The most successful method used for improving the growth rate of broilers is genetic selection. Improvements in nutrition, housing and disease resistance have been impressive, yet genetic selection is purported to have contributed the majority of the tremendous increase in growth rate that has taken place over the past 50 years (McKay, 2008). Many selection strategies are available, but not all are suitable, as the choice is dependent on the objective of the breeder. Selection strategies are bound to change over time as different traits become more important, and this has been the case in the broiler industry: focus was initially placed predominantly on growth rate, but the negative genetic correlation that exists between growth rate and reproductive and liveability traits has forced breeders to change their position, especially as growth rate has almost reached its upper limit and reproductive traits lag behind. This has resulted in a change from single trait to multiple trait selection. In the exercise reported here, four selection strategies commonly used for single trait selection, namely individual, between family, within family and family-index selection, were applied to a simulated broiler population using the Monte Carlo method of simulation, and constructed with the use of genetic parameters obtained from the literature. Theoretical and simulated methods of the four selection strategies were compared. A fifth selection strategy, index selection, was applied to represent multiple trait selection. The relative merit of each selection procedure was then compared, as well as the results obtained from the theoretical and simulated methods. Construction of the selection index was complex in comparison to single trait selection, as each trait included in the index had to be assigned an economic value. This value is representative of the relative importance of that trait to the overall profitability, or ability to save costs in the operation. Therefore traits favourable to profitability, or having the ability to reduce production costs, are given a heavier weighting and will consequently achieve a relatively larger improvement when applied to the selection index. A model was constructed using production rates, income and costs to represent the current overall economic situation in the industry. This was then used to determine cost economic values, which represent the saving in cost per unit improvement in each of the economically important traits, and revenue economic values, calculated as the value of each unit improvement attained in each of the economically important traits. Body weight remains the most profitable trait in a broiler enterprise; however breeder egg production is equally important as the industry would fail without sufficient day-old broilers. Therefore, it would be beneficial to determine whether current egg production levels could be maintained, or even improved, whilst improvement is made to the growth rate of the progeny. The above statement was found to be possible with the use of index selection. This multiple trait selection strategy proved capable of defying the negative genetic correlation that exists between body weight and egg production by improving egg production to 60 weeks by eight eggs, and body weight at 35 days by 259 grams. Furthermore, in some cases index selection was able to achieve improvements in some traits greater than those attained with single trait selection, whilst simultaneously improving certain negatively correlated traits. Index selection has illustrated its superiority over single trait selection strategies and its relative value to the poultry industry. / Thesis (M.Sc.Agric.)-University of KwaZulu-Natal, Pietermaritzburg, 2009.

Poultry have many behavioural, structural and biological features that are ideal for domestication and for meat and egg production (Appleby et al., 1992). Because of the importance of poultry meat and eggs to the human population, breeders and farmers are always looking for ways of improving these traits. Artificial selection is the primary method of trait improvement, and involves selecting individuals with the highest breeding values as parents in each generation. There are a number of different methods of artificial selection, including: individual selection, between family selection, within family selection, family-index selection and index selection. In order to maintain a good response to selection breeders are constantly striving to improve the effectiveness and accuracy of the different methods of artificial selection for traits of economic importance. One method of achieving this goal is the use of computer models. Computer models can be used to simulate selection strategies and to predict what strategy will be the most appropriate for the improvement of a particular trait. This is important as all traits are influenced by many different genetic and environmental factors (Falconer and Mackay, 1996). This investigation was designed to compare the effectiveness of five different artificial selection strategies, namely individual selection, between family selection, within family selection, family index selection and index selection. Five computer models were developed using Microsoft Excel 2000 and these models were then used to compare the efficiencies of the five selection strategies for four different traits. The selection techniques were applied to an artificially, randomly generated population of 500 chickens. The four traits were egg weight with a heritability of 0.51, egg production with a heritability of 0.22, age at first egg with a heritability of 0.41 and body weight with a heritability of 0.55. Firstly, each of these traits were selected for independently using the first four selection methods and secondly the traits were selected for two at a time using index selection. The most significant results obtained from the single trait simulations were that for all traits family-index selection produced the best response to selection in the initial generations and between family selection produced the best response in the later generations. The traits with a higher heritability (egg weight and body weight) responded better to individual selection than they did to within family selection and between family selection in the initial generations. However, within family selection and between family selection proved to be more effective for traits with a low heritability such as egg production. Individual selection and family-index selection resulted in a very rapid decline in the standard deviation of all the traits. Between family selection resulted in the slowest drop in the standard deviation of all the traits, which is why this technique produced the best responses to selection in the later generations. The impact of the correlations between the economically important traits were evident from the results of index selection. For example, egg production is negatively correlated with egg weight making it difficult to gain a correlated response in both these traits simultaneously. Furthermore, egg production is negatively correlated with age at first egg implying that early maturing birds will lay more eggs, however, these eggs will be lighter. The majority of the results obtained were to be expected. Family-index selection takes all the information about an individual's breeding value into account resulting in this method of selection consistently identifying the most desirable individuals being selected. It is therefore the preferred method of selection under all circumstances. It is, however, often not economically and practically efficient to incorporate this technique and the use of another method of selection usually proves to be more beneficial. Individual selection proved to be most effective when applied to traits with high heritabilities, due to the fact that this method selects individuals based on their own phenotypic values. For traits with a high heritability, an individual with a good phenotypic value will have a good breeding value. Between family selection and within family selection proved better for traits with lower heritabilities. For traits with a low heritability the phenotypic value of an individual is a poor indicator of its breeding value. Information from a number of relatives may thus improve the accuracy of prediction of the breeding value by accounting for the influence of environmental effects. The use of computer models to simulate the selection techniques proved very successful in illustrating the effectiveness of the different selection techniques under various genetic and environmental conditions. The models may also prove to be very effective from an educational perspective. / Thesis (M.Sc.)-University of Natal, Pietemaritzburg, 2003.