Journal of Scientific Research and Reports

Species delimitation in morphologically complex tropical plant groups is increasingly supported by digitised natural-history collections, quantitative morphometrics and machine learning. This methodological review evaluates how explainable artificial intelligence (XAI) and morphometric machine learning can contribute to species-hypothesis generation and specimen prioritisation in the pantropical genus Senna (Fabaceae: Caesalpinioideae), while preserving the central role of expert taxonomy. Peer-reviewed evidence on integrative taxonomy, herbarium digitisation, computer vision, geometric morphometrics, multimodal learning, uncertainty assessment and explainability was synthesised using targeted scholarly searches of publisher platforms and DOI-indexed literature. A seven-figure worked example is retained to illustrate a governed analytical architecture comprising specimen-quality screening, nomenclatural reconciliation, organ localisation, quantitative trait extraction, multimodal classification, uncertainty-aware review, drift monitoring and explainability. The worked-example values, including the 420-record corpus and classifier benchmark, are used to demonstrate reporting and governance requirements and are not treated as independent evidence of species boundaries or taxonomic novelty. The literature indicates that machine learning can scale specimen identification and trait extraction, but taxonomic inference remains vulnerable to label error, geographic and institutional bias, missing reproductive organs, domain shift, class imbalance and shortcut learning. Accordingly, classifier confidence should not be equated with taxonomic certainty, and low-confidence predictions should be interpreted as candidates for expert reassessment rather than as evidence of hybridisation, incomplete lineage sorting or undescribed species. A defensible computational taxonomy for Senna should therefore combine auditable morphology, rigorous validation, type-based nomenclature, geographic and ecological context, and, where required, molecular evidence within an integrative-taxonomy framework.