Determinação do efeito de ilhas pequenas a partir dos modelos bent-cable probabilísticos

Abstract

Ecological thresholds are change points or transitional zones within an ecological system signifying changes between different states. Detecting these thresholds is crucial for shaping conservation policies and sustainable practices. This research aims to spot random thresholds by employing probabilistic bent-cable models, a specialized type of segmented regression models adept at characterizing gradual transition zones. The study focused on modeling the Small Island Effect to identify the Area Threshold. To demonstrate this technique, data from Matthews et al. regarding the richness of plants in the archipelago of the Aegean Sea, Greece, were utilized. Statistical analysis utilized Bayesian methods, considering the non-normality of the data and how the area impacts the distribution and spread of species richness. Our results surpassed those of commonly used broken-line models for representing ecological thresholds. Additionally, the bent-cable model allowed for the retrieval of the underlying distribution of random thresholds, enabling a probabilistic interpretation of the Small Island Effect.