Using ensemble modeling to predict the current distribution of Pistacia atlantica Desf. in Algeria

Abstract

To understand the distribution of Atlas pistachio (Pistacia atlantica Desf.) in Algeria, we analyzed the environmental factors influencing its habitat. This study employs an ensemble modelling (EM) approach, a robust predictive technique in ecological niche modelling that enables us to identify critical environmental drivers affecting plant distributions across different ecosystems’ focal species. The EM incorporated four prediction algorithms (generalized linear model, boosted regression trees, random forest, and maximum entropy algorithms); we modelled Atlas pistachio’s niche with 2810 occurrence points and 32 environmental variables, including climatic, edaphic, topographic, and anthropogenic factors. The model demonstrated high accuracy, with an AUC of 0.97 and TSS of 0.88. Key factors influencing distribution were precipitation in the driest month (Bio14), soil bulk density (BD), cation exchange capacity (CEC), human modification, and average diurnal amplitude (Bio2), with a relative importance of 20.1%, 12.7%, 6.7%, 4.9%, and 3.1%, respectively. These findings underscore the utility of ensemble modelling to pinpoint specific environmental variables critical to the species’ presence and ecological adaptability, which has broader implications for other plant species in arid landscapes. Notably, the probability of Atlas pistachio occurrence increased with BD and decreased with CEC and human influence. Our results emphasize the EM approach as a versatile tool in ecological modelling, facilitating species-specific analyses that contribute to broader ecological restoration efforts, especially in degraded arid and semi-arid regions. This study advances our understanding of Atlas pistachio’s environmental requirements and highlights the importance of EM in developing targeted programs to restore degraded ecosystems.

Graphical abstract