While the breeding and wintering ranges of the eastern population of Barrow's Goldeneyes (Bucephala islandica) are generally described, molting locations and links among breeding, molting, and wintering areas are unclear, particularly for adult females. Incubating females from the same breeding location (n = 5) were equipped with satellite transmitters in June 2009. Four molting sites were identified over 2 years, spread broadly across Québec: an inlet in Ungava Bay 1,100 km from the breeding area, a lake 100 km south of Ungava Bay (880 km from breeding area), a lake near Hudson Bay (910 km from breeding area) and the mouth of the Rivière aux Outardes River in the St. Lawrence Estuary (165 km from breeding area). The distance between molting females averaged 755 km and two females molted in regions where males were known to molt. Of four birds with consecutive years of molt locations, three showed inter-annual fidelity to within 5 km of the previous molt sites and the fourth molted in sites that were 968 km apart. Females wintered in different locations within the St. Lawrence Estuary and moved widely throughout the area during winter. The south coast of the St. Lawrence Estuary was used during spring and fall staging, and the north coast during winter. There was not strong migratory connectivity among annual cycle stages in eastern adult female Barrow's Goldeneyes, indicating that they should be considered a single management unit that occurs over a broad range throughout the year.
The eastern North American population of Barrow's Goldeneyes (Bucephala islandica) was designated to be ‘of special concern’ by the Committee on the Status of Endangered Wildlife in Canada in 2001 (Robert et al. 2000a) due to low numbers (estimated to be 4,000 individuals at the time) and anthropogenic threats (Savard 1996; Savard and Dupuis 1999). Lack of knowledge about basic ecology and distribution also contributed to conservation concerns; for example, breeding areas were not fully determined until 1998 (Robert et al. 2000b). In recent years, movements and migratory connectivity of adult males have been studied (Benoit et al. 2001; Robert et al. 2002), the wintering distribution and ecology have been well documented (Bourget et al. 2007; Ouellet et al. 2010), and numbers have been estimated to be approximately 5,200 birds (Robert and Savard 2006). Several molting sites for males were located, all north of the breeding areas, including Hudson Bay, Ungava Bay, southern Baffin Island, along the Labrador Coast, and on northern inland lakes (Robert et al. 2000c, 2002). The large number, diversity and wide distribution of molting sites were unexpected because of the small size of the population and because in Iceland most birds molted in a few areas (Einarsson and Gardarsson 2004). These studies provided insight into the relationships among breeding, molting and wintering locations of males, but whether females followed similar patterns was unknown. Data from British Columbia suggested that females molted closer to the breeding areas than males (Eadie et al. 2000; S. Boyd, unpubl. data). Males outnumber females (Robert and Savard 2006), pair on wintering areas (Rodway 2007), and follow their mate to her natal area (Savard and Eadie 1989). Females are thus important in structuring populations. As the number of adult females is small in the eastern population (∼2,000 individuals; Robert and Savard 2006), their protection is essential and knowledge of their seasonal movements, molting areas, molting phenology and winter habitats is important to document. Also, it is important to determine whether there are subpopulations or management units within the eastern population as this would influence conservation strategies.
Our main objectives were to: (1) determine molting location (s) of adult female Barrow's Goldeneyes, (2) define migratory connectivity and population structure, and (3) determine whether females used the same molting location each year. Philopatry is important to assess as it may help define meaningful population subunits.
Five female Barrow's Goldeneyes were captured in nest boxes (Savard and Robert 2007) during incubation or laying on 9 and 10 June 2009 in the Martin-Valin and Chauvin Zecs located 60 km northeast of Tadoussac, Québec (48° 31′ 40.59″ N, 70° 14′ 16.28″ W; Fig. 1). Satellite transmitters were implanted in the abdominal cavities of females by experienced veterinarians (see Fitzgerald et al. 2001 for the detailed procedure). We used 26 g Argos PTT-100 implant transmitters (Microwave Telemetry) with external antennae. Transmitters were programmed to transmit for 5 hr every 3 days (5 hr on and 70 hr off) and constituted 3.5–3.8% of female mass. Females were released 1–2 hr post-surgery on the lake adjacent to their nest box. Bird movements were monitored using Argos (Harris et al. 1990; Service Argos 1996). The exact date on which an individual left or arrived at a given site could not be determined because locations were usually obtained at 2- to 3-day intervals. We assumed that a bird first arrived at a given site on the date of the first signal from that site and that it departed the site on the day following the last location at that site. We used a modified version of the Douglas-Argos Filter to remove implausible locations using minimum redundant distance and distance-angle-rate tests between consecutive location points (Douglas 2006). The distance between a location of quality 2–3 (< 350 m precision) and the next one of quality 2–3 (on a different day, usually 3 days later given our duty cycle) was measured to get an estimate or index (using the average) of the area used by a female when on the breeding, molting and wintering areas as well as on the fall and spring staging areas. Also, the coefficient of variation (CV) was calculated to portray individual variability.
Detailed movement characteristics of adult female Barrow's Goldeneyes from the same breeding area.
Females stayed on the breeding area (∼600 m elevation) until mid to late July (Table 1). Arrival on molting areas ranged between 5 and 15 August in 2009 and 31 July and 10 August in 2010. Molting locations were quite dispersed, ranging from 97 to 1,044 km from the breeding area (Table 1) in a variety of directions (north, west and south; Fig. 1). In 2009, three females molted in three different estuarine habitats, two in the St. Lawrence Estuary (one at the mouth of the Rivière aux Outardes near Baie-Comeau and one near Kamouraska), and one in Ungava Bay. Two others molted on inland lakes: one about 100 km south of Ungava Bay (370 m elevation) and one near the southeastern shore of Hudson Bay (4 m elevation; Fig. 1).
In 2010, two females returned to the same area where they molted in 2009: female 94768 to the exact same estuary in Ungava Bay and female 94770 molted on a smaller lake, 4 km from the 2009 molting lake located about 100 km south of Ungava Bay (Fig. 1). One female (94773) molted on a lake near Hudson Bay in 2009 (930 km northwest of the breeding area) and in the St. Lawrence Estuary in 2010 (132 km southeast of the breeding area and 960 km from the 2009 molting site). Female 94767 molted both years in the St. Lawrence Estuary but the precise location was unclear. Duration of stay at molting areas outside the St. Lawrence Estuary averaged 79.5 ± 5.3 days (±SD; Range = 75-87; n = 4). The last signal from the molting areas outside the St. Lawrence Estuary averaged 25 October (±6 days; n = 4). Females departed on average 46 days later than males from the breeding areas and spent 42.5 fewer days at their molting site (Table 2). Females departed later than males from molting sites, but there were wide ranges in departure time of individuals (20 October-2 November; Table 2).
Comparison of the timing of migration and molt of female and male Barrow's Goldeneyes in eastern North America. Only for birds that molted outside the St. Lawrence Estuary. Males identified per Robert et al. 2002.
Wintering, Staging and Breeding Areas
All five females wintered in the St. Lawrence Estuary within the known wintering range. Arrival in the St. Lawrence Estuary for the females that molted north ranged between 25 October and 8 November (n = 4). One female staged for a few weeks during the fall on the north shore of the estuary and three staged along the south shore (Fig. 2). Female 94768 staged on the south shore of the St. Lawrence Estuary in 2009 and 2010 but at sites 38 km apart. All wintered on the north shore along a 220-km stretch of shoreline (Fig. 3). In the spring, two females returned to the south shore of the estuary where they staged before departing for their breeding area and two remained on the north shore. None of the females staged at the same location. The closest spring staging sites were 40 km apart and the farthest 160 km apart (Fig. 4). Last signals from the spring staging areas ranged between 20 April and 3 May. For three females, the next signal was 3 days later on the breeding area. The breeding area averaged 118 km (Range = 82–188) from the spring staging areas. All females for which we had data (n = 4) returned to the same breeding area in 2010.
Both spring and fall migrations were quick with birds moving directly to breeding locations from their spring staging areas and from their northern molting locations to their fall staging sites. Our duty cycle did not permit a close monitoring of the migration, but birds spent < 5 days staging on lakes before returning quickly to the St. Lawrence Estuary. Two females (94770, 94773) spent < 5 days close to their breeding areas on their way back from their molting locations, but no inland stays > 4 days were detected during fall or spring migrations.
Females stayed within the smallest area on the molting areas (females averaged 1.5 ± 0.6 km between successive high quality locations) and the largest on the wintering areas (20.4 ± 11.8 km; Table 3). Females were more mobile during fall (5.6 ± 2.5 km) than during spring staging (1.4 ± 0.7 km) or breeding (2.6 ± 0.9 km; Table 3). Females used the smallest area during molt and during spring staging. Movements of individual females were more variable during fall staging (CV = 84%) than wintering (CV = 49%) but not as extensive (5.6 km vs. 20.4 km).
Average distance (km) ± SE (CV) between two Argos locations of quality 2 or 3 (distance between a location of quality > 2 and the next one of quality > 2 on a subsequent day) during different periods of the annual cycle. Breeding = while on the breeding area; Molting = while on the molting area; Fall = from arrival in St. Lawrence Estuary to settlement on wintering areas (for birds already in St. Lawrence Estuary, when moved >50 km from molt area); Wintering = mostly January and February; Spring = from first movement out of wintering area or in March and April. CV = coefficient of variation expressed in %; n = number of different female-seasons (number of different females involved).
Adult female Barrow's Goldeneyes from a single breeding area molted across a broad range of non-breeding sites. These findings indicate that there is little migratory connectivity among annual cycle stages, and thus there is little suggestion that subpopulation structure exists within the eastern North American population in relation to molting areas. A similar lack of relationship between breeding and molting locations has been found in Common Mergansers (Mergus merganser; Pearce et al. 2009) and in Steller's Eiders (Polystica stelleri; Dau et al. 2000). All females wintered within the St. Lawrence Estuary but not together. Thus, within the St. Lawrence Estuary (∼250 km long and 60 km at its widest part), there is no direct connection between a wintering and a breeding site. Robert et al. (2002) did not find any relationship between the wintering and molting locations of male Barrow's Goldeneyes. However, there was a high degree of interannual site fidelity at both breeding and molting sites, indicating that perturbations at one of these stages might lead to effects on a population subunit. Unlike in western North America where Barrow's Goldeneyes remain faithful to a wintering site in winter (Savard 1988; D. Esler, unpubl. data), female Barrow's Goldeneyes in the St. Lawrence Estuary move around, possibly in response to ice conditions.
In eastern North America, females, like males (Benoit et al. 2001; Robert et al. 2002), molt in estuarine settings as well as coastal and inland lakes, whereas in western North America they only molt on inland alkaline lakes (Van de Wetering 1997; Hogan et al. 2011). Three female Barrow's Goldeneyes were philopatric to their molting location and one was not. Hatton and Marquiss (2004) found similar patterns in female Common Mergansers with most females being philopatric to their molting location. Fidelity to molting locations has also been documented in Steller's Eiders (Flint et al. 2000) and Harlequin Ducks (Histrionicus histrionicus; Brodeur et al. 2002). Causes leading to the selection of a given molting site in a given year are unknown, but change of molting location between years has been documented in sea ducks but its frequency and causes are unknown (Hatton and Marquiss 2004; J.-P. L. Savard, unpubl. data).
Female Barrow's Goldeneyes likely abandoned their clutch following surgery as none remained on a single wetland for several weeks. As female sea ducks are known to remain on their breeding area even after nest failure (Eadie et al. 2000; Savard et al. 2007), it was impossible to clearly determine whether our females bred successfully or not. All females stayed within a few km during breeding suggesting that they do not prospect for nest cavities within a large area (Eadie and Gauthier 1985). Both female and male Barrow's Goldeneyes fly more or less directly to their molting site after leaving the breeding area (Robert et al. 2002). Male Barrow's Goldeneyes reach their molting location about a month before females (Robert et al. 2002).
Similar to males, females stay at their molting location until late fall, well after having regained their flying abilities. The molting area could be also considered as a first fall staging area as birds are flightless for about 4 weeks but remain there for over 10 weeks. Females left their northern molting areas and arrived in the St. Lawrence Estuary about a week after males but timing varied among individuals. Female King Eiders (Somateria spectabilis) also dispersed later from the molting areas than males (Phillips et al. 2006).
Females used the same fall staging areas as males. In the St. Lawrence Estuary, these sites are located mostly along the south shore. Birds are forced out of these areas by ice conditions (Bourget et al. 2007) as intertidal areas of the south shore of the St. Lawrence Estuary, unlike on the north shore, are icecovered during winter (Robert et al. 2003). However, they return to the south shore in the spring as intertidal areas become icefree. Food resources are likely greater there, as they are not exploited during winter. They used a very small area during spring staging, likely because females, likely already paired, focus on feeding at that time in preparation for the breeding season.
Female Barrow's Goldeneyes moved frequently during winter and each female used several sites throughout the winter. Also, all females did not winter in the same area. However, it seems that the estuary wintering population is a single one. We did not observe any movements between the St. Lawrence Estuary and the Gulf of St. Lawrence; neither did Robert et al. (2002) for males. Although sample sizes are relatively low, it may be prudent, from a conservation and habitat use perspective, to consider birds wintering in the Gulf of St. Lawrence and in Chaleur Bay (Robert and Savard 2006; Ouellet et al. 2010) distinct from those wintering in the estuary until shown otherwise. This should be kept in mind when planning and interpreting winter survey results.
The lack of a one-to-one relationship between breeding and molting locations implies that any disaster at a molting site will affect female Barrow's Goldeneyes from more than one breeding location as was found for males (Robert et al. 2002); inversely, any disaster at a breeding area would affect birds from several molting sites. Within the estuary (scale of 250 km long), females from the same breeding area did not winter together nor did they use the same spring or fall staging areas. Factors leading to the selection of wintering and staging areas are unknown in goldeneyes as is the case for most sea ducks. Possibly this selection occurs during the subadult years, which have been poorly studied in all sea ducks and could also have a genetic component (Berthold 1996). Philopatry to wintering areas has been documented in Barrow's Goldeneyes (Savard 1985) and other sea ducks (Iverson et al. 2004; Iverson and Esler 2006). Whether Barrow's Goldeneye pairs remain together for several years as was found for some pairs in western North America (Savard 1985) is unknown. Factors associated with winter movements are likely related to ice conditions and food resources but this remains to be confirmed.
We thank the team of veterinarians who performed the surgeries: S. Lair, S. Larrat, and I. Dubois. Thanks to C. Marcotte for assistance with logistics and capture. Thanks to the two anonymous referees for their pertinent comments. Thanks to A. Nadrau and M. Benoit for their help with the figures. The Animal Care (SCFQ2009-01) and scientific (SC-22) permits were issued by Environment Canada, Québec Region. Funding for the study originated from the Sea Duck Joint Venture and Environment Canada.