Managing Invasive Mammals to Conserve Globally Threatened Seabirds in a Changing Climate

Invasive mammals are an ongoing threat at many seabird breeding locations, while impacts from climate change can occur over broad time scales. Combining management strategies for invasive mammal and climate change impacts is important for mitigating current threats and maximizing seabird survival into the future. We assessed all 713 islands with threatened seabirds for conservation importance, immediate benefits of three invasive mammal management actions, and risk of climate‐change related flooding. Preventing invasions on the 397 islands without invasive mammals could benefit 72 seabird species. Invasive mammal eradication or localized action on 249 and 67 islands could benefit 71 and 46 seabird species, respectively. The long‐term risk of flooding on the 713 islands was low (69%). Low‐risk islands were concentrated where eradications or localized action were highlighted (75% and 100% of islands, respectively). These results inform management feasibility assessments and highlight rare opportunities to make significant contributions to seabird conservation.


Introduction
About 30% of seabird species are threatened with extinction, primarily due to invasive species (Towns et al. 2011;Croxall et al. 2012). Most threatened seabirds breed on islands, with greater than 90% co-occurring with a threatening invasive species on at least one island (Spatz et al. 2014). Some of these islands have low elevations, and are the most vulnerable areas globally to climate-related increases in sea levels and storms (Nicholls & Cazenave 2010). Impacts include flooded nests, reduced breeding site availability, and potentially negated benefits from conservation actions (Bellard et al. 2013;Courchamp et al. 2014;Reynolds et al. 2015). While sea levels will increase over time, the threat of invasive mammals to seabirds is immediate, requiring direct attention (Croxall et al. 2012). To address the threat of invasive mammals, eradications and localized control have occurred on islands globally, directly benefiting seabirds through positive demographic and distributional responses (Young et al. 2009;Jones et al. 2016;Luther et al. 2016). By removing a key threat, eradications can buffer seabirds from extinction through increasing seabird populations and conserving genetic diversity, which is a critical component of adaptability when facing global changes as well as for preserving global evolutionary history (Frankham 2005;Chambers et al. 2011;Jetz et al. 2014). Therefore, effective planning requires integration of invasive mammal management against the threat of longer-term exposure to climate change to ensure both immediate and long-term species persistence.
To determine the global scale of conservation opportunities for seabirds, Spatz et al. (2014) identified all islands on which globally threatened seabird species are known to breed and determined if invasive mammals occurred on these islands. Here, we identify invasive mammal management scenarios to conserve threatened seabirds (classified as Critically Endangered, Endangered, and Vulnerable by BirdLife International for the International Union for the Conservation of Nature [IUCN] Red List), and examine the potential exposure of islands to sea level rise impacts. To do this, we: (1) score islands for their conservation importance to threatened and evolutionarily distinct seabirds; (2) identify where exposure to coastal flooding may impact conservation efforts; and (3) place islands into one of three portfolios that identifies the most urgent invasive mammal management needed: prevention (where invasive mammals are absent), eradication (whole-island) removal, or localized action (subisland control or removal) of invasives. For islands in the eradication management portfolio, we created two ranked lists that (1) maximized seabird importance (highlighting the most globally important projects for conserving threatened species), or (2) minimized the operational complexity of an eradication (highlighting projects that could be implemented at faster time scales). Our goal is to prompt conservation planning for threatened seabirds at transnational scales, by identifying near-term conservation actions and recognizing potential longer-term threats from sea level rise.

Methods
Characteristics of islands (i.e., size, location, presence of human populations, and invasive mammals; Supporting Information) with globally threatened breeding seabirds (BirdLife International 2012), hereafter, "seabirds," are from the Threatened Island Biodiversity Database (Threatened Island Biodiversity Database Partners 2015) and described in Spatz et al. (2014).

Island exposure to flooding
The low-lying coastal zone, characterized by the topographic distribution of land adjacent to the sea, represents where physical exposure from climate change, such as flooding and sea level rise, will be most impactful (Woodroffe 2008;Nicholls & Cazenave 2010). This exposure potential on islands is hereafter referred to as "flooding risk" (Supporting Information). We used elevation data from Shuttle Radar Topography Mission with a resolution of 90 m horizontal and 1 m vertical (Rabus et al. 2003). We calculated the land below 5, 10 and 20 m elevation and compared it with total surface area for each island. Islands with >50% surface area below 5 or 10 m, or above 10 m, were classified as high, medium or low flooding risk, respectively. When elevation profiles were not available (150 islands), we found maximum elevations of each island, and classified the islands as high, medium or low flooding risk when maximum elevation was <10, 11-20 or >20 m, respectively. We tested for differences in island characteristics against the flooding risk categories using Kruskal-Wallis (KW) rank sum tests.

Important islands for seabird conservation
We examined all islands with extant seabirds (97 species as 1,078 confirmed, probable, or potential breeding populations; Supporting Information). A seabird species breeding on an island was considered a single population. For each species (s), we calculated a Threatened Seabird (TS) score, by finding the product of its probability of extinction ("extinction risk"; E), relative endemism ("irreplaceability"; I), and relative evolutionary distinctiveness ("distinctiveness"; D) Extinction risk (E) was scored as 0.005, 0.05, and 0.5 (Butchart et al. 2004), for the IUCN Red List categories of Vulnerable, Endangered, and Critically Endangered, respectively. Irreplaceability (I) is a relative measure of endemism (Margules & Pressey 2000), calculated as 1/the number of extant populations for each species. Evolutionary distinctiveness is a measure of contribution to global phylogenetic diversity (Isaac et al. 2007). We calculated relative distinctiveness (D), by dividing each species' evolutionary distinctiveness score (EDGE 2014) by the greatest evolutionarily distinct score of all seabirds in our data set (28.418 ED for Peruvian Diving-petrel [Pelecanoides garnotii]). We ran sensitivity analyses to assess the impact of each score on species priorities (and subsequently island priorities; Supporting Information).

Seabird island importance score
We calculated a Seabird Island Importance (SII) score as a metric of seabird conservation benefit on the 713 islands with extant seabirds and known information on invasive mammals (Supporting Information). Each island received a relative rank based on SII score. A TS score was applied to each seabird species' breeding population, then modified based upon that population's breeding status (P). We summed the modified TS scores for all the species (s) on each island (i) Each confirmed or probable breeding population received a full TS score. A potentially breeding population received 0.75 of its TS score to reduce weighting of populations with uncertain occurrence. We only included islands that supported extant seabird breeding populations but if these islands held other species that are now extirpated, the extirpated population received 0.5 of its TS score in recognition of the potential recovery of an extirpated population. We did not account for a nearby source population for that species.

Invasive mammal management portfolios
Each island was placed into an invasive mammal management portfolio, then examined for globally available island characteristics, including flooding risk. We used Kruskal-Wallis (KW) rank sums to test for differences in SII scores between portfolios.

Prevent invasions
We identified islands where invasive mammals are absent and policy actions (e.g., biosecurity) can proactively prevent mammal invasion (Broome 2007;Simberloff et al. 2013).

Eradication
We identified islands where ࣙ1 invasive mammal species are present and complete removal of an invasive mammal species may be achievable. This included islands where an eradication is underway, or managing repeated incursion responses, is ongoing (Supporting Information). To identify where eradication may be achievable, we examined thresholds of three globally recognized criteria (Supporting Information) that influence eradication operation complexity-invasive mammal type (e.g., cat, rat), island area and human population size (Oppel et al. 2011;Veitch et al. 2011)-referenced from successful mammal eradication efforts to date (DIISE 2015) and expert opinion.
Islands with ࣙ1 invasive mammal considered for eradication were subsequently scored in two ways, resulting in two ranked lists. First, we applied the SII score to each island to understand where eradication would benefit the most important seabirds, highlighting the most important projects at a global scale, regardless of operational complexity. The second score was based on basic island characteristics influencing complexity of eradication operations (Eradication Complexity [EC] score), for which we then ranked islands for minimized complexity, to identify projects that could proceed at faster time scales than others. The EC score was calculated as For each island (i), we calculated the product of the proportion of all invasive mammal species that could be potentially eradicated (ME), a weighted number of invasive mammal species remaining (MR), the number of human inhabitants present (log scale; HI) and island size relative to the smallest island (A; Table S2). We highlighted top-scoring priority islands (the upper 75 th percentile of island scores) in each list.

Localized action
We identified islands with invasive mammals where eradication thresholds were exceeded for all invasive mammal types present. These islands were considered for subisland management, such as local control or fencing, to eliminate or reduce threat at a colony scale (Parkes & Murphy 2003).

Seabird and island scores
Christmas Island Frigatebird (Fregata andrewsi) and New Zealand Storm-petrel (Oceanites maorianus) received the highest TS scores (Table S3; Data Supplement). Christmas (Australia) and Hauturu-o-Toi (New Zealand) islands received the highest SII scores ( Figure 1). Sixteen islands ranked in the upper 97 th percentile of SII scores (Table 1). All scores and sensitivity results are described in the Supporting Information and Table S3.

Figure 1
The worldwide distribution of islands with 97 threatened seabirds, scored by their IUCN Red List category, endemism, and contribution to global evolutionary distinctiveness (SII). Dots are scaled by the quantile distribution of scores whereby the largest dots represent the upper 75 th percentile of scores and the smallest dots represent the lower 25 th percentile of scores.

Figure 2
The percent of islands within each conservation management portfolio (prevention, eradication, localized action), broken up by high, medium, or low coastal flooding risk. "All Islands" represents the cumulative total of seabird islands analyzed (713) and the percent of islands within each flooding risk category.

Conservation management portfolios Prevention
Invasive mammals were absent on 397 (56%) islands that support 623 populations of 72 (74%) seabirds (10 species were restricted to these islands; Data Supplement).
Islands were in 31 countries, primarily New Zealand, the United Kingdom (Overseas Territories), the United States and Chile, and were mostly uninhabited (372 islands, 94%). Seventy-three percent of all higher flooding risk islands were concentrated in this portfolio: 163 islands (41%) were high-or medium-risk. Three species had ࣙ50% of their extant breeding populations on these islands: Ascension Frigatebirds (F. aquila), Cape Gannets (Sula capensis), and Bank Cormorants (Phalacrocorax neglectus).

Eradication
Invasive mammals were present on 249 (35%) islands that fell within our eradication thresholds, supporting 339 populations of 71 (73%) seabird species. Twentynine seabird species had half of their islands in this portfolio and seven were restricted to these islands. Islands were in 34 countries, primarily Italy, Japan, New Zealand, and the United States and were minimally inhabited (  almost exclusively (80%) on these islands (Bermuda Petrel, Pterodroma cahow). All islands scored and ranked for SII and operational EC are in the Data Supplement (see also Figure  3). Only three islands were identified as priority islands in both data sets (Table 3). The SII priorities included 65 islands (SII score ࣙ0.001) in 16 countries and supported 128 populations of 51 seabird species; 8 species were restricted to these islands. Islands were large and variable in size (mean ± SD = 76.5 ± 150.7 km 2 ), 54% uninhabited and 34% with 1-1,000 people, and had low flooding risk (77%). Based on eradication thresholds used, all invasive mammals could be potentially eradicated from 51 (80%) islands. Islands contained a median of two invasive mammal species (maximum = 12).
The EC priorities included 63 islands (EC score ࣙ0.005) in 21 countries and supported 67 populations of 22 740 Table 2 The 42 threatened seabird species on islands with high or medium risk of exposure to coastal flooding, including 3 Critically Endangered species and 7 species found only on these islands. seabird species. Islands were small (mean ± SD = 0.1 ± 0.1 km 2 ), uninhabited (100%), and 43% were at low flooding risk (33% of islands at high risk). Based on our eradication thresholds, all invasive mammals could be potentially eradicated from these islands. Islands contained a median of one invasive mammal species (maximum = 4).

Localized action
Invasive mammals were on 67 (9%) islands that did not meet eradication threshold criteria (Data Supplement). SII scores were highest in this portfolio (KW = 21.5, df = 2, P < 0.001). These islands supported 116 populations of 46 (47%) seabird species; 11 species from 12 islands  Information for more details). c An invasive species was considered as "present" if it was confirmed or suspected on an island (including if an eradication was underway or incursion responses were ongoing, see Supplemental Information for more details).

Figure 3
The two ranked lists of islands for invasive mammal eradication on islands with globally threatened seabirds. Dots are scaled by the quantile distribution of scores based on the (A) SII score that maximizes the conservation benefit of protecting seabird species that are most threatened and evolutionarily unique and the (B) EC score that minimizes the operational complexity of an eradication.

Discussion
We focused on immediate and long-term threats to seabirds on islands: invasive mammals, a driver of current seabird extinctions that can often be mitigated with demonstrated benefits for seabirds, and the exposure of islands to coastal impacts from climate change, a threat that can affect seabird conservation over longer time frames (Courchamp et al. 2003;Croxall et al. 2012;Jones et al. 2016). We investigated seabirds and islands at the global level, reflecting the multinational distributions of many seabirds, and a necessary scale for coordinated and effective conservation planning. Our results highlight important islands and species globally for conservation actions from which more detailed island or species-specific feasibility assessments can be built.

Important islands for seabird conservation
We highlighted the conservation importance of islands with the most threatened seabirds and with the highest potential for genetic diversity loss. Invasive mammals can have profound effects on both species survival and the overall maintenance of genetic diversity, which is critical for adaptability in the face of global changes (Frankham 2005;Gasc et al. 2010) as well as for preserving global evolutionary history (Jetz et al. 2014). Our results often highlighted Critically Endangered species with high evolutionary distinctiveness scores, such as the Christmas Island Frigatebird and New Zealand Storm-petrel (Critically Endangered, single-island endemic, and evolutionarily distinct species) and their islands, Christmas and Hauturu-o-Toi (previously, Little Barrier Island). In some cases, less threatened but highly endemic and evolutionarily distinct species, like the Endangered Abbott's Booby and Vulnerable Ascension Frigatebird, received high scores and their islands were also highlighted for their importance to global seabird conservation (Table S1). Prioritizing conservation of these threatened and evolutionarily distinct species can prevent imminent extinctions while maintaining important genetic diversity (Isaac et al. 2007;Jetz et al. 2014).

Island exposure to sea level rise
A fine-scale understanding of the risk to islands from climate change depends on many biophysical factors (Nicholls & Cazenave 2010;Slangen et al. 2014). However, these data are lacking or inconsistent for most of the world's islands. Our approach, using elevation as a proxy for risk of flooding from sea level rise (Woodroffe 2008), was a coarse first step toward inferring physical coastal impacts from climate change, and enabled the integration of potential long-term impacts from flooding alongside current immediate impacts to seabirds from invasive mammals. We found that most threatened seabirds breed on islands with sufficient elevation to be at low risk of exposure to impacts from sea level rise. This is encouraging for seabird conservation, suggesting that tackling invasive mammal threats can be a long-term solution for many threatened seabird populations. However, we did not account for where seabird colonies are located on islands, which could result in over-or underestimates of exposure. Some seabirds nest on high-elevation cliffs or in mountain forests, making them less sensitive to coastal impacts, while others nest along coastlines and can be sensitive to flooding (Schreiber & Burger 2002). To address these data gaps, next steps should include speciesspecific climate vulnerability assessments at the island or colony level, including consideration of species' sensitivity and adaptability to change (e.g., Foden et al. 2013). Twenty-five percent of threatened seabird populations in our study were on low elevation islands (223 islands) and may be exposed to sea level rise impacts (Table 3). Most (73%) of these islands were free of invasive mammals and the remaining were included in our eradication portfolio, typically ranked highly for minimized eradiation complexity. These islands may represent only interim refuges for threatened seabirds, yet are important short-term opportunities where managers can offset threats from invasive mammals and reduce extinction risk.

Prevention
Through a combination of biogeographic good fortune, successful invasive mammal eradication efforts on 82 islands (DIISE 2015), and ongoing biosecurity programs (e.g., Hauturu-o-Toi Island [New Zealand Department of Conservation 2014]), 55% of islands supporting extant threatened seabird populations were free of invasive mammals. These islands have ࣙ1 breeding population of 74% of threatened seabird species, including 10 species breeding exclusively on these islands. Biosecurity is a proactive approach to protecting islands from invasions (Broome 2007). If not already in place, risk assessments of both anthropogenic and natural reinvasion risk are critically important for conserving these islands, their breeding seabirds, and endemic species (Harris et al. 2012).

Eradication
Eradications on 249 islands (79% of islands with threatened seabirds and invasive mammals) would conserve populations of 73% of threatened seabird species. Our two ranked sets of islands, highlighting either globally important projects to conserve seabirds or islands with assumed lower eradication complexity, were inversely related. Many of the most important islands for conserving threatened seabirds tend to be larger and have more invasive mammal species and human inhabitants. When eradication complexity was minimized, higher-ranked islands were small islets offshore larger islands or mainland areas, uninhabited, and often low elevation with only one invasive mammal species.
While conservation efforts on the islands in the two ranked lists will likely require different approaches (i.e., due to increased operational complexity on larger islands or risk of reinvasion on small offshore islands), significant conservation opportunities exist. For example, three islands ranked highly in both sets conserve some of the world's most threatened seabirds (e.g., Critically Endangered Chinese Crested Tern [Sterna bernsteini]) and represent relatively straight forward eradication opportunities. As another example, four islands (Gough, Amsterdam, Plata, and Clarion) collectively support nine globally threatened, evolutionarily distinct seabird species. While these islands are relatively large, they are uninhabited by people and at low flooding risk, highlighting that benefits gained from eradicating invasive mammals on these islands may be sustained over long time scales.
Island area and human population size are two of the most useful criterion available globally for identifying potential eradication opportunities to conserve seabirds 744 at both global and regional scales (e.g., Brooke et al. 2007;Capizzi et al. 2010;Harris et al. 2012;Dawson et al. 2014). However, we recognize that these metrics are not the sole determinants of eradication feasibility, and that not all islands identified here will be deemed feasible with further assessment. Next steps include islandspecific and archipelago-wide conservation assessments, including evaluations of ecosystem and species-specific threats and benefits, social, political and economic eradication feasibility, reinvasion risks, stakeholder valuation, and design of eradication plans (Zavaleta et al. 2001;Courchamp et al. 2003;Murdoch et al. 2007).

Localized action
We found that 8% of islands with invasive mammals exceeded our conservative thresholds for achievable whole-island invasive mammal eradications. Because the distribution of seabirds here is not island-wide, localized action, such as through predator-proof fencing or invasive species control, may be appropriate (Parkes & Murphy 2003;Luther et al. 2016). These subisland invasive mammal management actions are critically important for several of the most threatened and evolutionarily distinct seabird species, including 11 restricted to these islands (e.g., Mascarene Petrel [Pseudobulweria aterrima] on Reunion Island). Yet, while these actions are effective, they often require ongoing resources (Hodges & Nagata 2001), thus long-term adaptive approaches, such as translocations to invasive-free islands, may be important to consider for some seabird species in this portfolio. Finally, several islands in this portfolio are currently undergoing invasive mammal eradication feasibility assessments (DIISE 2015; Supplemental Information), highlighting that advances in eradication techniques will open up new opportunities for seabird conservation in the future.

Conclusions
We identified 713 islands globally where invasive mammal prevention, eradication, or localized action can potentially conserve populations of Critically Endangered, Endangered, or Vulnerable seabirds. Our analysis also identified islands where coastal impacts from sea level rise, such as flooding, could impact the long-term effectiveness of these conservation actions. While some species are at risk of losing nests or habitat from flooding, sea level rise does not appear to be a significant threat for 75% of globally threatened seabirds. With appropriate climate adaptation plans in place alongside invasive mammal management, the long term conservation of seabirds represents a rare opportunity to make a significant contribution to global biodiversity conservation.

Supporting Information
Additional Supporting Information may be found in the online version of this article at the publisher's web site: Table S1. Area and human population size thresholds for invasive mammal eradications used in our analyses Table S2. Variables and metrics used in the EC score. Table S3. The 97 globally threatened seabird species ordered by their TS rank and the sensitivity of this rank to probabilities of extinction, relative endemism, and evolutionary distinctiveness Table S4. Sensitivity results: correlations between extinction risk probabilities against irreplaceability scores Data Supplement: SII scores for all 713 islands and results from the invasive mammal eradication ranking.