Non-destructive assessment of chloride in persimmon leaves using a miniature visible near-infrared spectrometer
New methods for the non-destructive, rapid and reliable assessment of leaf chloride (Cl−) are needed in many agricultural salt-threatened areas throughout the World. Although portable visible near-infrared spectrometers could be useful for this task, accompanying models for leaf Cl− assessment on...
| Autores principales: | , |
|---|---|
| Formato: | acceptedVersion |
| Lenguaje: | Inglés |
| Publicado: |
Elsevier
2020
|
| Materias: | |
| Acceso en línea: | http://hdl.handle.net/20.500.11939/6451 https://www.sciencedirect.com/science/article/pii/S0168169919302212?via%3Dihub |
| Sumario: | New methods for the non-destructive, rapid and reliable assessment of leaf chloride (Cl−) are needed in many
agricultural salt-threatened areas throughout the World. Although portable visible near-infrared spectrometers
could be useful for this task, accompanying models for leaf Cl− assessment on basis their visible near-infrared
(Vis-NIR) spectra had yet to be developed. To study the feasibility of such models, a persimmon orchard was
grown in E Spain and irrigated with eight waters ranging in Cl− concentration from 70 to 350 mg/L. Then, six
times throughout the growing season the Vis-NIR foliar reflectance spectra of various representative leaves per
Cl− treatment were recorded with a portable miniature Vis-NIR spectrometer. The leaf Cl− was subsequently
determined in the same leaves with a laboratory reference method. Next, both the Cl− and spectral data were
analyzed using descriptive statistics, mathematical data transformations, principal components analysis (PCA)
and partial least squares regression (PLSR). A wide, sufficiently normal-distributed and representative range of
leaf Cl− content from 0.17 to 3.14%Cl− was obtained. With correlation coefficients exceeding +0.65 between
390 and 472 nm and between 690 and 692 nm, the foliar reflectance was the spectral data correlated the most
with the leaf Cl−. Remarkably, both wavelengths intervals are where chlorophyll a has its two absorbance
maxima in the visible region. A PLSR model including all the wavelengths in the Vis-NIR reflectance spectra plus
the day of the year, attained values of R2=0.78, RMSE=0.34%Cl− and RPD=2.18 in the external validation.
However, for leaf Cl− below 1.5%Cl−, i.e., when Cl− assessment is most needed, this model presented
R2=0.55, but RMSE=0.08%Cl− and RPD=5.5, in addition to a bias of +0.04%Cl− and a standard error of
0.05%Cl−. These results show that portable Vis-NIR spectrometers can be useful for the non-destructive, rapid
and even reliable assessment of leaf Cl−. The potential of this methodology for other crop plants and inorganic
ions should be tested. |
|---|