AIS-based noise models underestimate vessel noise exposure risk to marine mammals.

Abstract

Vessels are the dominant source of anthropogenic underwater noise, with well-documented effects on marine mammal behaviour and physiology that may ultimately compromise individual fitness. The Automatic Identification System (AIS) provides standardized, global, high-resolution positional information for most commercial and passenger vessels, making AIS-based noise models widely used to assess exposure risk. However, their accuracy remains largely unvalidated. To evaluate their reliability, we deployed animal-borne GPS and biologging tags on 13 wild harbour porpoises, spanning shallow nearshore habitats and major shipping lanes. Over 60% of the tag-detected vessel noise exposures could not be attributed to AIS-tracked vessels within 4 km of the tagged animals, demonstrating that AIS-based impact assessments may severely underestimate true exposure rates in both small-vessel-dominated coastal areas and offshore waters. We further show that certain vessel types, such as pilot boats, contribute disproportionately to exposure rates relative to their abundance, indicating that vessel noise exposure is driven not only by traffic volume but also by vessel-type-specific operating patterns. Additionally, porpoises maintained greater distances from fast-moving, loud vessels, resulting in received noise levels comparable to those from slower vessels and suggesting vessel-type-specific avoidance responses. These findings reveal critical limitations in AIS-based approaches and underscore the need for integrated monitoring frameworks that combine AIS data with complementary tracking methods capable of capturing the full spectrum of vessel activity while accounting for vessel-type-specific risks. Improving exposure assessments is essential for protecting marine biodiversity in increasingly trafficked habitats and for accurate monitoring under the EU Marine Strategy Framework Directive.