Improving nitrogen use efficiency in rice by estimating leaf nitrogen content with near-infrared spectroscopy and chemometric modeling.

Abstract

Accurate nitrogen management in rice (Oryza sativa L.) is essential for optimizing both crop productivity and environmental sustainability. This study evaluated the potential of Near-Infrared Spectroscopy (NIRS) combined with chemometric modeling to classify leaf nitrogen content (LNC) in five rice genotypes (Nerica, Rufipogon, IR64, Ciherang, and Curinga) subjected to five nitrogen fertilization levels (0%, 25%, 50%, 75%, 100%). Spectral data (350-2500 nm) were processed using Principal Component Analysis followed by Linear Discriminant Analysis (PCA-LDA) to distinguish nitrogen treatments and explore genotype-specific spectral responses. The 1700-2200 nm spectral region yielded the highest classification accuracy, consistently exceeding 94%, indicating strong sensitivity to nitrogen-related biochemical variation. Compared to conventional destructive methods, NIRS provides a non-invasive, rapid, and scalable alternative for nitrogen assessment in field conditions. This is the first study to demonstrate NIRS-based discrimination of nitrogen levels across multiple rice genotypes, offering new avenues for genotype-informed fertilization strategies and improved nitrogen use efficiency (NUE). The results support the objectives of the Green Campus Initiative at the Alliance Bioversity International & CIAT and contribute to broader Sustainable Development Goals (SDGs 2, 12, 13, and 15), promoting data-driven, environmentally responsible nutrient management in rice production.

Plant phenotyping relevance

NIRSとケモメトリクスによりイネ葉窒素含量を非破壊推定・分類する手法が研究の中心であり、植物形質の取得と技術性能を評価している。

Code and data availability

The paper's spectral/phenotyping data are not publicly deposited; the Data availability statement says data are available only upon request from the corresponding author. No public code, models, or datasets are mentioned.

No evidence-backed public reproduction asset is currently recorded.