where sqrt(yijklmn) is the square root of observed harvest body weight
of the lth shrimp of the selection or control populations from each
generation; μ is the overall mean; Pondi is the fixed effect of the ith
pond (six ponds); Sexj is the fixed effect of the jth gender (male and
female shrimp); Popk is the fixed effect of the kth population (selection
and control populations); Agel(Pondi ∗Sexj) is the covariate of
age from the lth shrimp nested within the ith pond and the jth gender
combination fixed effect; Sirem(Popk) is the random effect of the mth
male parent nested within the kth population; Damn(Popk ∗ Sirem) is
the random effect of the nth female parent nested within the kth
Pop and the mth Sire combination subclass fixed effect; and eijklmn is
the random residual error of the lth individual.
The realised genetic gain for harvest body weight was estimated
as the difference between the least squares means of the selection
and control populations in each generation. This selection response
was also expressed as a percentage of the realised genetic gain to
the least squares means of the control population in each generation.
2.6.3. Predicted genetic gain based on estimated breeding value
Using the same model (formula 1) with the variance components
estimate, the EBVs of all animals for five generations were calculated
based on best linear unbiased prediction (BLUP) by ASReml software.
The predicted genetic gain in square root transformed harvest body
weight for each generation was computed as the difference in the
mean of the EBVs for square root transformed harvest body weight
of individuals between the current and the previous generations.
The genetic gain per generation was also expressed as a percentage
of the square root transformed least squares means for body weight
from the control population. The cumulative genetic gain was
obtained as the difference between the mean of EBV of the G4 and
G0 generations. To compare the difference in predicted genetic gain
based on the genetic parameters estimated from the within- and
across-generation datasets, two sets of genetic parameters were
used to predict the EBVs of square root transformed harvest body
weight for all animals in the five generations. The first (Set 1) was
obtained as the mean of the genetic parameters estimated from five
different within-generation datasets. The second (Set 2) was from
the genetic parameters estimated from the across-generation dataset.
The cumulative genetic gain from G0 to G4 was expressed as a percentage
based on the following formula:
Pc ¼ Πni
¼1 1 þ pi ð Þ−1 ð4Þ
where Pc is the cumulative genetic gain (%);pi is the genetic gain (%)
for the ith generation; i is the number of selection generation (i=1,
2, 3, 4).