Relationship between in situ degradation kinetics and in vitro gas production fermentation using different mathematical models

In vitro and in situ studies were conducted to evaluate the influence of different mathematical models, used to fit gas production profiles of 15 feedstuffs, on estimates of nylon bag organic matter (OM) degradation kinetics. The gas production data were fitted to Exponential, Logistic, Gompertz and a Sigmoidal model. Using only gas production parameters allowed poor prediction of in situ degradation. It was not possible to estimate the washout (W) and degradable (D) in situ fractions for all models, with the exception of the Sigmoidal model with which the D fraction was poorly estimated (R2 = 0.28). The Exponential model did not show any estimation capability, and the Logistic and Gompertz models best predicted in situ degradation rate of OM (kd) with R2 values of 0.65 and 0.62, respectively. The transformation of the in situ rate of degradation (kd) to its half-life value of degradation ((ln 2/kd)100) provided an improvement of kd prediction in the Sigmoidal model, with R2 changing from 0.35 to 0.64. As to kd and fermentable organic matter (FOM) all estimations improved upon inclusion of chemical composition characteristics, such as sugars, crude protein (CP), neutral detergent fibre (NDFom) and crude fat (CFat). The Logistic and Gompertz models continued to better predict kd, with R2 values of 0.79 and 0.88, respectively, while the Sigmoidal model showed a higher capability to estimate FOM (R2 = 0.90). It should also be noticed that the estimation of the washout fraction (W) estimation was obtained with only sugar and starch contents (R2 = 0.62). There were only moderate relationships between in situ and gas production indicating that the methods do not describe the degradation of these feedstuffs in a similar way.