Six-year quantitative NMR profiling of polyfloral honey from a single beekeeping area in Okayama: harvest-season-associated variation in minor components.

Forty-six polyfloral honey lots collected in 2018-2023 from four apiaries in one beekeeping area in Okayama, Japan (5 early-spring, 15 spring, 14 late-spring, and 12 summer) were profiled for 17 minor components by quantitative broadband-WET 1H NMR; conventional 1H NMR quantified major sugars in 2018-2019. Tryptophan, phenylalanine, and tyrosine were broadly spring-associated, whereas valine and ethanol were relatively high in summer; lactic acid was higher in spring and summer than in late spring. Each year included spring, late-spring, and summer lots, so year-aware validation used 41 lots; early-spring lots (2022-2023) were retained for component-wise comparisons. The harvest-season association persisted after adjustment for year and site (pseudo-F = 20.312, partial R2 = 0.5672, P = 0.0001). Nested leave-one-year-out PLS-DA classified 39/41 correctly (balanced error rate = 0.0460; permutation P = 0.0001). These findings characterize harvest-season-associated patterns within this geographically restricted dataset without establishing causality or external predictive validity.