| Abstrakt: |
Plant diseases are one of the major problems that directly influence the quality and quantity of agricultural production. This study investigates the ability of visible and near-infrared spectroscopy and aquaphotomics approaches to detect fungal diseases on durum wheat. Durum wheat leaf samples, healthy and infected with Puccinia striiformis, Puccinia graminis, Septoria spp. and Blumeria graminis, causing yellow rust, stem rust, leaf spot, and powdery mildew diseases, respectively, were investigated. Reflectance and absorbance spectra of wheat leaves were collected using two spectrometers with a reflection fiber-optics probe - USB4000 for 450-1100nm region and NIRQuest 512 for 900-1700nm region (OceanOptics, Inc.). NDVI values were measured using a portable instrument PlantPen model NDVI 300. Soft independent modelling of class analogies (SIMCA) models for classification of leaf samples were created using Pirouette software (Infometrix, Inc.). Additionally, aquagrams were calculated using spectral data in the region 1340-1520nm. Differences between the reflectance spectral characteristics of healthy and infected leaves in both spectral regions were obtained. The most important differences in the visible range were observed in the red region around 670 nm, which could be explained by the change in chlorophyll content in diseased plants. In the near-infrared range, the most significant differences were found between 1250-1300nm and 1370-1500nm. The biggest ones were observed between 1417 and 1444nm, related mainly to the first overtone of O–H stretch vibration of water and N-H bonds. This statement is confirmed by the information from the aquagram charts. The aquagrams showed that the pathogens have changed a water structure in the wheat plants in a unique way during disease development. There are differences between NDVI values of healthy and infected leaves. The highest values of NDVI were measured for healthy wheat samples, and the lowest for infected with Puccinia graminis leaves. The statistically different NDVI values were identified. The developed SIMCA models for classification had a high accuracy rate. The spectral information in the visible-short wave NIR region allowed a correct classification of all samples from different classes. Slightly less accuracy showed the developed models based on the spectral region 900-1700 nm. The study showed the potential of VIS-NIR spectroscopy to discriminate successfully fungal diseases on durum wheat plants. [ABSTRACT FROM AUTHOR] |
