¹H NMR metabolomic signatures associated with oil-related bunch traits in within-family oil palm (Elaeis guineensis) progenies

Background Oil palm ( Elaeis guineensis Jacq.) improvement relies on progeny testing, but bunch performance is expressed only after several years of field growth, and oil-related traits vary considerably even among genetically related progenies. Whether early-stage leaf metabolism is associated with this variation is unknown. We therefore compared the spear leaf metabolomes of top- and bottom-performing palms selected within the same biparental crosses. Results Unopened spear leaves (leaf 0) were collected from 36 palms, one top and one bottom performer from each of 18 biparental crosses, and profiled by ¹H NMR. Top-class palms had higher mesocarp-to-fruit (+ 6.4%), oil-to-dry mesocarp (+ 4.4%) and oil-to-bunch (+ 22.4%) ratios than bottom-class palms (all P < 0.001). Of 123 metabolites identified, those with the highest variable importance in projection (VIP) scores, including guanidinoacetic acid, xylitol, methylguanidine, threonic acid and lysine, contributed to the class separation seen in partial least squares discriminant analysis. Several metabolites correlated with oil-related traits, notably L-arabitol (positively) and glutamine (negatively), but none remained significant after false discovery rate (FDR) correction. Six pathways were enriched (FDR < 0.01), led by alanine, aspartate and glutamate metabolism (impact 0.849) and including arginine biosynthesis, glyoxylate and dicarboxylate metabolism, and the TCA cycle. Five metabolite ratios discriminated the classes (AUC 0.790–0.814), the best being β-hydroxyisovaleric acid/methylguanidine (AUC = 0.814) and guanidinoacetic acid/pimelic acid (AUC = 0.806). Conclusions Within biparental crosses, top- and bottom-performing palms differed in spear leaf profiles involving carbon-conserving and nitrogen-related (amino acid and guanidino compound) metabolism, a pattern consistent with coordinated carbon–nitrogen metabolism supporting oil accumulation. The candidate metabolite ratios are leads for non-destructive early selection, but the extreme-group design, the small sample size and the absence of FDR-significant individual correlations mean that they must be validated in independent, unselected populations and environments before use in breeding.