Predicting cetacean distributions in data‐poor marine ecosystems

Abstract

AbstractAimHuman activities are creating conservation challenges for cetaceans. Spatially explicit risk assessments can be used to address these challenges, but require species distribution data, which are limited for many cetacean species. This study explores methods to overcome this limitation. Blue whales (Balaenoptera musculus) are used as a case study because they are an example of a species that have well‐defined habitat and are subject to anthropogenic threats.LocationEastern Pacific Ocean, including the California Current (CC) and eastern tropical Pacific (ETP), and northern Indian Ocean (NIO).MethodsWe used 12 years of survey data (377 blue whale sightings and c. 225,400 km of effort) collected in the CC and ETP to assess the transferability of blue whale habitat models. We used the models built with CC and ETP data to create predictions of blue whale distributions in the data‐poor NIO because key aspects of blue whale ecology are expected to be similar in these ecosystems.ResultsWe found that the ecosystem‐specific blue whale models performed well in their respective ecosystems, but were not transferable. For example, models built with CC data could accurately predict distributions in the CC, but could not accurately predict distributions in the ETP. However, the accuracy of models built with combined CC and ETP data was similar to the accuracy of the ecosystem‐specific models in both ecosystems. Our predictions of blue whale habitat in the NIO from the models built with combined CC and ETP data compare favourably to hypotheses about NIO blue whale distributions, provide new insights into blue whale habitat, and can be used to prioritize research and monitoring efforts.Main conclusionsPredicting cetacean distributions in data‐poor ecosystems using habitat models built with data from multiple ecosystems is potentially a powerful marine conservation tool and should be examined for other species and regions.