Department of Agronomy, State University of Maringá, Maringá, Brazil.
Department of Agronomy, State University of Maringá, Maringá, Brazil.
Spectrochim Acta A Mol Biomol Spectrosc. 2022 Jun 5;274:121082. doi: 10.1016/j.saa.2022.121082. Epub 2022 Feb 26.
Tobacco genetic improvement programs, as well as the tobacco industry, require techniques that allow the estimation of its attributes in a fast and cheap way. The use of remote sensing through visible, near infrared and short-wave spectroscopy (Vis-NIR-SWIR) has been studied aiming to meet such demand. Thus, the aim of this work was to evaluate the use of Vis-NIR-SWIR spectroradiometer as a rapid tool to estimate alkaloids, sugars and yield of tobacco varieties. For that purpose, a study was carried out in a greenhouse with plants grown in pots (18 dm) containing nutrient solutions. The experimental design was completely randomized, with 30 treatments (tobacco varieties) and 10 repetitions. Tobacco leaf reflectance was collected at 13, 34 and 68 days after transplantation (DAT) with a plant-probe device connected to the spectroradiometer by an optical fiber. Subsequently, leaf analysis of alkaloids, sugars and yield were performed, and such attributes were estimated by using the Partial Least Squares Regression (PLSR), combined with the following pre-processing (PP) techniques: multiplicative scatter correction (MSC), Savitzky-Golay (SG) and standard normal variate (SNV). The results showed presence of typical inflections of chemical and structural components of the plants, which allowed obtaining PLSR models with R and RPD superior to 0.71 and 2.27, respectively, for all PP techniques and attributes evaluated. The most important wavelengths were well distributed within the three operating ranges of the spectroradiometer (Vis-NIR-SWIR). Thus, the methodology proposed by this research was able to simultaneously determine all the three attributes (alkaloids, sugars and yield) with excellent predictive capacity. This is a promising result for genetic improvement and processing of tobacco (as well as other crops), since it is necessary to evaluate a large number of samples within a short period and at a low cost.
烟草遗传改良计划以及烟草行业都需要能够快速、廉价地评估其属性的技术。利用可见近红外和短波光谱(Vis-NIR-SWIR)的遥感技术已经被研究用于满足这种需求。因此,本工作的目的是评估使用 Vis-NIR-SWIR 分光辐射计快速估计烟草品种的生物碱、糖和产量的方法。为此,在温室中进行了一项研究,在含有营养液的盆(18 dm)中种植植物。实验设计采用完全随机化,有 30 个处理(烟草品种)和 10 个重复。在移栽后 13、34 和 68 天(DAT),用与分光辐射计通过光纤连接的植物探头装置收集烟草叶的反射率。随后,对叶片的生物碱、糖和产量进行分析,并利用偏最小二乘回归(PLSR)结合以下预处理(PP)技术对这些属性进行估计:乘性散射校正(MSC)、Savitzky-Golay(SG)和标准正态变量(SNV)。结果表明存在植物化学和结构成分的典型拐点,这使得对于所有的 PP 技术和评估的属性,都能够获得 R 和 RPD 分别大于 0.71 和 2.27 的 PLSR 模型。最重要的波长在分光辐射计的三个工作范围内分布良好。因此,本研究提出的方法能够同时测定所有三个属性(生物碱、糖和产量),具有极好的预测能力。这是遗传改良和烟草加工(以及其他作物)的一个有希望的结果,因为需要在短时间内以低成本评估大量的样本。