Expected performance of selection for uniformity of birth weight in real populations of guinea pigs

The guinea pig is a species of significant economic and nutritional importance, especially in the Andean region of South America. Measuring within-litter variability of birth weight (BW) is a challenge in this species because there are few records per litter. The objective of this study was to evaluate the performance of heteroscedastic models under realistic pedigree structures through simulation, as well as to determine the potential of selection for uniformity of within-litter BW in guinea pig populations. Real pedigree and BW data from three guinea pig populations (P1, P2, and UNH) were analyzed. BW data were simulated based on prior estimates of variance components and by maintaining the actual mating scheme. Parameter estimation was performed by fitting a complete model (with litter effect) and an incomplete model (without litter effect) under Frequentist and Bayesian approaches. The results showed that the heteroscedastic model allows for a reasonable estimation of the variance components and prediction of genetic values. In population P1, which had greater pedigree depth and a larger number of records, the results were more consistent compared to populations P2 and UNH. Comparing the complete and incomplete models, the accuracy of the genetic values for the residual variance were similar. After five generations of selection, a reduction in BW variability is expected. Overall, despite their low litter size, our results suggest that selection for uniformity of BW in guinea pig populations is potentially viable, but its efficiency is contingent upon a sound pedigree structure and a sufficient number of records.