ABSTRACT
Remote sensing techniques have the potential to monitor physiological and biochemical changes caused by nutritional stress in plants. However, few studies have monitored the influence of foliar phosphorus levels on the spectral responses of plants, especially maize. This work aimed to assess phosphorus nutritional status in maize plants subjected to different sources and doses of phosphate fertilizer using leaf-based hyperspectral measurements and multivariate analysis. The specific goals addressed were a) the qualitative analysis of the spectral response, b) the identification of significant spectral wavelengths for differentiating doses of phosphate fertilizer, and c) the development of a spectral model for classifying the nutritional status of maize as an effective approach for crop nutritional assessment. Two experiments with different sources and doses of phosphate fertilizer were carried out: experiment 1 with monoammonium phosphate fertilizer (MAP) and experiment 2 with simple superphosphate fertilizer (SS). Hyperspectral reflectance (visible and near-infrared wavelengths) was collected at the V6 development stage of maize using an ASD Fieldspec 3 Jr Spectroradiometer, totalling 100 samples for each experiment. The collected data were submitted to principal component analysis (PCA) and linear discriminant analysis (LDA). The spectral signature of leaves showed higher reflectance values in the Vis spectral interval and lower reflectance in NIR under the absence of fertilization. For the other treatments, smaller differences in their spectral response were observed, especially for experiment 2. For experiment 1, the LDA achieved accuracy from 46 to 72% in the cross-validation step and from 50 to 75% in the external validation step. For experiment 2, the accuracy ranged from 25 to 82% in the cross-validation step and from 28 to 85% in the external validation step. In general, most misclassification was observed among treatments with similar fertilizer doses. These results suggest the potential of using hyperspectral data to monitor leaf P levels in maize, which contributes to the determination of crop management zones based on production variability under P limitation, in addition to optimizing the use of chemical inputs and guaranteeing greater sustainability of production systems.
Acknowledgements
The authors would like to thank the personnel of the Remote Sensing and Geoprocessing Laboratory of the Agronomy Department of the State University of Maringá.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The data that support the findings of this study are available from the corresponding author, K.M.O., upon reasonable request.