PS13-3. Accounting for Particle Size Distribution and Ensiling Time Improves Whole-Sample Starch Digestibility Estimates in High-Moisture Corn.

Abstract

Accurate prediction of starch digestibility in high-moisture corn (HMC) is essential for estimating available energy for ruminants. Whole-sample digestibility is commonly estimated through lab procedures by grinding to 6 mm prior to incubation, altering particle size distribution (PSD) relative to feed delivered to the animal. Analysis of sieved fractions assumes additive digestibility, thus not reflecting non-additive fermentation dynamics of the whole-sample. Objectives of this survey were to evaluate starch digestibility across incubation time points between laboratory and in situ analysis for whole-sample and particle sizes. Additionally, correction factors for PSD estimates of HMC were determined for PSD and incubation time interactions to improve biological accuracy relative to whole-sample in situ digestibility. Thirty HMC piles were sampled across dairies (n = 4) and feedlots (n = 26) at harvest and across ensiling durations. Whole samples were collected along with sieved samples obtained using the 8mm, 4mm, and bottom trays of the Penn State Particle Separator. Subsamples were analyzed by a third-party lab (Rock River Laboratories; Watertown, WI) by NIR. Additional subsamples of whole and sieved fractions were incubated in ruminally-cannulated beef steers fed a receiving diet for 7, 24, and 36 h. Differences between laboratory and in situ digestibility were evaluated using paired t-tests within fraction and incubation time. Weighted digestibility was calculated from fraction digestibility by proportion of total sample. Relationships between in situ whole sample digestibility and weighted predictions were evaluated using linear mixed-effects models in R (v. 4.5.2) with fixed effects of additive prediction, incubation time, PSD proportions, and time by PSD interactions. Random effects included ensiling time, sample, and steer. Laboratory 6-mm ground values exceeded in situ digestibility across fractions and incubation times (P < 0.01), with the greatest inflation observed in the 8-mm fraction (33.5–35.2 units), intermediate differences for whole samples (24.1–27.1 units), and minimal differences for the bottom fraction (1.6–5.2 units). Uncorrected additive estimates systematically overpredicted whole-sample digestibility by 2.12 units (P < 0.01). Whole-sample digestibility was strongly associated with additive prediction (P < 0.01); however, this relationship was significantly modified by PSD and its interaction with incubation time (P ≤ 0.01). Including PSD and PSD × time terms improved model fit (P < 0.01), indicating that particle size distribution altered fermentation dynamics rather than contributing additively to starch disappearance. Incorporation of PSD- and time-based correction eliminated systematic bias (mean bias ≈ 0), reducing prediction error relative to the uncorrected estimates across 720 observations. Whole-sample starch digestibility of HMC appears not to be additive across particle size fractions, and PSD seems to influence fermentation kinetics. Integrating PSD-sensitive correction and laboratory translation might improve the biological fidelity of starch digestibility estimation, possibly enabling supporting more accurate prediction of available energy in ruminant systems.

For image description, please refer to the figure legend and surrounding text.