Adsa Logo White Adsa Title White

An upgrade of the neutral detergent fiber characterization in NDS Professional.

E. Raffrenato

Events

06-23-2020

Join E. Raffrenato on this page for a live text chat!
6:00 PM - 8:00 PM GMT

Abstract:

T131
An upgrade of the neutral detergent fiber characterization in NDS Professional.
E. Raffrenato*, A. Ferrari, E. Melli. RUM&N Consulting Reggio Emilia, Italy.

Neutral detergent fiber degradation has recently been described by a 3-pool system, with the contemporary exponential decay of a fast (B3) and slow (B4) pool, their rates (kB3, kB4) and an indigestible pool (C). These fractions are now correlated with intake in ration evaluations and they will be implemented in CNCPS. Their accurate and precise characterization is therefore of utmost importance. The aim of this study was to validate and, possibly, improve NDF characterization done by a Vensim model (V) as described in Raffrenato et al. (2019), using 3 time points NDFd. A set of 500 fermentation profiles was randomly generated, each corresponding to a unique combination of B3, B4, C, kB3 and kB4, based on published ranges. An alternative algorithm was developed using the same 3 time points (M1). The true values were regressed on the ones predicted by V and M1 and RMSPE was calculated. The significance of the deviation of the intercept from 0 and the slope from 1 was analyzed by t-test. The V model resulted in weak 3-time point predictions, with the risk of picking a local, vs. a global, minimum payoff. We then allowed the addition of the 12-h NDFd to the model (M2). Both R2 and RMSPE indicated that both M1 and M2 performed better than the V model, with M2 performing better than M1. The payoff was numerically smaller for M1 and M2 in 482 samples, and very similar to the V payoff in the remaining cases. The V model had R2 always lower than 0.60 with M1 and M2 having R2 higher than 0.80 and 0.90, respectively. The V model had also slope larger than 1 (P < 0.05) for kB3. The V model consistently underestimated the aggregated kd, resulting in positive residuals, especially for the lower quality samples. The risk of having a false low value is then higher than when predicting a higher kd forage. M1 residuals were independent of the size of the kd and M2 residuals confirmed the drastic improvement of the prediction when including the 12 h time point NDFd in the prediction. The study clearly showed how relevant is NDFd before 30 h, especially for extreme cases of either very high or very low-quality forages. Both M1 and M2 are now implemented in NDS Professional (RUM&N, Italy).

Keywords: modeling, CNCPS, rate of degradation.