Modelling broiler populations for purposes of optimisation.

Abstract

With the narrow margin of profit in the broiler enterprise, how can producers increase profit potential? It is not an easy task to answer this question since the net financial return depends on many factors; some are related to the animal, some to the feed, some to the environment and others are outside the production system, like availability and cost of labour and capital. Many researchers have attempted to improve the efficiency of the system using alternative management strategies and to develop a unified theory that could simultaneously evaluate all the relevant factors and the interactions between them. Simulation models are seen as the most promising means of moving this subject forward. Geneticists are continually improving the potential growth rate of broilers, yet there has been little change in feed specifications for these birds over the past few decades. Only recently has it been possible to make use of simulation models to optimise the feeds and feeding programs of modern broiler strains at a commercial level, but little testing of these programs has been carried out. What is needed is a thorough investigation of these models, which at present are based on an individual, as opposed to a population response. Modelling plays an increasingly important part in animal science and research as a way of organizing and evaluating the large body of existing knowledge. With the use of an accurate description of the potential growth rates of broiler genotypes, it is possible to make more efficient use of growth models which are becoming more abundant in the industry and which, in turn, enable the nutritionist or producer to predict the performance of animals when subjected to a given feed or feeding programme. The predictions made by most of the growth models now available are based on individual animals, and the results obtained may be inadequate in optimising the nutrient requirements of a broiler population because of the variation that exists in these populations. Variation in performance traits in broilers may be the result of variation in the genotype, in the environmental conditions within the house, and in the composition of the feed offered to the birds, and these sources of variation cannot all be accommodated in a model that simulates the food intake and growth of just one bird. But if variation is to be incorporated into growth models, it is necessary to ascertain the effects of variation in the various genetic parameters on the mean response of the population. A sensitivity analysis is useful in accomplishing this objective. Similarly, it is important to know what the optimum size of a simulated population should be, that takes account both of the accuracy of the simulation and the time taken to complete the exercise. This is especially important when optimisation routines are followed, as such calculations are time consuming. As a means of addressing these issues, simulation exercises were conducted using EFG Broiler Growth Model version 6 and EFG Broiler Optimiser Model version 1 (EFG Software, 2006) to determine: (a) whether it is worth generating a population when optimising feeds and feeding programs for broilers, rather than using the average individual, (b) the size of the population required to obtain an accurate estimate of the population response when optimising the feeding program for different objective functions, (c) the effect of changing the value of genetic parameters such as mature protein weight, rate of maturing, feathering rate and the maximum lipid:protein ratio in the gain on the optimum amino acid contents and nutrient densities of broiler feeds, and (d) the effect of variation in nutrient composition of different batches of feed, which have the same nutrient profile but different qualities of the main protein source, on broiler performance. A review of sources of variation in the nutrient content of poultry feed was conducted, and simulation exercises were carried out to determine to what extent broiler performance is affected by the segregation or breakage of pellets into small pieces at the time of delivery and along the feed conveyor within the broiler house, by the change in nutrient quality that might occur along the conveyor, and by the microclimates that develop in a longitudinally ventilated broiler house. The tendency in broiler marketing in most parts of the world is to sell broilers cut up, as portions or deboned after evisceration, rather than selling whole birds. Estimation of the growth rates of carcass parts is therefore of considerable importance if simulation models are to be useful in optimising the feeds and feeding programmes of broilers under different conditions. Allometric equations are used in the EFG broiler growth model to predict the weights of these carcass parts from the weight of body protein at the time. These equations are based on data collected many years ago, and it would be useful to determine whether they are still relevant in the face of announcements by the major broiler breeding companies that tremendous strides have been made in improving breast meat yield, for example, by judicious selection. For the purpose of this investigation it was important to determine to what extent the weights of the physical parts varied at the same body protein weight, thereby enabling a more accurate estimation of the variation that could be expected in these weights when developing a population response model. Towards this end, experiments were conducted to determine the effect of dietary protein content on the performance of Cobb and Ross broilers, including mortality and uniformity, and on the allometric relationships between the physical and chemical components of the body and body protein. The overall objective of these exercises was to address issues relating to the use of simulation models in predicting food intake and growth of broilers, in optimising the amino acid contents and nutrient densities of feeds for broilers, and in representing a population of broilers when the performance of only one bird is simulated at a time.