A Machine Learning Approach to Assess the Relevance of Interspecific Acoustic Cetacean Co‐Occurrence When Modeling Species Presence

ABSTRACT Ecological models are essential for understanding cetacean distribution, particularly in marine ecosystems where direct observation is challenging. The present study applied passive acoustic monitoring (PAM) and artificial neural networks (ANN) to model the presence of dolphins (Family Delphinidae) in the Cerralvo Channel, Gulf of California, Mexico, incorporating 10 environmental descriptors and introducing the variable of interspecific acoustic co‐occurrence of cetaceans (IACC). The latter implies acoustic co‐detection regardless of proximity, interaction or mutual awareness among species. Data were collected from November 2022 to December 2023 using PAM, yielding 452 h of acoustic recordings. The best‐performing ANN achieved 81.7% accuracy, identifying IACC, wind speed, Chlorophyll‐a concentration, hydrophone distance‐to‐coast ratio, and diel pattern as the most influential descriptors. Variable contribution analyses revealed that IACC exerted the strongest positive influence on dolphin detection, with blue and sperm whale vocalizations being the most frequently detected with dolphins. These findings highlight PAM and ANNs as effective tools for capturing complex ecological relationships and emphasize IACC as a key driver shaping dolphin habitat use through the shared acoustic space.