Groundwater-dependent ecosystems (GDEs) are a fundamental part of groundwater flow systems, where natural flow paths discharge and support unique ecosystems. GDEs are at risk worldwide, particularly in Mexico, where groundwater abstraction and anthropogenic pollution are major issues. This study presents the first regional-scale GDEs potential mapping across the Cuitzeo Groundwater Flow System (GFS) located in Central Mexico. Traditional methods, including the Analytic Hierarchy Process (AHP) and Weights of Evidence (WoE) were implemented, as well as two machine learning (ML) methods: Logistic Regression (LR) and Random Forest (RF) using geospatial and remote sensing data. The GDEs potential maps were validated with GDEs locations using the receiver-operating characteristic (ROC) curve, area under the curve (AUC), and other evaluation metrics. The RF (AUC=0.82) achieved the highest prediction, outperforming the LR (AUC= 0.70), the WoE (AUC=0.61) and the AHP (AUC=0.59) models, and showed statistically significant superiority in specificity over all models. GDEs potential zones indicate that GDEs are widely distributed in the Cuitzeo GFS where groundwater abstraction and pollution threaten their existence. These results provide a strong robust spatial framework to deepen the study of GDEs in the area and support future conservation management practices. • The first groundwater-dependent ecosystems (GDEs) mapping in Mexico was conducted. • Four common approaches in GDEs mapping were compared. • The Random Forest outperformed other machine learning and traditional approaches. • Groundwater in the study area supports wetlands, springs, and riparian vegetation.
Salgado-Albiter et al. (Sun,) studied this question.