Multiblock Integrative Modelling of New Zealand Monofloral Honeys: Mapping the Multi-Constituent Drivers of Antioxidant and Anti-Tyrosinase Activity

Honey bioactivity is a multi-constituent property often oversimplified by single-marker analysis. This study applied an exploratory multiblock integrative framework to map the systems-level architecture of bioactivity in New Zealand thyme, mānuka, kānuka, and clover honeys. Sugars, minerals, targeted phenolics, and bioactive indices were profiled alongside a functional panel (ferric reducing antioxidant power, FRAP; cupric ion reducing antioxidant capacity, CUPRAC; 1,1-diphenyl-2-picrylhydrazyl radical scavenging, DPPH; and anti-tyrosinase). Univariate analysis revealed a distinct functional gradient (p ≤ 0.004), with thyme showing the highest mean activity across the four in vitro endpoints within this cohort. Beyond univariate comparison, data integration analysis for biomarker discovery using latent components (DIABLO), supervised by categorical honey type, and sparse partial least squares regression (sPLS-REG), using the continuous assay panel as the response, identified a bioactivity-associated signature spanning multiple chemical domains. The high-bioactivity axis co-varied with vitamin C, syringic acid, phosphorus, and sucrose; these associations do not establish biochemical synergy or soil causation. Botanical adjustment weakened or reversed several pooled associations; vitamin C–CUPRAC depended on the inference method and within-thyme support, and no pair passed within-type permutation BH correction. Lumichrome showed a kānuka-associated pattern within the present cohort, consistent with previous reports, but its specificity requires confirmation in larger, independently sourced cohorts. The contribution is the within-cohort integration of complementary compositional domains with multiple functional endpoints. The findings are hypothesis-generating and require external validation.