Multiobjective optimization of Murta (Ugni molinae Turcz) juice lactic acid fermentation with monocultures and cocultures of Lactobacillus acidophilus and Lactiplantibacillus plantarum

• TPC, LA, and viable lactic acid bacteria count were optimized simultaneously • Optimization used the global desirability (GDF) and genetic algorithm NSGA-II • A more complete Pareto front was obtained with RSM-NSGA-II compared to RSM-LH-GDF • MOO in lactic acid fermentation improves via multicriteria decision-making (MDM) • Applying different MCDM techniques is key to selecting the best compromise solutions Response surface methodology (RSM) and the desirability function (DF) approach have been widely used for multiresponse optimization in lactic acid fermentation processes. However, multiobjective optimization (MOO) can provide a broader range of optimal cultivation conditions that compromise between multiple objectives. This study applied multiobjective optimization of murta ( Ugni molinae Turcz) juice lactic fermentation under monoculture and coculture conditions. The total phenolic content (TPC), Lactic Acid Bacteria (LAB) count, and Lactic Acid (LA) concentration were optimized simultaneously. Decision variables were temperature, initial pH, and inoculation percentage. Several multicriteria decision-making (MCDM) methods were applied to select the optimal compromising responses, considering 80% priority for TPC, 15% for LAB, and 5% for LA. The highest TPC (723 ± 15 mg/L) was achieved at 35.3°C, pH 5.9, and 2% (v/v) inoculum in monoculture. The highest LAB count (6.7 ± 0.0 log CFU/mL) was obtained in monoculture at 34.6°C, pH 6, and 2.7% (v/v) inoculum. The highest LA (0.48 ± 0.03 g/L) was achieved in coculture at 35.6°C, pH 5.9, and 2.1% (v/v) inoculum. The simple additive weighting (SAW) method generated optimal solutions that aligned better with our expected priorities. Experimental validation confirmed that monocultures closely matched predicted optima, yielding higher TPC and LAB values than cocultures. While MOO and DF produced similar Pareto fronts, MOO required less computational time and generated more diverse solutions. These findings highlight the practical utility of MOO-MCDM for designing efficient multiobjective fermentation processes in functional food development.