P. Luiting et al., PIG BREED COMPARISON FOR BODY-COMPOSITION AT MAINTENANCE - ANALYSIS OF COMPUTERIZED-TOMOGRAPHY DATA BY MIXTURE DISTRIBUTIONS, Livestock production science, 43(3), 1995, pp. 225-234
Sixteen Duroc and 16 Norwegian Landrace pigs were individually fed to
maintenance (ARC norm) at +/- 65 kg for 8 weeks. All pigs were weighed
weekly and scanned at several positions on the body by computerized t
omography (CT) at the start and after 3, 6 and 8 weeks. Each scan is r
epresented by a large number of CT attenuation values, which are trans
formed into a frequency distribution. Because of accuracy and bias pro
blems with the generally used prediction equations for body compositio
n, an alternative statistical technique, fitting 'mixture distribution
s', was used to analyse the CT data in the present experiment. The met
hod is based on the assumption that the observed CT frequency distribu
tion is the sum of separate underlying normal frequency distributions
for fat and muscle tissue. The parameters of these underlying distribu
tions were estimated by maximum likelihood (EM algorithm). Body weight
s decreased during the first week and gradually increased afterwards t
o reach the start levels again. No significant body weight differences
were observed between the breeds. The fitted mixture distributions sh
owed no significant differences in parameter estimates between the bre
eds at the start of the experiment. During maintenance feeding, the am
ount of fat in the carcass gradually decreased. In Landrace pigs the f
at mobilization was significantly larger than in Duroc pigs. The fat t
issue had higher densities in Landrace pigs than in Duroc pigs at the
end of the experiment. As a consequence, carcass fat/muscle ratio and
energy content decreased stronger in Landrace than in Duroc pigs. The
indication of higher maintenance requirements in Landrace than in Duro
c pigs is discussed.