Population dynamic modeling of the MÄui dolphin based on genotype capture-recapture with projections involving bycatch and disease risk Report to Ministry for Primary Industries, Wellington, NZ
Justin Cooke, Rochelle Constantine. Rebecca M. Hamner. Debbie Steel, C. Scott Baker
The MÄui dolphin (Cephalorhyncus hectori maui) is endemic to the west coast of the North Island of New Zealand and is listed by IUCN as Critically Endangered. In this project an individual-based population model was fitted to genetic identification data collected during 2001-16 from both living and beachcast or entangled animals (Baker et al. 2016). Data on cause of death from the DoC website and for three individuals necropsied by Roe et al. (2013) were also used. The model fits were used to provide estimates of recent population size and trend, and to project the population into the future under a range of mitigation scenarios.
In the absence of assumptions regarding the natural growth rate or anthropogenic threats, the best-fitting model involved no individual heterogeneity and no sex difference in survival or sampling probability and indicates a population declining at the rate of 3-4% per year over the period 2001-16. The estimated population trajectory was insensitive to the assumed age at first reproduction. The results show that approximately 10% of Maui dolphin deaths are recovered
Additional model runs were performed combining four sets of assumptions estimating the effects of commercial fisheries bycatch and/or other anthropogenic sources of death. Estimates of annual commercial fisheries death (hereafter â€˜bycatchâ€™rates were generated externally, based on outputs of the spatially explicit multi-threat risk assessment described in Roberts et al. (2019). That assessment applied a spatially explicit Bayesian model fitted simultaneously to all commercial fisheries effort and fisheries observer data across the full geographic range of Hectorâ€™s dolphins (of which Maui dolphins are a subspecies); that external model estimated approximately 0.12 commercial fisheries deaths per year for Maui dolphins in the last five years. For runs including non-fisheries threats, non-fishery deaths were estimated directly in the current model, fitting to necropsy data and mark-recapture data for Maui dolphins only.
The first set of model runs (Group A) assumed that the time series of bycatch risk estimated in Roberts et al. (2019) is accurate, and that commercial fisheries are the only anthropogenic threat to the dolphins. These runs imply that if the population is not already near carrying capacity, then the population should be increasing, but the fit to the data is poor. Use of a lower value for the natural growth rate r0, instead of an externally derived prior from Edwards et al. (2018), improves the fit only slightly.
A second set of model runs (Group B) again assumes that commercial fisheries are the only anthropogenic threat affecting the dolphins, but this time treats the annual time series of fisheries risk by Roberts et al. (2019) as a relative index only -- i.e. the inter-annual pattern is considered accurate but the absolute magnitude of fisheries risk is unconstrained. Under these runs the absolute fisheries risk is estimated to be 15-20 times higher than was estimated by Roberts et al. (2019), (with a mean of 1.5-2.4 commercial fisheries deaths per year in the last five years, down from 3-6 per year in the early 2000s). Model fits are substantially improved. However, the plausibility of such a high fisheries risk is doubtful, in the absence of a priori reasons to expect that the catchability per encounter with fishing effort would be higher for MÄui dolphins than for Hector's. These runs project that the population has declined due to historical bycatch and will continue to decline if commercial fisheries deaths continue at this level. A reduction in fisheries risk of 50% relative to the 2016 level would be just enough to stop the decline, unless the lower r0 is assumed. A reduction of 75% would be sufficient in all the cases considered.
A third set of model runs (Group C) assumes that the time series of fisheries risk estimated in Roberts et al. (2019) is accurate and that the death rate from other (unspecified) anthropogenic threats has been constant over time. These runs estimate that approximately 2.8 â€“ 4.1 annual deaths from non-fisheries anthropogenic threats are required in order to fit the historical population trajectory; model fits are better than under the first set of runs but worse than under the second set.
A fourth set of runs (Group D) were fitted to limited data on deaths attributable to the parasite Toxoplasma gondii (which was the diagnosed cause of death for 2 out of 3 MÄui dolphin carcasses examined for this disease); outputs suggest that toxoplasmosis may account for all of the unexplained anthropogenic mortality (i.e. 2.8 â€“ 4.1 deaths per year, comparable to Group C runs above). Under these runs, if toxoplasmosis (or other non-fisheries) risk remains at the estimated level, the population is predicted to decline towards extinction. A reduction of anthropogenic risk by 50% per decade beginning in 2030 would reduce the probability of extinction, while a reduction of the anthropogenic risk at the rate of 50% per 5 years beginning in 2025 would virtually eliminate the likelihood of extinction.