Abstract
Understanding the complex patterns of habitat use by animal species and unravelling the processes that control these patterns are fundamental goals in the field of ecology. In cases where data on the wintering range of bird species are limited, species distribution models become essential tools for predicting distributions. This study investigates the winter habitat suitability of Marmora's Warbler (Curruca sarda Temminck, 1820), an endemic Mediterranean bird, for which research on habitat preferences during the wintering period is particularly limited in North Africa, especially in parts of northern Algeria and Tunisia. Initially, 96 occurrence records, ranging from 1970 to 2022, were extracted from biodiversity database platforms and local surveys. After filtering for spatial bias and removing duplicate records, 51 points were retained for modeling. Additionally, 1085 absence points were generated using eBird sampling event data, applying strict filters to ensure reliable presence-absence data for robust modeling. Advanced modeling techniques, including boosted regression trees and random forest (RF), were employed, with RF achieving the highest performance metrics, including an area under the curve of 0.956 and a true skill statistic of 0.78. Habitat suitability maps highlight Algeria (44,062 km2 ) and Tunisia (4464 km2 ) as critical wintering areas. These models identified key environmental predictors influencing habitat suitability, with annual mean temperature (11–13 °C), elevation (1024–1300 m asl), and precipitation seasonality (32–41%) emerging as the most significant factors. These findings provide essential insights for conservation planning, emphasizing the need to address habitat fragmentation and climatic variability to ensure the survival of this species.