Reagent-Free Prediction of Free-Ammonia Toxicity in Algal Systems Using Chlorophyll Fluorescence Transients and Interpretable Sparse Regression.

Accurate assessment of free ammonia (NH3) is critical for managing nutrient-rich effluents and algal-based treatment systems, but conventional assays are reagent-intensive and poorly suited to real-time monitoring. Chlorophyll a fluorescence offers a rapid, reagent-free proxy for photosynthetic health, yet widely used indices such as Fv/Fm respond nonspecifically to diverse stresses and rarely yield quantitative NH3 concentrations. Here, we develop a reagent-free framework that predicts NH3 toxicity in microalgae from fast chlorophyll fluorescence transients (O-J-I-P; OJIP) analyzed with interpretable sparse regression. Three species (Chlorella vulgaris, Acutodesmus obliquus, and Arthrospira platensis) were exposed to environmentally relevant NH3 levels, and OJIP responses were monitored for 35 h. Among several multivariate approaches, Lasso regression optimized by nested cross-validation provided the best compromise between accuracy (R2 = 0.93-0.98 within species) and transferability across taxa (R2 = 0.81-0.87 for green algae). The selected OJIP parameters captured conserved fluorescence signatures of NH3-induced stress and retained predictive power under light-induced photoinhibition that markedly depressed Fv/Fm. Incorporating model evaluation diagnostics enabled transparent uncertainty quantification and discrimination of extrapolative predictions. This workflow converts routine OJIP measurements into real-time, reagent-free, and quantitative information on ammonia toxicity in algal cultures and high-ammonia effluents, providing a transparent biosensing platform for environmental risk assessment.