This study evaluated whether the FieldScout CM 1000 NDVI and Yara N–Tester models can produce accurate and reliable estimates of nitrogen (N), buffer-soluble nitrogen (BSN), buffer-insoluble nitrogen (BISN), non-protein nitrogen (NPN) and in vitro ruminal nitrogen degradability after 3, 12 and 24 h incubation (ND3, ND12 and ND24) in three tropical grasses: Brachiaria hybrid, Megathyrsus maximus and Paspalum atratum. Correlation between the Yara N-Tester and N, BISN and in vitro ruminal N degradability of the Brachiaria hybrid and M. maximus were high (r 0.67–0.83). The Yara N-Tester accounted for 81% and 86% (p 0.000) of N variability in the Brachiaria hybrid and M. maximus, respectively. The Yara N-Tester prediction models explained 72% and 70% (p 0.000) BISN variability in the Brachiaria hybrid and M. maximus, respectively. In vitro ND24 of the Brachiaria hybrid (R 2 0.75) and M. maximus (R 2 0.75) was also best predicted with the Yara N-Tester. Model validation showed generally low (≤0.90) concordance correlation coefficients except for Yara N-Tester N and ND24 in M. maximus. Random error was the main source of error. We conclude that the accuracy of the Yara N-Tester prediction models was superior to that of the FieldScout CM 1000 NDVI models, and that the Yara N-Tester can produce accurate and reliable estimates of Brachiaria hybrid and M. maximus N and M. maximus ND24.
Read full abstract