Genetic correlations between first and later parity calving ease in a sirematernal grandsire model

1
Genetic correlations between first and later
parity calving ease in a sire-maternal grandsire
model G. R. Wiggans, C. P. Van Tassell, J. B.
Cole, and L. L. M. Thornton Animal Improvement
Programs Laboratory, Agricultural Research
Service, USDA, Beltsville, MD 20705-2350

• INTRODUCTION
• First parity cows have higher incidence of
  calving difficulty than later parity cows
• In US calving ease evaluations, all parity
  calvings assumed to be single trait
• Genetic differences may exist between calving
  difficulty in first and later parities

• RESULTS
• Computing
• AIREML converged quickly
• Gibbs Sampling took gt20,000 samples for threshold
  stabilization
• Estimates
• Bayesian and AIREML produced comparable genetic
  correlations between parity effects
• Considerable between sample variation for sire
  correlations (Table 1)
• Table 1. Genetic correlations between first and
  later parity calving ease effects
• MGS estimates were lower and more stable
• Avg correlation for MGS of 0.8 suggests
  substantial genetic differences by parity

• RESULTS (cont.)
• Table 3. Variances relative to parity 1 residual
• Thresholds lower for later parity which is
  consistent with lower variance (Table 4)
• Differences in variance by parity combined with
  correlation of approximately 0.8 indicate value
  in modeling MGS CE effect separately by parity
• Table 4. Mean thresholds across 5 samples

• OBJECTIVES
• Determine genetic correlation between calving
  ease in first and later lactations for sire and
  MGS effects
• Examine if differences warrant multi-trait
  analysis

• DATA METHODS
• Five samples of 250,000 from 13,000,000 Holstein
  calvings since 1980
• Limited to 2,600 most commonly occurring bulls as
  sire or MGS.
• Scores recorded on 1 to 5 scale (easy to hard)
• Herd-years required to have at least 20 calvings
• Number of herds ranged 721 860
• Compared threshold and linear model
• For linear model, scores transformed to linear
  scale
• Value was the standard normal deviate at midpoint
  of probability for category
• Probabilities calculated separately by sex of
  calf and for first and later parity
• Model
• Fixed
• Year-season of calving (two per year starting in
  April and October)
• Sex of calf within parity (1, 2, 3)
• Birth year for sire and MGS effects
• Random
• Herd-year
• Sire
• MGS

• CONCLUSIONS
• Correlations between first and later parity were
  high for sire effects, but around 0.8 for MGS
• Variances for later parity groups were less than
  first parity, especially for MGS
• Evaluations can be improved by treating first and
  later parities as correlated traits, particularly
  MGS
• Additional research is needed before
  implementation of separate genetic effects for
  first and later parities
• A bivariate threshold model should be considered
  for national evaluation
• A non-Markov Chain Monte Carlo implementation
  would be complicated
• A linear model on the transformed scale is
  possible, but approximations required for linear
  model may be unacceptable