Predicting Rice Yield Response to Midseason Nitrogen with Plant Area Measurements

Abstract

A simple method is needed to aid farmers with midseason nitrogen (MSN) decisions in dry‐seeded, delayed flood rice (Oryza sativa L.). This study was conducted to develop thresholds using visual and digital image measurements for predicting rice yield response to MSN. ‘Francis’ and ‘Cheniere’ rice were drill seeded on 19‐cm row spacing from 2004 to 2006 on silt loam and clay soils at Glennonville and Portageville, MO. Preflood N (PFN) was applied at rates of 0, 39, 78, 118, and 157 kg urea‐N ha−1 with and without two MSN applications of 34 kg N ha−1 at panicle differentiation (R1) and at R1 + 7d. Plant area observations were made 1 to 2 d before R1. In each plot, a yardstick was floated on floodwater positioned between two center rows. Visible inch numbers were counted while standing above the rows. A number was not counted when rice leaves obstructed the view of one or more digits in the whole number. Plant height was measured, and digital images of canopy were analyzed to determine percent green pixels. Highest rice yields on both soils were most often achieved with 78 kg N ha−1 with MSN or 118 kg N ha−1 without MSN. The PFN significantly affected visible yardstick numbers, plant height, and percent green pixels. Height was the least reliable indicator of rice N status. Using regression analysis, no rice yield increase from MSN was produced when fewer than 13 yardstick numbers were showing or more than 64% of image pixels were green.