Spatial distribution modeling and mapping of a wild mushroom (Lactarius deliciosus L. Gray) in the Refahiye and Tekçam forest planning units

In this study, habitat suitability for the Saffron milk-cap mushroom (Lactarius deliciosus) was investigated through GIS-integrated statistical and machine learning methods in the Refahiye and Tekçam planning units. For this purpose, the relationships between predictive variables at different scales, such as topography, stand structure, and climate, and the presence/absence of the species were analyzed. 173 temporary sampling plots in temperate coniferous forests were determined by a simple random sampling method. Empirical relationships between the presence of the relevant species and environmental factors were defined using species distribution models. The model success was evaluated according to the TSS (True Skill Statistics) and ROC curve (Receiver Operating Characteristic curve). The modeling results indicated that the machine learning models performed slightly better than the statistical ones, however, all the models obtained total accuracy values greater than 80%. The random forests (RF) model was the most sensitive and specific with the highest values of ROC (0.96) and TSS (0.91), thus identifying it as the best modeling method. According to the results, L. deliciosus is a species highly affected by interannual variations in temperature, precipitation, solar radiation, and stand variables such as canopy closure. Furthermore, these methods can be valuable tools for predicting the distribution of other NWFPs and serve as a basis for elaborating sustainable management plans.