A video and acoustic methodology to map bite placement at the patch scale

Abstract

Bite placement plays an important role in determining the mean weight of bites removed from a patch, and hence intake rate. This paper describes a video and acoustic methodology to map the sequential placement of bites on the sward surface. The methodology was evaluated for dairy heifers grazing small (0.34 m 2) patches of alfalfa ( Medicago sativa L.) to three levels of depletion (6, 18 and 30 bites). Grazing was recorded by a single video camera positioned 0.9 m in front of the animal and on average 1.67 m above the ground. Acoustic monitoring enabled the precise timing of biting jaw movements (bite or chew-bite). For each bite, a single video frame showing the position of the animal's head in the sward canopy was selected at the point in time corresponding to bite severance. Custom image-processing software was used to extract the image (apparent) coordinates of the tip of the muzzle, which was then converted geometrically to world (true) coordinates on the sward surface. A fixed displacement of 8 cm was applied to the world coordinates in order to determine the bite centre. The pattern of grazing impact across the sward surface was predicted by assuming a radius of impact around each bite centre. This pattern was compared with the measured status (grazed or un-grazed) of the herbage beneath each cell of a grid placed over the patch after grazing. Bite placement on the sward surface, as defined by the bite-centre coordinates, followed a progressing, side-to-side pathway. The median distance between the centres of consecutive bites was 12.1 cm, and was significantly less than expected by random bite placement. The orientation of consecutive bite pairs differed significantly from random, consistent with the side-to-side pathway of bite placement. Nevertheless, based on nearest neighbour analysis, the overall pattern of bite spacing on the surface of the sward did not differ significantly from random. These results suggest that the observed declining trend in mean effective bite area with increasing depletion ( P = 0.06) derived from inefficient exploitation of the sward surface. The assumption of a radius of impact of 6 cm revealed a broad correspondence between the measured and predicted patterns of grazing impact across the sward surface. Discrepancies most likely reflected sweeping actions of the tongue during bite formation, lateral head movements during bite severance and simplifying assumptions required by the single-camera approach. The results indicate that the methodology is potentially useful for investigating the rules governing bite placement in the horizontal plane.