Naslov
Covariance functions for litter size in pigs using a random regression model

Autori
Luković, Zoran

Vrsta, podvrsta i kategorija rada
Ocjenski radovi, doktorska disertacija

Fakultet
Biotehniška fakulteta

Mjesto
Ljubljana

Datum
10.07

Godina
2006

Stranica
111

Mentor
Kovač, Milena

Ključne riječi
livestock production; genetic evaluation; random regression models; longitudinal data; litter size; pigs

Sažetak
Litter records from three large farms were used. Each farm had two data sets. The first one included records from the first to the sixth parity, and the second data set included records from the first to the tenth parity. Sows had between 3.8 and 3.50 litters in the first data set, and in the second data set between 3.56 and 4.16 litters. Estimates of variances in repeatability model, covariance components for multiple-trait analysis and for random-regression coefficients were estimated by the REML method and the VCE5 software package. Fixed effects in model for the number of piglets born alive included sow genotype, parity, weaning to conception interval, mating season and service sire as class effects. In fixed part of the model previous lactation length as linear regression and age at farrowing as quadratic regression were included. Model for NBA included direct additive genetic, permanent environmental and common litter environmental effect, except in multiple-trait model that did not include permanent environmental effect. Orthogonal Legendre polynomials (LG) of different order were fit to random effects in random regression model. LG with four coefficients were found sufficient. Heritability estimates were around 10 % in repeatability model, and between 10 and 14 % in multiple-trait analysis and random regression model. In repeatability model the ratio for the common litter environmental variance with respect to the total variance were between 1 and 2 %, in multiple-trait analysis and random regression model between 1 and 5 %. The eigenfunctions and corresponding eigenvalues showed that including higher parities in data sets two increased percentages of the total variability which were explained with individual production curve. In the second data sets around 85 and 90 % of the total genetic variability was explained by the constant term in regression, while 10 to 15 % was genetic variability in the shape of litter size curve.

Izvorni jezik
Engleski

Znanstvena područja
Poljoprivreda (agronomija)

POVEZANOST RADA