The Sabine's Gull (Xema sabini) is a pelagic, Arctic-breeding species with a circumpolar breeding distribution. Little is known about migration routes for northern Alaska-breeding Sabine's Gulls. We tagged Sabine's Gulls on their northern Alaska breeding grounds to identify migration routes and wintering areas and compare geolocators and GPS pinpoint tags for use on small-bodied gulls. Twelve geolocators were deployed in northern Alaska in 2011 (Colville River delta) of which four were recovered, and five GPS pinpoint tags in 2021 (Qupaluk). Although the GPS pinpoint tags provided more accurate locations allowing for finer-scale habitat evaluation, and did not require recapture of birds, the overall coverage provided by geolocators was superior given constraints on the number of locations GPS pinpoint tags record. Broadly, Sabine's Gulls migrated away from the breeding grounds as expected, passing along the west coast of Alaska and south along the west coast of the Americas to winter in the Humboldt Current off the coast of Peru. Our tracked gulls used the same migratory staging and wintering areas as did Sabine's Gulls breeding in the Canadian Arctic (Davis et al. 2016). Such reliance on specific marine areas presents risks from climate-related changes or ecological damage to those areas.
The Sabine's Gull (Xema sabini) occurs throughout the circumpolar Arctic during the boreal summer, with some of the highest concentrations breeding in western and northern Alaska (Stenhouse et al. 2001; Stenhouse and Robertson 2005; Davis et al. 2016; Day et al. 2020). For most of its annual cycle, the species is highly pelagic and undertakes the longest annual migration (30–40,000 km) of any gull (Stenhouse et al. 2012), migrating from Arctic breeding sites to winter in the highly productive waters of the Humboldt Current (Pacific) or Benguela Current (Atlantic; Stenhouse et al. 2012; Day et al. 2020; Gutowsky et al. 2021). Canadian breeders from the same colony may winter in the Atlantic or Pacific (Davis et al. 2016). Individuals breeding in Arctic Alaska are believed to migrate along the west coast of the Americas to wintering areas within the Humboldt Current (Chapman, 1969; Gutowsky et al. 2021). Within this wintering area, Sabine's Gulls are commonly sighted from late November until early April around Lobos de Tierra (06°20′S) to Ilo (17°S) on the north coast of Peru, and between Pisco (14°S) and Camana (17°S) in southern Peru (Javier Antonio Quiñones Dávila, pers comm; eBird data). Sabine's Gulls breeding in Greenland and migrating through the Atlantic migrate down off the coast of Europe and Africa to winter in the Benguela Upwelling (Stenhouse et al. 2012).
The development of geolocators and more recently, GPS pinpoint tags, have allowed researchers to track smaller species such as Sabine's Gulls (158–214 g on average; Day et al. 2020), throughout the year. Geolocators are archival light-recording devices, weighing as little as 0.5 g, that register light levels in relation to an internal clock, from which latitude and longitude are estimated after it is retrieved. Their accuracy varies by bird behavior, geographic location, temporal proximity to the equinoxes, habitat, and weather but generally are accurate to 100–300 km of latitude and longitude, which is sufficient to investigate patterns of connectivity, migration timing, and large-scale habitat use (McKinnon et al. 2013). Argos Pinpoint tags relay GPS locations via Argos satellites, without the need to recapture birds to obtain location data. Tags are pre-programmed to record a finite number of GPS locations at specific times and dates. GPS locations are accurate to several meters.
Although fine-scale migratory and winter movements of Sabine's Gulls breeding in the Canadian Arctic and Atlantic have been described (Furness and Furness 1982; Stenhouse et al. 2012; Davis et al. 2016; Gutowsky et al. 2021), there has been little investigation of Alaska-breeding Sabine's Gulls. The goals of this study are to 1) describe the migration routes and wintering areas for Alaskan-breeding Sabine's Gulls, and 2) compare geolocators and GPS pinpoint tags for use on small-bodied gulls. By doing so we can compare the migratory patterns and important stopover areas of Alaskan and Canadian-breeding birds, as well as assess the conservation implications in light of climate change and related oceanic ecosystem change.
Methods
We deployed tags at two sites in northern Alaska; the Colville River Delta (Anachlik Island; 70°23′N, 151°29′W), and northeast of Teshekpuk Lake (Qupaluk; 70°40′N, 152°43′W). In the summer of 2011, we captured 12 adult Sabine's Gulls on Anachlik Island (the site of a homestead), using either a whoosh net or cage trap placed at a feeding station, and tagged them with Mk12 (British Antarctic Survey) light-level geolocators (<1 g). Before spring snow-melt the feeding station was a small platform baited with chopped whitefish and was kept supplied 24 hours a day. After break-up, the feeding station was by the edge of a lake and supplied twice a day with commercial duck food. An additional 13 Sabine's Gulls were banded with U.S. Geological Study (USGS) metal bands only. Geolocators were fixed to Darvic plastic leg flags (following Conklin and Battley 2010) and placed on the tibiotarsus above a small ‘spacer’ band to keep the geolocator positioned above the ankle and prevent any contact between the joint and the geolocator (<0.01% of total body weight). Tags recorded ambient light levels every 5 minutes. Tagged birds were recaptured in 2012 or 2013 for removal of their geolocator at the same feeding station.
In the summer of 2021, we captured five adult Sabine's Gulls on their nests with a bownet at Qupaluk (remote area with no permanent human presence) and equipped them with 4 g Lotek Argos PinPoint 75 tags using leg-loop harnesses with 1.9 mm Teflon ribbon, and copper crimps (<0.03% of total body weight). Tags were programmed to take a GPS point every two days from 15 August to 15 November, and then one point every five days until the batteries died. This tracking schedule was based on the likely phenology of fall migration. Individuals were also marked with a USGS metal leg band.
Geolocator migration tracks were generated from ambient light level readings using the TwGeos (v0.1.2; Lisovski et al. 2016) and FLightR (v0.5.2; Rakhimberdiev and Saveliev 2022) packages in program R (v4.2.1; R Core Team 2022). First, we used the ‘findTwilights’ function to identify the time of each sunrise and sunset (i.e., twilights; Lisovski et al. 2020), and used the ‘twilightEdit’ function to identify and discard incorrect twilight assignments due to periodic shading of the light sensor (e.g., if a gull roosted with the geolocator tucked among its body feathers). A twilight was considered incorrect and discarded if it was >45 minutes different than the corresponding twilights that occurred in the surrounding 4 days (2 days before and 2 days after; Lisovski et al. 2020). Second, because 24-hour sunlight precluded calibrating geolocators on the breeding grounds (i.e., the deployment site), we used the ‘find.stationary.location’ function to identify calibration locations on the wintering grounds (Rakhimberdiev et al. 2017). Identified calibration locations were within known wintering areas (Davis et al. 2016; Day et al. 2020).
Twilight periods and calibration parameters were then used in the state-space hidden Markov model in FlightR to generate twice-daily location estimates (Rakhimberdiev et al. 2015, Rakhimberdiev et al. 2017). Because Sabine's Gulls typically do not rest or forage on land while away from their breeding grounds (Stenhouse 2012, Davis et al. 2016), but may feed on beach-cast invertebrates (see Day et al. 2020), we parameterized the FLightR movement model so that location estimates over land or over ocean were equally likely if ambient light levels indicated an individual was in a migratory state but were weighted toward ocean if light levels indicated an individual was in a sedentary state. Parameter values for the movement model and the spatial behavioral mask were the same for all individuals. We discarded location estimates ≥65° N because solar geolocation performs poorly at high latitudes (Ekstrom 2004). Finally, we estimated individuals' wintering ranges, south migration staging areas, and north migration staging areas using the kernelUD function (smoothing parameter: reference bandwidth) in the R package adehabitatHR (v0.4.19; Calenge 2006). Kernel density estimates were averaged across individuals. Positive values occurring on land were set to 0 and remaining values were corrected to sum to 1.
The GPS pinpoint tags did not require the level of manipulation of the geolocator data. GPS location data uploaded to the Argos satellites from the tags was retrieved and downloaded via the Argos CLS website. The tags also generated PTT locations via the Argos satellites and depending on location class, these points were also used. Points generated with unsuccessful satellite fix were excluded. We only used PTT locations derived from ≥4 satellite transmissions (i.e., location classes 0, 1, 2, and 3; Douglas et al. 2012). Values are reported as means ± standard error.
Results
Ten of the 13 banded, but non-geolocator, birds returned to Anachlik (77% minimum return rate). Three geolocators were recovered in summer 2012 and one in 2013 at Anachlik Island for a total of four (33% minimum recovery rate).
Four of the five GPS pinpoint tags deployed at the Qupaluk field site transmitted GPS and PTT data for the duration of the fall migration (Table 1). An average of 34 successful satellite-quality points were collected from each functioning tag. We only used points with location classes of 0, 1, 2, or 3; excluding points with location classes of A, B, and Z (recorded points were at least 6-days apart limiting our confidence to verify these more uncertain location classes). Consequently, we only used 14 of the overall points – seven points in southward migration and seven points in the wintering area (Fig. 1).
Sabine's Gulls with geolocators departed the Colville Delta by late August in 2011. Individuals appeared to migrate west from the breeding grounds to the Chukchi Sea and south along the Bering Sea coast, crossing the Alaska Peninsula between 15–24 August (Fig. 1; Table 2). They then spent about a month in the waters off the coast of British Columbia, Washington and Oregon (north of 35° N), departing mid-September (range: 9–21 September; Table 2). They spent another 8–12 days north of 20° N, arriving at their wintering grounds in mid-October (range: 27 September–18 October; Table 2). They remained in the wintering area (<20° N & <90° W) until the end of March and returned to Alaska (north of 50° N) by the end of April (Table 2). Sabine's Gulls spent less than two weeks on migration between 20°N and 35°N, and only a week north of 35° N while migrating to the breeding grounds (Table 2; Fig. 2). On average Sabine's Gulls moved 15,888 km (SE = 817) in total distance between the breeding site and the winter area, and 14,749 km (SE = 1231) while northbound (including the straight-line distance between the first southbound and last northbound location as locations were not estimable north of 65° N).
Table 1.
Summary of geolocator and GPS pinpoint tag deployments on Sabine's Gulls at Anachlik Island and Qupaluk, Alaska.
GPS pinpoint-tagged birds departed Alaska prior to August 15th (2021) when the first satellite fix was acquired, and initial migration route and timing of the start of fall migration is unknown. On August 15, all four individuals were spread out along the Pacific coast of North America from Vancouver Island to northern California (Fig. 1). GPS data suggested one individual spent multiple days within the Juan de Fuca Eddy during the southbound migration. Two birds spent time inland from the coast in the western United States and Mexico (Fig. 1). By mid-September, three of the four individuals had moved south to the coast of Central America, with the remaining individual lingering near Monterrey Bay, California. Two of the more southerly individuals spent multiple days off the southeastern tip of Baja California Sur within the Gulf of California before continuing southward. By mid-October, all four individuals had settled in the Humboldt Current (near Chiclayo). We did not receive reliable GPS points after December 15th, 2021 (Table 1).
Discussion
The migratory behavior of our Alaska-breeding Sabine's Gulls was consistent with birds tracked from breeding sites in the Canadian Arctic to wintering areas in the Pacific (Davis et al. 2016; Gutowsky et al. 2021) and between Greenland and the south Atlantic (Stenhouse et al. 2012). All the tracked Alaskan Sabine's Gulls migrated south along the Pacific coast of North America. Birds lingered in several staging areas (i.e., sites with abundant, predictable food sources during migration; Warnock, 2010), taking an average of 35 days to move between Alaska (50° N) and Mexico (20° N; Table 2). These included the exceptionally productive Juan de Fuca Eddy (MacFadyen et al. 2008), and California Current System off the coast of Washington and California (Fig. 1). It appears that the majority of the Pacific wintering population stage in the Juan de Fuca Eddy and the northern end of the California Current System for a significant amount of time (this study, Davis et al. 2016; Gutowsky et al. 2021). Similarly, Sabine's Gulls migrating through the Atlantic used the highly productive Bay of Biscay on the southbound migration and the Canary Current on the northbound migration (Stenhouse et al. 2012).
Table 2.
Timing (mean and interquartile range) for arrival, departure, and duration of Sabine's Gulls to their migratory and wintering areas. Gulls were tagged with geolocators (n = 4) at Anachlik, Alaska, 2011, one geolocator recorded two years of data.
Birds wintered, as expected, in the highly productive Humboldt Current System (Chavez et al. 2008), primarily off the coast of northern Peru, with no difference between the birds we tagged with geolocators as compared to those with GPS pinpoint tags (Fig. 1). This pattern was also consistent with Sabine's Gulls breeding in the Canadian Arctic (Davis et al. 2016; Gutowsky et al. 2021). Alaskan-breeding birds arrived at the Pacific winter area (10 Oct; Table 2) on average a month earlier than did those breeding in the Canadian Arctic (11 Nov; Davis et al. 2016), however it is currently unknown if this is due to annual variation or breeding area. In contrast to most migratory seabirds, Sabine's Gulls on both the Pacific and Atlantic oceans appear to depend on one to two highly restricted wintering area (this study; Stenhouse et al. 2012; Davis et al. 2016). The northbound migration appears to follow a similar route (Fig. 2), but with different staging dynamics. Staging areas were further north than those used during the southbound migration, with less time spent off the coast of California and Oregon. Birds primarily staged in the Gulf of Alaska Current System off the west coast of Canada and into Alaska (Fig. 2). This pattern is similar to that found by (Stenhouse et al. 2012) where southbound and northbound migratory routes were not significantly different but different staging areas were used. However, the Atlantic birds' northbound staging area was further south than the southbound staging areas (Stenhouse et al. 2012), the opposite of the pattern observed in the Pacific. However, Gutowsky et al. (2021) showed more time spent off the coast of Mexico and California than we did during northbound migration. It is possible that timing and routes vary between years, as shown by Davis et al. (2016). It is important to remember that the patterns and conclusions in this study are based on a small sample of birds tagged with geolocators, and future studies should verify this assumption with larger sample sizes.
Figure 2.
Kernel density estimates of the wintering area and northbound migration staging areas, and migration tracks of Sabine's Gulls tagged with geolocators at Anachlik Island, Alaska, in 2011.
Gutowsky et al. (2021) also showed several Sabine's Gulls migrating large distances over land in the northbound migration. First, directly from the Pacific to Hudson Bay then north into the Arctic, and alternatively, from the Gulf of Alaska directly over the Yukon or Alaska into the Beaufort Sea and Canadian Arctic. However, there was no evidence that our tracked birds travelled over mainland Alaska from the Gulf of Alaska to the Beaufort Sea. Instead, our tracks suggest birds might cut west across the Alaska Range / Aleutian Range into the eastern Bering Sea and then north to the Arctic.
Despite the fact that we received more accurate location data from the GPS pinpoint tags, their lower number of locations, weight, and expense detracted from their use. For studies with the ability to return to the breeding site in subsequent years to recapture birds, we recommend geolocators for the time being. It is likely that future technological advances will improve the GPS pinpoint technology.
The low return rates of the geolocator birds in 2012 corresponded with overall lower return rates of banded adult birds after the winter of 2011/2012 at Anachlik (Jim Helmericks, pers comm). This unexpected mortality points to the value of monitoring returns and understanding migratory and wintering areas that link conservation concerns across regions. Sabine's Gulls forage in marine regions of high and relatively predictable productivity during migration and winter (Stenhouse et al. 2012), however years when that food source fails either due to overfishing (Crawford 1991, Crawford and Dyer 1995, Crawford et al. 2007) or oceanographic shifts (Fife et al. 2018) could negatively impact a large proportion of the population. Fife et al. (2018) found high adult mortality in years of extreme climatic events on the wintering areas for Sabine's Gulls, and as it appears almost all of the Sabine's Gulls breeding in Alaska and the central Canadian Arctic winter in the same general location (this study; Davis et al. 2016), this could impact a significant proportion of the population. Additionally, while oceanographic dynamics may be a factor, seabird bycatch in fishing gear has also been raised as a potential source of mortality. However, we found no information specific to Sabinés Gull bycatch for Peru north to Alaska, either in the literature or via local experts (Carlos Zavalaga and Javier Antonio Quiñones Dávila, pers comm.). Understanding the spatial and temporal use of marine areas by Sabine's Gulls is important for their conservation and for facilitating the identification of Important Bird Areas.
Data Availability Statement
Data are available from the Dryad Digital Repository: https://datadryad.org/stash/share/r4ersCOwlQ7mLZK3QQ-7WrI0ETnHiIOvPP3Cg9e2nkE (McGuire et al. 2024).
Acknowledgements
Derek Helmericks helped fund and capture the birds at the Anachlik site. An original review of the data was conducted by David Ward and John Reed (USGS), which helped inform this manuscript. Peter Detwiler and Anna Van Duyke assisted in the capture of birds at Qupaluk. Wilburforce Foundation, Bureau of Land Management, and Wildlife Conservation Society provided funding. Bird captures were under permit from the Alaska Department of Fish and Game and the USFWS Migratory Bird Permit Office. All applicable ethical guidelines for the use of birds in research have been followed, including those presented in the Ornithological Council's “Guidelines to the Use of Wild Birds in Research” (Fair et al. 2010).
