Knowledge of the distribution and abundance of most avian species nesting in the Canadian High Arctic remains patchy and incomplete in part because of the extreme remoteness and large size of the region as well as the logistical difficulties inherent in conducting surveys there. Some catalogues of large seabird colonies, particularly cliff-nesting species, in the Canadian Arctic are available (Nettleship 1973; Mallory & Fontaine 2004), but species nesting in low densities or only in isolated areas of suitable habitat have probably been overlooked (Gaston et al. 2012). Current estimates of ground-nesting seabird populations in the Canadian Arctic are particularly uncertain due to partial or incomplete survey coverage, extrapolation from small sample sizes, and a general lack of understanding of how variability in local habitat suitability may affect annual numbers of breeders (Gaston et al. 2012). In the Arctic, mammalian predators exert strong predation pressure on ground-nesting birds (Birkhead & Nettleship 1995; Smith et al. 2010), and most species have adopted a strategy of nesting in low densities over large areas. This tendency makes most populations difficult to survey and monitor effectively, a problem further complicated by unpredictable changes in local abundance (Clark & Shutler 1999; Egevang & Frederiksen 2011) and the reduced detectability of cryptic or easily overlooked species (Meltofte 2001).

The ocean passage that extends from Penny Strait through Queens Channel to MacDougall Sound (hereafter collectively referred to as Queens Channel) contains an extensive archipelago of small, low-elevation gravel islands. The shallow bathymetry and strong tidal currents in this area lead to the formation of several localized and highly productive polynyas (Hannah et al. 2009), which provide important foraging habitat for many bird and mammal species (Stirling 1997). This region is unique within the Canadian High Arctic and is used by a large number of bird species which exploit the rich foraging opportunities provided by polynyas early in the breeding season when most of the surrounding water remains frozen (Mallory & Gilchrist 2003; Mallory & Fontaine 2004; Maftei et al. 2012). This area had already been recognized as an important breeding area for Ross's gull (Rhodostethia rosea) after the first nests in North America were found on the Cheyne Islands in north-western Queens Channel (MacDonald 1978), and more recent surveys have identified other islands in the area used by this species for nesting (Maftei et al. 2012). A partial survey of Queens Channel in 2002 and 2003 (Mallory & Gilchrist 2003) revealed that substantial numbers of Arctic terns (Sterna paradisaea) and common eiders (Somateria mollissima) also nest throughout the islands in this archipelago. The northernmost known North American colonies of Sabine's gull (Xema sabini) and the largest known ivory gull (Pagophila eburnea) colony are also found in the Queens Channel region. A more thorough understanding of the size, distribution and dynamics of ground-nesting seabird populations in this area is of critical importance in developing practical management and conservation measures for the region at a time of increasing interest in industrial development of the Arctic.

The objectives of this study were to (1) update and augment the Mallory & Gilchrist (2003) survey to examine potential breeding sites for marine birds in Queens Channel, and (2) assess the numbers and distribution of ground-nesting marine birds in this poorly known region.

Methods

Study area

We defined a study area extending between 75° N and 76°55′N and 102°40′W and 93°47′W, encompassing an archipelago of approximately 120 islands which run from southern Penny Strait through Queens Channel and MacDougall Sound into northern Barrow Strait. This area is bound by Bathurst Island to the west, Devon Island to the north-east and Cornwallis Island to the south-east (Fig. 1). After excluding islands that were very large (i.e., capable of supporting resident mammalian predators), those that were effectively extensions of much larger islands (i.e., would be joined by ice bridges for much of the year), and those that were covered at highest tides, we had 80 islands to choose from. We used ArcGIS software (version 10, Environmental Systems Research Institute, Redlands, CA) to compile topographic, land cover and digital elevation model maps of the study area, and from these data we selected our target islands to survey. These comprised sites that Mallory & Gilchrist (2003) had identified as important as well as sites that had similar physiographic features to those in the earlier study. We determined mean island elevation (m above sea level [asl]) and island area (ha) for these sites. Although elevation and area were correlated (r_s31=0.89, p<0.001), we assessed these criteria separately. The region has numerous small polynyas in most years (Mallory & Gilchrist 2003), which are variable in annual size and shape (and which vary through the season), but islands in this study were generally 35±5 km (range 0–110 km) from typical polynya locations (polynyas from Hannah et al. 2009).

Surveys

We surveyed 30 islands in Queens Channel from a Bell 206 L4 helicopter on 19 and 23 June 2012 (Fig. 1). Seventeen of these islands had never been surveyed before, 10 had been previously surveyed on 16 July 2002 by Mallory & Gilchrist (2003), and one, Nasaruvaalik Island, has been surveyed and monitored with varying intensity of effort annually since 2002. While flying around the perimeter of small islands or transects over larger ones from an altitude of approximately 100 m, three observers independently recorded all birds seen on or in the immediate vicinity of each island, and the overall high count for each species was recorded. We also surveyed six islands on foot while searching for Ross's gull nests on the same dates as above. Observations from Nasaruvaalik Island were recorded on 25 June 2012 by three ground-based observers.

Assessment of regional populations

We assessed fluctuations in numbers of birds attending island colonies at the regional level by comparing the difference in observed numbers of birds at 10 islands surveyed in both 2002 and 2012 (Table 1). We also examined fluctuations in annual attendance at the colony level by comparing the variation in annual maximum and mean numbers for 25 regularly observed species over six years at Nasaruvaalik Island (Table 2). We calculated mean values based only on July observations to more accurately reflect the numbers of breeding birds which would be present and incubating at that time. The yearly maximum is a more accurate indication of the total number of birds (including non-breeders) using the island. For each year, we determined the total annual abundance representing the sum of the maximum high count for each of the 25 species as well as a total mean representing the mean of daily counts in July for each species (n=31 for all years). We then compared these values between years and across all six years. We also compared the per-year variation in annual maximum and mean observed numbers for Arctic terns and common eiders separately. Comparisons were conducted using Pearson r or Spearman r_s. All means are presented±SE.