Phenotyping the nutritional status of crops using proximal and remote sensing techniques

Understanding the nutritional needs of crops is crucial for ensuring their health and maximising yield. However, the capability to accurately measure relevant physical characteristics (phenotypes) of important crops in response to complex nutrient stresses is limited. For crop breeders and researchers, the existing capacity to characterise crops with adequate precision, detail and efficiency is hindering significant progress in crop development. In this PhD thesis, the use of advanced sensing techniques to assess the nutritional status of African crops was explored, focusing on three main objectives.
First, the use of a handheld proximal sensor was investigated to evaluate the spectral properties of quinoa and cowpea crops grown under different N and P supplies in controlled glasshouse conditions (Chapter 3). By analysing these spectral properties, the aim was to identify spectral indices that could show early signs of N and P stress separately in the plants. These stress indicators were related to the overall performance of the crops. Spectral indices were found that could distinguish between N and P stress at the early growth stage of the crops. However, identifying spectral indices for P stress was limited, particularly in cowpea due to the shorter wavelength range of the handheld device. The results showed significant relationships between the spectral indices and traits related to the morphology, physiology and agronomy of the crops.
Second, it was demonstrated that different levels of N impact the drought responses of spring wheat (Chapter 4). By evaluating morpho-physiological changes in the plants under high N and low N conditions, an understanding of how spectral reflectance measured at the leaf level could help distinguish between combined and complex stresses such as drought and nutrient deficiency was investigated. The results showed a greater amplitude of drought response in plants that were supplied with high N compared to low N levels, with interactive effects on many morphological and physiological traits. Out of a group of 39 different SRIs, only the Renormalised Difference Vegetation Index (RDVI) and the Red Difference Vegetation Index (rDVI_790) showed better accuracy in detecting drought stress. The results also revealed that indices sensitive to chlorophyll levels, such as the chlorophyll Index (mNDblue_730), Greenness Index (G) and Lichtenthaler Index (Lic2), as well as red-edge indices like Modified Red-Edge Simple Ratio (MRESR), chlorophyll Index Red-Edge (CIrededge) and Normalised Difference Red-Edge (NDRE), were more accurate in detecting N stress.
Lastly, the effectiveness of using spectral information from images collected from a drone and spectral reflectance measured with proximal sensors on the ground were compared for detecting N stress in winter wheat under field conditions (Chapter 5). By comparing these two sensing methods, it was assessed which approach is more accurate, reliable and cost-effective for assessing the N nutritional needs of the crop in real-world agricultural settings. The results indicated that the NDVI measured on the ground at the leaf level could accurately detect the small changes in N levels earlier compared to the drone NDVI and canopy level NDVI and for assessing the agronomic performance of winter wheat. Overall, this PhD research sheds new light on the potential of advanced sensing techniques to improve crop management practices and enhance agricultural productivity by providing timely and accurate information about the nutritional status of the studied crops.