ABSTRACT
Firstly, the different definitions of longevity and the methods to estimate genetic
parameters and to predict breeding values associated with the different definitions are
reviewed. Afterwards the Survival Analysis methods are described in detail, since they
are used extensively in this thesis. Finally, in the introduction, a bibliographic review
has been done on the estimated genetic parameters for longevity in different livestock
species.
Afterwards the two main objectives of the thesis are raised, which are on one
hand to define a model for the analysis of longevity in rabbit does for meat production,
including the estimation of genetic parameters for this trait, and on the other hand the
evaluation of a maternal line (L-P), which has been found by selection of outstanding
females in longevity and with a prolificacy around the mean of the population from
commercial farms.
Regarding the first main objective Survival Analysis techniques have been used
to infer genetic parameters in a subpopulation of the V line from a selection nucleus in
which no culling based on production performances took place. Initially the trait was
defined as the number of days between the first mating and the death/culling or
censoring of the female. With this definition it was observed that the model most
commonly employed to fit longevity data in livestock species (Weibull model) was not
totally adequate for rabbits for meat production. For a second definition of the trait, the
number of days between the first positive pregnancy test and the death/culling or
censoring date, the result of the test to determine whether the Weibull model fit properly
our longevity data was also negative. Thus, in order to study how different factors
(including the genetic) determine longevity it was decided to use semiparametric
proportional hazard models.
For the first definition of the trait it was observed that the year-season (defined
every six months) showed a significant effect over the risk of culling or death.
Moreover, in general, it was observed that those levels of factors relating to sick does
(very low number of young born alive) were the ones with higher associated risk of
culling or death. Regarding the reproductive rhythm, it was observed that does delayed
at first parturition had a lower survival probability than non delayed females during
their whole lives. A doe delayed at second parturition showed a lower risk of
dying/being culled than a non delayed female, when the comparison is done at time of
vii
the parturition of the latter. This is explained by the high risk associated with the
parturition, especially at these ages, when the does are still growing. If the comparison
is done at the parturition date of the former the differences disappear. Delays at late ages
had a negative but negligible effect. The age of the female at first mating showed a
significant effect – it was more likely for a young doe to die or to be culled than an
older one. An heritability of 0.053, on the log-g(t) scale, was estimated using a sirematernal
grandsire model.
With the second definition of the trait the year-season (defined every three
months) also showed a significant effect. The reproductive rhythm of the female it was
more properly defined by considering the physiological status of the female during its
whole life (Lactating, Pregnant, Empty or Pregnant-Lactating). Regarding the
systematic effects in the model, in general, the same conclusion as with the first
definition was achieved: Those levels relating to sick does (very low number of young
born alive and infertility) were the ones with higher associated risk of culling or death.
The estimated heritability, on the log-g(t) scale, was 0.095. This increase in the
estimated value for this parameter could be due to a better fitting of systematic effects
and because an animal model was used.
Regarding the estimation of the correlations between functional longevity (using
the second definition) and prolificacy, it was observed that neither the number of young
born alive nor the number of young at weaning showed a genetic correlation statistically
significant from zero with functional longevity. The only significant environmental
correlation was obtained between functional longevity and the number of young born
alive (-0.11 ± 0.016). This estimated value was very low but favourable.
Immediately afterwards, the process of foundation of the new Long-lived –
Productive line (L-P) is described, which was similar to the one applied in the
foundation of hyper-prolific lines of pigs and rabbits. Moreover, the statistical power
achieved in the experiment to evaluate this line is studied. The evaluation was carried
out by comparing this line for longevity and for productive performances with the V
line, which has a recognized high prolificacy and medium longevity. The process
founding the L-P line has been successful, since it showed higher survival ability
especially late in life than the V line. Regarding prolificacy traits the line L-P had a
lower performance than the V line, especially during the first parturitions. If the
prolificacy between both lines was compared during all the parturitions the V line
showed 0.5 young more than the L-P. In spite of differences in longevity and prolificacy
viii
no differences in cumulative production of weaned young were observed between the
two lines. Differences in fertility were not observed either.