Random regression models for genetic analysis of body weight in crossbred pigs

Abstract

Body weight of an animal is represented by a continuous function of time (longitudinal trait) and can be characterized by a trajectory with number of measurements. The present study was carried out to determine heritability estimates of body weight in crossbred pigs (75% Landrace + 25% Bareilly local) using random regression model with Legendre polynomials of quadratic power. Data of 9044 records of 1,292 crossbred piglets, progeny of 86 boars and 98 sows; born in 184 parities between 5 years from 2013–17 was used for the study. Records on weight at birth and at 1 week interval up to 6 week were used. Model included sex, year of birth, season of birth and parity as fixed effect, age of dam at farrowing as co-variable and direct additive genetic effect and maternal permanent environmental effect as random regression. There was a steady increase in body weight over the age from birth (0.96 kg) to 6th week (9.0 kg). Direct additive genetic (0.006 to 7.37 kg2), maternal permanent environment (0.053 to 70.07 kg2) and total phenotypic (0.18 to 77.56 kg2) variance increased continuously from birth to 6 week of age. In general, heritability estimates of body weight at different ages of pre-weaning stage were low ranging from 0.031 to 0.12. The estimate increased up to 1st week (0.119±0.025) with very low value at birth (0.031±0.015) and decreased thereafter to 0.095±0.022 at 6 week. Relative importance of each order of Legendre polynomials showed that quadratic Legendre polynomials with three regression coefficients were enough to capture almost all variability in the model to explain all additive genetic and maternal permanent environment variability. Hence, use of random regression model with quadratic Legendre polynomials was suggested for genetic analysis of pig data for growth.