A. Reverter et Cj. Kaiser, THE ROLE OF DIFFERENT PEDIGREE STRUCTURES ON THE SAMPLING VARIANCE OFHERITABILITY ESTIMATES, Journal of animal science, 75(9), 1997, pp. 2355-2361
A computer-intensive process was performed to simulate 12,600 data set
s each with n = 5,000 individuals from distinct pedigree structures to
assess the effect of pedigree information on the sampling variance of
heritability (h(2)) estimates. Pedigree structures were determined by
varying the proportion of foundation animals (PF), percentage replace
ment rates for males (RM) and females (RF), and ratio of females to ma
le (F2M). A 2(3) factorial design was modeled; levels of RM and RF wer
e 10 and 20%, and levels of F2M were 10 and 20. For each of the eight
cells, 60 foundation animals were simulated, each with 10 replicates.
The required mating seasons (MS) to obtain the number of individuals w
as simulated based on PF and F2M. A REML algorithm was used to estimat
e h(2) and its associated SE. The effect of all factors was analyzed i
n a regression model with linear and quadratic components for PF. An a
lternative model with MS replacing PF was also investigated. There was
a non-monotonic association (P < .01) between PF and h(2) SE. The min
imum h(2) SE occurred when PF ranged from 20 to 40%. Here, the proport
ion of first-generation progeny was near its maximum with rapid increa
ses in the proportion of subsequent descendants. Among the class effec
ts, F2M yielded the highest mean square (P < .001). When considering m
ore than one MS, h(2) SE was positively associated(P < .01) with RF an
d F2M and negatively associated with RM. Results suggest that h(2) is
most accurately estimated when there is performance information on man
y animals closely related to foundation animals.