Modelling digestion in calves fed milk replacer and solid feed

In young calves, intake of milk replacer is combined with feeding of solid feed. The initiation of solid feed intake stimulates rumen development, and over time nutrient yield per unit of solid feed intake increases. However, the interactions between nutrient intake from milk replacer and from solid feed in terms of implications for growth as well as animal health and rumen development have only recently been investigated (e.g., Berends et al. 2012). Of particular interest was the observation that NDF digestibility, but not the digestibility of other components of the diet, increased with age and BW when intake level (g kg 1 BW d 1) was held constant. In order to better understand these interactions, an existing dairy cattle rumen fermentation model (Dijkstra et al. 1992), previously modified for use with finishing beef cattle (Ellis et al. 2014), was adapted to simulate fermentation and digestion in calves. Areas of the model evaluated included prediction of rumen volume (VRu), rumen fractional passage rate (kp), pH and nutrient digestibility. The evaluation database size varied depending on the predicted variable of interest due to inconsistent availability of observed data. However, the average BW of calves in the full dataset (41 treatments from seven studies) was 143 (99.3) kg, and DMI was 1.1 (90.15) kg d 1. Model predictions were evaluated by concordance correlation coefficient (CCC) analysis. In order to represent the development of the rumen microbial population upon initiation of solid feeding, the cellulolytic enzyme activity data of Sahoo et al. (2005) was used to scale the maximum fractional rate of fibre fermentation, relative to its value for a mature ruminant, via an equation based on cumulative DMI. This modification allowed the model to predict the directional change in NDF digestibility observed by Berends et al. (2012) (46% at 108 kg and 56% at 164 kg BW), although the magnitude of change was less (47 and 50%, at 108 kg and 164 kg BW, respectively). To improve simulated survival of fibrolytic microbes within the rumen, kp(solids) was divided into kp(forage) and kp(concentrate), with the corresponding division of substrate pools. The variable VRu was predicted with a CCC of 0.69, kp(forage) with CCC0.49 and kp(concentrate) with CCC0.61. The model over-predicted pH with a CCC0.33. Lack of representation of the development of rumen motility and presence of lactic acid may account for biased predictions of kp and pH, respectively. Overall, the model performed reasonably well on these data, but results also highlight areas for further developments.