Predictive modeling of Cedrus atlantica natural regeneration: grazing intensity and bioclimatic extremes as the primary drivers in the Ain Leuh Forest, Middle Atlas of Morocco

North African forests are experiencing a significant lack of woody species’ natural regeneration as a result of global change factors like climate change and human pressure. Natural regeneration is a crucial process in forest ecosystem dynamics, providing valuable information on the future of forest stands. A lack of recruitment can jeopardize the sustainability of forests and impair the many ecosystem services they supply. The Atlas cedar (Cedrus atlantica (Endl.) Manetti ex Carrière) is a major woody species in North Africa and the Mediterranean region in general. Modeling this species’ natural regeneration occurrence and determining the underlying causes of its variability at the forest scale is critical for ecological, management, and restoration objectives. In this study, two predictive models were used to investigate the natural regeneration occurrence of Atlas cedar in mixed stands of the Ain Leuh Forest, which is located in the central Moroccan Middle Atlas. Data from 1885 plots was used to create a standard Binary Logistic Regression model (BLR) and a Random Forest machine learning model (RF), with predictors chosen from a consistent dataset of 55 explanatory variables such as bioclimatic, edaphic, topographical, stand and anthropogenic factors, and vegetation indices. Exploratory data analysis, feature selection, and multicollinearity analysis were carried out utilizing point biserial correlation, polychoric correlation, and Spearman’s rank correlation. To train models and avoid overfitting, a tenfold cross validation was utilized. The models were then assessed during the testing phase using the area under the curve metric (AUC). According to the findings, grazing intensity and bioclimatic extremes, such as precipitation and temperature, have the greatest impact on natural regeneration occurrence. The BLR model outperformed the RF model (AUC = 0.659) with an AUC of 0.704. Spatial prediction using BLR shows that extremely appropriate locations for Atlas cedar natural regeneration, with a probability of occurrence more than 0.60, are relatively limited, accounting for 6.8% of the total surface area. The findings of this study demonstrate that: (1) grazing intensity and bioclimatic extremes are the primary factors influencing the spatial variability of Atlas cedar natural regeneration occurrence in the Ain Leuh Forest; edaphic and topographical factors have a secondary influence; (2) the current suitable areas are very limited in space, primarily located in the high altitude stands of the forest; and (3) the predicted future climate shift into drier conditions in North Africa may worsen the current situation by impeding recruitment and regeneration processes.