Selection for production and reproduction traits in pigs

English. IntroductionReproduction traits are important for piglet production, whereas production traits are important for fattening. Pig breeding organizations improve both groups of traits by selection in nucleus populations. Optimization of selection in these nucleus populations is important, because these populations determine the performance in all levels of the breeding pyramid, including the piglet production and fattening herds.Optimization of selection in pigs requires a careful study, because many complicating aspects have to be dealt with (multi- trait and multi-stage selection, overlapping generations, inbreeding, reductions in selection intensity due to small numbers, continuous selection process).Production traits can be measured on young boars and gilts during a performance test. Reproduction traits can be measured only on sexually mature sows. The difference in expression between the two groups of traits form an additional complication for the optimization of pig breeding programmes.Most of the previous optimization studies for pig breeding have focussed on selection for only one group of traits. However, an important question at the moment, for the majority of pig breeding organizations, is how to improve production and reproduction traits simultaneously.The objective of this study is optimization of combined selection for production and reproduction traits in pig breeding programmes. The research is focussed on genetic response to selection and rate of inbreeding.Definition of the breeding goalThe first two chapters of this thesis deal with definition of the breeding goal. In Chapter 1, a deterministic model was developed to estimate economic values of traits in pig breeding programmes. The model describes efficiency of pig meat production as a function of breeding goal traits. Traits in the breeding goal were: oestrus and litter traits, mature weight, feed requirements and longevity of sows, growth rate and daily feed intake of young pigs and of fatteners, mortality rate of pigs, carcass traits and incidence of PSE- meat.The model was applied to the Dutch situation. Economic values (Dfl. per slaughter pig) of most important traits were:-0.09 (per day) for age at first oestrus,-0.32 (per day) for interval weaning- oestrus,8.90 (per pig litter -1) for litter size born alive,-1.10 (per %) for mortality rate of piglets in suckling period,2.30 (per farrowing) for longevity of sows,0.26 (per g day -1) for growth rate of fatteners,-0.06 (per g day -1) for daily feed intake of fatteners,3.10 (per %) for lean content of the carcass.Sensitivity of economic values was tested to changes in production circumstances (changes in feed prices, price of replacement gilts, labour and management costs and technical performance).After estimation of the economic values of traits, the breeding goal can be defined. However, direct use of the economic values as weights in the breeding goal is not always optimal. One of the aspects that can play a role is the competitive position of a breeding organization (i.e. the performance of its breeding stock relative to other organizations). This aspect is dealt with in Chapter 2.The value of improvement of a trait for a breeding organization is determined by its impact on saleability of the breeding stock. This impact is influenced by the competitive position of the organization. This is especially relevant for the optimal balance of selection between production and reproduction traits, because breeding stock needs to be acceptable for piglet production as well as for fattening herds. No method could be found in literature to quantify effects of competitive position on values of traits.A generally applicable method was developed to take effects of competitive position into account. With an example it was shown that modification of the breeding goal can be necessary (for the short-term benefits of a breeding organization) when performance levels of traits deviate widely from competitors. Traits with a relatively low performance would need a higher weight in the breeding goal than traits with a high performance.In the General Discussion of the thesis, attention was given to aspects that can affect the breeding goal. It was concluded that, in addition to competitive position, biological interactions between traits can be important (e.g. when the level of feed intake capacity becomes a limit for genetic improvement of protein deposition rate). Genotype x Environment interactions do no affect the breeding goal, but they may change the optimal balance between selection for production and for reproduction traits.Evaluation of alternative breeding programmesChapters 3, 4 and 5 deal with the optimization of selection for production and reproduction traits in dam lines of pig breeding programmes. For this purpose, an existing stochastic simulation model for sire lines was adapted for darn lines. Several factors of the breeding programme that determine the way of selection were studied.With the simulation model, effects of selection over 25 years were evaluated. Attention was focussed on changes in production and reproduction traits and on increase of inbreeding coefficient. Traits were assumed to be affected by many unlinked loci, each of small additive effect. Selection of boars and gilts was on an index that combined estimated breeding values for production and reproduction traits. Estimated breeding values for production traits were based on individual performance data, whereas estimated values for reproduction traits were based on family information, using a multi-trait animal model.Effects of size of the nucleus population and sow/boar ratio were examined in Chapter 3. Population size was varied between 50 and 400 sows, and annual number of boars varied between 10 and 40.Increasing the number of sows had a large positive effect on selection response: an increase from 200 to 400 sows gave 11% more response. For most breeding organizations, this might be high enough to offset the extra costs for sow and test places.Variation in annual number of boars had a small influence on selection response, especially for large populations. A high number of boars was needed to keep the rate of inbreeding acceptably low. Therefore, use of a high annual number of boars (40) is recommended for dam lines.The objective of the study in Chapter 4 was to evaluate alternative selection and testing systems in dam lines. The stochastic simulation model was used to study effects of alternative systems on variances in family size, rate of inbreeding and response to selection.Two alternative testing systems were evaluated. A system of one boar tested per litter gave about 10% lower response to selection than a system of two boars tested per litter. The only advantage of the first system is that testing costs are lower.Differences in selection response between alternative selection systems were small. A restriction on the number of boars selected per litter (within full-sib family selection) had little influence on rate of inbreeding and on selection response. A restriction on the number of boars per sire (within paternal half -sib family selection) gave a small reduction in rate of inbreeding and in response to selection. Based on these results and those in Chapter 3, it could be concluded that increasing the number of boars is a better option for limiting the rate of inbreeding than within family selection.Multi-stage selection in dam lines is dealt with in Chapter 5. The simulation model was used to study effects of time of selection on accuracy of selection, response to selection and rate of inbreeding.First stage selection of boars was before the performance test. The proportion of boars selected in the first stage (p 1 ) was varied between 100% and 25%. From p 1 =100% to p 1 =50% the reduction in overall response was on average 3.5%, while from p 1 =50% to p 1 =25% the reduction was an additional 6%. The optimum of p 1 depends on the costs for testing, and on the size of the nucleus population relative to the total breeding pyramid. With a relatively large nucleus breeding herd, a low proportion of boars tested can be justified.Breeding schemes with sequential culling of sows (weaned sows competing with replacement gilts) were compared to schemes without sequential culling (no genetic culling after weaning). Sequential culling gave on average 2-3% extra response.Most pig breeding organizations have a nucleus with sire as well as dam lines. In the short term, total capacity of the nucleus herds (number of sow places) and testing capacity are fixed, but the distribution over lines can be varied. Therefore, optimization of population size and testing capacity must be done simultaneously for sire and dam lines. The objective of the study in Chapter 6 was to optimize distributions of nucleus and testing capacity over lines in various situations.Effects of alternative distributions of nucleus places for sows and testing capacity for boars on total selection response were studied with an approach that might be referred to as semi -deterministic. The distributions were optimized with a deterministic model, whereas parameters used in this model were derived from stochastic simulation.Conclusions (for a four-way crossbreeding system) were as follows:- The optimum ratio of sow places for sire lines to sow places for dam lines was about 1 : 2.- The optimum ratio of boar testing capacity for sire lines to boar testing capacity for dam lines was about 1 : 1.- Reductions in total selection response at suboptimal distributions were limited as long as no extreme values were chosen.- Optimum ratios depended on total testing capacity relative to total number of sow places in the nucleus. Optimum ratios were also sensitive to testing system (maximum number of boars tested per litter). Culling rate of boars after test (for conformation or semen quality) and crossbreeding system (three-way vs. four-way cross) had only slight influence.Main conclusionsFrom the studies in this thesis (and from some related studies), the following main conclusions could be drawn for current pig breeding programmes:- The breeding goal should be based on economic efficiency of piglet production and fattening herds; the model in Chapter 1 can be used for this purpose. For the short-term benefits of a breeding organization, some modification of the breeding goal can be necessary when performance level of a trait deviates much from that of competitors. For the long-term benefits, modification can be necessary when the level of a trait would become a biological limit for further improvement of other traits.- Selection response for production and reproduction traits in dam lines can be much increased by enlargement of the nucleus. Additional nucleus sows do not necessarily have to be accompanied by additional test places for boars, because efficient selection of young boars (on pedigree index) is possible before the test. From a large proportion of the litters in dam lines, no boars have to be tested.- For dam lines, a high turn-over of breeding boars is necessary. This is a better option for limiting rate of inbreeding than restrictions on family size in selection and testing. With 40 boars per year, rate of inbreeding can be limited to 0.5% per year.- In a pig breeding programme with specialized sire and dam lines, sire lines can be much smaller (± 50%) than dam lines. Testing capacity should be equally distributed over sire and dam lines.

Dutch; Flemish. Het doel van deze studie is de optimalisatie van de gecombineerde selectie-strategie op produktie- en reproduktiekenmerken in varkensfokkerijprogramma's. Het onderzoek is met name gericht op genetische vooruitgang en inteelttoename