Evaluation of Alfalfa-Grass in New York.

Approximately 85% of the alfalfa in New York is sown with perennial grass. Alfalfa-grass stands can be heterogeneous, particularly in research plots, making sampling crucial. Samples can be separated for individual evaluation of alfalfa and grass nutritive value, but the ratio of alfalfa to grass may not be accurately represented in a small sample. Digital imaging analysis of photos has been able to successfully estimate alfalfa:grass ratio, but this technique is not as effective with grasses that are heading. Our objective was to evaluate whether visual photo evaluation can effectively estimate the alfalfa:grass species ratio in mixed stands. In spring and early summer of 2015 we acquired samples (n=207) of alfalfa-grass stands in farmers’ fields, and determined alfalfa and grass dry matter proportions for each sample. A camera was used to capture a digital image (5-Megapixels) of the sampling area. Representative samples were selected and delineated using a round hoop (66-cm diameter), which was rested on the vegetative canopy. Three individuals visually rated photographs for alfalfa percentage, and individual ratings were relatively consistent. A set of calibration photographs was identified that covered the range of alfalfa percentage in hand-separated samples, selecting photographs that visually represented a decreasing alfalfa percentage, and also agreed with hand separation results. Two individuals also rated photos using a point-count system. On each photo, 100 random points were categorized as alfalfa, grass or unknown. Calibrated visual estimates (y = 13.3 + 0.833x; R² = 0.70) were better than point-count estimates (y = 18.6 + 0.826x; R²= 0.61). Both systems tended to overestimate alfalfa when the alfalfa percentage of the stand was low. Visual and point-count estimates were well correlated (r = 0.88), with point-count estimates of alfalfa percentage about 10% higher than calibrated visual estimates.