Machine learning integrated visible diffuse reflectance spectroscopy for in-situ analysis of phosphorus status in Arabidopsis plants under soilless culture.

Abstract

Phosphorus (P) is a vital macronutrient for plant growth, but its limited availability in soil due to fixation renders up to 80 % of fertilizers ineffective. Visual symptoms for P deficiency appear late or remain inconclusive, complicating timely intervention. The conventional methods are often time-consuming, costly, and labour-intensive. Diffuse reflectance spectroscopy offers a rapid, label-free alternative, though its application is challenged by weak P spectral response. In this study, Arabidopsis thaliana (Col-0) plants were subjected to soilless culture under controlled phosphorus-sufficient (P+) and deficient (P-) conditions. Leaf reflectance spectra were analyzed using Linear Discriminant Analysis (LDA), and the selected wavelengths were used to train three machine learning classifiers such as Support Vector Machine (SVM), Random Forest, and K-Nearest Neighbors (KNN). Among these, the SVM model demonstrated best performance, achieving a classification accuracy of 97.78 %. Independent validation using biochemical, morphological, and combined datasets, yielded classification accuracies of 100 %, 71.88 %, and 100 %, respectively. This approach offers a rapid, and non-destructive alternative to conventional techniques for sustainable nutrient management in agriculture.

Plant phenotyping relevance

植物のリン栄養状態を可視拡散反射分光と機械学習で非破壊推定し、分類性能を検証する方法研究であり、表現型取得・推定手法が中心である。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.