Abstract
Species recovery is often the goal of conservation management, usually achieved by reducing threats throughout its range. For management to be effective, it needs to be based on accurate information on the ecology and biology of the focal species. Unfortunately, we often lack the necessary information to design effective protection for many species. As a result, many management actions are based on a variety of untested assumptions. The endangered Hector’s dolphin (Cephalorhynchus hectori) is the only endemic cetacean in Aotearoa New Zealand. It is a species of high conservation concern as it has suffered widespread declines throughout its range due to bycatch in gillnet and trawl fisheries. To date, management has focused on reducing the interaction between dolphins and fisheries, by implementing area-based prohibitions on gillnet fisheries and restrictions on trawl fisheries, as well as, establishing fisheries related mortality limits (FRMLs). In 2020 a major review of the management of Hector’s dolphin was conducted, and the resulting changes were based largely on the results of a spatially explicit fisheries risk assessment (SEFRA). The SEFRA used a species distribution model (SDM) built from occurrence data collected during nationwide aerial surveys at a large spatial resolution and was then used to predict densities across fisheries management areas. Several assumptions were inherent with this management approach, including: 1) large-scale distribution data are sufficient to understand the suitability of habitat, 2) the designated population boundaries are appropriate management units, and 3) the estimates of demography used in management decisions were accurate. For this thesis I chose to explore these three assumptions by testing the spatial transferability of SDMs, investigating connectivity of sub-populations, and provide an updated estimate of fecundity. I then used simulations to investigate the impact that different levels of management may have on dolphins.
Between 2021 and 2024, I completed over 8,200 km of active survey effort to collect distribution data (440 presence and 399 absence locations) and environmental DNA (eDNA) samples (n = 129) from dolphins living at three areas: Banks Peninsula (Horomaka), Timaru (Tihi-o-Maru), and Otago (Ōtākou)/Dunedin (Ōtepoti). In addition, I completed a further 1,000 km of survey effort at Banks Peninsula to help continue the longitudinal photo-identification project. I showed, that when SDMs were transferred to new areas they were unable to predict dolphin occurrence any better than random, likely resulting from different patterns of habitat use. Even when model interpolation (validated within the same area) was high, or data were combined from multiple sites to build the models, spatial transferability was poor. It is unlikely that large-scale distribution patterns will reflect local habitat use. Understanding the distribution of Hector’s dolphins likely requires data collected at local scales.
I successfully amplified mtDNA from eDNA samples and was able to isolate known haplotypes at every site. Haplotype frequencies at Timaru and Banks Peninsula were similar to those previously described, with a dominance of a common east coast population haplotype. The haplotype frequency at Dunedin, was significantly different to both those at Timaru and Banks Peninsula, with a dominance of haplotype S that has been previously described from the south coast population. Dolphins in Dunedin have been considered as part of the east coast population; a designation that is unlikely to be correct. I provide insights into low connectivity between sub-populations, even when geographically close (i.e. < 100 km). The current population designations for Hector’s dolphins are unlikely to contain a series of interacting sub-populations. As such the assumption that a management action distributed throughout the wider population (e.g. on the East Coast) would provide adequate protection for all sub-populations within, is unlikely to be true.
Using almost four decades of photo-identification data from Banks Peninsula, I provided a new estimate of fecundity. I estimated a much lower fecundity for Hector’s dolphins (0.298, 95% CI: 0.296 – 0.302) than had previously been described (0.409, 95% CI: 0.267 – 0.635). A result that would lower the estimated capacity for the population to increase. As such, FRMLs may exceed sustainable levels of bycatch.
Finally, I built an individual based model and simulated population trajectories using three different levels of protection (no protection, current management, and protection to the 100 m isobath). Simulation results indicated that the current level of protection was unlikely to result in population growth at any of the three areas. The most optimistic realised population growth rate (λ) was at Banks Peninsula, where protection is greatest, resulting in an on average stable population over three generations, and a relative increase in λ (compared to no protection) of approximately 4.7%. In Dunedin, the area where current protection is the lowest, the outlook was the worst with only a 3.0% increase in λ after the introduction of protection. When protection was extended to the 100 m isobath, λ had the largest relative increase in Timaru (6.0%), while Banks Peninsula and Dunedin were similar (5.5%). Even with increases in λ, population growth remained slow (<1% per annum).
Over the course of this thesis, I have provided new insights into the habitat use, connectivity, and reproduction of Hector’s dolphins. I developed individual based models, which can be used to explore trends and the likely impact of management actions. Many of the assumptions that current management is based on are inconsistent with my results and I recommend that future assessments should prioritise collecting data on and managing sub-populations at a local level.