The grasshopper Trimerotropis pallidipennis (Orthoptera: Acrididae) exemplifies the challenges of species delimitation in widely distributed taxa with low morphological differentiation. This species extends from North to South America, frequently inhabiting high-altitude Andean environments. Although only four species have been formally recognized, cytogenetic and molecular evidence suggest that T. pallidipennis represents a complex of multiple evolutionary lineages.

To investigate this, I reanalyzed genome-wide SNP datasets generated through ddRAD-seq using unsupervised machine learning (UML) approaches. Multiple clustering algorithms and dimensionality-reduction methods were implemented to evaluate whether UML can improve species delimitation compared with traditional model-based validation methods.

Those analyses revealed that UML consistently identified lineage-level divergences and signals of admixture, supporting the existence of cryptic diversity within T. pallidipennis complex. In contrast to model-based approaches, which tended to oversplit taxa, UML successfully grouped samples according to major evolutionary lineages.

Overall, this study highlights the complexity of species boundaries in the T. pallidipennis complex and underscores the importance of integrative frameworks that combine morphology, cytogenetics, and genomic data. By demonstrating the capacity of UML to recover cryptic diversity while avoiding artificial over-partitioning, our findings provide methodological and conceptual advances for species delimitation in complex taxonomic systems.