Acoustic monitoring of short-term ingestive behavior and intake in grazing sheep

Abstract

Acoustic monitoring of the ingestive behavior of grazing sheep was used to study the determinants of intake rate and to estimate dry matter intake (DMI) based on biting and chewing sounds. Each of three crossbred ewes (85±6.0kg body weight) were tested in 16 treatments resulting from the factorial combination of two forage species (orchardgrass and alfalfa), two levels of biomass depletion (tall=30±0.79cm and short=14±0.79cm) and four numbers of bites (20, 40, 60 and 80 bites). During each grazing session biting and chewing sounds were recorded with a wireless microphone placed on the ewe's forehead and connected to a digital video camera for synchronized audio and video recording of ingestive behavior. Dry matter (DM) intake rate was higher for alfalfa than orchardgrass (9.4±0.64 vs. 7.8±0.58g/min, P<0.05) because of lower fiber content (434±14 vs 558±6.6g/kg DM, P<0.01) and consequently shorter chewing time and fewer chews per unit DM (11±1.0 vs. 14±1.0 chews, P<0.05) in alfalfa than in orchardgrass. There were no differences in DMI rate between tall and short plants (8.7±0.67 vs. 8.5±0.68g/min, P>0.05), because sheep increased biting rate (from 17±1.6 to 28±1.6 bites/min, P<0.01) as bite mass declined from tall to short plants (from 0.54±0.02 to 0.31±0.01g DM, P<0.01). Sheep compensated for the reduction in bite mass by allocating fewer chews per bite (from 6.0±0.46 to 3.8±0.47, P<0.05) and increasing total jaw movement rate (from 95±6.3 to 122±6.3 movements/min, P<0.05). Compound jaw movements (chew-bites) were observed in every grazing session. The number of chew-bites was higher for tall than short plants (0.52±0.05 vs. 0.25±0.04 chew-bites/bite, P<0.05). The total amount of energy in chewing sound in a grazing session was linearly related to DMI (root mean square error=6.1g, coefficient of variation=27%); 79% of the total variation in total amount of energy in chewing sound was due to DMI. Dry matter intake was estimated accurately by acoustic analysis. The best model to predict DMI from acoustic analysis had a prediction error equal to 4.1g (coefficient of variation=18%, R2=0.92). Chewing energy per bite and total amount of energy in chewing sound were the most important predictors because they integrate information about eating time and intake rate of forages. The results demonstrate that ingestive sounds contain valuable information to remotely monitor feeding behavior and estimate dry matter intake in grazing ruminants.