Effects of selection and pre-weaning growth on the mean and variability of post-weaning growth and carcass performance of French Large White pigs

Get Complete Project Material File(s) Now! »

Use of Best linear unbiased prediction (BLUP) methodology

Best Linear Unbiased Prediction (BLUP) has been the most widely used methodology to estimate breeding values over the last 25 years. BLUP is Best in the sense that it is the Predictor which minimizes the prediction error variance (and maximizes the correlation between true and predicted breeding values) in the class of Linear (it is a linear combination of observations) Unbiased (i.e. its expectation is equal to the true values of the parameters) predictors. Its main advantage over standard selection indexes lies in the fact that it allows to simultaneously account for all pedigree information and environment effects in the estimation of breeding values. BLUP animal model estimates of breeding values (BLUP-BV) have several other desirable properties. First, BLUP-BV have been shown to be unbiased by selection, provided that the model used to describe the data is correct, that the true genetic parameters are being used and that it includes all the data on which selection decision were based since the beginning of the selection process (Henderson, 1975; Sorensen and Kennedy, 1984). Under these same assumptions, BLUP-BV remain unbiased when replacing true genetic parameters by their restricted maximum likelihood estimates (e.g. Gianola et al., 1989; Juga and Thompson, 1989) and BLUP-AM accounts for genetic drift, assortative mating and inbreeding (Sorensen and Kennedy, 1983). Finally, BLUP animal models easily provide estimates of genetic trends, which are computed as changes in average BLUP-BV over time. For instance, yearly genetic trends can be obtained at any moment by computing average BLUP-BV according to animals’ year of birth. BLUP-AM is thus a formidable tool which potentially allows an almost real-time control of the efficiency of selection in selected populations. Things may be somewhat less simple in real situations, as the above-mentioned assumptions of data exhaustiveness may not be fulfilled, which may result in potential biases due to errors in genetic parameter estimates and an incomplete description of the selection process. Moreover, genetic trends can be obtained only for traits that are regularly measured during the period considered.

Genetic trends in pigs

The objective of this third part of chapter on is to provide results on estimated genetic trends (EGT) in the population investigated, i.e. French Large White dam breed, in order:
• to put the experiment analyzed in this thesis in its context,
• to analyze the consistency of the results obtained in this experiment with other estimates of genetic trend in this same Large White population,
• to compare the results obtained in Large White breed to available literature results in other populations.
An additional objective will be to present and discuss potential adverse effects of selection in order to identify traits which should receive a particular attention.

Estimates of genetic trends in pig populations

As mentioned above, several results of the current experiment have already been published. Yet, most of them (Canario et al., 2007a; Canario et al., 2007b; Foury et al., 2009; Canario et al., 2014b) concern traits that were not previously measured, and thus cannot be compared. In fact, the results obtained on standard production traits (Tribout et al., 2010) are the only traits that can be compared with either BLUP estimates of genetic changes or with results from previous experiments based on the use of frozen semen (Molénat et al., 1986; Ollivier et al., 1991; Bazin et al., 2003).
BLUP genetic trends estimated over successive periods of time for 4 major production traits have been compiled and compared with those obtained by Tribout et al (2010) in figure 1.4. Trends from both methods are rather consistent, except for average backfat thickness where a larger trend is obtained by Tribout et al (2010) as compared with BLUP estimates.
The situation is less consistent when comparing the results of Tribout et al (2010) with previous estimates based on frozen semen (table 1.3). Differences are particularly important with the results of Molénat et al. (1986) and, to a lesser extent, Ollivier et al. (1991), who report surprinsingly large estimated genetic trends. As discussed by Ollivier et al. (1991) , these results are likely to be oversestimared due to a strong disequilibrium in sire progeny size, to the limited period of time considered and to changes in the breeding goal, with the successive introduction of meat quality (1980’s) and litter size (1990’s) in the breeding goal. The results are more consistent with those reported by Bazin et al. (2003) during the same time interval (1977-1998) as Tribout et al. (2010), except for growth rate, and, to a lessser extent food conversion ratio. In spite of these differences, it can be argued that the current experiment provides rather relevant estimates of genetic trends and can be used to measure genetic trends for a larger number of traits.
Estimates have also been compared with those published in other populations in different countries, mainly Europe and America. Results are shown in table 1.4 and 1.5 for production and reproduction traits, respectively. It should be pointed out that most estimates from the scientific literature are rather old, i.e. from the 1990’s, and that the recent results have been obtained from technical reports. They are mainly originating from “national” organisations, as most private breeding companies do not publish their results, presumably for fear of unfavourable comparisons with competitors. Large variations are observed between breeds, the moment and the length of the period considered. They reflect differences in breeding goals over time and between populations, as well as potential differences in selection efficiency. Average genetic trend are 6.4%, -7.1%, -5.7% and -5.8% of trait phenotypic standard deviation for average daily gain, feed conversion ratio, average backfat thickness and carcass lean content, respectively. Recent estimates tend to be larger than old ones, indicating an increased efficiency of selection. The majority of estimates in French populations exceed the above-mentioned average values, even in dam lines. Average trends for reproduction traits are 0.11 (total number born), 0.12 (number born alive), 0.10 (number weaned) and 0.06 (teat number). Genetic trends in French populations have been limited until the mid-nineties and the inclusion of litter size in the breeding goal of Large White (LW) and Landrace (LR) populations. Since 1995, annual trends have been 0.23 and 0.17 piglet born alive/litter, respectively, in LW and LR breeds.
Yet trends have not been linear (see figure 1.5.). After a steep increase until 2002, accompanied by an important degradation of piglet survival at birth, it was decided to change total number born in the breeding goal by a combination of number born alive and teat number. Both traits have increased until 2013, but at a lower rate than in the previous decade. In 2013, competition with Danish and Dutch population has led French breeding organisations to increase the economic weight of number born alive and decrease that of teat number, resulting in a large increase in NBA from 2013-2015. The 1977-1998 period considered in this study thus corresponds to beginning of the increase in prolificacy.

READ  EEG signals and motor imagery 

Undesired effects of selection

Production levels in pigs have considerably increased over last decades as a result of both selections, the generalisation of crossbreeding and improved management practices. As mentioned above, selection does not only affect selected traits, but also a much larger number of traits genetically correlated to the traits of the breeding goal. Correlated trends can be favourable – e.g. selection for feed efficiency results in decreased nitrogen and phosphorus excretion (Shirali et al., 2012; Saintilan et al., 2013) -, but is also sometimes detrimental. For instance, including the total number of piglets born in the breeding goal of pig dam lines has often resulted in an increase in the number of stillbirths in nucleus herds (Canario et al., 2006b; Su et al., 2007; Silalahi et al., 2016) as well as at the commercial level (see table 1.6.). The objective of this chapter is to review potentially undesirable effects of selection on major economically important traits.

Table of contents :

1. Evaluation of the efficiency of breeding schemes
1.1. Organization of pig breeding schemes
1.1.1. Some elements on pig production
1.1.2. Pig breeding goals and breeding programmes
1.1.3. Genetic variability of traits of interest
1.1.4. Selection criteria and genetic evaluation
1.2. Measuring the efficiency of pig breeding schemes
1.2.1. Evaluation at commercial level
1.2.2. Evaluation of genetic trends in nucleus herds
1.2.2.1. Use of control population
1.2.2.2. Use of frozen material (semen or embryos)
1.2.2.3. Use of best linear unbiased prediction (BLUP) methodology
1.3. Genetic trends in pigs
1.3.1. Estimates of genetic trends in pig populations
1.3.2. Undesired effects of selection
1.3.2.1. Piglet survival
1.3.2.2. Sow lifetime
1.3.2.3. Sow behaviour
1.3.3. Robustness
2. Estimation of the effects of selection in French Large White pigs from 1977 to 1998 p30
2.1. Experimental design
2.2. Estimation of the effects of selection on French Large White reproductive performance using frozen semen
2.3. Estimation of the effects of selection on French Large White sow and piglet performance during the suckling period
2.4. Effects of selection and pre-weaning growth on the mean and variability of post-weaning growth and carcass performance of French Large White pigs
2.5. Influence of selection on sow macro-environmental effects on their offspring performance
3. General discussion
4. Conclusion
5. References
6. Curriculum vitae
7. Acknowledgments

GET THE COMPLETE PROJECT

Related Posts