Detection of mosaic virus disease in cassava plants by sunlight-induced fluorescence imaging: a pilot study for proximal sensing

Cassava mosaic disease CMD is a prominent virus infection that causes considerable crop damage and yield reduction. Early detection of crop damage by remote sensing could be a useful tool for initiating remedial measures to reduce further crop damage. This article presents a non-destructive method for detection and classification of CMD infection, based on the red:far-red chlorophyll chl fluorescence image ratio. This pilot study was carried out in 14 varieties of potted cassava plants Manihotesculenta Crantz with a multispectral imaging system MSIS consisting of an electron multiplying charge coupled device EMCCD camera. Sunlight-induced chl fluorescence SICF images of plant leaves were recorded using the MSIS at the Fraunhofer lines of O2-B at 687 nm and O2-A at 759.5 nm and their off-lines at 684 and 757.5 nm. The recorded images were analysed using the Fraunhofer line discrimination FLD technique to extract the SICF from the solar reflectance in the recorded images. The chl fluorescence image ratio red:far-red, F687:F760 was computed and correlated with the laser-induced chl fluorescence LICF ratio F685:F735 determined by point monitoring, chl content variation, and the net photosynthetic rate Pn. The scatter plot of the F687:F760 image ratio showed good discrimination between different levels of CMD infection as evidenced by the high sensitivity and specificity values. It is observed that the fluorescence image ratio F687:F760 has a good correlation with Pn coefficient of determination R2 = 0.85, chl content R2 = 0.82, and the LICF ratio F685:F735R2 = 0.80, thereby highlighting the potential of the SICF image ratio in the discrimination of CMD infection. The results clearly indicate that changes in the red:far-red fluorescence image ratio due to CMD stress can easily be detected at an early stage and the technique has great potential for monitoring the health of crops and vegetation from proximal sensing platforms.