Search

In the recent years, the reproductive performance of sows has been drastically improved. Nowadays, in many farms, with the use of hyperprolific sows, the average litter size exceeds 15 piglets at farrowing and 13 at weaning. These changes in performance have been the major drivers for the evolution of sow's nutritional requirements during pregnancy and lactation. At the same time, a large amount of scientific knowledge has been generated over the past thirty years allowing, with the help of mathematical modelling, a holistic approach of energy, amino acids and minerals supplies to sows. The aim of this chapter is to describe the evolutions in the approach of sows' feeding over the past decades and show how the current state of knowledge on sow nutrition can be used for developing models and decision support tools and improving feeding strategies in practice. Different examples are shown to illustrate how the use of such tools can help in optimizing the productivity of sows. Until recently, the maximization of reproductive performance of sows and their litters has been the main objective considered when defining nutritional supplies. Nowadays, new objectives in relation with the increasing societal concern for animal welfare and environment must also be considered. Pig production systems, as well as other animal production systems, are also questioned by the society at a global level in terms of their overall sustainability and acceptability, and this will affect feeding strategies in the future.

A selection experiment on litter size in the pig was carried on for seventeen generations in an Inra experimental herd. The founder population was made up of 10 males and 120 females from the Large White breed. Selection was first performed for ten generations in a closed line, compared to a control line derived from the same founder population. Selection was carried on within sire family on the total number of piglets born in the first two litters of the dam (TB1 + TB2). After ten generations, the selection criterion became dam TB1 only. The control line was then discontinued and a panel of frozen semen from the 11th generation boars was created for later comparisons. The selected line was opened to gilt daughters of hyperprolific boars and sows, at a rate of 1/8 per generation, and the same selection procedure was applied irrespective of the origin of the gilt. During the whole experiment, the number of ova shed (OS) and the number of live embryos (LE) at 30 days in the 3rd pregnancy were recorded. These two parts of the experiment were analysed using REML estimation of genetic parameters and a BLUP-Animal Model in order to estimate the responses to selection. Significant heritabilities for TB1, TB2, OS and LE were obtained, i.e. 0.10, 0.05, 0.43 and 0.19, respectively. Significant common environment variances and covariances were estimated for nearly all traits. Significantly positive BLUP responses per generation were observed from G0 to G17 for TB1 (+0.086), TB2 (+0.078), OS (+0.197) and LE (+0.157). However, the responses were 3- to 4-fold higher in the G12–G17 interval compared to G0–G11, and they were also in fair agreement with previous estimates based on standard least-squares procedures, using the control line and the control frozen semen panel. Since G11, the selection intensity was increased by nearly 80 p. cent compared to the previous generations, and the proportion of hyperprolific ancestry increased up to 65 p. cent in the sows of the last generation. The total genetic gain of about 1.4 piglets at birth per litter could be shared between a gain due to immigration, of about 0.8 piglets per litter, and a within-line selection gain of about 0.6 piglets. Thus by combining selection and immigration in the second part of the experiment, advantage could be taken from both the genetic superiority of the immigrants and the higher internal selection intensity made possible by immigration.