Banks Peninsula, New Zealand has been the epicentre for both the research and conservation of Hector’s dolphins (Cephalorhynchus hectori), an endemic and taonga species. After research in the 1980s detected unsustainable rates of bycatch in gillnets here, the first area-based protection for the species (the Banks Peninsula Marine Mammal Sanctuary, or BPMMS) was established in 1988. To assess the effectiveness of the BPMMS, as well as additional restrictions enacted in 2008, this thesis leverages long-term, photo-ID monitoring of this population, which has been ongoing since 1986.

In addition to refining capture-recapture models to update estimates of survival rate and trends in population size, I also aimed to improve the data quality and management practices of this long-term dataset. Therefore, Chapter 2 details the evolution of data collection and management practices of the photo-ID programme, including the development of a new relational database in Microsoft Access. Chapter 3 investigates the impact of changes in photo-ID protocol (mainly, data management practices and thresholds for photo quality and mark distinctiveness) on past estimates of survival rate estimated from photo-ID data collected from 1986–2006. Results from the analyses of Chapter 3 indicated that a landmark finding by Gormley et al. (2012), which provided the first evidence that the BPMMS increased survival rate, was robust to the changing photo-ID protocol during the study. Point estimates of survival were essentially unaltered, but their annual process variation was markedly improved from σ_φ = 0.39 (SD=0.35) to σ_φ = 0.13 (SD=0.16), suggesting the new, stricter protocol reduced “noise” in the dataset.

Building upon this prior research that indicated survival improved after the BPMMS, Chapter 4 incorporates additional years of photo-ID data (encompassing 1986–2020), as well as updated capture-recapture models, to also assess the effectiveness of the 2008 restrictions. An assortment of hierarchical Bayesian capture-recapture models (including the model used by Gormley et al., 2012), each using a different strategy to account for heterogeneity of capture, were applied to the dataset and compared using WAIC. A new model, that incorporated additional covariates for an assigned geographic “zone” in which an individual was usually sighted, survey effort to that area, and the interaction between an individual’s assigned zone and the number of times it was seen the previous year, was favoured. This model suggested a 77% probability that survival increased after the BPMMS, but only a 43% probability survival improved after the 2008 restrictions. Survival rate for the most recent management period (2008–2020) is estimated as 0.91 (95% CI: 0.87 – 0.95), indicating a notable increase (by 0.023) compared to the estimate of survival before protection (0.89).

To examine trends in abundance, Chapter 5 applied even stricter criteria for data inclusion to reduce bias caused by variation in survey effort and sampling intensity. Using photo-ID data collected from the southern portion of Banks Peninsula between 2001– 2020, population growth rate was estimated as 0.31% (95% CI: -1.08 – +1.82%) per annum. While this is the first evidence that the Banks Peninsula population (at least the southern subset) is stable or slowly increasing, the rate of growth is too slow to indicate a recovering population.

Overall, these analyses align with previous assertions that while the BPMMS has improved survival, historic levels of area-based protection have not been sufficiently comprehensive to result in genuine recovery (i.e., an increasing population size). Emphasising the importance of long-term monitoring, these findings underscore the need for more extensive protection measures, particularly against gillnets and trawling, to ensure a sustainable Hector’s dolphin population in Banks Peninsula.