Prediction models of reticulorumen particles and solutes passage rate in growing goats

International audience

English. The reticulorumen (RR) fractional passage rate (kp; /h) of particles and solutes plays an important role in fiber digestion, methane production, and microbial yield. However, none of the available models for predicting RR kp consider individuals' characteristics of growing goats. The objective was to develop empirical models for predicting the RR kp of particles and solutes in growing goats. Our database involved 175 individual records of castrated males (n = 61), females (n = 57), and intact males (n = 57) growing Saanen goats fed ad libitum, 75% or 50% of ad libitum. Goats were slaughtered around 15, 22, 30, 37, or 45 kg BW. We used Akaike’s information criterion to select the best prediction models. We evaluated the predictive ability of these models using Lin’s concordance correlation coefficient (CCC) and RMSE of prediction (RMSPE) in a 4-fold cross-evaluation. The DM intake (DMI; kg/day), potentially digestible NDF intake (pdNDFI) level (g/kg BW), and RR wet pool size (kg) demonstrated similar importance in predicting RR kp of solutes (CCC = 0.59; RMSPE = 0.050 /h or 34.43%). However, when RR wet pool size was not included in the model, RR kp of solutes could still be precisely and accurately predicted using only DMI level (g/kg BW) (CCC = 0.47; RMSPE = 0.053 /h or 36.58%). The RR wet tissues and wet pool size (kg), NDF intake (NDFI) (kg/day), and indigestible NDFI (iNDFI):NDFI ratio were important predictors of RR kp of particles (CCC = 0.51; RMSPE = 0.0064 /h or 25.43 %). However, when RR wet tissues and wet pool size were not included in the model, iNDFI:NDFI ratio, NDFI level (g/kg BW), and RR kp of solutes presented greater importance in predicting RR kp of particles (CCC = 0.20; RMSPE = 0.0074 /h or 29.55%). Sex was not a significant predictor variable for the selected models. In summary, the RR kp of solutes was more dependent on feed intake level while the RR kp of particles was more dependent on diet composition and RR kp of solutes. Our models were precise and accurate for predicting RR kp of solutes (CCC = 0.57 and 0.47; RMSPE = 0.051 and 0.054 /h) and particles (CCC = 0.48 and 0.17; RMSPE = 0.0066 and 0.0076 /h) after cross-evaluation. This suggests that our models can be integrated into feeding systems with mechanistic approaches that simulate other reticulorumen functions, such as digestion, microbial growth, and methane emission.