Modelos preditivos das características quantitativas da carcaça de cordeiros pantaneiros por ultrassonografia e analise de imagem in vivo com diferentes pesos corporais

Abstract

The aim of this study was to determine predictive models of carcass traits and comercial cuts yield (neck, palette, fixed ribs, floating ribs, loin, shank and low) of the carcasses of Pantaneiro lambs slaughtered in different body weights through real time ultrasound images. Data were obtained from slaughter weight the collection of ultrasound images of 45 lambs randomly divided into 5 groups according to slaughter weight (15, 20, 25, 30 and 35 kg) with 9 reps for each weight group. The images were collected the day before slaughter between 12 ° and 13 ° ribs after shearing the area to be assessed. Later images were analyzed by measuring the loin eye area (ULMA) and backfat thickness (UFAT). Statistical analysis was performed using Pearson correlation analysis between body weight (BW), ultrasound measurements and carcass characteristics and linear regression analysis with the statistical package XLSTAT. Three regression models between body weight , ultrasound measures (independent variables) and production cuts (dependent variable), three simple models, one for BW and one for ULMA and one for UFAT, and three models with use of a combination BW and ULMA or BW and UFAT and then the combination of BW + ULMA+ UFAT. There was a relationship between the animals weight estimated by ULMA (R ² = 0.82), thus showing increased muscle development linearly between 15 and 35 kg. The BW showed high correlation with commercial cuts. The leg presented higher correlations (0.96) and fixed ribs lower (0.79).The BW were regression equations dominant variable for generating predictive models, when added to ULMA and / or UFAT, little they influenced prediction models. The use of BW, ULMA and/or UFAT were only able to predict the performance palette. The BW was the independent variable that best explained the weight and muscle tissue percentage, fat and bone carcass half, and the inclusion of ULMA and/or UFAT improved little predictive models.