Non-Linear Mixed Models for the Characterization of Growth Trajectory Measured by Ultrasound of ‘San Pedreño’ Creole Pig Breed From Colombia

The aim was to identify non-linear regression models, under the context of mixed models, to describe the growth trajectory of ‘San Pedreño’ Creole pigs based on two muscular measurements obtained using ultrasonography. The growth rates of 30 animals (9 barrows and 21 gilts) from 3.5 to 24 months of age, raised on traditional farms in different municipalities of Meta (Colombia), were evaluated over three replicates. In vivo ultrasonographic images for back fat thickness (BFT) and longissimus dorsi muscle width (LMA) were obtained for each animal. Ultrasound measurements of LMA and BFT were performed at a single anatomical point of the live pigs, in the dorsal midline region (left/right). Comparison between sexes was done with a one-way PERMANOVA. Correlations were done with the Spearman's r_s coefficient. The traits body weight (BW), BFT and LMA were calculated as a function of each animal age (days) using each of the following non-linear models: quadratic, exponential, von Bertalanffy, Michaelis–Menten, Logistic, Gompertz, Gaussian and Hill's equation. Traits were not significantly different between males and females (F = 0.437; p = 0.578). The variability of BFT and LMA tended to decrease with age (r_s = 0.265 and 0.317 for BFT and LMA, respectively, p < 0.001). Correlation between both carcass traits was high (r_s = 0.702, p < 0.001). The BFT growth in ‘San Pedreño’ Creole pigs can be described by applying the Hill equation and LMA growth by the quadratic model, applied to field data in live pigs.