Non-destructive in Situ Measurement of Aquaponic Lettuce Leaf Photosynthetic Pigments and Nutrient Concentration Using Hybrid Genetic Programming

Abstract

Phytopigment and nutrient concentration determination normally rely on laboratory chemical analysis. However, non-destructive and onsite measurements are necessary for intelligent closed environment agricultural systems. In this study, the impact of photosynthetic light treatments on aquaponic lettuce leaf canopy (Lactuca sativa var. Altima) was evaluated using UV-Vis spectrophotometry (300-800 nm), fourier transform infrared spectroscopy (4000-500 per cm), and the integrated computer vision and computational intelligence. Hybrid decision tree and multigene symbolic regression genetic programming (DT-MSRGP) exhibited the highest predictive accuracies of 80.9%, 89.9%, 83.5%, 85.5%, 81.3%, and 83.4% for chlorophylls a and b, β-carotene, anthocyanin, lutein, and vitamin C concentrations present in lettuce leaf canopy based on spectro-textural-morphological signatures. An increase in β-carotene and anthocyanin concentrations verified that these molecular pigments act as a natural sunscreen to protect lettuce from light stress and an increase in chlorophylls a and b ratio in the white light treatment corresponds to reduced emphasis on photon energy absorbance in chloroplast photosystem II. Red-blue light induces chlorophyll b concentration while white light promotes all other pigments and vitamin C. It was confirmed that the use of the DT-MSRGP model is essential as the concentration of phytopigment and nutrients significantly change during the head development and harvest stages.

Keywords

• computer vision
• leaf nutrient level prediction
• leaf pigment prediction
• lettuce
• machine learning