Our objective was to develop a model to predict the acidosis potential of barley based on the in vitro batch culture incubation of 50 samples varying in bulk density, starch content, processing method, growing location, and agronomic practices. The model was an adaptation of the acidosis index (calculated from a combination of in situ and in vitro analyses and from several components of grain chemical composition) developed in Australia for use in the feed industry to estimate the potential for grains to increase the risk of ruminal acidosis. Of the independent variables considered, DM disappearance at 6 h of incubation (DMD6) using reduced-strength (20%) buffer in the batch culture accounted for 90.5% of the variation in the acidosis index with a root mean square error (RMSE) of 4.46%. To evaluate our model using independent datasets (derived from previous batch culture studies using full-strength [100%] buffer), we performed another batch culture study using fullstrength buffer. The full-strength buffer model using in vitro DMD6 (DMD6-FS) accounted for 66.5% of the variation in the acidosis index with an RMSE of 8.30%. When the new full-strength buffer model was applied to 3 independent datasets to predict acidosis, it accounted for 20.1, 28.5, and 30.2% of the variation in the calculated acidosis index. Significant (P < 0.001) mean bias was evident in 2 of the datasets, for which the DMD6 model underpredicted the acidosis index by 46.9 and 5.73%. Ranking of samples from the most diverse independent dataset using the DMD6-FS model and the Black (2008) model (calculated using in situ starch degradation) indicated the relationship between the rankings using Spearman’s rank correlation was negative (ρ = −0.30; P = 0.059). When the reduced-strength buffer model was used, however, there were similarities in the acidosis index ranking of barley samples by the models as shown by the result of a correlation analysis between calculated (using the Australian model) and predicted (using the reduced-strength buffer DMD6 model) acidosis index (ρ = 0.67; P < 0.001). Results suggest that our model, which is based on a reduced-strength buffer in vitro DMD6, has the potential to predict acidosis risk and can rank barley samples based on their acidotic risk. Nonetheless, the model would benefit from further refinement by expanding the database.
- Dry matter disappearance