Scholarlink

Early prediction of biopolymer content in plants is crucial for efficient orchard management. Traditional laboratory methods are reliable but time-consuming and require skilled technicians. Portable near-infrared (NIR) spectrometers offer a faster alternative for screening. This study evaluates the effectiveness of wavelength-selective partial least squares regression (WS-PLSR) compared to conventional partial least squares regression for predicting protein, cellulose, lignin, and starch content in plant leaves. The analysis used reflectance spectra from 70 leaf samples, covering over 50 species, including gymnosperms, monocotyledons, and dicotyledons. Laboratory analysis of 120 samples, harvested in July and September 1993, was conducted using standard wet chemical methods in France and Belgium. WSPLSR achieved R2 values of 0.8 for protein and lignin and 0.7–0.8 for cellulose, with corresponding RPD values of 2–3 and 2–2.5, respectively. Starch prediction showed lower R2 values of 0.7 and RPD values between 1 and 2. These results highlight the potential of NIR spectroscopy as a reliable screening tool for rapid biopolymer analysis in agriculture.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Wavelength-Selective Partial Least Squares Regression of Near-Infrared Spectra for Biopolymer Prediction in Leaves

  • 章节

  • 发表日期：

    2026年

查看全文
  • Mohammad Uzair Ali,
  • Saurav Bharadwaj

摘要

Early prediction of biopolymer content in plants is crucial for efficient orchard management. Traditional laboratory methods are reliable but time-consuming and require skilled technicians. Portable near-infrared (NIR) spectrometers offer a faster alternative for screening. This study evaluates the effectiveness of wavelength-selective partial least squares regression (WS-PLSR) compared to conventional partial least squares regression for predicting protein, cellulose, lignin, and starch content in plant leaves. The analysis used reflectance spectra from 70 leaf samples, covering over 50 species, including gymnosperms, monocotyledons, and dicotyledons. Laboratory analysis of 120 samples, harvested in July and September 1993, was conducted using standard wet chemical methods in France and Belgium. WSPLSR achieved R2 values of 0.8 for protein and lignin and 0.7–0.8 for cellulose, with corresponding RPD values of 2–3 and 2–2.5, respectively. Starch prediction showed lower R2 values of 0.7 and RPD values between 1 and 2. These results highlight the potential of NIR spectroscopy as a reliable screening tool for rapid biopolymer analysis in agriculture.