Do Harbour Porpoise Mortality Records Reflect Living Population Structure? A Matrix Population Model Diagnostic

Effective conservation of marine mammals depends on reliable demographic information, yet acquiring such data for highly mobile cetaceans is challenging. Harbour porpoises (Phocoena phocoena) are widely used as sentinel species, but much of what is known about their demography comes from stranding and bycatch records. While invaluable, these data may be filtered by drift, detection, reporting effort and age-specific vulnerability. We developed a matrix population model (MPM)-based diagnostic framework to quantify how observed stage compositions in mortality datasets deviate from asymptotic demographic expectations. Using a stage-structured MPM parameterised from published vital rates, we derived the stable stage distribution (SSD) and compared it with age-class distributions from 10,863 harbour porpoise strandings (of 16,181 total records) in Danish and North Sea waters, bycatch and mixed-mortality records, and hunting captures from Greenland and Denmark. Deviations from SSD, and from the composition expected among deaths, were assessed using goodness-of-fit tests, distance metrics, and tests of the juvenile proportion. Strandings showed strong and consistent departures from SSD expectations, with juveniles markedly over-represented across spatial and temporal scales. Juveniles remained over-represented relative to the composition expected among deaths under baseline rates, and the departure from the living population was robust to variation in vital rates. Divergence from the asymptotic stage structure was moderate to strong, varied across space, and increased significantly over time in 2 of 6 regions. Bycatch, mixed-mortality, and hunting captures showed comparable SSD divergence, indicating that juvenile-heavy mortality composition is not an artefact of stranding sampling alone. Our results show that harbour porpoise stranding, bycatch, and even hunting data should not be assumed to be demographically representative of living populations. Separating sampling bias from higher juvenile mortality needs independent estimates of juvenile survival.