Living on the edge of a shrinking habitat: the ivory gull, Pagophila eburnea, an endangered sea-ice specialist

1. Introduction

In recent decades, Arctic sea ice has decreased dramatically in extent and thickness, with important ecological consequences on Arctic biota, especially for marine mammals and seabirds that use the highly productive ice-edge through their life cycles [1–3]. Besides being one of the very few seabird species specialized on sea-ice habitats, the ivory gull (Pagophila eburnea) is a rare (8000–11 500 breeding pairs worldwide) and endangered (red listed in all countries where it breeds) Arctic endemic [4].

Although a few of its colonies have recently been found on sea ice [5], the ivory gull usually breeds on land but uses sea ice to feed and rest during most of the year and over most of its distribution range [6,7]. Little is known about its offshore ecology but the species is known both to feed on small fish and invertebrate prey found by hovering a few metres over sea ice and open leads, and to scavenge on larger food items like whale or seal carcasses (e.g. remains of polar bear kills [8]).

Improving our understanding of the links between ivory gull and sea ice has been identified as a research prerequisite before a circumpolar conservation strategy can be implemented [4]. Empirical observations suggest that the ivory gull extensively uses the ice-edge, but the strength of this association and its spatio-temporal variability have never been properly assessed.

Between 2007 and 2013, the movements of 104 ivory gulls belonging to the four existing breeding origins (Greenland, Svalbard, Arctic Canada and Arctic Russia) were monitored with Argos-compatible bird tracking PTTs (platform terminal transmitters) [6,7]. For each of the gull positions, ice concentration and distance to the nearest ice-edge were estimated using satellite microwave radiometers [9].

Based on this unique dataset, this study addressed the following questions: (i) Are ivory gulls really associated with the ice-edge ecotone?; and Which parameters best explain differences (ii) in their use of sea-ice concentrations in time and space or (iii) according to their geographical origin?

2. Material and methods

(a) Ivory gull positions

The 104 ivory gulls were monitored using different types of PTTs and attachments [6,7] and produced 268 772 positions between 2007 and 2013 (electronic supplementary material, table S1).

We used PTT data from the Argos positioning system, restricting data to those points in location classes LC3, LC2 and LC1 with accuracies of less than 250 to less than 1500 m [10]; we also included LCA, which has an accuracy similar to LC1 [6,11]. This resulted in 123 787 positions although, for analyses involving ice concentration data, we only used the 78 254 offshore positions that were farther than 20 km from the nearest coast (figure 1; electronic supplementary material, table S1). Indeed, terrestrial positions are not relevant for this study and ice concentration estimated close to land (i.e. 7.5% of all LC321A, of which 73% were produced in June–August close to the colonies and during commuting flights between colonies and feeding grounds) are not reliable owing to mixed coastal pixels [9].

• Download figure
• Open in new tab
• Download PowerPoint

3. Results and discussion

Most ivory gull positions collected during this study were located in highly concentrated sea ice (79.3% have IC3x3 > 30%) and 75% were within 100 km of the ice-edge (62% over ice and 13% over water; figure 1b). Interestingly, ivory gulls use different ice concentrations during the year, with more than 80% of the positions found in high ice concentrations (IC3x3 > 30%) between May and September compared with fewer than 50% during most of the migrating and wintering seasons (figure 1c). These differences cannot be explained only by differences in availability of sea ice in the Arctic (figure 1d). Indeed, if we compare the frequency distribution of the ivory gull's most used sea-ice concentrations (i.e. IC3x3 between 40 and 90%; figure 1b) with its availability at larger scales, it appears that the birds are primarily using areas where ice concentration is higher than what they would randomly find at these regional and circumpolar scales (figure 1d). The typical feeding habitat of the species, in small leads that freeze over in winter, is more likely to explain these seasonal changes.

The best retained models to explain d2edge (table 1) included IC3x3, month in interaction with country (figure 2a), d2col and sex; with IC3x3 having a positive effect on d2edge (β ± s.e. = 7.36 × 10⁻¹ ± 7. 74 × 10⁻³), d2col having a low negative relationship with d2edge (β = −5.83 × 10⁻³ ± 6. 051 × 10⁻⁴) and sex having no effect (contrast analyses between female and male = 16.68 ± 13.96). The main driver of d2edge was IC3x3, but this link can vary spatially and temporally.

• Download figure
• Open in new tab
• Download PowerPoint

Table 1.Model selection of mixed-effects models explaining distance to ice-edge in ivory gulls. and represent marginal and conditional pseudo-R²-values, respectively (see the electronic supplementary material). Only the best models are shown.

Collapse

models	d.f.	log-likelihood	AICc	ΔAICc	weight
country × month + IC3x3 + d2col	53	−27 162.22	54 430.5	0.00	0.71	0.42	0.71
country × month + IC3x3 + d2col + sex	54	−27 162.11	54 432.3	1.79	0.29	0.42	0.71
country × month + IC3x3	52	−27 171.15	54 446.4	15.85	0.00	0.42	0.71
country × month + IC3x3 + sex	53	−27 171.04	54 448.1	17.64	0.00	0.42	0.71
month + IC3x3 + d2col	17	−28 523.16	57 080.3	2649.82	0.00	0.42	0.69
month + IC3x3 + d2col + sex	18	−28 523.16	57 082.3	2651.81	0.00	0.42	0.69
country + month + IC3x3	20	−28 522.53	57 085.1	2654.57	0.00	0.43	0.69
month + IC3x3	16	−28 526.91	57 085.8	2655.32	0.00	0.42	0.69

Except during active post-breeding migration between October and December, when they cross open water along their flyways in the Davis Strait and Labrador Sea [6,7], Canadian and Greenland birds were found closer to highly concentrated sea ice than to open water (figure 2a; electronic supplementary material, table S3). For these birds, d2water and d2edge increased markedly in summer, when they got closer to their remote breeding sites, far from open water and ice-edge. For Svalbard birds, the situation differed on at least two points: (i) because birds migrate later and partly winter off SE Greenland [6], there was no peak in d2ice in November; (ii) in July, during breeding, d2ice increased while d2water decreased (figure 2a). This latter observation is a clear signature of the shift in feeding grounds recently evidenced in Svalbard [14]: when the ice-edge was too far and had to be reached by flying over open water, some ivory gulls started using nearby marine glacier fronts more frequently.

As d2edge was positively correlated with ice concentrations (figure 1b), seasonal changes were also obvious for the ice concentration used by ivory gulls (figure 2b). Furthermore and regardless of their origin, from May to September (when close to breeding grounds) ivory gulls tended to be found (IC3x3) in areas of higher ice concentrations than what was found at a larger (IC50) scale, while in March the trend was opposite (electronic supplementary material, table S4). This changing pattern is easily explained for the summer period because most positions are produced on the inner, icy side of the ice-edge (hence IC50 < IC3x3 because it includes pixels of open water as far as 50 km). In March, which is often the coldest month in the Arctic, we can assume that ivory gulls have no other choice than to feed on the outer, open side of the ice-edge when all leads are frozen (in this case IC3x3 includes more open water than IC50).

Although very few other published studies used Argos tracking data from more than 100 birds [15], our sample was still too small for Russian breeding birds and for the winter period (solar-powered PTTs produce few positions at high latitudes from October to February; electronic supplementary material, table S5).

Another reason to continue to monitor movements and habitat use of ivory gulls in the long-term is to document species’ responses to the ongoing decline of Arctic sea ice and, eventually, to its total disappearance during summer. Our study clearly showed the strong affinity of ivory gull to sea ice and especially to the productive ice-edge where they feed [1]. It has already been shown that sea-ice decline can strongly affect the distribution, behaviour and fitness of marine mammals [1–3,16,17]. Unlike marine mammals, seabirds like the ivory gull cannot fast for long; thus, we assume that, unless they can rapidly and significantly change their foraging ecology, sea-ice decline could impact pagophilous seabirds like the ivory gull dramatically in the future.

While sea-ice decline is a serious threat to the ivory gull population, contaminants [18] and the important overlap between the ivory gull's distribution range (figure 1a) and offshore oil exploration activities [19] are other major concerns. Ivory gulls consume ice-associated prey and scavenge marine mammal carcasses. Owing to the latter, they have egg concentrations of far-transported industrial contaminants, especially mercury (Hg) and some organochlorines (DDE), that exceed published thresholds known to disrupt the reproductive success of avian species [20]. In fact, the ivory gull have the highest egg Hg concentrations of any Arctic bird, and concentrations of methyl mercury (MeHg) found in feathers have increased by a factor of 45 during the industrialized period from 1877 to 2007 [21]. Oil exploration is potentially adding to these pressures. In the Greenland, Barents and Kara seas, important breeding and summering areas for ivory gulls [6,22], several oil consortiums were granted exploration and exploitation licences in 2013–2014, while in the Labrador Sea, a key winter habitat for ivory gulls [6,7], oil exploration has already started. Thus, further monitoring of the ivory gull population is critical to assess their responses to these and other new stressors (e.g. shipping routes, mining, infrastructure) and to develop adapted conservation strategies.

Ethics

All research was conducted following relevant regulations. For Canada: Canadian Wildlife Service Banding Permit 10694, Canadian Wildlife Service Scientific Permit NUN-SCI-09-02 and Nunavut Wildlife Research License WL2010–032. For Norway: Governor of Svalbard and Norwegian Animal Research Authority permit ID 3260. For Greenland: Danish Polar Center permit C-07-207 and Danish Agency for Science, Technology and Innovation permits C08-211 and C09-116.

Data accessibility

Data can be found in the electronic supplementary material: rs.figshare.com.

Authors' contributions

O.G., G.H. and G.Y. conceived the idea. O.G., H.S., M.V.G., M.L.M, G.G, A.E., B.S. and G.Y. captured and tagged the birds. L.I., G.H. and M.H. extracted the sea-ice data. O.G. and G.Y. performed analyses and prepared the first draft. A.M. contributed to the Discussion. All authors gave final approval for publication and agreed to be accountable for all aspects of the content herein.

Competing interests

We have no competing interests.

Funding

This work was supported by, in Canada: Kenneth Molson Foundation (58-0-205568), Environment Canada (K4E21-12-0850), Natural Sciences and Engineering Research Council of Canada (41-0-205561); in Greenland: National Geographic Society, Prix Gore-Tex initiative, Fondation Avenir Finance, Arctic Ocean Diversity Census of Marine Life Project, CNES, CLS, Sagax expeditions, Magasins Intermarché, Société Henry Maire, Lestra, MSR, Vitagermine, Moulin des Moines, GREA, F. Mariaux, R. Cuvillier, E. Bourgois, F. Paulsen; in Norway and Russia: Norwegian Ministry of Climate and Environment, Norwegian Polar Institute (Centre for Ice, Climate and Ecosystems; ICE), Statoil, Arctic and Antarctic Research Institute. Fieldwork conducted in 2007 and 2008 was part of the Joint Norwegian–Russian Commission on Environmental Protection and Russian national programme of the IPY 2007/08.