Prospectivity maps generated by both Random Forest and Neural Networks delineate highpermissivity domains spatially associated with the batholith. Several of these domains coincide with known porphyry–skarn deposits, supporting the geological coherence of the approach. In addition, new permissive areas were identified outside historically explored corridors, highlighting zones that warrant further investigation. The integration of the Mineral Systems approach with supervised machine learning provides a transparent and reproducible tool for regional-scale exploration targeting. However, results are dependent on dataset quality, proxy selection, and spatial scale, and should therefore be interpreted as relative indicators of permissivity rather than deterministic predictions. Incorporating higher-resolution geophysical, geochemical, and radiometric datasets would improve model refinement and reduce geological uncertainty in future applications. 5. RECOMMENDATIONS Advancing regional exploration in southern Peru requires moving beyond permissivity mapping toward a more discriminative understanding of mineral system efficiency. Future models should prioritize quantitative fertility indicators capable of distinguishing productive magmatic suites from barren intrusions. Geochemical ratios such as Sr/Y, La/Yb, Eu/Eu*, trace-element systematics, and zircon-based fertility proxies represent critical next-generation inputs that would substantially refine current lithology-based representations of magmatic potential. Equally important is the improvement of crustal architecture characterization. Integrating regional geophysical datasets and more precise structural interpretations would allow a transition from simplified buffer-based structural influence models to physically constrained representations of magma ascent corridors and fluid focusing pathways. This step is essential for reducing spatial overgeneralization and improving geological realism. At the geodynamic scale, incorporating quantitative plate-tectonic parameters—such as convergence rate variability, slab geometry evolution, and obliquity—would enable dynamic rather than static representations of tectonic controls on mineralization. Such integration would align regional prospectivity modeling with time-dependent mineral system evolution. Furthermore, incorporating hydrothermal fluid characterization, including fluid inclusion studies and trace-element signatures, would allow discrimination between permissive magmatic corridors and systems that achieved effective metal transport and precipitation. This represents a critical shift from identifying where mineral systems could form to understanding where they were most efficient. Ultimately, the evolution of Andean exploration strategies depends on integrating tectonics, structural architecture, magmatic geochemistry, and predictive modeling within a coherent mineral systems framework. Strengthening geological realism in model inputs will enable a transition from probabilistic permissivity maps to more robust assessments of mineral system fertility and efficiency, increasing the likelihood of identifying concealed or underexplored deposits.
RkJQdWJsaXNoZXIy MTM0Mzk2