Modelagem individual e em tempo real das exigências nutricionais de suínos em crescimento

• Autor(a) principal: Hauschild, Luciano
• Data de Publicação: 2010
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Manancial - Repositório Digital da UFSM
• Texto Completo: http://repositorio.ufsm.br/handle/1/4315
• topic: Exigências nutricionais; Modelos de simulação; Variabilidade animal; Suínos; Animal variability; Nutrient requirements; Pigs; Simulation models; CNPQ::CIENCIAS AGRARIAS::ZOOTECNIA

Two studies were realized with the objective (i) to describe and to establish the limitations of the empirical and factorial methods used to estimate nutritional requirements and (ii) develop a mathematic model able to estimate the daily nutrient requirements of individual pigs in real time. For this purpose, in the first study a systematic analysis was carried out on the lysine - net energy (Lys:NE) ratios estimated through the empirical and factorial methods with the help of the InraPorc® growth model. Based on analysis of data on 68 pigs, Lys:NE ratios were estimated for three feeding phases. In the empirical method, population response was determined on eleven different Lys:NE ratios. Response was evaluated with ADG and FCR being used as variables. In the factorial method, the Lys:NE ratio for each animal was estimated through model inversion. In the second study, it was developed a model which includes an empirical and mechanistic component. The empirical component estimates the daily feed intake (DFI) and BW based on individual pig information collected in real time. The mechanistic component estimates the optimal concentration of amino acid that must be offered to pigs to meet their requirements. In establishing limitations, it was observed that the factorial method, when applied to the mean animal, does not allow for estimation of the Lys:NE ratio to maximize response in a heterogeneous population which is to be fed over a long period of time. In the empirical method, while these aspects are considered, the estimated response is fixed and cannot be used for other intervals or populations. The empirical component results indicate that the DFI and BW trajectories of individual pigs fed ad libitum could be predicted 1 (DFI) or 7 days (BW) ahead with an mean absolute percentage error of 12 and 2 %, respectively. The mechanistic component estimates the optimal lysine to net energy ratio with reasonable between ‐ animal (average CV = 7 %) and over ‐ time variation (average CV = 14 %). Thus, the nutrient requirements estimated by the model are animal and time dependents and they follow in real ‐ time the individual DFI and BW growth patterns. The results from first study confirm that the two methods needs lead to different recommendations and have important limitations that should be considered when used to optimize the response of an individual or a population of pigs. The model developed is part of the development of precision feeding techniques allowing the right amount of feed with the right composition at the right time to be provided to each pig in the herd.