Prediction of Digestibility and Intake of Roughage Diets Fed to Ruminants

Abstract

Predicting intake of roughage diets by ruminants uses simulation modelling. This study takes advantage of studies (e.g. Illus and Gordon 1991) that generalize across ruminants. The aim was to test the suitability of Illius and Gordon’s model, and modify it for predicting digestibility and intake of roughages in tropical ruminants. Data from seven studies on roughages fed to tropical ruminants were used. These studies had 235 intake and 70 digestibility measurements on diets of cattle (n=43), goats (n=5), and sheep (n= 22), with at least 5 animals.diet-1. Animals in all degradation studies, but one (study 2), were provided with protein supplement; roughages were fed non-supplemented except in one (study 3). The model comprises 13 digesta pools in the rumen and six postruminal pools. Fluxes among pools were modelled by a system of ordinary differential equations. Ingesta were partitioned into long and short particles, chemically into cell content (CC) and cell wall (CW), and biologically into digestible CW (DNDF) and indigestible CW (INDF). Once consumed, the CC in the fine particle pool is instantly digested; other particles are retained in the rumen for the duration of the lag phase. All solid matter in the rumen contributed to the level of fill. Maximum rumen load was determined as a function of body weight. Ingesta disappeared from the rumen at a rate determined by fractional rates (FR) of degradation (kdig or k2) and passage of particles (kp) for short and long particles. The model allows a maximum of eight meals/ day. A re-run of the model was needed after re-calibration: kp (/h) = 0.0907 x NDF-0.373 x kdig 0.40 (n=72; R2=0.37); and rumen DM load (RL) as a function of live weight (W) in proportion to the mature live weight (MW): RL (g/kg W) = e3.38*(W/MW)-0.27*CP-0.17 (n=26, R2=0.31), with CP in g.kg-1. Observed vs predicted digestibility and intake were respectively 0.52±0.080 vs 0.53±0.132 and 60±0.10.5 vs 59±0.12.9 for sheep, 0.69±0.057 vs 0.61±0.101 and 69±8.3 vs 63±20.8 for goats; and 0.59±0.097 vs 0.56±0.080 and 79±0.15.6 vs 66±0.12.0 for cattle. Relationships between observed and predicted digestibility (R2=0.17) and intake (R2=0.13) were poor. A re-run of the model increased the variation accounting for digestibility (R2=0.32) and intake (R2=0.49); showing two studies (R2=0.66-0.83 for intake; or 0.78-0.85 for digestibility) with good precision. Both rates of passage and digestion are largely properties of feeds, but rumen load is affected by animal and plant factors. Modified particulate rate of passage and rumen load increased predictions substantially. Consequently, more research is required to in this area to refine these parameters.