Physicochemical, mineral, and metabolic variation across coconut varieties and coastal landscape types

Abstract

Coconut (Cocos nucifera L.) supports coastal value chains, yet compositional variation across varieties and production landscapes remains insufficiently characterized. This study evaluated physical, physicochemical, mineral, and metabolic variation across two coastal landscape types (alluvial and fluviomarine plains) and three coconut varieties (Manila Red, Hybrid, and Alto Pacífico) using a 2 × 3 factorial sampling design (n = 36 biological replicates; each replicate comprising six fruits from six different palms). Coconut water was analyzed for pH, titratable acidity, soluble solids, and mineral composition; kernel physical characteristics, moisture, and total oil content (dry basis) were determined; and coconut milk was characterized physicochemically and profiled by untargeted UPLC–ESI–QTOF–MS. Kernel circumference differed among varieties (p = 1.21 × 10⁻⁶), and the kernel-to-water ratio differed between landscapes (p = 0.0401), whereas no significant landscape, variety, or interaction effects were detected for kernel moisture, total oil content, or the evaluated coconut-water and coconut-milk physicochemical characteristics. Mineral composition showed analyte-specific variation: Ca was associated with variety (p = 1.04 × 10⁻⁴), whereas Mg and K exhibited significant variety × landscape interactions (p = 4.88 × 10⁻⁷ and p = 0.00214, respectively). Untargeted metabolomics retained 709 features after QC filtering; PCA showed no evident clustering by landscape or variety, and no valid PLS-DA model was obtained. Overall, variation was trait-specific, with limited detectable effects for most physicochemical characteristics and more pronounced variation in coconut-water mineral composition.