Crecimiento corporal de novillas de reemplazo en fincas lecheras de Costa Rica. 1.Tipificación de curvas de crecimiento en novillas Holstein y Jersey
Archivos
Fecha
1997
Autores
Vargas Leitón, Bernardo
Solano Patino, Cesar
Título de la revista
ISSN de la revista
Título del volumen
Editor
Asociación Latinoamericana de Producción Animal
Resumen
El estándar de crecimiento para las hembras de reemplazo de lecherías de Costa Rica fue calculado a través del ajuste de las curvas promedio de crecimiento de las razas Holstein y Jersey a seis modelos matemáticos, cinco de los cuales fueron del tipo exponencial: Brody (Br), Richards (Ri), Bertalanffy (Be), Gompertz (Go) y Logístico (Log) y uno intrínsicamente lineal (Men), utilizando mínimos cuadrados con el método de Marquart. Las curvas reales fueron obtenidos a través del cálculo de la media de peso por cada semana de edad desde 1 hasta la 104, utilizando 20 310 observaciones de peso de 4 264 animales de raza Holstein en 80 fincas y 9 520 observaciones de 1 689 animales Jersey en 51 fincas recolectadas entre 1988 y 1995 en tres zonas geográficas de Costa Rica correspondientes a cuatro zonas de vida: Bosque húmedo (bh), Bosque seco (Bs), Bosque muy húmido (bmh) y Bosque pluvial (bp). La selección del modelo con mayor potencialidad fue realizada a través de la comparación de cinco estadígrafos de bondad de ajuste: El error medio de predicción (EMP), la desviación estándar de los residuos (DE), el rango de residuos (RA), el estadístico estándar de los residuos (DE), el rango de residuales (RA), el estadístico de Durbin-Watson (DW) y el coeficiente de determinación (R2). Una vez seleccionado el modelo de mejor ajuste, fueron calculados los parámetros de las curvas de crecimiento para cada zona de vida y grupo de año de nacimiento (1988-1991 y 1992-1995). Los resultados del ajuste de los modelos a las curvas promedio mostraron que los seis modelos poseen una alta bondad de ajuste, siendo los valores de R2 superiores a 0.98, valores de EMP inferiores a 6.7% y una DE máxima de 4.5 kg. Las estadísticas RA Y DW mostraron una gran variación, oscilando entre 12 y 22 kg y entre 0.38 y 2.32, respectivamente. Los modelos de Br y Ri mostraron ser los más eficientes en la descripción del crecimiento de ambas razas con una pequeña ventaja del primero, por lo que fue seleccionado para representar el crecimiento de ambas razas. En el calculo de los parámetros de las curvas por zonas de vida y años de nacimiento se obtuvieron estadígrafos satisfactorios. Estas curvas mostraron grandes variaciones en el crecimiento relacionadas con las condiciones ambientales y poca variación entre los grupos de años de nacimiento.
The growth standard for replacement females from Costa Rican dairies was calculated by fitting the average growth curves of the Holstein and Jersey breeds to six mathematical models, five of which were of the exponential type: Brody (Br), Richards (Ri), Bertalanffy (Be), Gompertz (Go) and Logistic (Log) and one intrinsically linear (Men), using least squares with the Marquart method. The actual curves were obtained by calculating the mean weight for each week of age from 1 to 104, using 20 310 weight observations of 4 264 Holstein animals in 80 farms and 9 520 observations of 1 689 Jersey animals in 51 farms collected between 1988 and 1995 in three geographic zones of Costa Rica corresponding to four life zones: humid forest (bh), dry forest (Bs), very humid forest (bmh) and rainforest (bp). The selection of the model with the highest potential was made through the comparison of five goodness-of-fit statistics: Mean prediction error (MPE), the standard deviation of the residuals (SD), the range of residuals (RA), the standard statistic of the residuals (SD), the range of residuals (RA), the Durbin-Watson statistic (DW) and the coefficient of determination (R2). Once the best-fit model was selected, the parameters of the growth curves were calculated for each life zone and birth year group (1988-1991 and 1992-1995). The results of the fit of the models to the average curves showed that all six models had a high goodness of fit, with R2 values greater than 0.98, MPE values less than 6.7% and a maximum SD of 4.5 kg. The RA and DW statistics showed a large variation, ranging from 12 to 22 kg and from 0.38 to 2.32, respectively. The Br and Ri models were shown to be the most efficient in describing the growth of both breeds with a small advantage of the former, thus was selected to represent the growth of both breeds. In the calculation of the parameters of the curves by life zones and years of birth, satisfactory statistics were obtained. These curves showed large variations in growth related to environmental conditions and little variation between birth year groups.
The growth standard for replacement females from Costa Rican dairies was calculated by fitting the average growth curves of the Holstein and Jersey breeds to six mathematical models, five of which were of the exponential type: Brody (Br), Richards (Ri), Bertalanffy (Be), Gompertz (Go) and Logistic (Log) and one intrinsically linear (Men), using least squares with the Marquart method. The actual curves were obtained by calculating the mean weight for each week of age from 1 to 104, using 20 310 weight observations of 4 264 Holstein animals in 80 farms and 9 520 observations of 1 689 Jersey animals in 51 farms collected between 1988 and 1995 in three geographic zones of Costa Rica corresponding to four life zones: humid forest (bh), dry forest (Bs), very humid forest (bmh) and rainforest (bp). The selection of the model with the highest potential was made through the comparison of five goodness-of-fit statistics: Mean prediction error (MPE), the standard deviation of the residuals (SD), the range of residuals (RA), the standard statistic of the residuals (SD), the range of residuals (RA), the Durbin-Watson statistic (DW) and the coefficient of determination (R2). Once the best-fit model was selected, the parameters of the growth curves were calculated for each life zone and birth year group (1988-1991 and 1992-1995). The results of the fit of the models to the average curves showed that all six models had a high goodness of fit, with R2 values greater than 0.98, MPE values less than 6.7% and a maximum SD of 4.5 kg. The RA and DW statistics showed a large variation, ranging from 12 to 22 kg and from 0.38 to 2.32, respectively. The Br and Ri models were shown to be the most efficient in describing the growth of both breeds with a small advantage of the former, thus was selected to represent the growth of both breeds. In the calculation of the parameters of the curves by life zones and years of birth, satisfactory statistics were obtained. These curves showed large variations in growth related to environmental conditions and little variation between birth year groups.
Descripción
Palabras clave
NOVILLOS, GANADO DE LECHE, MILK CATTLE, COSTA RICA, HOLSTEIN, JERSEY, PRODUCCIÓN LECHERA, MILK PRODUCTION