A National Sire Fertility Index

1
A National Sire Fertility Index


2
Bull fertility (phenotypic ranking)

• Estimated relative conception rate (ERCR)
• 70-day nonreturn rate (NRR)
• Source
• DRMS, Raleigh, NC, 1986-2005
• USDA, Beltsville, MD, 2006-present
• Western Bull Fertility Analysis
• 75-d veterinary-confirmed conception rate (CR)
• Source AgriTech, Visalia, CA, 2003 -present

3
Sire conception rate (SCR)

• New USDA service-sire phenotypic fertility
  evaluation
• Based on CR rather than NRR
• More accurate
• Inseminations from most of the United States
• Services 17 (not just first)
• Additional model effects included
• Implemented August 2008

4
Data included

• Only AI inseminations with pregnancy status
  confirmation (success or failure)
• Inseminations 17 for cows in lactations 15
• Lactation length at breeding limited to 30365
  DIM
• Cow age of 215 yr
• Standardized milk yield
• gt10,000 lb for Holsteins
• gt8,000 lb for Brown Swiss
• gt6,000 lb for all other breeds

5
Data included (cont.)

• Most recent 4 yr of breeding records
• Inseminations 70 d before data submission
  deadline
• 6 traditional U.S. dairy breeds
• Ayrshire
• Brown Swiss
• Guernsey
• Holstein
• Jersey
• Milking Shorthorn

6
Data excluded

• Embryo-transfer donors
• Sexed semen
• Heifers
• Consecutive services within 10 d of each other
• Only information from later service kept
• Earlier service not considered when assigning
  subsequent service numbers for same lactation

7
Data excluded (cont.)

• Herd with 50 of milking cows without recorded
  breeding
• Herd CR lt10 or gt90
• Service sire lt0.8 yr old

8
Data sources (August 2008)

• 3 dairy records processing centers
• AgriTech Analytics
• AgSource Cooperative Service
• DRMS
• gt99 of data
• 46 States and Puerto Rico

9
Development of SCR

• 4-year research effort primarily by
• Dr. Melvin Kuhn
• Bull variables (expanded service-sire effect)
• Cow (nuisance) variables

10
Bull variables

• Inbreeding
• Service sire
• Embryo
• Bull age
• AI organization combined with mating year
• Bull

11
Cow variables

• Combined herd, mating year, cow parity, and cow
  registry status
• Combined mating month, year, and State
• Cow parity
• Service number
• Short interval between matings
• Cow age
• Cow standardized milk yield
• Cows permanent environment
• Cows genetics

12
SCR model

• Categorical effects
• Individual parities for lactations 15
• State-year-month of insemination group
• 6 standardized milk yield groups
• Service number for inseminations 17
• Cow age
• Herd-year-season-parity-registry status class
• Covariate (linear regression) effects
• Service-sire inbreeding coefficient
• Mating inbreeding coefficients
• Random effects
• Service-sire age group
• AI organization-insemination year group
• Individual service sire
• Cows genetic ability to conceive
• Cows permanent environmental effect
• Residual

The most complex model that I know of to
evaluate animal performance Bennet Cassell,
VPISU, 2008
13
Variances

• Service-sire age 0.00014
• AI organization-insemination year 0.00011
• Service sire 0.00054
• Cow 0.00294
• Cows permanent environment 0.00533
• Residual 0.19697

14
SCR accuracy

• Reliability (R) n/(n 260)
• n number of inseminations
• Constant 260 derived by including all random
  effects in expanded service sire term
• Confidence interval (CI)
• 0.02313 true standard deviation
• 1.282 standard normal variate from normal
• distribution for an 80 CI

15
Relationship of R and 80 CI
Inseminations R () 80 CI
200 43 2.2
300 54 2.0
500 66 1.7
1,000 79 1.3
2,000 88 1.0
5,000 95 0.7
10,000 97 0.5
15,000 98 0.4
20,000 99 0.3
16
Proposed CI table
Bull name SCR () R () 80 CI
A 1.6 99 1.3 to 1.9
B 0.8 98 0.4 to 1.2
C -0.4 90 -1.3 to 0.5
D 1.1 82 -0.2 to 2.4
E -3.8 77 -5.2 to -2.4
F 2.3 59 0.4 to 4.2
17
SCR release

• Released 3 times a year in conjunction with USDA
  national genetic evaluations
• January
• April
• August
• Only AI bulls 15 yr old
• Active AI
• Progeny test

18
SCR release (cont.)

• Overall matings
• Holstein 300 in 10 herds
• Ayrshire, Brown Swiss, 200 in 5 herds
• Guernsey, Jersey
• Milking Shorthorn 100 in 5 herds
• Matings during current 12 mo
• Holsteins, Jersey 100
• Ayrshires, Brown Swiss, 30
• Guernsey
• Milking Shorthorn 10

19
Interpretation of SCR

• Phenotypic predictor of bull fertility
• Expressed as relative CR
• Reported as a percentage
• Average bull has SCR of 0.0
• Standard deviation for August 2008 SCR was 2.4

20
Examples

• Bull with SCR of 3.0 expected to have 3 higher
  CR than average bull and 6 higher CR than bull
  with SCR of -3.0
• Bull with SCR of 2.0 expected to have CR of 32
  in herd that normally averages 30 and
  historically has used bulls with average SCR

21
Impact of individual effects

• Individual effects sequentially removed from full
  model to test alternative models
• Service-sire inbreeding
• Mating inbreeding
• Service-sire age
• AI organization-insemination year
• Each effect added back to the model and another
  effect removed

22
Correlations of alternatives with full model
Alternative model AI organization AI organization AI organization AI organization AI organization AI organization AI organization
Alternative model All A B C D E F
No mating inbreeding 1.00 1.00 1.00 1.00 1.00 1.00 1.00
No service-sire inbreeding 1.00 1.00 1.00 1.00 1.00 1.00 1.00
No AI organization- insemination year 0.98 1.00 1.00 1.00 1.00 1.00 1.00
No service-sire age 0.95 0.95 0.95 0.95 0.94 0.93 0.92
Interpolated age 0.99 0.99 0.99 0.99 0.99 0.98 0.98
23
Service-sire age effect

• Greatest impact on SCR prediction across and
  within AI organization
• Interpolated age expected to provide most
  consistent evaluations across time
• Not intended for comparison of rankings at a
  common age
• Provide more accurate representation of
  phenotypic value of CR for a bulls semen at this
  point in his life

24
Maximum absolute change

• Individual bull
• Comparison with January 2008 full-model
  evaluation
• Change
• Alternative models (percentage units)
• No AI organization-insemination year 2.2
• No service sire-age 1.9
• Interpolated age 0.9
• No service-sire inbreeding 0.8
• No mating inbreeding 0.2

25
Prediction effectiveness

• July 2006 Holstein SCR from alternative models
• Average CR for later (July 2006 January 2008)
  inseminations
• Deviation of outcome for each later insemination
  from average for all inseminations in same
  herd-year-season
• Herd fertility differences removed
• 300 inseminations for each bull SCR and either
  100 or 300 inseminations for later CR

26
Bull correlations of SCR with later CR
Model Bulls with ?100 inseminations Bulls with ?300 inseminations
Full model 0.6213 0.6526
No mating inbreeding 0.6222 0.6536
No service-sire inbreeding 0.6189 0.6497
No AI organization- insemination year 0.6179 0.6488
No service-sire age 0.6089 0.6326
Interpolated age 0.6238 0.6549
27
Optimal AI organization-insemination year

• AI industry concern
• NAAB code used to assign bulls to AI
  organization-insemination years
• Not as effective in predicting future CR as
  assigning all bulls to most recent AI
  organization-year
• Assigning bulls to AI organization-year just
  prior to most recent also of considerable value

28
Optimal AI organization-year (cont.)

• Additional studies applied multiple-regression
  methods
• Prediction of future CR most improved by
  including 2 most recent AI organization-years
• 60 weighting for most recent year
• 40 weighting for previous year

29
Herd fertility

• Relationship between bull SCR and fertility of
  herds for which bull was service sire
• Herd-years stratified into 3 equally sized groups
  by CR
• 27.3 Low fertility
• 27.4 to 33.9 Medium fertility
• 34.0 High fertility
• Bulls stratified into 3 equally sized groups by
  SCR
• -0.9 Low fertility
• -0.8 to 1.0 Medium fertility
• 1.1 High fertility

30
Herd CR (August 2008)
Service-sire fertility Herd fertility Herd fertility Herd fertility
Service-sire fertility Low Medium High
Low 20.3 27.4 35.3
Medium 22.6 30.0 38.7
High 24.8 32.4 41.4
Difference 4.5 5.0 6.1
31
Conclusions

• New SCR evaluation
• Based on confirmed pregnancies
• Measures phenotypic service-sire fertility
• Expressed as a relative CR (average bull has SCR
  of 0.0)
• Standard deviation of 2.4 in August 2008

32
Conclusions (cont.)

• First official SCR evaluations released in August
  2008 for active-AI and progeny-test bulls
• Data from gt80 of DHI herds that collect breeding
  information
• Most States and Puerto Rico represented for 6
  dairy breeds

33
Conclusions (cont.)

• SCR more accurate than ERCR because of data from
  3 times more inseminations
• More DHI herds (Western herds added)
• Extra services (27)

34
Female fertility evaluations

• Genetic evaluations to be implemented in 2009
• Heifer conception rate (HCR)
• Percentage of inseminated heifers that become
  pregnant at each service
• Cow conception rate (CCR)
• Percentage of inseminated cows that become
  pregnant at each service
• Similar to reporting for daughter pregnancy rate
  (DPR)
• Will be reported to Interbull

35
Acknowledgments

• Reproductive records supplied by AgriTech
  Analytics, AgSource Cooperative Service, and DRMS
• Willingness of U.S. dairy producers to record
  their management data essential for continuation
  of effective fertility evaluation
• Suggestions provided by the National Association
  of Animal Breeders Fertility Committee
  beneficial in development of SCR