Statistical And Machine Learning Models Applied To Species Distributions Under Climate Change Scenarios: A Systematic Review

Abstract

Objectives: Climate change is altering species distribution worldwide, causing shifts in geographic ranges and disruptions in ecosystems. To understand these changes and develop conservation strategies, species distribution models (SDMs) have become increasingly relevant. Methods: This systematic review analyzes 47 studies published from 2015 to 2024 that have applied statistical models and machine learning techniques to model species distribution under climate change scenarios. The study examines the methodological approaches used, the main climatic factors considered, and the most studied taxonomic groups. Results: The results highlight the growing adoption of machine learning techniques, such as MaxEnt and Random Forest, due to their ability to capture complex relationships between environmental variables. Temperature and precipitation emerge as key climatic factors, while the area under the curve (AUC) is the most used validation metric. Conclusions: Despite advancements in modeling, gaps remain in the representation of certain geographic regions and underrepresented taxonomic groups, emphasizing the need to expand future research to enhance the understanding of climate change impacts on biodiversity.