Czech J. Anim. Sci., 2021, 66(11):441-449 | DOI: 10.17221/98/2021-CJAS

Modelling the growth of rearing cattleOriginal Paper

Hanna Unterauer, Norbert Brunner, Manfred Kühleitner ORCID...^*

Department of Integrative Biology and Biodiversity Research (DIBB), University of Natural Resources and Life Sciences (BOKU), Vienna, Austria

Scientific growth literature often uses the models of Brody, Gompertz, Verhulst, and von Bertalanffy. The versatile five-parameter Bertalanffy-Pütter (BP) model generalizes them. Using the least-squares method, we fitted the BP model to mass-at-age data of 161 calves, cows, bulls, and oxen of cattle breeds that are common in Austria and Southern Germany. We used three measures to assess the goodness of fit: R-squared, normalized root-mean squared error, and the Akaike information criterion together with a correction for sample size. Although the BP model improved the fit of the linear growth model considerably in terms of R-squared, the better fit did not, in general, justify the use of its additional parameters, because most of the data had a non-sigmoidal character. In terms of the Akaike criterion, we could identify only a small core of data (15%) where sigmoidal models were indispensable.

Keywords: Akaike information criterion; Bos taurus; Bertalanffy-Pütter differential equation; simulated annealing

Published: November 5, 2021  Show citation