Abstract

Rapid determination of pasture quality using near infrared reflectance spectroscopy (NIRS) makes it feasible for farmers to incorporate forage composition in feed budgeting and feed supplementation decisions for grazing animals. In grazed pastures, heterogeneity in the distribution of the mass and quality, necessitates a sampling strategy to accurately predict the diet. A single paddock was systematically sampled by selecting 60 sites according to a pre-determined grid. At 12 of these sites, 4 separate but adjacent samples of 150 x 150 mm were taken from within a 300 x 300 mm quadrat. To characterise vertical heterogeneity, the herbage at each of the 60 sites was cut sequentially into 4 equal strata, the depth of each being one-quarter of the canopy surface height at that site. These samples were used to derive a statistically-based 3-dimensional characterisation of variability of herbage mass and of several parameters of herbage quality determined using NIRS. This description was used to calculate the number of samples necessary to estimate parameters to arbitary, pre-determined levels of accuracy of ± 50 kg DM/ha for herbage mass in each stratum, ± 50 g/kg protein (PROT), acid detergent and neutral detergent fibre (ADF; NDF), ± 20 g/kg for soluble carbohydrate (CARB), ± 25 g/kg for in-vitro dry matter digestibility (DIG) and ± 0.5 MJ/kg for metabolizable energy (ME). A minimum of six samples bulked together is necessary to estimate PROT, CARB, NDF and ADF to the specified level of accuracy, and this number was similar for each stratum. Twelve samples are required to estimate ME and DIG for each of the 3 upper strata, but in the lowest stratum with greater variability, 48 and 19 samples would be required for ME and DIG, respectively, to reach the same accuracy. A sampling strategy should take the suggested number of samples by systematically covering the paddock area. Uncertainty in the depth of sampling because of the necessity to sample before grazing has little effect on the number of samples required or on the accuracy of estimates, providing the lowest stratum is not penetrated during either sampling or grazing.

GP, Cosgrove, K Betteridge, VJ Thomas, and DC Corson

Proceedings of the New Zealand Society of Animal Production, Volume 58, , 25-28, 1998