on type classification decreased numbers to a lesser degree. Mean indices suggested that cows classified "Excellent" received preferential treatment.
Introduction
Methods of indexing dairy cows for milk production have been i n use in the United States, New Zealand, Canada, and possibly other countries for several years. Canadian procedures involve the computation of an index for milk yield based on the cow's own records, each expressed as a difference from a contemporary average and the first-lactation progeny proof of the cow's sire. Breeders and artificial insemination sire-purchasing personnel consider each cow's type classification and milk fat per cent as well as milk yield in searching for potential dams of young bulls. The purpose of this paper is to describe a method used to index cows for milk yield in Canada and to show the number of cows with high milk indices which survive culling when restrictions are set on other traits. Data and Methods
Breed (Breed Class Average, or BCA). To be included in this study, a cow must have completed one or more records before IV[arch 2, 1967, and, in addition, have calved as recently as September 1, 1965. b) Sire progeny proofs for milk. These were used in the calculations when based on five or more "effective daughters." Sire proofs for milk production were taken directly from the latest official ROP milk ratings (3) which were based on first-lactation contemporary comparisons. Although a cow's first record with contemporaries contributed to her site's proof, for simplicity index weights for the cow's records and the site's proof were computed on the assumption that her first record was excluded from the proof. Sires were required to have at least five effective daughters in their proofs, thus variation in weights nnder the alternative assumptions would be small in any case. Type classification information, if available for the cow, was taken from the latest official type classification of the cow in the period January, 1953, to June, :1967. Type classification information was not considered in the calculation of the milk index. Computation of the index. The difference between the cow's records and the average of contemporaries (i.e., daughters of other bulls calving in the same herd-year-season) was computed. Each cow's first-lactation record was compared to the average of her first-lactation contemporaries, and second and later lactations were compared to the average of second and later lactation contemporaries. The difference between the cow's record and the first or second and later lactation contemporary average was adjusted for varying nmnbers of contemporaries by the formula: Adjusted contemporary difference ~-
Data. A total of 5,704 Guernsey cows was indexed using : a) Official Record of Performance (I~OP) production records. These production records were expressed on a twice-daily 305-day basis and then were age-adjusted by expressing each record as a percentage of the appropriate ageof-calving average production for the Guernsey Received for publication August 11, 1969.
Number of cont [Cow's BCA for milk - - ~ Number of c o n t + 1 \ C o n t avg BCA for m i l k ] The basis for this adjustment is the weighting of the actual difference for the inverse of its variance. This adjustment regresses differences based on small numbers of contemporaries towards zero to a more pronounced degree than the adjustment for number of herdmates used in the USDA Cow Index procedure (1).
195
196
BURNSIDE
720
600 -0
480 N ~ 360
240 --
120--
-20
,
-15
I
-10
-5
0 5 10 MILK INDEX Fro. 1. D i s t r i b u t i o n o f m i l k index of 5,704 Guernsey cows.
This difference was further adjusted for level of production in the herd in which the cow made each record as follows: F i n a l contemporary difference ---- adjusted contemporary difference 1~ Cont avg BCA for milk - • ~Breed season avg BCA for m i l k / This adjustment is more important in indexing cows than in dealing with sire proofs, since there is much greater variation in the contemporary average between cows in different herds than between sire proofs, even in different areas. No estimate of the regression of contemporary average on breed season average was available from Canadian data, and a value of 0.1 was assumed. The milk index was computed by the formula : Index ---- W I Avg of cow's final cont differences + W2 Proof of the cow's sire: where weights W1 and W2 are functions of a) the number of records the cow has completed b) the number of first lactation daughters included in the sire's progeny average c) the heritability of milk records (assumed = .25) d) the repeatability of milk records (assumed = .45) JOURNAL OF DAIRY SCIENCE VOL, 53, NO. 2
15
20
The formulae for W1 and W2 were derived by the method of path coefficients and are similar to those derived by Miller (4). I f important environmental correlations exist among paternal half-sibs (2, 5, 6), the weights may be in error, particularly in cases where the sire of the cow had a nonartificial insemination or single herd proof. Results and DiscussionM Examination of the Index
The distribution of the index, expressed in units of BCA for milk, is shown (Fig. 1). TABLE 1. Table of means and standard deviations. All cows calving since September 1, 1965 (5,704 cows) Mean SD Milk index (BCA) .95 5.78 Per cent fat 4.86 0.41 Type-classified cows calving since September 1, 1965 (3,677 cows) Mean Milk index (BCA) 1.27 Per cent fat 4.88 Type class score 81.1
SD 5.95 0.42 4.2
197
COW I N D E X I N G A N D C U L L I N G
TABLE 2. Milk index and type score. Final type class Poor Fair Good Good plus Very good Excellent Not classified
Number of cows 3 164 1,073 1,519 827 91 %027
Mean milk index (BCA) --1.67 0.43 0.32 1.26 2.26 5.06 0.38
Means for the index and milk fat per cent and standard deviations were computed for the 5,704 Guernsey cows indexed, as well as for the 3,677 of these cows that had also been type classified (Table 1). The mean index was slightly positive for both groups of cows, indicating that breeders were putting some emphasis on milk yield in culling cows or in selecting sires. The mean milk indices and milk fat per cent of classified cows were slightly but not significantly higher than those of all cows. This could easily be attributed to chance or to herd differences. Mean milk indices were computed by classification rating for the 3,677 cows classified and for the 2,027 cows not classified (Table 2). The number of "Poor" cows was insufficient to set a mean index. The index which, at least in part, is dependent on a within-herd comparison of
the cow's records with contemporaries, did rise slightly (average of +.61 BCA/interval) as classification increased from "Fair" through "Very Good." The mean change in the index was 2.80 BCA points in going from "Very Good" to "Excellent" cows. This increase may be due in part to special treatment of "Excellent" cows, an area that requires more study by researchers. Adjustment of indices of "Excellent" cows may be justified, if part or all of the difference observed can be shown to be due to preferential treatment. Genetic progress for milk yield is quite dependent on the genetic superiority of cows selected to be the dams of bulls. The effects of various restrictions for type classification and fat per cent on very high producing groups of cows were studied (Table 3). I t was assumed that the cows should have an index two standard deviations above the average index for milk, if they were to be considered as dams of bulls. Accordingly, a minimum index of +12.3 BCA points was required of all 5,704 cows indexed and 144 cows survived this initial restriction as prospective dams. Few artificial insemination studs in the United States or Canada would consider entering the son of an unclassified cow. Requiring the cow to be classified caused a drop from 144 to 119 prospects. When the final grade of type classification was required to be "Good Plus" or higher, 105 cows survived, as compared to 53 surviving prospects when the cow was required to grade "Very Good" or
TABLE 3. Effect of restrictions on type and test on cows indexing two or more standard deviations above breed average for milk. Average Restrictions
Number of cows
Index (B CA)
Type score
All cows All type-classified cows Type GP or better % F a t 4.10 % F a t 4.50 % F a t 4.90
144 119 105 96 59 21
15.9 16.0 16.0 16.0 15.4 t5.4
4.57 4.58 4.57 4.63 4.83 5.13
...... 83.6 84.5 84.6 84.5 85.1
Type YG or better % F a t 4.10 % F a t 4.50 % F a t 4.90
53 48 27 12
16.9 17.0 16.3 16.2
4.53 4.59 4.82 5.07
87.1 87.2 87.4 87.1
Type E X % F a t 4.10 % F a t 4.50 % F a t 4.90
10 9 6 2
17.7 18.0 17.5 17.7
4.53 4.60 4.82 5.20
90.6 90.7 90.6 90.8
Fat
(%)
JOURNAL OF D ~ Y
S C ~ N 0 ~ ~trOn. 53, NO, 2
] 98
BURNSIDE
TABLE 4. Effect of restrictions on type and test on cows indexing one or more standard deviations above breed average for milk. Average Restrictions
Number of cows
Index
All cows All type-classified cows Type GP or better % F a t 4.10 % F a t 4.50 % F a t 4.90 Type VG or better % F at 4.10 % F a t 4.50 % F a t 4.90 Type E X % F a t 4.10 % F a t 4.50 % F a t 4.90
818 587 454 433 329 164 215 205 160 89 36 35 30 14
10.5 10.7 10.9 10.8 10.3 9.8 11.3 11.2 10.6 10.2 12.1 12.0 11.5 10.9
Fat
Type score
(%)
higher and only ten prospects when the cow was required to grade "Excellent." Restrictions on average milk fat per cent of the cow were set at the breed average (4.9%), at one (4.5%), and at two (4.1%) standard deviations below the breed average. I f one decided that the dam of a bull must grade "Good Plus" or "Better" and, of course, be two standard deviations above the breed average in milk index, 105 cows were available. This number dropped to 96 if an average milk fat per cent of 4.1% was desired, to 59 i£ an average of 4.5% was desired, and to 21 if the cow was required to have an average milk fat test equal to the breed average. The limits set on milk fat per cent were relatively low, yet they removed large numbers of cows. A negative genetic correlation between milk fat per cent and milk index could be a major cause of these results. Similar results were obtained when the milk index was set at -b6.6 B C A points, one standard deviation above the breed average, and the other traits were studied as before (Table 4). A lower limit of one standard deviation above breed average for the milk index of dams of bulls would not lead to rapid genetic change, but it may be higher than that currently set by many artificial insemination studs. Studs tend to set fairly high limits of milk fat per cent on the dams of bulls, as well as reasonably high limits on milk, milk fat per cent, and type of the dam of the cow and other female relatives in the pedigree. In view of the recent results which suggest that relatively low emphasis should be JOURNAL OF I)AIlC¥ SCIEI~CE VOL. 53, NO. 2
4.68 4.69 4.70 4.74 4.88 5.12 4.72 4.76 4.89 5.10 4.73 4.75 4.84 5.09
.
.
.
.
.
.
82.5 84.3 84.4 84.5 84.7 87.0 87.0 87.1 86.9 90.0 91.0 91.0 91.3
given to milk fat per cent in selecting cows and sires (7, 8), it is essential that dairy cattle breeders use their influence to convince sire analysts and sire-purchasing committees: a) to avoid putting undue emphasis on milk fat per cent in selecting dams of bulls; b) to start with lists of potential dams that index high f o r milk (preferably two or more standard deviations above breed averages) and select cows only from these lists; c) to carefully explore unusual herd management practices which pertain to cows grading "Excellent" in final classification. Preferential treatment may have biased the average index values of "Excellent" cows in this study upward by as much as 0.4 of a standard deviation. Acknowledg ments
The author acknowledges the critical comments of Dr. J. W. Wilton, the programming assistance of Miss J. Longmuir, the technical assistance of Mr. John Creeden, and the financial assistance of the Ontario Department of Agriculture and Food. References
(1) Agricultural Research Service, USDA. 1969. D I t I A Cow Performance Index List. ARS 44-209. (2) Bereskin, B., and J. L. Lush. 1965. Genetic and environmental factors in dairy sire evaluation. III. Influence of environmental and other extraneous correlations among the daughters. J. Dairy Sci., 48: 356. (3) Canada Department of Agriculture. 1967.
COW I N D E X I N G The seventeenth report on sire appraisal. Production and Marketing Branch, Ottawa. (4) Miller, R. H. 1964. Indexing Progeny of A.L Sires. Unpublished mimeograph. (5) Thomson, G. M. 1969. Environmental correlations and selection of sires for use in artificial breeding. J. Dairy Sol., 5 2 : 9 2 7 . (6) Van Vleck, L. D. 1966. Environmental cor-
AND
CULLING
199
relations in sire evaluation. J. Dairy Sei., 49 : 86. (7) Wilton, J. W., a n d L. D. V a n ¥1eck. 1968. Selection of dairy cows for economic merit. J. Dairy Sci., 51: 1680. (8) Wilton, J. W., and L. D. Y a n ¥1eck. 1969. Sire evaluation for economic merit. J. Dairy Sci., 52 : 235.
JOURNAL 0¥ DAIRY SCIENOE rOD. 5~, NO.