128 Reviewing the Definition of Mortality in Broiler Chickens

Abstract

Abstract Mortality has been a trait of importance in livestock species, especially for poultry, as birds continue to become more efficient and grow larger, and mortality rates have shown an increase at the phenotypic level. Mortality at different stages may be due to different genetic, environmental, and biological factors; however, in most cases, selection can be based on overall mortality. This study aimed to investigate alternative models and trait definitions of mortality by exploring the classification of mortality into different life stages and including a maternal effect in the model. Data provided by Cobb-Vantress, Inc. included mortality records from three separate lines of chicken making up about three generations worth of data. The datasets included records for mortality (0-1) and weekly mortality (0-10) recorded on individuals during their first ten weeks of life. Due to low incidence levels of weeks seven through ten for weekly mortality, those birds were combined with those that survived the first six weeks. Three models with different trait definitions were considered in this study. They were compared against a benchmark model with overall survival as trait definition (M0). Each line was evaluated separately for each model investigated, and the traits under different definitions were treated as binary traits. Thus, threshold models were used to estimate variance components under a Bayesian approach via the Gibbs sampling algorithm implemented in Gibbsf90+ software. The first model and trait definition (M1) investigated was a single trait model with weekly mortality classified into a binary response to whether the bird died within the first six weeks of life (BMORT). The second model and trait definition (M2) investigated was a bivariate model that classified weekly mortality into early (EMORT; weeks 1 to 3) and late (LMORT; weeks 4 to 6) mortality. The heritabilities ranged from 0.11 to 0.19 for M1, 0.10 to 0.21 for EMORT, and 0.15 to 0.17 for LMORT. For both models investigated, heritabilities were greater than that of M0, ranging from 0.03 to 0.09. The second model was additionally run with the maternal effect (M3) composed of the additive maternal and maternal permanent environmental (mpe) effects to investigate any potential effect on the estimated variance components. The maternal heritability and the percentage of phenotypic variance explained by mpe ranged from 0.02 to 0.03 and 0.01 to 0.04, respectively. Including the maternal effect considerably decreased direct heritability for EMORT but not for LMORT, indicating the maternal effect plays an important role in early mortality. The genetic correlations between EMORT and LMORT were low, ranging from 0.23 to 0.44; therefore, it may be necessary to include both traits in the selection index to make better-informed decisions. Selecting against early and late mortality may help improve overall livability in broiler chickens.