Presentation given at 2015 ISVEE14 conference in Merida, Mexico.
Relationship between longevity and health or reproductive performance at the cow level is well demonstrated in the dairy literature but this association might not hold true at the herd level. Many herd-, farm-, and cow-level components are part of the culling decision. The objectives of this study were to: 1) quantify culling rates of Québec dairy herds; and 2) investigate if Québec dairy farms could be differentiated based on herd-level factors such as management, reproduction, production and health indices, and explore their relationship with herd culling rate.
A retrospective study was conducted on data from dairy herds in Québec, Canada, by extracting their health and production data. Data were extracted for all lactations taking place between 2001 and 2011. A total of 432,733 lactations records (from 156,409 cows; 763 herds) were analyzed. Thirty herd-level variables were aggregated for each herd/year of follow-up and their relationship investigated by Multiple Factor Analysis (MFA). The overall culling rate was 31.6% with a 95% confidence interval of [31.2, 32.1]. The explained variance for each axis from the MFA was very low (first and second axis: 13.7 and 12.6%, respectively) suggesting that there was no relationship among the groups of variables. Associations were found between culling rates and herd-level variables such as seasonality, proportion of primiparous cows, calving intervals, 21-day pregnancy rates, days to first service, and average age at first calving.
Pregnancy is a known cow-level protective factor against culling and herd reproductive performances were found in this study to be associated with culling rates. However, these were the only herd-level factors associated with culling while there are many cow-level risk factors. This stresses the importance of acknowledging the discrepancy between herd- and cow-level associations. Inferences at the group level should not be based on individual-level data.
Davis plaque method.pptx recombinant DNA technology
Herd-Level Management Factors and Culling Rates in Québec Dairy Herds (ISVEE14, Merida, MX, 2015)
1. HERD-LEVEL MANAGEMENT FACTORS AND
CULLING RATES IN QUÉBEC DAIRY HERDS
ISVEE14—Merida, MX—November 6, 2015
Denis Haine1
H. Delgado2
R. Cue2
A. Sewalem3
K. Wade2
R. Lacroix4
D. Lefebvre4
J. Arsenault1
É. Bouchard1
J. Dubuc1
1Université de Montréal
2McGill University
3Agriculture and Agri-Food Canada
4Valacta
3. Introduction Material & Methods Results Conclusion
CULLING: RISK, ANALYSIS & INFERENCE
Objectives
To quantify culling rates
To determine profiles of herds based on herd-level
factors
To describe relationship between these factors and
herd culling rate
4. Introduction Material & Methods Results Conclusion
DATASET
DHI and health data, 01/01/2001—12/31/2010:
432 733 lactations,
156 409 cows,
763 herds
Management indicators
Herd size,
% first lactation,
% livestock sales,
% fall calvings
Reproduction indicators
Calving interval,
Days to first service,
21-d pregnancy rate,
Age at first calving
Production indicators
305-d milk, fat, protein,
For heifers and cows:
Peak milk,
Persistency,
Peak variation,
Persistence variation
Health indicators
MF, RP, metritis, DA,
COD,
Lameness, Mastitis,
Dystocia,
Mortality,
Udder health index
5. Introduction Material & Methods Results Conclusion
DATA ANALYSIS
Relationship between groups of herd indices:
Multiple Factor Analysis (MFA)1
Test-value2 for culling rate as supplementary
variable
Culling incidence: GEE model
All analyses with R version 3.2.2; packages
geepack3 and FactoMineR4
1
Escofier and Pagès, 1994.
2
Morineau, 1984.
3
Højsgaard, Halekoh, and Yan, 2006.
4
Lê, Josse, and Husson, 2008.
7. Introduction Material & Methods Results Conclusion
MULTIPLE FACTOR ANALYSIS
q
management
reproduction
productionhealth
culling0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
Dim. 1 (13.7%)
Dim.2(12.6%)
Contribution of groups of indicators to the 2 first dimensions of the MFA,
according to their squared loadings
8. Introduction Material & Methods Results Conclusion
PRINCIPAL COMPONENT ANALYSIS
livestock sales
proportion of
first lactation
herd size
proportion of calvings in the fall
−1.0
−0.5
0.0
0.5
1.0
−1.0 −0.5 0.0 0.5 1.0
Dim. 1 (30.9%)
Dim.2(26.7%)
Management
9. Introduction Material & Methods Results Conclusion
PRINCIPAL COMPONENT ANALYSIS
livestock sales
proportion of
first lactation
herd size
proportion of calvings in the fall
−1.0
−0.5
0.0
0.5
1.0
−1.0 −0.5 0.0 0.5 1.0
Dim. 1 (30.9%)
Dim.2(26.7%)
Management
% cull rate
% sales
% L1
& autumn
& herd size
& cull rate
& sales
& L1
10. Introduction Material & Methods Results Conclusion
PRINCIPAL COMPONENT ANALYSIS
days to
first service
calving
interval
21d pregnancy
rate
age at
first calving
−1.0
−0.5
0.0
0.5
1.0
−1.0 −0.5 0.0 0.5 1.0
Dim. 1 (53.9%)
Dim.2(21.8%)
Reproduction
11. Introduction Material & Methods Results Conclusion
PRINCIPAL COMPONENT ANALYSIS
days to
first service
calving
interval
21d pregnancy
rate
age at
first calving
−1.0
−0.5
0.0
0.5
1.0
−1.0 −0.5 0.0 0.5 1.0
Dim. 1 (53.9%)
Dim.2(21.8%)
Reproduction
% cull rate (and vice versa)
% calving interval
% days to 1st service
& pregnancy rate
& age at 1st calving
13. Introduction Material & Methods Results Conclusion
PRINCIPAL COMPONENT ANALYSIS
metritis
cystic
ovaries disease
milk fever,
L3+
milk fever,
all parities
retained placenta
mastitis
lameness
dystocia
displaced abomasum
udder health
index
mortality
−1.0
−0.5
0.0
0.5
1.0
−1.0 −0.5 0.0 0.5 1.0
Dim. 1 (29.2%)
Dim.2(12.3%)
Health
14. Introduction Material & Methods Results Conclusion
PRINCIPAL COMPONENT ANALYSIS
metritis
cystic
ovaries disease
milk fever,
L3+
milk fever,
all parities
retained placenta
mastitis
lameness
dystocia
displaced abomasum
udder health
index
mortality
−1.0
−0.5
0.0
0.5
1.0
−1.0 −0.5 0.0 0.5 1.0
Dim. 1 (29.2%)
Dim.2(12.3%)
Health
medium cull rate
% MF
15. Introduction Material & Methods Results Conclusion
CONCLUSION
Herds could not be clustered based on multiple
herd-level variables
Each group of variables have to be considered
separately
Herd management and dynamics, reproduction
indicators, 305-d and peak productions + MF
incidence
cow-level risk factors 6= herd-level
contextual variables and multilevel analysis