ABSTRACT
 T
he general aim of this thesis has been to evaluate the productive performance of four maternal lines of rabbits as well as their corresponding crossbreds, and in addition to estimate the crossbreeding genetic parameters, from a complete diallel cross. The lines were A, V, H and LP, founded on different criteria but all of them selected for litter size at weaning since their foundation until present, for 41, 37, 20 and 7 generations, respectively. 
  In chapter 3, comparison between lines was made at the foundation time of the lines, using mixed animal models including additive and permanent random effects. In this analysis all the process of selection was considered by including the complete data set (recorded from June 1980 to February 2009) and the full pedigree. A second comparison was made at fixed times when does from different lines were bred in the same farm and under the same feeding and management conditions: i) from March 1997 to August 1998, for lines A, V and H, and ii) from September 2007 to February 2009 for lines A, V and LP. The models used for these second analyses did not include the additive genetic effects and only consider the records of each period; therefore line comparisons were not dependent on the genetic model. The raw means for 47,132 parities produced from 12,639 does were 9.80, 9.07, 7.79 and 6.95 rabbits for total born, number born alive, number weaned and number marketed per litter, respectively, and 49.80 days for kindling interval, demonstrating high levels of productivity of these lines. At their respective times of foundation, line A showed the lowest litter size, being important the difference between this line and the average of the other lines: 1.39 rabbits per litter for total born, 1.20 for number born alive, 0.84 for number weaned and 1.06 for number marketed. Lines V and H did not show significant differences for litter size traits, but for kindling interval the contrast was 3.30±0.72d, which was significant and favourable to line H. LP line exceeded V line by approximately one rabbit for all litter size traits. The differences between these lines for kindling interval were negligible. Some interactions between lines and farm-year-seasons were important. Regarding the comparison of lines A, V and H from March 1997 to August 1998, the pattern of the differences between the line A and the others was similar to those observed at the origin, and the only significant difference between lines V and H was found for the kindling interval (4.62 d in favour of line V) . The comparison between the lines A, V and LP from September 2007 to February 2009 indicates that differences in reproductive performance between lines were lower than at the origin. In general, good agreement was observed between the comparisons of lines for litter size traits at fixed times, using a model without genetic effects and data recorded during the comparison periods, and the predictions derived from the model with genetic effects and the complete data set.. Another point is the importance of the criteria used in each line to select the founders in determining the initial performance levels. 
  The aim of chapter 4 was to use demographic and litter size data of the previous lines, as a case study, in order to: i) estimate their effective population size, in order to monitor the rate of increase of inbreeding with selection and ii) study whether the inbreeding effect on litter size traits depends on the pattern of its accumulation over time. The lines were kept closed at the same nucleus of selection under the same program of selection and management. Some practices in mating and selection management, such as avoiding matings between animals sharing grandparents and making that each sire contributed with a son to the next generation, allowed an increase of the inbreeding coefficient lower than 0.01 per generation in these lines of around 25 males and 125 females. The effective population size (Ne) for them was around 57.3, showing that the effect of selection on increasing inbreeding, has been counterbalanced by the management practices which try to reduce increase of inbreeding. The inbreeding of each individual was partitioned in three components: old inbreeding (inbreeding accumulated between the foundation of the line and generation 15th for animals born after generation 30th), intermediate (inbreeding accumulated between generations 15th to 30th for animals born after generation 30th or the inbreeding accumulated in the first 15generation for animals born before generation 30th) and new (the rest of situations). The coefficients of regression of the old, intermediate and new inbreeding on total born (TB), number born alive (NBA) and number weaned (NW) per litter, showed a decreasing trend from positive to negative values. Regression coefficients significantly different from zero were those for the old inbreeding on TB (6.79±2.37) and BA (5.92±2.37). The contrast between the regression coefficients associated to old and new inbreeding were significant for the three litter size traits: 7.57±1.72 for TB; 6.66±1.73 for BA and 5.13±1.67 for NW. These results have been interpreted as the combined action of purging of unfavourable alleles and artificial selection against these alleles which was favoured by inbreeding along generations of selection through the increase of homozygotes frequency. 
  The previous maternal rabbit lines were used in a complete diallel cross to produce sixteen genetic groups (four maternal lines and twelve single crosses). The objective of this experiment was to evaluate reproductive traits and functional longevity of crossbred and purebred does. The sixteen genetic groups were distributed in four Spanish farms but only one genetic group (V line) was present in all farms to connect records among these farms and to be used as the reference genetic group. The records of these traits were obtained from January 2009 to October 2011. The differences between all genetic groups and V line were estimated. Then, direct and maternal genetic effects of the lines, and direct heterosis between the lines were estimated according to Dickerson’s model. 
  Chapter 5, aimed to study ovulation rate (OR), number of implanted embryos (IE), total born (TB), embryo survival (ES), fetal survival (FS) and prenatal survival (PS). An animal model was used. The means for all traits were higher than the ones obtained in previous studies in these lines. ??Differences in OR and IE between lines were not significant despite A line showed relevant lower OR than the other lines, whereas the differences were significant and relevant for TB. Regarding the differences between the crossbred groups and V line, only significant differences were observed in favour of HxV. In general, the positive effect of crossbreeding on IE is primarily due to the lower pre-implantation loss and the observed differences were not significant. Regarding TB, the obtained differences were important in two cases: favourable in the case of AxH and unfavourable in the case of LPxV. In general, relevant but no always significant differences between lines in direct genetic effects were found although the magnitude of the differences was high for some traits. Line LP presented larger direct genetic effects than any other line, being significantly different to line A. Differences in maternal effects were not significant, except those between the LP and V lines. High and positive heterosis effect was found between lines A and H. The cross between lines LP and V had a negative heterosis for all traits with a highly negative effect on TB (16% of the mean).
  In Chapter 6, litter size (total born (TB), number born alive (NBA) and number weaned (NW)) and kindling interval (KI) were analyzed with the aim to get estimates of crossbreding genetic parameters. Only, LP line presented a higher direct genetic effect than V line being significant the effect on NBA. Between the other lines no significant differences in direct and maternal genetic effects for TB, NBA and NW were found but there were significant differences for KI. A relevant and large heterosis effect was found for TB in the cross HxV, this parameter for the crosses AxH, AxV and LPxH was lower. For NBA, significant heterosis was found in HxV and AxV and for NW in AxV, LPxH and LPxV. Favourable and significant heterosis for KI was found in AxV and LPxV, whereas in AxLP and LPxH the heterosis was unfavourable and significant. Crossbred does, generally, showed a higher reproductive level than V females and the differences between the average of all crosses and line V were important, being 0.45 for TB, 0.57 for NBA, 0.75 for NW and -2.22 d for KI.The contrast between every two lines showed a similar performance of the lines and we did not find significant differences among them for litter size. For KI, significant differences with relevant values were found between lines and it reflected the significant differences between direct and maternal genetic effects.
  Last chapter of this thesis has aimed to study the functional longevity, defined as the number of days between the first positive palpation and the death or culling of the doe for reasons other than production; it represents the ability to delay involuntary culling. Records from 7,211 does were analysed with the Survival kit 6.0 using a Cox proportional hazard model of fixed effects. The model incorporated time-dependent factors, such as group-farm-year-season (GFYS), number born alive (NBA), group-order of positive palpation (GPPO) and physiological status of the female (PS).Lines A, H and V had a similar ability to avoid culling or death and they show higher risk of being culled of dead than LP line. Line LP had the lowest associated hazard with important differences at later ages and the probabilities to be replaced were 0.39, 0.49 and 0.53 times the replacement rate of lines A, H and V, respectively. We did not find significant differences between all crossbred groups and line V except for the cross between A and H lines in favour of line V. The difference between a cross and its reciprocal, generally, was not significant except between VxH and HxV, favourable to HxV and between LPxH and HxLP, in favour of HxLP. Line V had the highest associated risk due to the direct genetic effects and these differences were significant with respect to those for lines H and LP. The differences in maternal genetic effects were small and not significant except between lines H and V, favouring V line. The estimated heterotic effects do not follow the same direction but they showed, in some cases, the importance of the crosses between specialized lines to produce crossbred does for intensive meat rabbit production. Thus, it was shown that at the early productive cycles, the risk associated to purebreds was higher than the risk of crossbreds, when the cost of the does has not been recovered yet. The contrary happened at late productive cycles (fifth cycle or more), when the cost of the does was already recovered.
  There are some crosses that could be specially recommended to be used in commercial production like HxLP, HxV and AxV or VxA. Some of these recommendations took into account the fact that in commercial production the fostering after birth is a common practice. This fact gives an added value to crosses with high average TB or NBA but having lower performances for NW. The AxH cross could be recommended for farms where the longevity is not a problem. 

XXXII
Mohamed Ragab

  
Resumen
XXXII