Effects of dietary brewery spent grain inclusion on egg laying performance and quality parameters of bovans brown chickens

Poultry rations are composed mainly of conventional cereal grains and proteins. Using non-conventional agro-industrial by-products may reduce the cost of production and thereby improve productivity. A study was conducted to evaluate the effect of dietary brewery spent grain inclusion on egg laying performance, quality parameters of bovans brown and profitability of the rations. A total of 200 pullets with similar body weight and age were randomly distributed to five treatments with four replications. The treatments were brewery spent grain (BSG) inclusion of 0, 10, 20, 30 and 40% levels for T1, T2, T3, T4 and T5, respectively. The CP and ME contents of treatment rations were 16.54-17.04% and 2871-2903 kcal/kg DM, respectively. Inclusion of the BSG in the layers ration did not affect dry matter intake, feed conversion efficiency and hen-day egg production (58.79, 57, 56.11, 55.11 and 54.13% (SEM=0.92)). Likewise, the inclusion of the BSG in the diets did not affect (albumen, yolk, shell) quality. However, feed cost decreased with the increasing level of the BSG in the rations because of its lower purchasing price. To conclude, a 40% inclusion of BSG in the diet of layers does not affect the production and the quality of the eggs and it can be economically profitable.(AU)

Effects of dietary brewery spent grain inclusion on egg laying performance and quality parameters of bovans brown chickens

Poultry rations are composed mainly of conventional cereal grains and proteins. Using non-conventional agro-industrial by-products may reduce the cost of production and thereby improve productivity. A study was conducted to evaluate the effect of dietary brewery spent grain inclusion on egg laying performance, quality parameters of bovans brown and profitability of the rations. A total of 200 pullets with similar body weight and age were randomly distributed to five treatments with four replications. The treatments were brewery spent grain (BSG) inclusion of 0, 10, 20, 30 and 40% levels for T1, T2, T3, T4 and T5, respectively. The CP and ME contents of treatment rations were 16.54-17.04% and 2871-2903 kcal/kg DM, respectively. Inclusion of the BSG in the layers ration did not affect dry matter intake, feed conversion efficiency and hen-day egg production (58.79, 57, 56.11, 55.11 and 54.13% (SEM=0.92)). Likewise, the inclusion of the BSG in the diets did not affect (albumen, yolk, shell) quality. However, feed cost decreased with the increasing level of the BSG in the rations because of its lower purchasing price. To conclude, a 40% inclusion of BSG in the diet of layers does not affect the production and the quality of the eggs and it can be economically profitable.

GERENCIAMENTO DE ÍNDICADORES DE PRODUÇÃO EM LOTES DE MATRIZES DE POSTURA COMERCIAL UTILIZANDO REDES NEURAIS ARTIFICIAIS

Este trabalho teve por objetivo utilizar redes neurais artificiais para predizer indicadores de produção dos lotes de matrizes de postura comercial em uma empresa matrizeira de aptidão postura comercial. Foram utilizados dados de 2007 a 2014, usando 51 lotes das linhagens Isa Brown e Bovans Withe totalizando 405.511 aves avaliadas. A empresa participante do experimento forneceu um banco de dados com informações de vários lotes pelo período de 7 anos. Para a construção das redes neurais artificiais foram utilizados os programas computacionais NeuroShell®Classifier e NeuroShell® Predictor, desenvolvidos pela Ward Systems Group. O programa identificou as variáveis escolhidas como entradas para cálculo do modelo preditivo e variável de saída aquela a ser predita. Neste trabalho foram utilizados 2.370 linhas para o treinamento, outras 593 serviram como testes para validação das predições, para obtenção dos resultados as redes neurais artificiais passaram por duas fases, na primeira foram apresentados o treinamento para as redes utilizando -se todas as variáveis de entrada que antecederam as variáveis de saída de cada rede. A segunda fase destinou-se à validação dos modelos com todas as variáveis de entrada que antecederam as variáveis de saída como: ovos produzidos semanal, consumo real de ração semanal, ovo/ave/alojada, ovo comercial semanal, ovo incubável semanal, peso de ovo real, produção de ovos real e viabilidade. Em todos esses oito modelos as redes neurais artificiais foram bem ajustadas, apresentando um Coeficiente de Determinação Múltipla (R²) próximo de um (1), salientando que R² quanto mais se aproximar de 1 maior precisão. De acordo com os dados apresentados pelas RNAs conclui se que foi possível predizer as informações alocadas dos lotes de matrizes de aptidão postura comercial, gerando predições úteis nas oito redes apresentadas além de apontar falhas nos dados da planilha de produção.

The objective of this study was to use artificial neural networks to predict zootechnical indexes of layer parent stocks flocks in a genetic company. Data from 2007 to 2014 were used considering about 51 flocks of Isa Brown and Bovans White lineages, totaling 405.511 birds evaluated. The company which participated in the experiment provided a database with information on several flocks over a period of 7 years. For the construction of artificial neural networks were used the computer programs NeuroShell®Classifier and NeuroShell® Predictor, developed by the Ward Systems Group. The program identified the variables chosen as inputs to calculate the predictive model and the output variable to be predicted. In this study 2.370 lines were used for training, others 593 served as tests to validate predictions. To obtain the results, the artificial neural networks went through two phases. In the first one, the training for the networks was presented using all the input variables that preceded the output variables of each network. The second phase was focused on validating the models with all input variables that preceded the output variables such as viability, fertile eggs production, utilization, egg / bird / housed and feed intake. In all of these models the artificial neural networks were well adjusted, presenting a high Multiple Determination Coefficient (R²), emphasizing that R² the closer it comes to 1, higher is the precision. According to the data presented by the RNAs, it is concluded that it was possible to predict the information allocated from the commercial posture aptitude matrices, generating useful predictions in the eight networks presented in addition to pointing out flaws in the data of the production.