Acoustic monitoring of chewing and intake of fresh and dry forages in steers

The goal was to evaluate acoustic analysis as a means to monitor and quantify chewing behaviour, and to estimate DM intake of forages with a wide range of water and fibre content. Each of three Angus steers (284-316 kg of body weight) was fed 32 treatments resulting from the factorial combination of four amounts (i.e. 75, 150,225 and 300 g dry matter, DM), two particle lengths (i.e. chopped or whole), two water contents (i.e. fresh or hay), and two species (i.e. grass or alfalfa). Behaviour was videotaped and chewing sound was recorded using a microphone attached to the steers foreheads. Intake rate of DM was higher for alfalfa versus grass higher (33.0 +/- 2.03 versus 28.7 +/- 2.04 g DM/min), and lower in fresh forages versus hays (28.0 +/- 2.04 versus 33.7 +/- 2.04 g DM/min). Fresh forage was chewed more than hay (3.1 +/- 0.19 versus 2.4 +/- 0.19 chews/g DM) demonstrating that chewing was primarily affected by water content of forage. Alfalfa hay was chewed less than grass hay, but fresh grass and fresh alfalfa did not differ from each other and required more chewing than fresh forages. The impact of forage internal water in chewing was clear, since chews per g of neutral detergent fibre (NDF) differed among the forages, but for fresh alfalfa (the forage with the most water and the least NDF), it was much higher than for the rest. There was little evidence of an impact of particle size on chewing behaviour, probably because all forages had very long particles. Of the variation of energy in chewing sound (E), 66% was explained by DM intake (DMI). The relationship between E and DMI was linear, and the slope was 28% higher for fresh forages versus hays (P = 0.033). Volume of chewing sound explained the most variation in E/DMI, and it was higher for fresh forage versus hays, whereas chews/DMI was the component of E/DMI most affected by forage characteristics. The DMI was estimated accurately based on easily observable behavioural and acoustic variables. Number of chews, cumulative sound energy and forage moisture (i.e. fresh versus hay) were the most important predictors. The prediction error in cross-validation was 27.3 g DM (coefficient of variation = 16%, R-2 = 0.89). Acoustic analysis yields more useful information than simple the number and rate of chewing, and should be further developed as method to research and monitor chewing behaviour of ruminants. (c) 2005 Elsevier B.V. All rights reserved.