Chemical diversity and species differentiation in Brazilian Vanilla: insights from LC-HRMS/MS metabolomics

TL;DR

This study uses advanced metabolomics to distinguish between three Brazilian Vanilla species by their unique chemical signatures.

Contribution

The study introduces a chemometric strategy for species-specific metabolic fingerprinting in Vanilla using untargeted LC-HRMS/MS.

Findings

Supervised PLS-DA models achieved robust species classification with Q² scores of 0.74–0.90.

17 core biomarkers were identified, including ferulic acid, which is a key precursor of vanillin.

Biome origin had limited influence on leaf metabolomes compared to species identity.

Abstract

Vanilla species represent a taxonomically complex and economically important group of orchids, yet species discrimination and chemotaxonomic characterization remain challenging due to high intrageneric metabolic variability and limited molecular insights.

We aimed to apply an untargeted metabolomic approach to characterize the metabolic signatures of three Brazilian Vanilla species (V. pompona, V. phaeantha, and V. calyculata) and to evaluate the relative contributions of species identity and biome origin to metabolic diversification.

The leaf metabolome of 102 Vanilla were profiled using high-resolution LC-HRMS/MS. Multivariate analyses including PCA and PLS-DA served to explore variance structures and build discriminant models. Discriminant features were selected using integrated criteria (VIP scores and volcano-plot significance). Annotation leveraged GNPS molecular networking and in silico databases to classify key metabolite classes.

Unsupervised PCA revealed pronounced intraspecific metabolic heterogeneity, preventing spontaneous grouping by species or biome. Supervised PLS-DA models, however, provided robust species classification (Q² = 0.74–0.90), while biome-based models lacked predictive power, indicating limited environmental influence on leaf metabolomes. Consensus selection between models identified 17 core biomarkers, including phenolic acids, cinnamic acid derivatives, C-glycosylated flavonoids, lipids, terpenoids, and N-containing compounds, with ferulic acid (a key precursor of vanillin) emerging as a prominent discriminating metabolite.

Untargeted LC-HRMS metabolomics coupled with chemometric modelling delineates species-specific metabolic fingerprints within Vanilla, offering a generalizable strategy for chemotaxonomy and species authentication. The conserved metabolic features also spotlight biologically meaningful pathways for biodiversity assessment and valorization of native Vanilla resources.

The online version contains supplementary material available at 10.1007/s11306-026-02422-8.

Figures