Integration of Biotechnological Intervention and Machine Learning Approaches for Enhancing Phosphorous Acquisition Efficiency in Plants

Abstract

Pakistan's economy is significantly dependent on agriculture, as it is an agricultural country. The agriculture industry requires continuous improvement to satisfy the increasing demand for food, which is a result of the increasing population. It is important to optimize the efficiency of synthetic fertilizers in order to increase crop yield. The most frequently applied synthetic phosphorous fertilizer to soils is Di-Ammonium Phosphate (DAP). This fertilizer is used to increase the phosphorus content of the soil, a significant portion of which is lost in the soil and is not accessible for plant uptake. This research study concentrates on the integration of biotechnological interventions and machinelearning strategies to improve the uptake of phosphorus by plants. The Microbial Strain Bacillus velezensis FB2 and Polyvinyl alcohol solution were coated to the DAP fertilizer. Bacillus velezensis FB2 can solubilize unavailable phosphorous in soil and convert it to available phosphorous, while PVA serves as a barrier for the effective slow release of nutrients in soil. The coating was applied using a fluidized bed coater with a solution of 0.5% PVA and 4% PSB in water. The surface morphology of the developed product was evaluated using scanning electron microscopy (SEM). The presence of functional groups and crystallinity of coated granules were analyzed using X-ray diffraction techniques and Fourier Transform Infrared spectroscopy (FTIR). UV-Vis Spectroscopy was employed to analyze the release rate of phosphorous and nitrogen in water, and the ability of the coated product to resist applied force was assessed using crushing strength. The product that was developed was subjected to pot trials to evaluate the impact of various treatments on plant yield. The impact of various treatments on the height, diameter, number of leaves, area of leaves, soil EC, soil pH, fresh matter yield, dry matter yield, quantity of available phosphorus, and change in phosphorus and nitrogen uptake of the plants were assessed after they were at full growth. Based on soil and plant analysis, it was determined that DAP coated with both PSB and PVA was the most effective fertilizer in terms of plant growth and the quantity of nutrients in the soil and plants. This is due to the ability of microbial strains to solubilize phosphorus and the effective release of nutrients as a result of PVA coating. A machine learning model was developed to XVII predict changes in the amount of soluble phosphorus caused by the use of a microbial strain. Data was obtained from the literature and utilized to train and test a number of models, such as Ensembled Learning Tree (ELT), Guassian Process Regression (GPR), Decision Tree (DT) based on Genetic Algorithm (GA), and Particle Swarm Optimization (PSO). The GA-based ELT model demonstrated the highest performance among all developed models with an R2 value of 0.7938.