Ecological drivers of seabird recovery after the eradication of introduced predators

Eradication of introduced mammalian predators to restore island ecosystems has become increasingly common, with over 800 successful projects around the world. Historically, introduced predators depleted populations of many native species, leading some, especially seabirds, to extinction or local extirpation. Island restoration is dependent on seabird population recovery, due to these birds’ indispensible role as ecosystem engineers. However, natural population responses are rarely documented and recovery dynamics are poorly understood. New Zealand holds the world’s highest diversity of seabirds and has completed more introduced predator eradication projects than any other country, making it an ideal location to study seabird population processes and island restoration. In this thesis I outline a model of seabird population growth, test the importance of key ecological variables in driving the recovery of burrow-nesting seabirds (order: Procellariiformes) on islands in New Zealand, and examine a possible method of enhancing population recovery. Finally, I summarize an effective post-eradication monitoring scheme that could provide information to improve the model of seabird population growth and facilitate priority setting for New Zealand’s seabirds. My thesis begins by constructing a generic conceptual model of seabird colony growth to identify key predictor variables relevant to recovery and re-colonization (Chapter 2). I tested the importance of these variables in driving seabird population responses after introduced-predator eradication on islands around New Zealand. The most influential variable affecting re-colonization of seabirds was the distance to a source population, with few cases of re-colonization without a source population ≤ 25 km away. Colony growth was most affected by metapopulation status; there was little colony growth in species with a declining metapopulation. I conclude that these characteristics can help guide the prioritization of newly predator-free islands for active management. The distribution of burrow-nesting seabird colonies is thought to be partly regulated by the availability and quality of suitable breeding habitat, which may limit colony growth after predator removal. I used a Bayesian hierarchical modelling approach to examine how nest-site selection differs among recovering procellariiform seabird communities after eradication of Pacific rats (Rattus exulans; Chapter 3). I found that soil depth was the most important predictor of burrow presence, abundance, and occupancy in plots among islands, with more burrows found in deeper soil. There was a striking linear relationship between burrow density and time since rat eradication (P < 0.01, R2 = 0.37) and birds showed weaker nest-habitat selectivity with increasing time since rat eradication (P = 0.02, R2 = 0.47). Results suggested that selection of particular nesting habitat may be more important in small recovering populations. Thus, colony expansion immediately after introduced-predator removal may be limited by nesting habitat quality, namely the availability of deep soil. When selecting nesting habitat, colonial animals can also use social cues provided by breeding conspecifics. Conservation practitioners have used seabirds’ affinity for conspecific cues to establish colonies at abandoned sites using decoys or call playback. However, success rates of these projects have varied. In Chapter 4 I tested the attraction of three sympatric petrel species to social cues in the form of vocalization playback. I then examined whether the size of breeding colonies within 1 km of playback locations (found to be important in Chapter 2) affected the strength of attraction. Grey-faced petrels (Pterodroma macroptera gouldi) were attracted to conspecific vocalization playbacks at all sites, fluttering shearwaters (Puffinus gavia) were only attracted at two of three locations, and flesh-footed shearwaters (P. carneipes) were not attracted at all. Response to playback increased with increasing densities of nearby breeding conspecifics (all P < 0.002). For some species, such as grey-faced petrels, vocalization playbacks may represent a cost-effective alternative to other restoration approaches. However, their effectiveness for individual species at different sites should be assessed before embarking on restoration initiatives. Although the recovery of burrow-nesting seabird communities is complex, driven in part by nesting habitat and social cues, the ultimate outcome will be limited by underlying assembly rules. Assembly rules are mechanisms governing colonization and eventual community structure, including inter-specific competition and facilitation. To draw inference on how inter-specific interactions regulate petrel community ‘reassembly’ after eradication, I investigated species co-occurrence using null model testing (Chapter 5). Community structure of six petrel species on six islands in north-eastern New Zealand provided evidence that reassembly is influenced predominantly by inter-specific facilitation. However, exclusion between petrel species increased as time since rat eradication increased, suggesting that interactions between species may be competitive at certain stages of recovery. I demonstrate how co-occurrence analysis can aid in understanding and managing recovery of communities, rather than single species, after predator eradication. Density dependence is important for managing recovery because it governs the rate and dynamics of population growth. In Chapter 6, I demonstrate how both positive and negative density dependence operate during seabird colony expansion after rat eradication. Using Bayesian hierarchical modelling of burrow density as a proxy for relative abundance, I tested whether petrel colonies increase in density or area on islands after rat eradication. I found that mean burrow density increased (mean effect size 0.05; 95% credible intervals 0.01 – 0.11), burrows remained clustered (i.e. spatially structured), but colony extent increased with time since rat eradication, with colonies filling over 70% of the island’s surface by 25 years after eradication. In Chapter 7, I outline a potentially effective petrel monitoring programme using a power analysis of simulated monitoring data based on burrow densities from Chapters 3-6. To detect levels of change in burrow density of interest to conservation managers with >80% power, I found that at least 80 plots of 3-m-radius on 15 islands must be monitored annually, with more plots required in less-preferred habitat (shallow slopes at low elevations). As large-scale changes continue to alter populations and conservation priorities for seabirds, data collected using robust monitoring approaches can be used to revise the model of population recovery and prioritize management interventions. Finally, I discuss the broader ecological implications of my thesis (Chapter 8). This includes my finding that, despite the central dogma in seabird ecology that species are strictly philopatric, strong evidence supports metapopulation connectivity as a driver of population recovery. Furthermore, my data suggest that positive and negative density dependence operate during recovery, meaning that remnant colonies facilitate initial colony growth, whereas competition may eventually reduce growth or encourage the spatial spread of a colony. I synthesize these ecological drivers of recovery into a set of management recommendations for island restoration, and suggest future research directions