Canopy reflectance informs in‐season malting barley nitrogen management: An ex‐ante classification approach

Abstract

Malting barley (Hordeum vulgare) requires precise nitrogen (N) fertilizer management to achieve a narrow range of grain protein content (∼9–10.5%) while maintaining yields, but practical tools to accomplish this are lacking. This study hypothesized that canopy reflectance (Normalized Difference Vegetation Index (NDVI)) measured at tillering (Feekes 2–3) and expressed as a sufficiency index (SI), can estimate the likelihood of a site‐specific response to in‐season N fertilizer in malting barley. Canopy reflectance was measured from plots at tillering with a GreenSeeker and unmanned aerial vehicle (UAV) borne multispectral cameras in trials across heterogeneous California agroecosystems. Field experiments included a range of N fertilizer application rates (0–168 kg N ha⁻¹) and timings (pre‐plant, tillering, or evenly split), and resulted in a range of crop N sufficiency/deficiency. NDVI‐based SI measurements were categorized into one of three quantitative categories (low, medium, and high) without additional experimental context using Gaussian mixture modeling. Despite that 85% of variation in protein yield was due to site‐year, the reflectance‐based categories indicated whether N fertilizer applied in‐season would increase protein yield (p < .01). Nitrogen application at tillering increased yield and protein for plots in the “low” and “medium” SI categories (45 and 4% for yield and 16 and 12% for protein, respectively) (p < .05), while “high” SI plots had neither yield (p = .23) nor protein (p = .26) increases. Importantly, the broader agronomic conditions of a site primarily determined whether response to in‐season N manifested as increased yield or protein.

Plant phenotyping relevance

圃場の作物キャノピー反射をGreenSeekerおよびUAVマルチスペクトルカメラで取得し、NDVI由来の指標を混合モデルで分類して、作物の窒素充足状態と施肥応答を推定・検証している。反射計測と解析ワークフローが研究の中心であり、単なる日常的形質測定ではない。

Code and data availability

The article's data availability statement explicitly deposits the data and code used to produce the manuscript at a public DOI (https://doi.org/10.25338/B8633H), which is an allowed URL. This covers the paper's NDVI/SI measurements, yield/protein outcomes, and mixture-model analysis. A GitHub reference (Nelsen 2019, 'D