Total volatile fatty acid concentrations are unreliable estimators of treatment effects on ruminal fermentation in vivo.

Volatile fatty acid concentrations ([VFA], mM) have long been used to assess the effect of dietary treatments on ruminal fermentation in vivo. However, discrepancies in statistical results between [VFA] and VFA pool size (VFAmol) possibly related to ruminal digesta liquid amount (LIQ, kg) indicate potential issues with the use of [VFA]. We investigated relationships among [VFA], VFAmol, and LIQ measured 2 h postfeeding using individual lactating cow data (n=175) from 7 separate feeding studies. Regression analyses were performed using mixed models with "study" as a discrete random variable. The mean across studies and average range of values within studies, respectively, were 151 and 75 for [VFA], 11.2 and 9.8 for VFAmol, 73.3 and 41.0 for LIQ, and 289 and 83 mmol/kg for rumen fluid osmolality. Liquid amount changed with VFAmol (3.76 VFAmol+31.2; average within-study R2=0.69), but the relationship was weak between [VFA] and LIQ (0.524 LIQ+112.8; average within-study R2=0.12). The relationship between LIQ and VFAmol was likely a function of the osmotic gradient between rumen liquid and blood. The VFA are a major ruminal solute; VFAmol amounts can affect water flux in the rumen as similar tonicities of rumen fluid and blood are maintained. This also has a damping effect on ruminal solute concentration, creating the weak relationship between [VFA] and LIQ. Within studies, similar [VFA] were found in LIQ differing by 30 kg or more. The difference between minimum and maximum LIQ within cow within study was 12.7 kg (standard deviation=7.1), so inclusion of "cow" in analyses did not correct for the variation in LIQ. To allow valid comparisons of experimental treatments, responses must be on an equivalent basis; concentrations in different LIQ are not on an equivalent basis and so are not valid to use for comparing treatment effects. The [VFA] changed with VFAmol (5.80 VFAmol+86.3; average within-study R2=0.56). However, the ratio of [VFA] to VFAmol ranged from 9.0 to 24.1 as a function of 1,000/LIQ; this reflects the inherent calculated relationship among the variables. The varying relationship of [VFA] to VFAmol further indicates that [VFA] is not an appropriate measure to evaluate the progress or effect of treatments on ruminal fermentation. Predictions of LIQ and VFAmol using cow and ruminal measures were insufficiently precise to be used in research. Previously drawn conclusions based on [VFA] need to be reevaluated, and alternate evaluations for in vivo ruminal fermentation are needed.