MODELOS NÃO LINEARES PARA CURVA DE CRESCIMENTO DE FÊMEAS BOVINAS JOVENS

Abstract

Brazil ranks third worldwide in milk production, with 35.4 billion liters produced in 2023, highlighting the expansion of the national dairy sector. To support this growth, herd management practices based on zootechnical indicators—especially during the rearing phase— are essential. Growth curves are valuable tools to describe animals’ body development over time and to estimate biological parameters such as mature weight, growth rate and age at first calving. This study evaluated growth patterns of ½ Holstein × ½ Gir heifers raised at IF Goiano – Campus Morrinhos, monitored monthly from birth to 690 days of age. Seven heifers were included and monthly body weights were fitted to several nonlinear models (Gompertz, von Bertalanffy, Meloun I, Meloun II and Weibull). Model performance was assessed using percentage of convergence, coefficient of determination (R²), mean prediction error (MPE), Akaike information criterion (AIC) and Bayesian information criterion (BIC). Results indicated poor performance for Meloun I and Weibull models (low convergence and high prediction errors). The Gompertz model provided the best fit (R² = 0.99; MPE = 29.8; AIC = 196.7; BIC = 196.5) with 100% convergence, while von Bertalanffy showed full convergence but inferior fit by other criteria. We conclude that the Gompertz model is the most suitable to describe growth of ½ Holstein × ½ Gir heifers under the studied conditions, offering reliable estimates to support rearing management, genetic selection and productivity improvements in dairy herds.