Tropical Fruit Seeds as Coproducts for Food Upcycling: Integrating Composition, Bioactives, and Chemometrics to Assess Their Potential as Functional Flours

Fruit processing generates substantial seed-rich side streams: reported fractions may reach about 24% of seriguela fruit mass, 40% of acerola processing residues, 30–50% of jabuticaba peel/seed residues, and 11.8–31.8% of pitanga fruit mass as seeds. This study evaluated seriguela, acerola, jabuticaba, and red and black pitanga seeds as candidates for upcycling into fiber-rich food ingredients and for cascading recovery of bioactive substances. Dried, milled seeds were characterized for proximate composition, water activity (aw), 168-h hygroscopicity, and total extractives. Hydroethanolic extracts were assessed for total phenolics and antioxidant capacity by DPPH, ABTS, FRAP, and ORAC, and profiled by LC–ESI–IT3D–MS/MS in positive and negative ionization modes. Nutritional/technological and phenolic/antioxidant blocks were autoscaled, block-balanced, and integrated byexploratory principal component analysis (PCA) and Ward hierarchical cluster analysis (HCA).Total dietary fiber ranged from 56.5 to 85.9 g/100 g dry basis, with seriguela showing the highest content. Protein ranged from 3.12 to 9.05 g/100 g, with the highest value in jabuticaba, whereas lipids remained low (0.97–2.60 g/100 g). Non-fiber carbohydrates ranged from 7.52 to 30.73 g/100 g. All flours showed low aw (0.322–0.356) and 168-h hygroscopicity of 6.74–7.96 g/100 g, supporting their evaluation as dry ingredients. Total extractives varied markedly, from 21.9% in seriguela to 41.7% in black pitanga. Jabuticaba showed the highest total phenolics (29.76 mg GAE/g dry biomass) and ABTS response (238.1 µmol TE/g), seriguela the highest DPPH response (223.3 µmol TE/g), and black pitanga the highest FRAP (262.0 µmol TE/g) and ORAC (78.0 µmol TE/g). The first two principal components explained 80.6% of total variance (PC1 54.4%; PC2 26.1%). Seriguela was associated with the fiber/DPPH axis, jabuticaba with protein, phenolics, and ABTS, and black pitanga with extractives and FRAP–ORAC responses. HCA identified acerola/red pitanga and jabuticaba/black pitanga as the closest pairs, while seriguela remained the most distinct matrix. MS/MS supported tentative annotation of an HHDP-hexoside/ellagitannin-related feature in jabuticaba ([M−H]− m/z 481, fragmentation m/z 301, 275), an ellagic-acid-related feature shared by jabuticaba and red pitanga ([M−H]− m/z 301, fragmentation m/z 257, 229, 201, 185), and a myricitrin-like feature in black pitanga ([M−H]− m/z 463, fragmentation m/z 316/317). These complementary profiles indicate matrix-specific upcycling routes, from fiber-rich flours to bioactive recovery, supporting cascading valorization and circular food bioeconomy strategies. Technological performance, safety, and sensory suitability should be validated before direct food application.Acknowledgments: CNPq (PIBIC / PIBITI – 2026 and 400122/2025-1), FAPESP(2025/11374-6) and PIDA/PRPI/USP 1602025/01.