Kinetic and Multivariate Optimization of Azolla filiculoides Biomass Production in Semi-Closed Bioreactors for Biorefinery-Oriented Bioprocessing

Abstract

A. filiculoides is a fast-growing aquatic fern with potential for laboratory-scale biomass production, nutrient recovery and biorefinery-oriented bioprocessing. However, its cultivation in controlled bioreactors remains limited by insufficient integration of treatment formulation, physicochemical monitoring and predictive optimization. This study evaluated A. filiculoides biomass production for 30 days in semi-closed 5 L glass bioreactors under three cultivation conditions: Hoagland-type mineral solution (T1), Murashige and Skoog (MS) medium (T2), and an aqueous growth-regulator treatment containing 6-benzylaminopurine and indole-3-acetic acid (BAP–IAA; 1 mg L−1 each) (T3). The initial biomass was standardized at 0.10 g FW L−1. By day 30, T3 showed the highest final fresh biomass concentration 1.208 ± 0.043 g FW L−1, followed by T1 0.948 ± 0.025 g FW L−1 and T2 0.538 ± 0.033 g FW L−1. Principal component analysis and k-means clustering showed that dissolved oxygen, oxidation–reduction potential, electrical conductivity, resistivity, pH and water temperature structured the cultivation environment. The physicochemical-modulated Gompertz model showed high internal predictive performance, with training R2 = 0.997, RMSE = 0.016 and MAE = 0.013, and cross-validation R2 = 0.986, RMSE = 0.033 and MAE = 0.024. Model-based optimization identified T3 at day 30 as the optimal condition, with predicted biomass of approximately 1.224 g FW L−1. The associated operating window corresponded to pH 7.34–7.52, dissolved oxygen 5.70–6.30 mg L−1, ORP 75.40–102.00 mV, EC 983.90–1077.80 µS cm−1 and resistivity 0.928–1.016 kΩ cm. These results support the use of coupled temporal monitoring, multivariate analysis and kinetic modeling for laboratory-scale optimization of Azolla biomass production.