Ruminal pH Sensing for Monitoring Volatile Fatty Acid Concentrations in Response to Short-Term Dietary Disruption

The purpose of this study was to investigate the potential of using ruminal pH measurements to track time-series, ruminal volatile fatty acid (VFA) concentrations occurring in response to short-term dietary disruption. Four ruminally cannulated, dry, Holstein dairy cows were individually housed and assigned to 4 treatments in a Latin Square design. Treatments differing in forage-to-concentrate (F:C) ratio (100:0 to 55:45) were used because they were expected to result in large differences in VFA concentration, over which the relationships between pH and VFA could be robustly evaluated. Each sampling period lasted 36 h. Animals were removed from pasture and fasted for 24 h, after which time, they were fed their treatment ration for 2 h and sampled for rumen fluid hourly for 12 h. Rumen fluid samples were analyzed immediately for pH, frozen, and subsequently analyzed for VFA concentrations using gas chromatography. Animals were returned to pasture for 7 d between sampling periods. To confirm that the short-term dietary disruptions resulted in expected variation in VFA concentrations, mean VFA concentrations during each animal-period (n = 16) were analyzed using a linear-mixed effects model with fixed (linear and quadratic) effects for F:C ratio and random effects for animal and period. Results indicated significant changes in VFA concentration across F:C ratio, but no significant shifts in VFA molar proportions, perhaps due to the short-term nature of the feeding protocol. To explore opportunity to use pH measurements to explain variability in VFA concentrations in real-time across dietary conditions, a linear, mixed-effect model was used to link the time-series measurements (n = 207). The VFA concentrations were analyzed with linear-mixed effect models using linear and quadratic terms for pH, and random effects for animal and period. These models had poor accuracy, with residual error variance ranging from 21 to 38%, and residuals patterning significantly with F:C ratio. The data suggest that pH may lack reliability for VFA prediction in short-term feeding scenarios differing considerably in F:C ratio.