Elsevier

Ecological Modelling

Volume 393, 1 February 2019, Pages 66-75
Ecological Modelling

Macquarie Island’s northern giant petrels and the impacts of pest eradication on population abundance

https://doi.org/10.1016/j.ecolmodel.2018.11.004Get rights and content

Highlights

  • Our model explained observations of northern giant petrel breeding pairs well.

  • Up to 1400 petrels estimated killed over two years from non-target poisoning.

  • The population is estimated to have a high probability of recovery.

  • The approach can be applied to other populations affected by sudden mass mortality.

Abstract

Pest eradication conducted over the years 2010 to 2014 at Macquarie Island successfully eradicated introduced rabbits, rats and mice from this sub-Antarctic island. The initial aerial baiting phase in the winters of 2010 and 2011 resulted in significant mortality of several native seabird species through primary and secondary ingestion of brodifacoum bait. A species of key concern is the northern giant petrel (Macronectes halli), which, although relatively abundant and increasing on Macquarie Island, is listed as threatened under Australian legislation and was one of the species most affected by poisoning. We use a Bayesian approach to estimate the total mortality and the response of the population to the poisoning event over the short- to medium-term. We then considered how population abundance might respond over the ensuing years. Projections of population trajectories suggest a greater than 50% probability of recovery to the pre‐poisoning levels of 2009 breeding pairs by 2017. This modelling approach could be applied to future planned eradications to quantify the mortality and recovery of incidentally affected populations.

Introduction

Macquarie Island (54°30' S, 158°57' E) is a World Heritage Site located approximately 1500 km south east of Tasmania, Australia (Fig. 1). The island is uninhabited except for a permanent research base. It is under Tasmanian state jurisdiction and is recognised for its wildlife, including numerous populations of seabirds and marine mammals (Parks and Wildlife Service, 2006). Rats and mice flourished on the island soon after their inadvertent introduction from European ships in the early 1800′s and rabbits were introduced on the island as a food source for sealers around 1870 (Terauds et al., 2014).

The increasingly destructive impact of rabbits on the native vegetation (Scott and Kirkpatrick, 2013), and the consequent impacts on seabird nesting habitat, combined with direct predation by rodents of burrowing seabird eggs and chicks (e.g. Brothers and Bone, 2008), led to a joint Tasmanian and Australian Government program, the Macquarie Island Pest Eradication Project (MIPEP). At 12,785 ha, this was to be the largest eradication program for rabbits, rats and mice attempted anywhere in the world at the time (Parks and Wildlife Service, 2014). The operational phase of the MIPEP commenced in June 2010 and was declared successful in April 2014. Springer (2016) describes the full technical and chronological account of the MIPEP implementation.

The initial helicopter aerial-baiting phase, in which brodifacoum poison was to be spread across the island via two whole-of-island bait drops, was specifically timed for the austral winter, when the majority of the islands’ wildlife had departed, to minimise non-target impacts (Parks and Wildlife Service, 2009a, b). A delayed arrival to the island, coupled with sustained inclement weather, resulted in limited flying opportunities in 2010. Only a small proportion of the island (8%) had been baited by the end of June 2010, and a decision was made to suspend activity until the following winter (Parks and Wildlife Service, 2014). In the months that followed, large numbers of native seabirds, including kelp gulls (Larus dominicus), black duck (Anas superciliosa), subantarctic skua (Catharacta skua), northern giant petrel (Macronectes halli) and southern giant petrel (M. giganteus) were reported as killed. Depending upon species, poisoning occurred through direct ingestion of toxic baits and/or, in the case of scavenging species, through secondary ingestion by feeding on toxic rabbit and seabird carcasses. Giant petrels are scavengers, and thus it took a period of weeks to months, depending upon the composition of individual birds’ diet, before lethal doses of brodifacoum accumulated in their systems.

As both giant petrel species are listed as threatened under the Australian Environmental Protection and Biodiversity Conservation Act 1999 (EPBC Act), the mortality associated with the 2010 secondary poisoning was considered a matter of national environmental significance and triggered a review of the eradication by the Australian Commonwealth government (Anon, 2010). Given the clear ecosystem benefits of a successful eradication to the island, the review recommended that the eradication proceed with additional measures aimed at reducing the volume of toxic carcasses available to scavenging seabirds. The principal mitigation measure was the introduction of rabbit calici-virus (Rabbit Haemorrhagic Disease Virus) in the summer months to reduce the rabbit numbers on the island prior to baiting (Cooke et al., 2017). This measure was extremely effective, and was estimated to have reduced the rabbit numbers by 80–90% (Parks and Wildlife Service, 2014) prior to the second attempt at aerial baiting. A second mitigation measure was the allocation of additional staff-time to conduct thorough ground searches and dispose of poisoned target and non-target carcasses before they could be ingested by scavenging seabirds.

The second season of baiting occurred throughout May and June of 2011. Over May to November 2011, ground teams recovered approximately 1,450 individual dead birds. This represented an approximately 50% increase in documented non-target native seabird mortality compared to the previous baiting attempt in 2010 (n∼960). It should, however, be noted that significantly more poison bait was broadcast in 2011, with 100% of target bait volume delivered compared to <5% in 2010. In addition, there was increased search effort dedicated to locating and estimating non-target mortality in 2011, so estimates for that year are more reliable, whereas the number killed in 2010 is likely underestimated. The baiting was followed by an on-ground hunting phase, during which teams of hunters and specifically trained dogs conducted comprehensive ground searches for any remaining rabbits and rodents. The ground team found the last rabbit in November 2011 and the Tasmanian Government declared the island free of rabbits and rodents in April 2014 (Parks and Wildlife Service, 2014); the vegetation has been recovering ever since (Whinham and Alderman, 2016).

Despite the mitigating actions, the combined observed non-target mortality across the two baiting periods was substantial for several native seabird species, particularly for kelp gulls, northern giant petrels and skuas (observed mortalities of 989, 692 and 512 respectively across the two winters), noting that the observed mortalities are minima, as many birds will have died at sea or been otherwise undetected by ground personnel. While the true scale of the mass mortality event is unknown, it is likely to have been detrimental to many of the affected seabird populations.

Here, we focus on estimating the scale and impact of this mortality event on one of the most affected species, the northern giant petrel. Given the conservation status of the species (Vulnerable and Rare under Commonwealth and State legislation respectively), and the global significance of this population with an estimated 1,500–1,800 annual breeding pairs prior to the mortality event, representing approximately 15% of the global population, the large number of individuals killed was of concern both nationally and internationally (ACAP, 2016; Phillips et al., 2016).

Single, or rare, mass mortality events have been documented for fish, seabirds and other wildlife, occurring as a result of both human activities and as natural events. These have included mortality due to algal blooms, fire, flood, sea-ice expansion, changes in ocean conditions, pollution and poisoning (Winters et al., 1986; Baduini et al., 2001; Lyon and O’Connor, 2008; Richlen et al., 2010). If the species impacted by a mass mortality event has threatened population status (either before or potentially after the event) or some commercial value, then an assessment of their post-event status and projected recovery time may be valuable for the responsible managers. For example, Punt and Wade (2012) considered the consequences of a catastrophic mortality event on population abundance for the eastern north Pacific stock of gray whales. Over the years 1999–2000 this population experienced an unusually large number of strandings of emaciated adult whales and a marked reduction in calf production, possibly due to heavy ice cover in those years. Punt and Wade (2012) explicitly included an additional mortality parameter in a population model for this population to account for the mass mortality event and applied a Bayesian approach to parameter estimation to predict stock status relative to management reference points.

Assessments of seabird population status have largely been driven by the need to understand the impacts of additional anthropogenic mortality through bycatch associated with industrial scale fishing. Industrial fishing, by longline and trawl gear, has been implicated in numerous seabird population declines (e.g. Weimerskirch et al., 1997; Gales, 1998; Phillips et al., 2006; Tuck et al., 2011). Assessments of bycatch often assume that annual fishing mortality is a function of fishing effort (numbers of hooks set or trawl hours), and as such is an ongoing mortality source, rather than a single event (Tuck et al., 2001, 2015; Zador et al., 2008; Robinson et al., 2015; Thomson et al., 2015). For example, Zador et al. (2008) consider the impacts of the Alaskan trawl fishery on the endangered short-tailed albatross (Phoebastria albatrus), and provide probabilities of achieving management goals (such as population recovery) according to assumptions regarding future fishing activity and uncertainty in growth rates.

Following Zador et al. (2008), we use a Bayesian approach to quantify parameter uncertainty. The advantage of quantifying uncertainty in this manner and providing probabilistic decision tables for managers is that a more honest and informed assessment of the uncertainty inherent in the biological and management systems is provided, thereby allowing managers the opportunity to debate potential risk trade-offs and learn more about their system (Zador et al., 2008; Bunnefeld et al., 2011; Robinson et al., 2015).

We use a Bayesian approach to estimate the population consequences of the major mortality event on the Macquarie Island population of northern giant petrels that occurred over 2010 and 2011. The true scale of the mortality is not known as observations of mortality are minima, therefore our objectives were to (1) use known or inferred population parameters and observations of annual breeding pairs within a population model that can predict a time-series of population size, (2) estimate a probability distribution for the magnitude of mortality as a consequence of the non-target poisoning and (3) evaluate the most likely future population trajectories to inform conservation management decisions. We show how a model like this can be used to evaluate the potential scale and impacts of mortality events, such as island pest eradications that may result in secondary mortality of threatened species, both prior to and after the event. This can assist managers in decisions regarding threatened status and recovery planning.

Section snippets

Population monitoring

Northern giant petrels nest singly or in small, loose groups on most coastal areas of Macquarie Island. A small number breed inland, including on ridgelines and headlands and occasionally on the highland plateau. Egg laying commences in mid-August and chicks fledge throughout February. Giant petrel populations on Macquarie Island have been studied sporadically since the 1950s. These studies have mostly focused on banding individuals to investigate distribution at sea, with occasional population

Results

The marginal posterior distributions for the four estimated parameters showed that the data substantially updated the uniform priors, with the posterior modes well determined for each of the parameters (Fig. 2; Table 3). The three chains overlap considerably, indicating support for the conclusion that the MCMC sampling process has converged to the posterior distribution. The correlation plots (Fig. 3) show a strong positive correlation between the initial number of age 1 birds and juvenile

Discussion

A primary motivation for developing this model was to quantify the impact of the Macquarie Island Pest Eradication Project operations on the status and trends of the islands’ northern giant petrel population. Our analyses suggest that the mass mortality, resulting from secondary poisoning during the baiting operations in 2010 and 2011, had a considerable immediate impact, killing up to 1,400 individuals over two years, resulting in a decline in breeding pairs by approximately 33% (posterior

Conclusion

Assessing the population impacts of mortality events, whether a single mass mortality event or ongoing additional mortality, is an important consideration for managers of natural resources. Relatively simple models, such as that presented here, can be used both before and after a mass mortality event to inform decision makers. Prior to the event, the model can be used to anticipate likely population impacts, if estimates of mortality can be quantified or provided in a scenario testing

Acknowledgements

The authors thank the Australian Bird and Bat Banding Scheme (ABBBS), Dr. Rosemary Gales, Dr. Aleks Terauds, Nigel Brothers, Keith Springer and the many field staff and volunteers for their contribution to both the long-term monitoring of northern giant petrels on Macquarie Island and to MIPEP. Helen McConnell reported the banded giant petrels on Enderby Island and facilitated the collection and analysis of samples. The authors also thank the editor and reviewers for their helpful comments and

References (52)

  • C.L. Baduini et al.

    Mass mortality of short-tailed shearwaters in the south-eastern Bering Sea during summer 1997

    Fish. Oceanogr.

    (2001)
  • A.D. Black et al.

    Non‐target mortality in South Georgia rodent eradication

    Final Report to GSGSSI. Report to the Government of South Georgia and the South Sandwich Islands

    (2017)
  • N.P. Brothers et al.

    The response of burrow-nesting petrels and other vulnerable bird species to vertebrate pest management and climate change on subantarctic Macquarie Island

    Pap. Proc.

    (2008)
  • B. Cooke et al.

    Rabbit haemorrhagic disease: macquarie Island rabbit eradication adds to knowledge on both pest control and epidemiology

    Wildl. Res.

    (2017)
  • S. Copello et al.

    Spatio-temporal overlap between the at-sea distribution of Southern Giant Petrels and Fisheries at the Patagonian Shelf

    Polar Biol.

    (2009)
  • M.C. Downes et al.

    Banding of giant petrels at heard and Macquarie Island

    Emu.

    (1954)
  • P. Gandini et al.

    Spatial and temporal patterns of seabirds by-catch in the Argentinean Longline Fishery

    Fish. Bull.

    (2006)
  • R. Gales

    Albatross populations: status and threats

  • A. Gelman et al.

    A single sequence from a gibbs sampler gives a false sense of security

  • A. Gelman et al.

    Bayesian Data Analysis

    (2004)
  • J. Gonzalez-Solis

    Sexual size dimorphism in northern giant petrels: ecological correlates and scaling

    Oikos

    (2004)
  • J. Gonzalez-Solis et al.

    Offshore spatial segregation in giant petrels Macronectes spp.: differences between species, sexes and seasons

    Aquat. Conserv. Mar. Freshw. Ecosyst.

    (2008)
  • J. Gonzalez-Solis et al.

    Foraging partitioning between giant petrels Macronectes spp. And its relationship with breeding population changes at Bird Island, South Georgia

    Mar. Ecol. Prog. Ser.

    (2000)
  • M. Haddon

    Modelling and Quantitative Methods in Fisheries

    (2011)
  • S. Hunter

    Breeding biology and population dynamics of giant petrels Macronectes at South Georgia (Aves: procellariiformes)

    J. Zool. (Lond.)

    (1984)
  • F. Hurtado-Ferro et al.

    Looking in the rear-view mirror: bias and retrospective patterns in integrated, age-structured stock assessment models

    Ices J. Mar. Sci.

    (2015)
  • Cited by (7)

    • Quantitative evaluation of the effects of bycatch on native species using mathematical models

      2022, Ecological Modelling
      Citation Excerpt :

      To prevent the unintended death of native species, it is necessary to evaluate bycatch by conducting detailed surveys, examining control plans, and predicting outcomes (Cromarty et al., 2002; Courchamp et al., 2003). Bycatch has been quantified by field monitoring (Taylor and Thomas, 1993; Dowding et al., 1999; Empson and Miskelly, 1999; Powlesland et al., 1999; Eason et al., 2002) and statistical estimation (Armstrong et al., 2002; Watari et al., 2013; Alderman et al., 2019), but there are few options available for theoretical evaluation. Some studies have proposed that it is necessary to kill all invasive species to prevent a negative impact on native species (e.g., Cromarty et al., 2002), whereas others have argued that invasive species should merely be minimized in consideration of bycatch (e.g., Eason et al., 2002; Courchamp et al., 2003; Howald et al., 2007).

    View all citing articles on Scopus
    View full text