Estimation of herbage mass in a bahia grass (Paspalum notatum) and a centipede grass (Eremochloa ophiuroides) pasture using a capacitance probe, a sward stick and a rising plate

The accuracy of 3 non-destructive techniques, an electronic capacitance probe, a sward stick and a rising plate in estimating herbage mass (HM) in a bahia grass (Paspalum notatum, BG) and a centipede grass (Eremochloa ophiuroides, CG) pasture from late spring to late autumn in 2 grazing seasons was determined. Two studies were conducted: in Experiment 1, stratified samplings investigated 10 locations (50 cm x 50 cm) whose HM covered the range of HM in each pasture at almost constant intervals; and in Experiment 2, systematic samplings investigated 40 locations along line transects in each pasture. With stratified samples from BG, all techniques consistently resulted in similar, significant linear relationships (r² = 0.86-0.98, P<0.001) between HM and capacitance or height readings. By contrast, for CG, the stick (r² = 0.57-0.98, P<0.05 or 0.001) and the plate (r ² = 0.59-0.96, P<0.01 or 0.001) tended to be less accurate than the probe (r² = 0.83-0.98, P<0.001). With systematic samples, the 3 techniques also produced significant linear relationships between HM and capacitance or height readings for both BG (r² = 0.47-0.82, P<0.001) and CG (r² = 0.28-0.74, P<0.001), although r ² values were lower than those from stratified samplings. The probe and the stick showed relatively poor performance for BG and CG, respectively. In both stratified and systematic samplings, HM-capacitance and HM-height relationships for BG in autumn underestimated actual HM in high HM locations, due to high amount of stems and dead materials in these overgrown areas. Quadratic regressions were more accurate in predicting actual HM in these situations. It was concluded that, although the rising plate was the most precise technique for estimating HM on the tropical grass pastures across seasons, further work is needed to overcome the bias of high HM locations on BG.