Modelling individual uncertainty and population variation in phenotypical traits of livestock

Filipe, JAN; Kyriazakis, I

doi:10.3920/978-90-8686-859-9

Modelling individual uncertainty and population variation in phenotypical traits of livestock

Lookup NU author(s): Dr Joao Filipe, Professor Ilias Kyriazakis

Downloads

Full text for this publication is not currently held within this repository. Alternative links are provided below where available.

Abstract

Characterising between-animal variation and its population-level consequences is key to effectivelivestock management and selective breeding. The aim may be to predict trait development (e.g.performance) from early growth or estimate unobserved traits (e.g. maximum growth or maturityparameters). A usual sequence of steps is:(1) to develop a mathematical model of chosen animal-level traits; (2) to estimate individualparameters across a sample of animals; (3) to estimate a population distribution of parameters; (4) togenerate a population distribution of traits by simulating the model across distribution 3. Thechallenge is in the parameter estimation given typical limitations. We use a Bayesian inferencemethodology to estimate the population distribution of predicted traits. The approach: (i) accountsfor individual-level uncertainty in parameters (2) due to their correlation and data limitations, e.g.short growth span or infrequent records, and (ii) does not invoke distributional assumptions andestimation of variance-covariance parameters (3). We present results derived from individual datawith usual limitations. Results include distribution of growth parameters within breeds and acrossspecies (pigs, chicken, rabbits); they extend the literature by showing the extent of uncertainty andvariation in parameters and by comparing variation not only across species but against that withinbreeds. We show distributions of protein and lipid growth parameters and metabolic heat production(HP) estimated across animals and species and predicted distributions of dynamic body composition.Literature on body composition estimates usually condition on input of average HP data; by estimatingboth jointly, their individual-level correlation is included and no metabolic data is needed. HPestimates distributed about 0.7 MJ/kg/d in pigs in line with literature, and body fat content variationwas much larger than that of body protein. We suggest this approach has general application in modelparameterisation and prediction of trait development in populations using limited individual data. Thisstudy is part of the Feed-a-Gene project and received funding from the European Union’s H2020program under grant agreement no. 633531.