ABSTRACT

Background and Aim: Guinea pig production has considerable economic, nutritional, and cultural importance in Andean food systems. However, no mechanistic dynamic model is currently available to simultaneously predict growth and chemical body composition in this species. This study aimed to develop and externally validate a mechanistic nutrient-partitioning model for predicting live body weight (BW), empty BW, and chemical body composition in male guinea pigs (Cavia porcellus) of the Peru genotype.

Materials and Methods: A mechanistic dynamic model integrating dry matter intake, maintenance requirements, protein and energy utilization, nutrient partitioning, and tissue deposition was developed and implemented in R. External-validation was performed using two independent datasets generated in Colombia. The first comprised longitudinal BW records from 184 male guinea pigs raised under commercial conditions, whereas the second included whole-body chemical composition measurements from 24 male guinea pigs. Predictive performance was assessed using Lin’s concordance correlation coefficient (CCC), coefficient of determination, root mean square error (RMSE), relative root mean square error, mean absolute error, and mean bias (MB).

Results: The model predicted live BW with almost perfect agreement between observed and predicted values (mean CCC = 0.931; RMSE = 149.6 g; MB = −1.39 g). Final BW and empty BW were predicted with high accuracy (CCC = 0.99 and 0.98, respectively). Total body protein and ash showed almost perfect agreement (CCC = 0.91 for both), whereas body water showed substantial agreement (CCC = 0.70). In contrast, body fat was poorly predicted (CCC = 0.04; RMSE = 121.9 g) and was systematically underestimated. Prediction errors increased in animals weighing >1000 g, indicating reduced accuracy during the later growth phase.

Conclusion: This is the first mechanistic dynamic nutrient-partitioning model developed specifically for guinea pigs. The model provides a biologically interpretable framework for simultaneously predicting growth and chemical body composition and may support nutritional assessment, precision-feeding, and decision-making in intensive production systems. However, parameters governing lipid accretion require further refinement before the model can be broadly applied across genotypes, sexes, and production environments.

Keywords: body composition, dynamic modeling, growth prediction, guinea pig, mechanistic model, nutrient partitioning, Peru genotype, precision-feeding.