Progress in Oceanography 136 (2015) 162–174

Contents lists available at ScienceDirect

Progress in Oceanography

journal homepage: www.elsevier .com/locate /pocean

Limits to benthic feeding by eiders in a vital Arctic migrationcorridor due to localized prey and changing sea ice

http://dx.doi.org/10.1016/j.pocean.2015.05.0140079-6611/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 307 399 7441; fax: +1 618 453 2806.E-mail address: lovvorn@siu.edu (J.R. Lovvorn).

James R. Lovvorn a,⇑, Aariel R. Rocha a, Stephen C. Jewett b, Douglas Dasher b, Steffen Oppel c,Abby N. Powell d

a Department of Zoology and Center for Ecology, Southern Illinois University, Carbondale, IL 62901, USAb Institute of Marine Science, University of Alaska Fairbanks, Fairbanks, AK 99775-7220, USAc RSPB Centre for Conservation Science, Royal Society for the Protection of Birds, The Lodge, Sandy, Bedfordshire SG19 2DL, UKd Alaska Cooperative Fish and Wildlife Research Unit, U.S. Geological Survey, University of Alaska, Fairbanks, AK 99775-7020, USA

a r t i c l e i n f o

Article history:Available online 12 May 2015

a b s t r a c t

Four species of threatened or declining eider ducks that nest in the Arctic migrate through the northeastChukchi Sea, where anticipated industrial development may require prioritizing areas for conservation. Inthis nearshore corridor (10–40 m depth), the eiders’ access to benthic prey during the spring is restrictedto variable areas of open water within sea ice. For the most abundant species, the king eider (Somateriaspectabilis), stable isotopes in blood cells, muscle, and potential prey indicate that these eiders ate mainlybivalves when traversing this corridor. Bivalves there were much smaller than the same taxa in deeperareas of the northern Bering Sea, possibly due to higher mortality rates caused by ice scour in shallowwater; future decrease in seasonal duration of fast ice may increase this effect. Computer simulationssuggested that if these eiders forage for >15 h/day, they can feed profitably at bivalve densities>200 m�2 regardless of water depth or availability of ice for resting. Sampling in 2010–2012 showed thatlarge areas of profitable prey densities occurred only in certain locations throughout the migration cor-ridor. Satellite data in April–May over 13 years (2001–2013) indicated that access to major feeding areasthrough sea ice in different segments of the corridor can vary from 0% to 100% between months and years.In a warming and increasingly variable climate, unpredictability of access may be enhanced by greatereffects of shifting winds on unconsolidated ice. Our results indicate the importance of having a rangeof potential feeding areas throughout the migration corridor to ensure prey availability in all years.Spatial planning of nearshore industrial development in the Arctic, including commercial shipping,pipeline construction, and the risk of released oil, should consider these effects of high environmentalvariability on the adequacy of habitats targeted for conservation.

� 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Climate change in the Arctic includes altered physical factorssuch as temperature, wind, and ice cover, with subsequent effectson marine organisms and their interactions. However, effects willalso include increased shipping and commercial fishing allowedby reduced ice cover, and industrialization encouraged by warmerconditions and expanding development of oil, gas, and mineralresources (Brigham, 2011; Smith and Stephenson, 2013;Christiansen et al., 2014). Both climatic and direct human impactsmust be considered in predicting the trajectory of important habi-tats, so that areas identified for conservation now will remain

adequate into the future (Speer and Loughlin, 2011; Reeves et al.,2012).

At high latitudes, access to foraging habitats by marine birdsand mammals can be constrained by too much or too little seaice (Barber and Iacozza, 2004; Heide-Jørgensen and Laidre, 2004;Bajzak et al., 2011). This factor is especially important forbottom-feeding animals such as sea ducks (Mergini) and walruses(Odobenus rosmarus), for which viable prey densities are oftenlocalized (Nelson et al., 1987; Jay et al., 2014; Lovvorn et al.,2014). Although increased open water relative to ice cover mayexpand accessibility to patchy benthic foods, some ice cover isimportant to provide platforms for resting (Lovvorn et al., 2009;Jay et al., 2012). Climate warming may cause greater variabilityin the locations of different ice concentrations, because less consol-idated ice is more susceptible to redistribution by wind (Zhanget al., 2012). In shallow nearshore areas, decreased seasonal

J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174 163

duration of fast ice, with greater movement and gouging of sedi-ments by unconsolidated ice, can also affect the density and sizeof prey organisms (Toimil, 1978; Barnes and Souster, 2011).Greater spatial and temporal variability in the locations of suitablehabitat calls for larger conservation areas, thereby increasing con-flicts with alternative, commercial uses. For marine birds andmammals important to subsistence hunting, adequate considera-tion of the probable range of future habitat conditions is alsoneeded to promote sustainability of indigenous human economiesand cultures (Wenzel, 2009).

These multiple concerns are converging in the nearshore zoneof the northeast Chukchi Sea (10–40 m depth, Fig. 1). This zone isan essential corridor for spring migration of a large fraction of mar-ine birds and mammals destined for regions of the western NorthAmerican Arctic. These animals migrate through the variable flawleads (open water) that develop between landfast ice and movingpack ice, in a region that earlier studies suggested had low avail-ability of prey for some bottom-feeding endotherms (sea ducksand walruses) (Feder et al., 1994a,b). Native villages along thiscoastline depend on access to these animals for subsistence hunt-ing (George et al., 2004). Plans for possible extraction of oil and gasoffshore (Fig. 1) include laying one or more major pipelinesbeneath the sediments from offshore drill rigs across the shallownearshore zone to onshore terminals (BOEM, 2014). In this region,deep ice gouging of nearshore sediments can occur during springice breakup, and if a pipeline were ruptured there is currently littlecapability for removing spilled oil from areas of broken ice (Grantzet al., 1982; AMAP, 2007; NRC, 2014). Locations where marinebirds and mammals are likely to concentrate during spring migra-tion should be considered in pipeline placement. However,

Fig. 1. Sampling stations in the Chukchi Sea migration corridor in August–September 201indicated are offshore areas leased for oil and gas exploration by the U.S. Bureau of Odesignated under the U.S. Endangered Species Act. Shaded depth contours annotated at 10of Barrow. Four segments of the corridor used for prey and ice analyses, delineated by linBay, and Point Hope.

nearshore habitats for benthic-feeding birds, as a function of bothprey dispersion and variable ice cover, have not been assessed.

Two species of eider listed as threatened under the U.S.Endangered Species Act (Steller’s eider, Polysticta stelleri; specta-cled eider, Somateria fischeri), and two eider species whose num-bers appear to have declined by over 50% since the 1970s (kingeider, S. spectabilis; common eider, S. somateria) (Suydam et al.,2000), nest along the Arctic coasts of Alaska and western Canada.All of these eiders that winter in the North Pacific region, includingabout 700 Steller’s eiders, 6500 spectacled eiders, 450,000 kingeiders, and 113,000 common eiders, pass through the Chukchinearshore corridor in the spring (Quakenbush et al., 2009; Stehnet al., 2013). Thus, well over a half million eiders depend on thiscorridor, not only the federally threatened species but also thecommon eiders and especially king eiders which are importantfor subsistence hunting (Braund, 1993; Byers and Dickson, 2001).

As a key habitat bottleneck for a range of migrating species thatlater disperse over a huge area, the Chukchi migration corridor isamong the most important habitats for eiders in the westernArctic (Petersen, 2009; Oppel et al., 2009; Sexson et al., 2014).For large eiders that nest in the Arctic, breeding propensity andsuccess in a given year are thought to be strongly affected by feed-ing conditions during spring migration that allow buildup of ade-quate body fat (Coulson, 1984; Oosterhuis and van Dijk, 2002;Bentzen et al., 2008). Moreover, it appears that for spectacledeiders and possibly king eiders, an important fraction of pair bond-ing may occur during spring migration and staging (Lovvorn et al.,2012). Although king eiders carrying satellite transmitters werelocated throughout this migration corridor, in the spring about80% of radio locations were in the Ledyard Bay segment (Fig. 1;

0–2012. Inset at lower left shows location of the area in northwest Alaska, USA. Alsocean Energy Management (BOEM), and critical habitat for spectacled eiders (SPEI)

-m increments do not show bathymetry >50 m deep in the submerged canyon westes perpendicular to shore, are (from north to south) Skull Bay, Wainwright, Ledyard

164 J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174

Oppel et al., 2009). Ledyard Bay is also heavily used by spectacledeiders during fall migration, and during molt of wing and bodyfeathers in late summer (BOEM, 2014; Sexson et al., 2014).Consequently, Ledyard Bay has been designated critical habitatfor spectacled eiders under the U.S. Endangered Species Act (Fig. 1).

During migration and staging in this corridor in April–May, seaice may constrain access to profitable feeding sites. A past benthicstudy in this region (Feder et al., 1994a,b) delimited large areaswith different invertebrate assemblages, but did not investigatepatch structure at smaller scales. These large-scale delineationssuggested that densities of the typical prey of eiders were on aver-age quite low. However, areas with high enough prey density tosupport profitable foraging can be a relatively small fraction oftotal habitat (Lovvorn et al., 2009, 2014), so assessing the adequacyof a larger area to support eiders requires finer-scale analyses. Ifhigh prey densities are localized, they are more prone to being cov-ered by ice (Barry et al., 1979; Eicken et al., 2006). For example, inthe wintering area of spectacled eiders in the north-central BeringSea, high ice concentrations can exclude the eiders from the betterfeeding areas, leading to very low fat levels at the end of winter(Lovvorn et al., 2014). Such effects can result from redistributionof ice concentrations by changing winds, regardless of trends intotal ice extent. In the northeast Chukchi Sea, reductions in foodavailability may be amplified by local feeding by tens of thousandsof walruses that now haul out along the coast in August–September, because the summer ice cap they need for resting hasreceded beyond the depth to which walruses normally dive (about100 m) (Jay et al., 2012).

In this study, we explored seasonal and among-year variationsin the location of open water relative to recent prey concentrationsin the Chukchi migration corridor. Based on stable isotope valuesin potential prey and in red blood cells and muscle of king eidersjust after they passed through this region, we inferred their dietsduring the preceding weeks of migration and staging in theChukchi Sea. We used benthic samples to corroborate the isotoperesults in terms of the relative availability of different potentialprey, and to map prey densities. A computer simulation model(Lovvorn et al., 2009) was used to estimate threshold prey densi-ties required by king eiders to balance their energy budgets. For13 years of available remote sensing data (2001–2013), we gener-ated maps of open water at intervals of 1–2 weeks in April and Mayof each year. Based on those analyses and maps of recent prey dis-persion, we estimated total areas of profitable feeding locationsthat were accessible through the ice at 2-week intervals in April–May of each year. Given the variability among months and yearsand projected climatic trends, we then asked how areas identifiedfor conservation might be distributed to ensure availability of ade-quate feeding areas for eiders in all years.

2. Methods

2.1. Stable isotopes in king eiders

We used stable carbon and nitrogen isotopes in king eider bloodand muscle to infer their diets during spring migration through theChukchi Sea. All king eiders that use the Chukchi Sea in spring andbreed near the Arctic coasts of Alaska or Canada migrate past thevillage of Barrow (Fig. 1) after passing through the Chukchi Sea(see Oppel et al., 2009). Breast muscle was sampled from 8 adultfemale king eiders shot during 15–25 May 2003 by subsistencehunters near Barrow (these studies were performed under the aus-pices of University of Alaska IACUC protocol 05-29). These sampleswere dried and ground, and lipids were extracted by repeatedrinses in 1:2 chloroform:methanol (Oppel et al., 2010). Also,between arrival and nest initiation (10–22 June 2006–2007), blood

samples (1 mL) were taken by jugular venipuncture from 30 maleand 21 female king eiders at two sites on the Arctic coastal plain ofAlaska, and centrifuged to separate red blood cells from plasma(for details, see Oppel and Powell, 2010). These birds had stagedmainly in the Chukchi Sea. However, some individuals may havespent up to several weeks (mean 15 days) staging in the easternBeaufort Sea after passing through the Chukchi migration corridor(Phillips et al., 2007). Both lipid-extracted muscle tissue and redblood cells were analyzed for d13C and d15N as described inOppel et al. (2010).

The turnover half-life of d13C is similar between whole bloodand blood cells of birds (Evans Ogden et al., 2004). Turnover ratecan vary both with diet (Haramis et al., 2001) and among bird spe-cies (11–23 days, mean 15 days, among non-passerines; Carletonand Martinez del Rio, 2005). For a moderately large diving duck(canvasback, Aythya vallisneria) eating bivalves (Macoma balthica),the half-life in whole blood was 23 days for d13C and 25 days ford15N (Haramis et al., 2001). Thus, isotope values in blood cellsshould reflect diet over the last 3–4 weeks. In Japanese quail(Coturnix japonica), d13C turnover took 9% longer in breast musclethan in whole blood (Hobson and Clark, 1992a).

Isotopic enrichment from diet to the blood or tissues of birdscan be quite variable, depending on type and nutrient content offood, nutrient and energy status of the bird, and other complexitiesof metabolism and physiological routing (Hobson et al., 1993;Podlesak and McWilliams, 2007). In captive spectacled eiders feda mixture of pelleted food and fish, fractionation between dietand blood cells was 2.0 ± 0.2 (SD)‰ for d13C and 4.0 ± 0.2‰ ford15N (Federer et al., 2010). These mean fractionations for bloodcells in eiders were at the upper end of those reported for bloodof 10 bird species (range �0.8‰ to 2.3‰, mean 0.8‰, for d13C;range 2.1–4.2‰, mean 2.9‰, for d15N; review in Cherel et al.,2005). Estimates of enrichment from food to breast muscle in gulls,quail, cormorants, grebes, and dunlins (Calidris alpina) ranged from0.3‰ to 2.1‰ d13C and 1.4‰ to 4‰ d15N (Mizutani et al., 1991;Hobson and Clark, 1992b; Evans Ogden et al., 2004; Paszkowskiet al., 2004). For both blood cells and muscle of king eiders, weassumed fractionation of 0–2‰ for d13C and 3–4‰ for d15N.

2.2. Sampling and mapping prey dispersion

We sampled 31 stations in the southern portion of the studyarea from 23 August to 3 September 2010, 30 stations in the north-ern part of the area from 5 to 16 September 2011, and 6 stations invarious parts of the area from 5 to 15 August 2012 (Fig. 1). Benthicinvertebrates were sampled with a van Veen grab (3 replicate sam-ples per station, area of each sample �0.1 m2, depth of sedimentssampled �10 cm). Grab samples were rinsed over a 1-mm sieveand preserved for later sorting and identification. Shell length ofeach bivalve in the samples was measured to the nearest0.1 mm. Epibenthic invertebrates were sampled (one trawl per sta-tion) with a 3.05-m plumb-staff beam trawl with 7-mm mesh inthe body, 4-mm codend liner, double tickler chain, and 16-cmlengths of chain attached to the footrope at 16-cm intervals.Time at the bottom for each trawl was 2–5 min at a speed relativeto water of 1–1.5 kt. Trawl samples were sorted by taxa andcounted on deck.

Areal densities of major macroinvertebrate taxa in grab andtrawl samples were interpolated among sampling stations andplotted on a map of the study area with a GeographicInformation System (ArcGIS 10.2, ESRI, Redlands, CA). We usedinverse distance weighting (IDW) to interpolate densities of differ-ent taxa from values at the nearest five sampling stations (IDW set-tings: power = 5, standard neighborhood, four sectors, major andminor semiaxes = 124,000). For bivalves, only those with shelllength P3 mm were included, based on size frequencies in the

J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174 165

esophagi and gizzards of king eiders in West Greenland (Frimer,1997). This size limit excluded many recently recruited bivalvesthat cause variation in bivalve populations between late summerand early spring.

2.3. Stable isotopes in potential prey

Results from grab and trawl sampling identified taxa with highenough densities to be potentially important prey for eiders. Forcomparison with stable isotope values in red blood cells and mus-cle of king eiders, we used published d13C and d15N values for can-didate prey collected in August 2004 near Cape Lisburne or PointHope (mostly in Alaska Coastal Water; Iken et al., 2010), or inSeptember 1983 between Cape Lisburne and Wainwright(Dunton et al., 1989) (Fig. 1). Of the few taxa included in both stud-ies, differences between studies in mean d13C values were 60.3‰

for a polychaete (Gattyana ciliata), two crabs (Chionoecetes opilio,Hyas coarctatus), and a brittle star (Stegophiura nodosa); and 0.8‰

for a gastropod (Buccinum sp.). The latter differences suggest thatchanges in isotope values of the same taxa over two decades wereminimal.

Fractionation of d13C during lipid synthesis can result in valuesfor body lipid that are 6–8‰ lighter (more negative) than for pro-teinaceous tissues. To avoid confounding effects of variable lipidcontent when tracing carbon sources, lipid is often extracted fromtissue samples before isotopic analysis. Iken et al. (2010) did notextract lipid from invertebrate samples because data from anotherstudy indicated that resulting d13C values changed by only about0.9‰ in a bivalve, 1.1‰ in a gastropod, and 0.6‰ in a polychaete.Dunton et al.’s (1989) samples also were not lipid-extracted.

Post et al. (2007) recommended that lipids be extracted if lipidcontent of an aquatic animal is quite variable and >5%, and if thedifference between end members (prey to be discriminated) is lessthan 10–12‰. Lipid content of different bivalve size classes eatenby spectacled eiders ranged from 14% to 20% in Macoma, 11% inEnnucula, and 11–15% in Nuculana (Richman and Lovvorn, 2003).Lipid content can also vary greatly among seasons in bivalves(review in Richman and Lovvorn, 2009), and we were interestedin d13C differences between prey end members of <3‰. Thus, forbivalves we used lipid-extracted values from the northern BeringSea (Lovvorn et al., 2014), which for Yoldia averaged 0.6‰ heavierthan the non-extracted value of Dunton et al. (1989). These bivalvetaxa included several potentially important prey in the Chukchimigration corridor (Macoma, Ennucula, Nuculana, Yoldia) for whichno lipid-extracted values from that area have been reported. Owingto their apparent trophic dependence on a long-term pool of sedi-ment organic matter, d13C and d15N values of Macoma, Ennucula,and Nuculana showed very little seasonal variation fromSeptember to early June in the northern Bering Sea (North et al.,2014).

2.4. Simulation model of eider foraging

A detailed simulation model of the energy balance of spectacledeiders wintering in the northern Bering Sea was presented byLovvorn et al. (2009). Based on a combination of biomechanicalmodels, respirometry of captive ducks in dive tanks, measure-ments of intake rates at a range of bivalve densities (functionalresponses), and remote sensing data on ice conditions, this modelestimates eider energy balance for varying prey densities, ice con-ditions, and water depths (eiders can readily dive to all depths inour study area, Fig. 1). We adapted this model for the slightly largerking eider, and incorporated elements of another model for divingducks (Lovvorn et al., 2013), to estimate threshold prey densitiesneeded to achieve positive energy balance. Because the bivalveprey of eiders in our Chukchi Sea study area were much smaller

than in the Bering Sea (see Section 3), we used the functionalresponse for smaller prey measured by Richman and Lovvorn(2004). These models were described in detail, and their sensitivityto a range of variables thoroughly explored, in the original papers.

To convert intake of bivalve prey to intake of energy in the sim-ulation model, we multiplied the estimated number of bivalvesconsumed at a given bivalve density (according to the functionalresponse) by the estimated energy assimilated per individualbivalve. In 2010–2012, mean shell length (±1SD) for all bivalvespecies combined in our study area (including only individuals3–60 mm; Frimer, 1997) was 7.4 ± 7.0, n = 3199. Energy contentat a given shell length varies widely among bivalve species andacross locations, seasons, and years in the same species (see reviewin Richman and Lovvorn, 2009). Based on values from our own andother studies (Richman and Lovvorn, 2003, 2004, 2009 and refer-ences therein), we modeled the consequences for profitable preydensities of energy content per clam of 200–700 J.

We also modeled effects of time spent feeding per day, waterdepth, and percent of non-feeding time spent resting on ice on preydensities needed for profitable foraging. The period of daylight inthe northeast Chukchi region (68.3–71.2�N) averages 17 h inApril and almost 22 h in May. King eiders wintering in ArcticNorway (69.8�N) foraged only about 19% of 16-h days (Systadet al., 2000), whereas common eiders in Newfoundland, Canada(46.3�N) foraged a maximum of 60% (mean 56%) of average10-h days (November–March; Goudie and Ankney, 1986). We eval-uated effects of daily foraging times of 6, 9, 12, and 15 h on thresh-old prey densities needed for profitable foraging.

2.5. Measurement and analyses of ice conditions

We evaluated a variety of satellite imagery for delineating openwater at spatial and temporal scales relevant to eiders. Passivemicrowave sensors, which measure only naturally occurring radia-tion, include the Advanced Microwave Scanning Radiometer-EarthObserving System (AMSR-E), Scanning Multichannel MicrowaveRadiometer (SMMR), and Special Sensor MicrowaveImager/Sounder (SSMI/S); the spatial resolution of these instru-ments is only 6.25–25 km. Digital ice charts from the NationalIce Center, which integrate a range of data types to delineate iceconcentrations (percent areal coverage) at 10% increments, areavailable every few days. However, these ice charts do not capturelead features at small scales. Synthetic Aperture Radar (SAR) hashigh spatial resolution (12 m) and, by transmitting microwavesand measuring their reflection, operates regardless of darkness orcloud cover. However, identification of leads with SAR was compli-cated by a number of factors, including the difficulty of distin-guishing sea ice from wind-disturbed open water. The AdvancedVery High Resolution Radiometer AVHRR/3 measures reflectancein 6 wavelength bands, with a number of scans of our study areaeach day at a resolution of 1.1 km. Being a passive sensor, AVHRRcannot penetrate clouds. However, for our long-term analyses,AVHRR provided the best balance of spatial resolution with tempo-ral and spatial coverage.

We downloaded AVHRR/3 scenes from NOAA’s ComprehensiveLarge Array-data Stewardship System (CLASS, www.class.noaa.gov).High Resolution Picture Transmission (HRPT) data were processedby an open-source remote sensing program (VISAT 4.11, EuropeanSpace Agency, Geesthacht, Germany). The VISAT program canprocess data from NOAA satellites 15 through 19, but not theprevious series of satellites. We found that the first two years ofspringtime data from NOAA-15 (1999 and 2000) were ofdiminished quality, so our analyses began with 2001. WithinVISAT, visible (channel 1) and thermal (channel 4) data werecalibrated to surface reflectance or brightness temperature,geometrically transformed, and radiometrically corrected.

166 J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174

We inspected each scene to identify images with relatively littlecloud cover over the study area. Optically thin and low clouds weredistinguished from a background of ice (or overlying snow) inchannel 3, and cirrus clouds by the difference between channels3 (T3) and 4 (T4) (Massom and Comiso, 1994). We eliminatedimages with >10% of their surface obscured by clouds. After pro-cessing in VISAT, images were imported as GeoTIFF files into aGIS (ArcDesktop, ESRI 10.2, Redlands, CA). They were thenre-projected into a polar stereographic coordinate system (NAD1983) and cropped to include only the study region.

To define open water, we looked for bimodal distributions inthe temperature or albedo of open water versus ice (or overlyingsnow). However, a mix of open water and ice within the same pix-els yielded continuous, rather than bimodal, averages for entirepixels. Consequently, we instead defined leads in terms of theirobserved thermal or brightness levels (Mahoney et al., 2012,2014). We calculated the proportion of open water cA within eachpixel (Lindsay and Rothrock, 1995; Mahoney et al., 2012) based onthe albedo of the entire pixel (a), of open water (aw = 0.1), and ofsurrounding thick ice (ai):

cA ¼ai � aai � aw

ð1Þ

Surrounding ice was assumed to have a surface albedo withinthe upper quartile of albedos of the entire scene. Each pixel wasdesignated as open water if cA P 0.66 (66% open water) or as iceif cA < 0.66. We selected a cA value smaller than 1.0 (i.e. a cell com-posed entirely of open water) to account for increased reflectancesdue to the presence of some floating ice or atmospheric effects suchas vapor plumes. We compared results from using a cutoff for openwater of cA P 0.66 versus cA P 0.33 (Eicken et al., 2006), and foundin the latter case that areas designated as leads were often overes-timated and included pixels with albedo values well above 0.2.

-20.0 -19.5 -19.0

10

11

12

13

14

15

Ampelisca

Anonyx

Macoma

Nuculana

Ennucula

Yoldia

P

Pectinaria

NemerteaKIEI RBC

Yoldia(Dunton)

13C only

BoreogadusEualus

15N

Fig. 2. Mean (±1 SE) stable isotope values (d13C, d15N) of lipid-extracted breast muscle o2003; of the red blood cells (RBC) of 21 female and 30 male king eiders during the pre-nesof potential prey including bivalves (Nuculana radiata, Ennucula tenuis, Yoldia spp., Macom(Ampelisca spp., Anonyx spp.), hermit crabs (Paguridae), snow crabs (Chionoecetes opili(Nemertea, mainly Rhynchocoela sp.), and Arctic cod (Boreogadus saida) from marine feewere collected near Cape Lisburne or Point Hope (Fig. 1) in August 2004 (Iken et al., 20Yoldia only). Taxa with underlined names were from the northern Bering Sea (May to earet al., 2014) whereas Arctic cod were not (Cui, 2009).

Based on 21 cloud-free scenes from three years (2003, 2008,2013), a threshold of 66% open water yielded a mean area desig-nated as leads (±SE) of 9,146 ± 1940 km2, whereas a threshold of33% yielded 11,988 ± 1824 km2 (31% higher). For a threshold of66% open water, the mean albedo of leads (±SE) was 0.10 ± 0.01,whereas for 33% the mean albedo of leads was 0.14 ± 0.01.

Resulting binary images (open water versus ice) were thenoverlaid on a map (GIS layer) of prey densities high enough to sup-port profitable foraging by eiders in the Chukchi migration corri-dor. The dispersion of eider prey can shift over several years(Lovvorn et al., 2014), so profitable feeding areas in 2010–2012may have changed over the period of ice analyses (2001–2013).Our question was, for a given prey distribution, how would thepossible range of variability in ice conditions affect access to prof-itable areas? Four segments of the corridor were delineated (Fig. 1)based qualitatively on areas of benthic faunal assemblages identi-fied by Feder et al. (1994a,b), overall oceanographic features, andthe distribution of king eiders carrying satellite transmitters(Oppel et al., 2009). Total areal extent of profitable prey densitiesaccessible through the ice in each of the four segments of the cor-ridor were calculated in ArcGIS.

3. Results

3.1. King eider diet and prey dispersion

Stable isotopes in red blood cells and muscle reflect diet overthe previous 2–4 weeks (see Section 2), and the mean length ofstay in the Chukchi migration corridor by king eiders nesting inAlaska or Canada was 21 days (Oppel et al., 2009). Turnover ofd13C can take 9% longer in breast muscle than in whole blood(Hobson and Clark, 1992a), but our blood samples were collectedlonger after passage through the migration corridor than were

-18.5 -18.0 -17.5 -17.0

Praxillella

CrangonSpirontocaris

Argis

aguridae

Nephtys

KIEI mus

Chionoecetes

13C

Bivalves

f 8 female king eiders (KIEI) collected as they migrated past Barrow (Fig. 1) in Mayting period (May–early June, 2006–2007) on breeding sites southeast of Barrow; anda spp.), polychaetes (Pectinaria hyperborea, Praxillella spp., Nephtys spp.), amphipods

o), shrimp (Crangon spp., Argis spp., Spirontocaris spp., Eualus spp.), ribbon wormsding areas. Invertebrate taxa (not lipid-extracted) whose names are not underlined10), or from Cape Lisburne to Wainwright in September 1983 (Dunton et al., 1989;ly June 2006–2007); for these samples, invertebrates were lipid-extracted (Lovvorn

Fig. 3. Abundance (no. m�2) in the eastern Chukchi Sea in late August to mid-September 2010–2012 of crabs including Oregoniidae (Chionoecetes opilio, Hyascoarctatus), Lithodidae, and Cheirogonidae; hermit crabs (Paguridae) and shrimps(Crangonidae, Pandalidae, Hippolytidae); amphipods (major taxa were Ampeliscaspp., Cheirimedeia spp., Protomedeia spp., Corophium spp., and Desdimelita des-dichada); and polychaetes including Oweniidae, Maldanidae, Capitellidae,Cirratulidae, Sabellidae, Ampharetidae, and Spionidae. Segments of the migrationcorridor used in our analyses are delineated by lines perpendicular to shore.

Fig. 4. Abundance (no. m�2) of all bivalve taxa combined (P3 mm shell length) inlate August to mid-September 2010–2012. Areas of high use (>50% kernel density)by satellite-tracked female king eiders during spring migration 1998–2008 (Oppelet al., 2009) are circumscribed. Segments of the migration corridor used in ouranalyses are delineated and labeled. The area of Barrow Canyon >85 m deep isexcluded from Skull Bay.

J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174 167

muscle samples. We did not use a mixing model (Phillips et al.,2014), because such models assume knowledge of fractionationfactors from food to eider tissue for each potential food type. Asreviewed in the Methods, the variability in such factors can begreater than the differences in d13C and d15N observed among mostpotential prey in the Chukchi Sea (Fig. 2), and we chose not tointroduce this additional error. Based on published ranges of frac-tionation (mostly 0–2‰ for d13C, 3–4‰ for d15N, see Methods), iso-topic values of potential prey relative to king eider red blood cells

(Fig. 2) indicate that bivalves were the main food during this periodwith little contribution from other taxa. Interpretation for d13C inbreast muscle is unclear, as its heavier value could have beenderived from a variety of foods (Fig. 2). However, d15N values inbreast muscle indicate that the eiders ate mostly bivalves duringthe weeks preceding their migration past Barrow.

Both amphipods and polychaetes were very abundant at somestations in the nearshore Chukchi corridor (Fig. 3). Nevertheless,king eiders appeared to favor bivalves which reached high densi-ties over large areas (as detectable by our sample spacing) at a lim-ited number of locations at the scale of the entire corridor (seebelow). Near the site where spectacled eiders were collected inthe Bering Sea in March 2009, ash-free dry mass of polychaetesavailable in the sediments was much greater than that of bivalves,but the eiders appeared not to eat polychaetes even though theeiders’ body fat had declined to critical levels (Lovvorn et al.,2014; but see Merkel et al., 2007). In the Ledyard Bay segment ofthe Chukchi migration corridor, where most use bysatellite-tracked king eiders occurred (Oppel et al., 2009), low den-sities of snow crabs (Chionoecetes opilio), other decapods, andamphipods (Fig. 3) corroborated isotopic evidence for their negligi-ble role in the eiders’ diet.

We lacked isotope data for lipid-extracted tissues from severalbivalve taxa that were relatively abundant in the nearshore corri-dor (e.g. Cyclocardia sp., Astarte borealis, Liocyma fluctuosa,Serripes groenlandicus, Megangulus lutea). Moreover, fractionationbetween prey and predator tissues often varies among speciesand within the same species among periods or areas (seeSection 2). Thus, we were uncertain which bivalves were eatenby the eiders. On their wintering area in the northern Bering Sea,spectacled eiders appeared to consume only one or two bivalvespecies that were most abundant, despite the availability of other

Bivalve energy content (Joules)200 300 400 500 600 700

Thre

shol

d de

nsity

(no.

m-2

)

200

300

400

500

6 h

9 h

15 h12 h

Fig. 5. Threshold densities (no. m�2) of all bivalve taxa combined (P3 mm shelllength) needed for profitable foraging by king eiders for time spent foraging per dayof 6, 9, 12, and 15 h. Mean hours of daylight were 17 h in April and 22 h in May.

Table 1Areal extent of the Chukchi Sea nearshore corridor studied, and area and percent oftotal area with densities of bivalves high enough for profitable foraging (>200 m�2) byking eiders in each of the four corridor segments (Fig. 4).

Segment Total area Profitable area

(km2) (km2) (%)

Skull Bay 2966 1665 56Wainwright 9412 4291 46Ledyard Bay 9673 1619 17Point Hope 8434 351 4

168 J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174

bivalve species that were suitable prey (Lovvorn et al., 2003, 2014).Bivalves in the spectacled eiders’ esophagi were 6–36 mm long, butmostly 12–30 mm. The diet of king eiders in Greenland includedabout 95% bivalves, but they consumed a wider variety of bivalvetaxa from 1 to 70 mm long, mostly 1–50 mm (Frimer, 1997). Themean shell lengths (for individuals P3 mm) of three dominantbivalve species were only 26–39% as long in the eastern ChukchiSea as in the northern Bering Sea – Nuculana spp.: 4.9 versus19.1 mm, Macoma spp.: 5.8 versus 17.3 mm, Ennucula tenuis: 5.1versus 13.1 mm. These small bivalve sizes greatly increase thethreshold densities needed for profitable foraging (Richman andLovvorn, 2003, 2009), and likely mandated opportunistic consump-tion of most bivalve taxa of a wide range of sizes. Consequently, weassumed that king eiders in the nearshore Chukchi Sea would eatall bivalve taxa with shell length P3 mm.

In the designated critical habitat for spectacled eiders (Fig. 1),which includes the Ledyard Bay segment where mostsatellite-tracked king eiders occurred, high densities (>200 m�2)of all bivalve species combined had relatively limited extent basedon the spatial scale of our sampling stations (Fig. 4). Although prof-itable prey patches probably occurred at smaller spatial scales, ourdata nevertheless indicate areas where such smaller patches werecommon and predictable.

3.2. Simulation and dispersion of threshold prey densities

We evaluated the joint effects of daily foraging times of 6, 9, 12,and 15 h, and a range of energy content per clam of 200–700 J, onthreshold prey densities needed for profitable foraging (Fig. 5).Given the small size of these bivalves and the eiders’ strong needto deposit fat stores for breeding, we assumed they would feedfor 15 h day�1 (68% of daylight hours during May when most eiderstransit the area) on bivalves with mean energy content of 590 J.These values yield a conservatively low threshold prey density of

200 m�2 (Fig. 5), which should provide upper estimates of theextent of viable habitat. For this combination of parameters, varia-tions in water depth from 10 to 40 m, and percent of non-feedingtime spent resting on ice from 0% to 80%, had negligible effect onthreshold prey density.

For each segment of the migration corridor, we calculated thetotal area of the polygon that encompassed all sampling stationsfrom 2010 to 2012 (Fig. 4). We also calculated the areal extent ofthe densities of all bivalve species combined (P3 mm) that couldsupport profitable foraging by king eiders (>200 m�2) (Fig. 4 andTable 1). The percentage of total area with profitable prey densitiesranged from 4% in the far south to 56% in the far north, with 17%and 56% in the two segments heavily used by satellite-tracked kingeiders (Ledyard and Skull Bays, respectively; Fig. 4).

3.3. Ice effects on access to profitable patches

The resolution of ice cover in our analyses can be seen in theimage for 17 April 2009 (Fig. 6). At a spatial resolution of 1.1 km,many small leads scattered throughout the pack ice were delin-eated. Our question was, for a given prey distribution (2010–2012), how would the possible range of variability in ice conditionsaffect access to profitable areas? Large differences in ice cover bothbetween years and between April and May of the same years hadmajor effects on access to profitable feeding areas through theice, with almost all such areas accessible on 22 May 2008 and onlyvery small areas accessible on 23 May 2009 (Fig. 6). When theviable (profitable) foraging area exposed by leads is averaged over2-week periods for all four segments combined, the huge variabil-ity among years belies any apparent pattern (Fig. 7). Early access tofeeding areas in late April was sometimes reversed in early May(2001, 2004, 2007), and sparse access through April was some-times followed by exceptionally large open areas by early May(2002, 2008, 2010). When the viable area exposed was averagedfor April versus May for each Chukchi segment separately, the dis-tribution of exposed feeding sites among segments also variedgreatly among months and years (Fig. 8). Although lower extentof accessible feeding areas in particular segments was often offsetby higher extent in other segments, the total area exposedthroughout the corridor was very low in both April and May 2006.

4. Discussion

During spring migration through the nearshore corridor of thenortheast Chukchi Sea, king eiders appeared to eat mostly bivalves.In 2010–2012, extensive patches with densities of bivalves highenough to support profitable foraging by king eiders occurredmainly in localized areas, even within regions of the corridor thathad received the most use by satellite-tracked king eiders in1998–2008 (Fig. 4). During April and May over 13 years (2001–2013), spatial and temporal patterns of ice cover in this corridorwere highly variable. Even if the dispersion of profitable feedingareas shifted during the 9 years before or the year after our sam-pling period, access through the ice to those areas would have var-ied appreciably. At the spatial and temporal scales of this analysis,no particular segments of the corridor stood out where profitablefeeding areas were dependably accessible in all years and months.Much of the area of Ledyard Bay that received high use by kingeiders in spring 1998–2008 did not overlap areas of profitable preydensities at the scale of our sampling in 2010. This discrepancysuggests that the spatial resolution of our sampling was inade-quate to detect all profitable patches. However, in the northernBering Sea, benthic samples at comparable spatial resolution weresufficient to detect effects of ice cover on access to prey withimportant consequences for fat levels of spectacled eiders

100 km

23 May 2009

22 May 200815 April 2008

17 April 2009

Study siteViablefeeding areaLeads

Fig. 6. Ice thickness (1.1 km resolution) in mid-April and late May in successive years with very different ice conditions (2008 and 2009), with leads (areas of open water)delineated by red lines. Areas with profitable densities of all bivalve taxa combined, interpolated from samples in late August to mid-September 2010–2012, are delineated bywhite dashed lines. Segments of the migration corridor used for analyses (Fig. 4) are delineated by black lines.

J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174 169

(Lovvorn et al., 2014). Analyses presented here suggest that inyears with ice conditions as in 2006 (Figs. 7 and 8), completingmigration in physiological condition adequate for breeding mightdepend on a few limited areas whose locations are difficult to pre-dict. Thus, for over a half million eiders of four species that use thiscorridor, maintaining a range of habitat areas throughout thisregion is an important concern in planning for industrialdevelopment.

4.1. Eider diets

Diets of both king and common eiders have often been studiedin cobble, rocky, or mixed rock and sand habitats, which differ inprey assemblages from the mostly soft sediments of the northeastChukchi Sea. In fjords of northern Norway and Greenland, diets ofthese two eider species generally included the same taxa, but indifferent proportions that appeared to reflect differences in typical

foraging depths (<10 m for common eiders, >20 m for king eiders;Bustnes and Erikstad, 1988; Bustnes and Lønne, 1997; Merkel et al.,2007). King eider diets contained a smaller mass fraction ofbivalves and a larger fraction of urchins and brittle stars than didcommon eiders, with similar fractions of polychaetes and amphi-pods if they occurred. Prey availability was not measured. Ourstudy indicates that in areas where bivalves are very abundant insoft sediments, king eiders will eat mainly bivalves while avoidingbrittle stars, polychaetes, and amphipods, as do spectacled eidersunder comparable circumstances in the northern Bering Sea duringwinter (Lovvorn et al., 2003, 2014). Common eiders using the samehabitats in the Chukchi Sea might respond similarly.

4.2. Length of stay, and fat gain versus loss

In 2006 and 2007, 17 adult female king eiders carrying satellitetransmitters were monitored as they passed through the Chukchi

Fig. 7. Temporal change in April through May of the total viable (profitable) feedingarea for king eiders that was accessible through the ice for the entire Chukchi Seamigration corridor in each year from 2001 to 2013, based on ice cover in each yearand densities of all bivalve taxa combined in August–September 2010–2012 (Fig. 4).

170 J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174

Sea nearshore corridor headed for breeding areas in Alaska (Oppelet al., 2009; S. Oppel and A.N. Powell, unpubl. data). Mean (±SD)length of stay in the corridor was 27 ± 14 days, ranging from 3 to

0

500

1000

1500

2000

2500

3000

3500

4000

4500

2001 2002 2003 2004 2005 2006

Apri

0

500

1000

1500

2000

2500

3000

3500

4000

4500

2001 2002 2003 2004 2005 2006

May

Number of cloud-free images c

Skull Bay 8 9 4 4 7 Wainwright 7 8 4 2 9 3 Ledyard Bay 7 9 4 1 9 4 Point Hope 6 8 4 1 7 1 Total 28 34 16 8 32 8

Number of cloud-free images co

Skull Bay 5 8 4 2 4 1 Wainwright 5 10 4 3 2 1 Ledyard Bay 7 10 4 4 5 2 Point Hope 5 9 4 4 4 2 Total 22 37 16 13 15 6

Viab

le a

rea

expo

sed

by le

ads

(km

2 )Vi

able

are

a ex

pose

d by

lead

s (k

m2 )

Fig. 8. Mean (±1 SE) viable (profitable) feeding area for king eiders that was accessible thMay of each year from 2001–2013.

45 days. Mean dates of arrival and departure were 4 and 25 May,respectively, with earliest arrival on 18 April and latest departureon 8 June.

How much body mass must eiders gain per day while migratingthrough the Chukchi migration corridor? For spectacled eiders, themean (±SD) body mass of 26 adult females collected on their win-tering area soon before spring migration (19–22 March 2001 and2009) was 1558 ± 96 g, ranging from 1272 to 1789 g (Lovvornet al., 2009, 2014). Farther south on the Yukon-Kuskokwim Delta,mean (±SD) body mass of 11 adult female spectacled eiders soonafter arrival was 1623 ± 153 g (Petersen et al., 2000), implying amean mass gain since leaving the wintering area of about 65 g.

No data are available on the body mass of king eiders at specificareas in late March before migration. However, mean body mass(±SD) of 60 adult females soon after arrival (18–22 June 2002–2007) at two breeding sites on the Arctic coastal plain of Alaskawas 1690 ± 222 g, range 1150–2140 (Oppel and Powell, 2009).Upon arrival, the mean body mass of adult female king eidersappears to be about 6% greater than in spectacled eiders. If weassume that adult female spectacled eiders must on average gain50 g of body fat during passage through the Chukchi migration cor-ridor, a comparable value for king eiders is 53 g. Thus, for a meanperiod of passage through the Chukchi corridor of 27 days (see

2007 2008 2009 2010 2011 2012 2013

l Skull BayWainwrightLedyard BayPoint Hope

2007 2008 2009 2010 2011 2012 2013

Skull BayWainwrightLedyard BayPoint Hope

ontributing to analyses

7 8 4 4 3 5 13 6 6 6 3 5 5 13 4 5 9 2 3 10 3 2 5 1 3 3 10 20 21 24 8 13 16 46

ntributing to analyses

8 6 9 2 4 2 10 6 9 2 4 1 15 3 10 3 1 1 10 5 9 2 43 20 37 9 9 3 1

rough the ice in each segment of the Chukchi migration corridor (Fig. 4) in April and

J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174 171

above), king eider females must on average gain about 2 g of fat perday while in this habitat. Large eiders have relatively low restingmetabolic rates in cold water (Richman and Lovvorn, 2011), andit is likely that heat produced by exercising muscles offsets muchof the heat lost to water during dives (Lovvorn, 2007). For a bodymass (Mb) of 1640 g, associated field metabolic rate (FMR) of about1227 kJ d�1 (FMR = 4.797 Mb

0.749, Nagy, 1987), and energy contentof body lipid of 37 kJ/g, king eiders would lose about 33 g of bodyfat for each day without food. Thus, the required average fat gainwhen profitable feeding areas are accessible is relatively easy toachieve (2 g per day), but days when such areas are inaccessible(loss of 33 g per day) can quickly increase the profit (and associ-ated prey density) needed on other days when feeding is possible.

4.3. Drivers of prey dispersion

In the vicinity of Ledyard Bay (Fig. 1), a southward-flowingcountercurrent often develops inshore of the predominantnorthward-flowing Alaska Coastal Current (Weingartner, 1997;Winsor and Chapman, 2004). This countercurrent with associatededdies, and possibly eddies generated by northward flow pastCape Lisburne, can create various depositional regions where finesediments, organic matter, pelagic larvae of benthic invertebrates,and re-suspended post-settlement juveniles can be retained (cf.Fig. 4). The size and location of depositional areas will vary withchanges in the strength and direction of wind, the associatedstrength of the main current and its countercurrent, and the partic-ular taxon of interest (Weingartner et al., 2013; Lovvorn et al.,2014; Kolts et al., 2015). In the Wainwright sector north of IcyCape, the area of high prey density (Fig. 4) may correspond to ahydrographic front that was previously identified as contributingto high benthic densities (Feder et al., 1994a,b). In the outer regionof Skull Bay, upwelling and other processes along the edge ofBarrow Canyon may explain areas of high benthic density in thatsegment (Figs. 3 and 4). Throughout this region, time series datafrom stations with comparable or finer spacing are needed toassess the consistency of high-density patches that might resultfrom these processes.

4.4. Effects of ice scour in the nearshore zone

For three dominant bivalve species, we were able to comparemean shell lengths in the nearshore Chukchi Sea with those forthe same taxa in the northern Bering Sea. Percentages of total num-bers of bivalves (P3 mm long) in our Chukchi Sea study area were29% for Ennucula tenuis, 22% for Macoma spp., and 3% for Nuculanaspp. For these taxa, shell lengths were only 39%, 34%, and 26%,respectively, of those of the same species at depths of 30–96 min the northern Bering Sea (see Lovvorn et al., 2003, 2014).Growth rates of bivalves are not expected to be lower in the north-east Chukchi Sea, where bottom water temperatures during sum-mer are up to 8 �C warmer than in areas where we collectedbivalves in the northern Bering Sea (cf. Weingartner, 1997; Lauth,2011). Samples in late August (Chukchi) might include more small,recently recruited bivalves than samples taken following winter inMay (Bering). However, mean shell length of 2-year-old Macomacalcarea in the Bering and Chukchi Seas was 3.9 mm (range 3.4–4.3 among areas; Stoker, 1978), so our exclusion of bivalves<3 mm probably eliminated most young-of-the-year recruits. Thelower mean size of the same bivalve species in the Chukchi Seawas driven mainly by the scarcity of larger specimens.

Thus, in the shallow nearshore zone of the Chukchi Sea, itappears that regular mortality events shift the population structureto younger, smaller bivalves. Factors such as foraging by walrusesor gray whales (Oliver et al., 1983; Klaus et al., 1990; Born et al.,2003), or disturbance of sediments by intense storm events

(Harris and Hughes, 2012), might contribute to this effect.However, we believe the main factor is ice scour, which is knownto have major impacts on survival, abundance, and age structureof benthic communities in shallow polar habitats (Conlan andKvitek, 2005; Barnes and Souster, 2011). Based on studies in the1970s, extensive ice gouging of sediments to depths of 1–2 moccurs regularly throughout the nearshore zone of the northeastChukchi Sea (Toimil, 1978; Grantz et al., 1982; BOEM, 2014). Themagnitude of this effect can increase strongly with decreasingduration of the fast ice period, when ice is locked into place ratherthan drifting more freely (Barnes and Souster, 2011). Thus, aswarming climate reduces the period of fast ice, increased move-ments of unconsolidated ice may further reduce the size and bio-mass of nearshore bivalves both here and elsewhere in polarregions (Smale et al., 2008).

Ice scour probably also increases spatial heterogeneity andreduces the scale of benthic patchiness in the Chukchi migrationcorridor. Patches created by ice scour at scales of tens to hundredsof meters (Toimil, 1978; Conlan and Kvitek, 2005) may confoundour mapping of prey densities based on sampling stations severalkilometers or more apart (Fig. 1). The limited extent of high preydensities in the area of heavy use by migrating king eiders inLedyard Bay (Fig. 4) suggests that profitable feeding areas existat scales smaller than our sampling. However, high spatial hetero-geneity might greatly increase the costs of finding those patches,especially within variable ice cover. Our maps indicate areas wherethe probability of locating accessible prey patches is predictablyhigh. Sampling to characterize smaller-scale patch structure oversuch a large area would be quite challenging.

4.5. Consequences of prey inaccessibility

What are the effects on eider populations when access to prof-itable feeding areas is low, as in April 2005 or in April and May2006 (Figs. 7 and 8)? In the nearshore Beaufort Sea, mortality oftens of thousands of migrating king eiders has occurred when openwater in spring staging areas re-froze (Barry, 1968; Fournier andHines, 1994). Although such massive mortality events have notbeen reported along the Chukchi Sea coast, smaller die-offs haveoccurred such as for king eiders near Barrow in May 2011. Low per-centages of hatching-year males (3–5%) among spectacled eiders inlate winter (Lovvorn et al., 2012) indicate that either first-year sur-vival is low, adult survival is high, or both. For king eiderssatellite-tagged on breeding areas, annual survival rate ofhatch-year birds was 0.67 and of adults 0.94 (Oppel and Powell,2010). For these species with typically long lives and delayedbreeding, instances of high mortality could have important effectson populations that use the Chukchi migration corridor in thespring.

Even without direct mortality, decreased fat stores caused byexclusion from suitable feeding areas during spring migrationmay have important effects on reproduction (Lehikoinen et al.,2006). Both king and spectacled eiders rely heavily on storedreserves during egg-laying and incubation (eggs are incubated onlyby females). On the Yukon-Kuskokwim Delta, spectacled eiderfemales lost about 33% of body mass (530 g) from arrival to hatch-ing (Petersen et al., 2000); with 90% incubation constancy, they lostabout 26% of body mass from beginning to end of incubation (Flintand Grand, 1999). King eider females rely on both stored reservesand local foods for egg-laying and incubation (Bentzen et al., 2008;Oppel et al., 2010), but still lost 26–35% of body mass (425–631 g)from arrival to late incubation (Bentzen et al., 2010). Incubatingking eiders spent only 1–3% of their time away from the nest to for-age (Bentzen et al., 2010), and at another site lost about 30% ofbody mass (482 g) during incubation alone (Kellett andAlisauskas, 2000). In common eiders, propensity to breed, clutch

172 J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174

size, nest success, brood attendance by hens, and resulting survivalof ducklings all depend on adequate prebreeding reserves(Coulson, 1984; Bustnes and Erikstad, 1991; Erikstad et al., 1993;Bustnes et al., 2002; Oosterhuis and van Dijk, 2002).

Given that fat reserves of spectacled eiders at the end of wintercan be quite low (Lovvorn et al., 2014), and the period betweendeparture from the wintering area and arrival at Arctic breedingareas may be up to 8 weeks (Lovvorn et al., 2003), the ability ofspectacled eiders to feed profitably during migration and stagingin the Chukchi Sea appears critical. For king eiders, data are lackingon late-winter body mass at specific locations, and some individu-als may spend about 2 weeks staging in the eastern Beaufort Seaafter migrating through the Chukchi corridor and before movingonto nesting sites (Phillips et al., 2007). Although better informa-tion on patterns of body mass and reserves of king eiders duringmigration is needed, it is likely that food availability in theChukchi Sea migration corridor is quite important to the propen-sity to breed and breeding success of both king and spectacledeiders.

Based on model estimates of threshold prey densities neededfor profitable foraging, maps of prey densities, and estimates ofthe numbers and lengths of stay of various eider species, one mightattempt to estimate the carrying capacity of the Chukchi migrationcorridor for eiders under given ice conditions (cf. Blicher et al.,2011). However, such estimates must also account for consump-tion of the same prey by invertebrates, fish, and perhaps increasingnumbers of walruses in this area. Moreover, estimates of thresholdprey densities based on energetics models often substantiallyunderestimate the giving-up densities at which predators actuallyabandon a particular feeding area (Lovvorn et al., 2013). As a result,such calculations might seriously underestimate the area and qual-ity of foraging habitat needed.

4.6. Potential effects of declining ice cover on the benthos

South of St. Lawrence Island in the north-central Bering Sea, themajor species of bivalve prey vary greatly in abundance and dis-persion among decades and even shorter periods (Lovvorn et al.,2009, 2014). Shifts in temperature, sediment characteristics, orother conditions for larval dispersal, settlement, or survival mayfavor one species over others at relatively small scales. In the shal-low Chukchi corridor, greater fetch and wave action in areas thatare increasingly ice-free in the summer (Overeem et al., 2011)might increase grain size and decrease the organic content of sed-iments, with resulting effects on biotic communities.

A potentially important indirect effect of decreased ice cover isthe need for walruses to haul out on land when the summer icepack recedes beyond depths accessible to foraging walruses(<100 m). Walruses, especially females with young, require seaice as a platform for resting, and with receding ice in recent yearshave hauled out by the tens of thousands on shorelines of thenortheast Chukchi Sea. Satellite telemetry suggests that some ofthe walruses, which have very large food requirements (Bornet al., 2003), fed within the nearshore migration corridor (Jayet al., 2012). It is unknown what these walruses were eating.Studies in other areas have shown walruses to eat mainly bivalvesof species or sizes that are usually not eaten by eiders, and that areoften buried deeper in the sediments than our van Veen grab pen-etrates (>15 cm, Oliver et al., 1983; Fay et al., 1984; Born et al.,2003). However, even if walruses do not compete directly witheiders for the same taxa or sizes of prey, massive disturbance ofsediments by walruses may affect the diversity, abundance, anddispersion of eider foods (Oliver et al., 1985).

4.7. Implications for industrial development

If oil and gas development proceeds in the Chukchi Sea, currentplans call for merging pipelines from various platforms into a sin-gle pipeline, which will run underneath the sea floor across thenearshore zone and then overland to connect with the existingpipeline from Prudhoe Bay to Valdez, Alaska (BOEM, 2014). Basedon studies in the 1970s, the nearshore zone along this entire coast-line is regularly gouged by drifting ice in the spring, as pressureridges that form along the outer edge of landfast ice break upand drag their keels through the sediments (Grantz et al., 1982).Most fresh gouges occurred at water depths of 6–30 m, with den-sities of 100 to 300 gouges per km2 (Toimil, 1978). Gouges weremostly 1–2 m deep (maximum 4.5 m) and 10–20 m wide (maxi-mum �100 m). In this region, a pipeline must be buried at least6 m beneath the sea floor to avoid ice gouging, often in bedrockwhich is commonly exposed or <3 m beneath the sediment surface(Grantz et al., 1982). As local currents that commonly reversecourse could carry spilled oil in various directions (Weingartneret al., 2013), locating a pipeline away from important habitatswould be little hedge against catastrophic spills. However, spatialplanning might be effective in the case of minor spills that couldbe contained in the vicinity, once adequate technology is devel-oped (NRC, 2014). Moreover, rapid growth of shipping traffic inthe Russian portion of the Chukchi Sea (Arctic Council, 2013)may also increase the risk of spills, and the need for advance prior-itization of areas to protect during emergency responses (Eickenet al., 2011).

From the perspective of migrating eiders, our data suggest thatthere may be areas that consistently support prey densities ade-quate for profitable foraging. However, the locations of such areasmay change over time (Lovvorn et al., 2009, 2014), and time seriessamples are needed at stations spaced closely enough to detectsuch areas and their spatial variations. Even if certain feeding areasare relatively stable, access to such areas through the ice can behighly variable and sometimes quite limited. Consequently, arange of areas likely to support adequate prey densities shouldbe maintained throughout the migration corridor. Improved mod-els of nearshore hydrography, which largely controls accumulationof organic matter and dispersing invertebrates, may help in pro-jecting the location, extent, and longevity of profitable feedingareas (Young et al., 1998; Hunt et al., 2009). Suchphysical-biological linkages might be especially useful in anticipat-ing how climate change will affect the consequences of industrialdevelopment.

Acknowledgements

We thank the Captains and crews of the R/V Norseman II forexcellent logistical support. We also thank Max K. Hoberg of theInstitute of Marine Science at the University of Alaska Fairbanksfor his laboratory analyses (taxonomy, abundance, biomass, andbivalve size) of the benthic samples from 2010 to 2012. This studyis part of the Synthesis of Arctic Research (SOAR) and was fundedin part by the U.S. Department of the Interior, Bureau of OceanEnergy Management, Environmental Studies Program throughInteragency Agreement No. M11PG00034 with the U.S.Department of Commerce, National Oceanic and AtmosphericAdministration (NOAA), Office of Oceanic and AtmosphericResearch (OAR), Pacific Marine Environmental Laboratory (PMEL).Additional funding was provided by the National ScienceFoundation’s program in Arctic Science, Engineering andEducation for Sustainability Grant 1263051 to JRL and M. L.Brooks, and North Pacific Research Board Grant 820 to JRL (this isNPRB Publication #521). Funding for benthic field sampling andanalysis was provided by the Alaska Monitoring and Assessment

J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174 173

Program (AKMAP) and the Coastal Impact Assistance Program ofthe U.S. Fish and Wildlife Service. Critical support for analyzinginvertebrate samples was facilitated by C.V. Jay through the U.S.Geological Survey and M. Macrander through Shell Alaska. Anyuse of trade, firm, or product names is for descriptive purposesonly and does not imply endorsement by the U.S. Government.

References

AMAP (Arctic Monitoring and Assessment Programme), 2007. Arctic Oil and Gas2007. Arctic Monitoring and Assessment Programme, Oslo, Norway.

Arctic Council, 2013. Arctic Resilience Interim Report 2013. Environment Instituteand Stockholm Resilience Centre, Stockholm, Sweden.

Bajzak, C.E., Hammill, M.O., Stenson, G.B., Prinsenberg, S., 2011. Drifting away:implications of changes in ice conditions for a pack-ice-breeding phocid, theharp seal (Pagophilus groenlandicus). Canadian Journal of Zoology 89, 1050–1062.

Barber, D.G., Iacozza, J., 2004. Historical analysis of sea ice conditions in M’ClintockChannel and the Gulf of Boothia, Nunavut: implications for ringed seal andpolar bear habitat. Arctic 57, 1–14.

Barnes, D.K.A., Souster, T., 2011. Reduced survival of Antarctic benthos linked toclimate-induced iceberg scouring. Nature Climate Change 1, 365–368.

Barry, R.G., Moritz, R.E., Rogers, J.C., 1979. The fast ice regimes of the Beaufort andChukchi Sea coasts, Alaska. Cold Regions Science and Technology 1, 129–152.

Barry, T.W., 1968. Observations on natural mortality and native use of eider ducksalong the Beaufort Sea coast. Canadian Field-Naturalist 82, 141–144.

Bentzen, R.L., Powell, A.N., Williams, T.D., Kitaysky, A.S., 2008. Characterizing thenutritional strategy of king eiders Somateria spectabilis in northern Alaska.Journal of Avian Biology 39, 683–690.

Bentzen, R.L., Powell, A.N., Phillips, L.M., Suydam, R.S., 2010. Incubation behavior ofking eiders on the coastal plain of northern Alaska. Polar Biology 33, 1075–1082.

Blicher, M.E., Rasmussen, L.M., Sejr, M.K., Merkel, F.R., Rysgaard, S., 2011.Abundance and energy requirements of eiders (Somateria spp.) suggest highpredation pressure on macrobenthic fauna in a key wintering habitat in SWGreenland. Polar Biology 34, 1105–1116.

BOEM (Bureau of Ocean Energy Management), 2014. Chukchi Sea Planning Area, Oiland Gas Lease Sale 193 in the Chukchi Sea, Alaska. Draft Second SupplementalEnvironmental Impact Statement. OCS EIS/EA BOEM 2014-653. U.S. Departmentof the Interior, Bureau of Ocean Energy Management. Office of Environment,Alaska OCS Region.

Born, E.W., Rysgaard, S., Ehlmé, G., Sejr, M., Acquarone, M., Levermann, N., 2003.Underwater observations of foraging free-living Atlantic walruses (Odobenusrosmarus rosmarus) and estimates of their food consumption. Polar Biology 26,348–357.

Braund, S.R., 1993. North Slope subsistence study, Wainwright, 1988 and 1989.Technical Report 147, Outer Continental Shelf Study MMS 91-0073.

Brigham, L., 2011. Marine protection in the Arctic cannot wait. Nature 478, 157.Bustnes, J.O., Erikstad, K.E., 1988. The diets of sympatric wintering populations of

common eider Somateria mollissima and king eider S. spectabilis in northernNorway. Ornis Fennica 65, 163–168.

Bustnes, J.O., Erikstad, K.E., 1991. Parental care in the common eider (Somateriamollissima): factors affecting abandonment and adoption of young. CanadianJournal of Zoology 69, 1538–1545.

Bustnes, J.O., Lønne, O.J., 1997. Habitat partitioning among sympatric winteringcommon eiders Somateria mollissima and king eiders Somateria spectabilis. Ibis139, 549–554.

Bustnes, J.O., Erikstad, K.E., Bjorn, T.H., 2002. Body condition and broodabandonment in common eiders breeding in the High Arctic. Waterbirds 25,63–66.

Byers, T., Dickson, D.L., 2001. Spring migration and subsistence hunting of king andcommon eiders at Holman, Northwest Territories, 1996–98. Arctic 54, 122–134.

Carleton, S.A., Martinez del Rio, C., 2005. The effect of cold-induced increasedmetabolic rate on the rate of 13C and 15N incorporation in house sparrows(Passer domesticus). Oecologia 144, 226–232.

Cherel, Y., Hobson, K.A., Hassani, S., 2005. Isotopic discrimination between food andblood and feathers of captive penguins: implications for dietary studies in thewild. Physiological and Biochemical Zoology 78, 106–115.

Christiansen, J.S., Mecklenburg, C.W., Karamushko, O.V., 2014. Arctic marinefishes and their fisheries in light of global change. Global Change Biology 20,352–359.

Conlan, K.E., Kvitek, R.G., 2005. Recolonization of soft-sediment ice scours on anexposed Arctic coast. Marine Ecology Progress Series 286, 21–42.

Coulson, J.C., 1984. The population dynamics of the eider duck Somateria mollissimaand evidence of extensive non-breeding by adult ducks. Ibis 126, 525–543.

Cui. X., 2009. Climate-driven Impacts of Groundfish on Food Webs in the NorthernBering Sea. Ph.D. Thesis, University of Tennessee, Knoxville.

Dunton, K.H., Saupe, S.M., Golikov, A.N., Schell, D.M., Schonberg, S.V., 1989. Trophicrelationships and isotopic gradients among Arctic and subarctic marine fauna.Marine Ecology Progress Series 56, 89–97.

Eicken, H., Shapiro, L.H., Gaylord, A.G., Mahoney, A., Cotter, P.W., 2006. Mapping andCharacterization of Recurring Spring Leads and Landfast Ice in the Beaufort andChukchi Seas. Final Report Outer Continental Shelf Study MMS 2005-0678.

Eicken, H., Ritchie, L.A., Barlau, A., 2011. The role of local and indigenous knowledgein Arctic offshore oil and gas development, environmental hazard mitigation,and emergency response. In: Lovecraft, A.L., Eicken, H. (Eds.), North by 2020.University of Alaska Press, Fairbanks, pp. 577–603.

Erikstad, K.E., Bustnes, J.O., Moum, T., 1993. Clutch-size determination in precocialbirds: a study of the common eider. Auk 110, 623–628.

Evans Ogden, L.J., Hobson, K.A., Lank, D.B., 2004. Blood isotopic (d13C and d15N)turnover and diet-tissue fractionation factors in captive dunlin (Calidris alpinapacifica). Auk 121, 170–177.

Fay, F.H., Bukhtiyarov, Y.A., Stoker, S.W., Shults, L.M., 1984. Foods of the Pacificwalrus in winter and spring in the Bering Sea. In: Fay, F.H., Fedoseev, G.A. (Ed.),Soviet-American Cooperative Research on Marine Mammals, Volume 1 –Pinnipeds. NOAA Technical Report NMFS 12, pp. 81–88.

Feder, H.M., Foster, N.R., Jewett, S.C., Weingartner, T.J., Baxter, R., 1994a. Mollusks inthe northeastern Chukchi Sea. Arctic 47, 145–163.

Feder, H.M., Naidu, A.S., Jewett, S.C., Hameedi, J.M., Johnson, W.R., Whitledge, T.E.,1994b. The northeastern Chukchi Sea: benthos-environmental interactions.Marine Ecology Progress Series 111, 171–190.

Federer, R.N., Hollmén, T.E., Esler, D., Wooller, M.J., Wang, S.W., 2010. Stable carbonand nitrogen isotope discrimination factors from diet to blood plasma, cellularblood, feathers, and adipose tissue fatty acids in spectacled eiders (Somateriafischeri). Canadian Journal of Zoology 88, 866–874.

Flint, P.L., Grand, J.B., 1999. Incubation behavior of spectacled eiders on the Yukon-Kuskokwim Delta, Alaska. Condor 101, 413–416.

Fournier, M.A., Hines, J.E., 1994. Effects of starvation on muscle and organ mass ofking eiders Somateria spectabilis and the ecological and managementimplications. Wildfowl 45, 188–197.

Frimer, O., 1997. Diet of moulting king eiders Somateria spectabilis at Disko Island,West Greenland. Ornis Fennica 74, 187–194.

George, J.C., Huntington, H.P., Brewster, K., Eicken, H., Norton, D.W., Glenn, R., 2004.Observations on shorefast ice dynamics in Arctic Alaska and the responses ofthe Inupiat hunting community. Arctic 57, 363–374.

Goudie, R.I., Ankney, C.D., 1986. Body size, activity budgets, and diets of sea duckswintering in Newfoundland. Ecology 67, 1475–1482.

Grantz, A., Dinter, D.A., Hill, E.R., Hunter, R.E., May, S.D., McMullin, R.H., Phillips, R.L.,1982. Geologic Framework, Hydrocarbon Potential, and EnvironmentalConditions for Exploration and Development of Proposed Oil and Gas LeaseSale 85 in the Central and Northern Chukchi Sea. U.S. Geological Survey Open-File Report 82-1053.

Haramis, G.M., Jorde, D.G., Macko, S.A., Walker, J.L., 2001. Stable-isotope analysis ofcanvasback winter diet in upper Chesapeake Bay. Auk 118, 1008–1017.

Harris, P.T., Hughes, M.G., 2012. Predicted benthic disturbance regimes on theAustralian continental shelf: a modeling approach. Marine Ecology ProgressSeries 449, 13–25.

Heide-Jørgensen, M.P., Laidre, K., 2004. Declining extent of open-water refugia fortop predators in Baffin Bay and adjacent waters. Ambio 33, 487–494.

Hobson, K.A., Clark, R.G., 1992a. Assessing avian diets using stable isotopes I:turnover of 13C in tissues. Condor 94, 181–188.

Hobson, K.A., Clark, R.G., 1992b. Assessing avian diets using stable isotopes II:factors influencing diet-tissue fractionation. Condor 94, 189–197.

Hobson, K.A., Alisauskas, R.T., Clark, R.G., 1993. Stable-nitrogen isotope enrichmentin avian tissues due to fasting and nutritional stress: implications for isotopicanalyses of diet. Condor 95, 388–394.

Hunt, H.L., Fugate, D.C., Chant, R.J., 2009. Modeling bedload transport of juvenilebivalves: predicted changes in distribution and scale of postlarval dispersal.Estuaries and Coasts 32, 1090–1102.

Iken, K., Bluhm, B., Dunton, K., 2010. Benthic food-web structure under differingwater mass properties in the southern Chukchi Sea. Deep-Sea Research II 57,71–85.

Jay, C.V., Fischbach, A.S., Kochnev, A.A., 2012. Walrus areas of use in the Chukchi Seaduring sparse sea ice cover. Marine Ecology Progress Series 468, 1–13.

Jay, C.V., Grebmeier, J.M., Fischbach, A.S., McDonald, T.L., Cooper, L.W., Hornsby, F.,2014. Pacific walrus (Odobenus rosmarus divergens) resource selection in thenorthern Bering Sea. PLoS One 9 (4), e93035.

Kellett, D.K., Alisauskas, R.T., 2000. Body-mass dynamics of king eiders duringincubation. Auk 117, 812–817.

Klaus, A.D., Oliver, J.S., Kvitek, R.G., 1990. The effects of gray whale, walrus, and icegouging disturbance on benthic communities in the Bering Sea and Chukchi Sea,Alaska. National Geographic Research 6, 470–484.

Kolts, J.M., Lovvorn, J.R., North, C.A., Janout, M.A., 2015. Oceanographic anddemographic mechanisms affecting population structure of snow crabs(Chionoecetes opilio) in cold waters of the northern Bering Sea. MarineEcology Progress Series 518, 193–208.

Lauth, R.R., 2011. Results of the 2010 Eastern and Northern Bering Sea ContinentalShelf Bottom Trawl Survey of Groundfish and Invertebrate Fauna. U.S.Department of Commerce, NOAA Technical Memorandum NMFSAFSC-227.

Lehikoinen, A., Kilpi, M., Öst, M., 2006. Winter climate affects subsequent breedingsuccess of common eiders. Global Change Biology 12, 1355–1365.

Lindsay, R.W., Rothrock, D.A., 1995. Arctic sea ice leads from advanced very highresolution radiometer images. Journal of Geophysical Research 100, 4533–4544.

Lovvorn, J.R., Richman, S.E., Grebmeier, J.M., Cooper, L.W., 2003. Diet and bodycondition of spectacled eiders wintering in pack ice of the Bering Sea. PolarBiology 26, 259–267.

Lovvorn, J.R., Grebmeier, J.M., Cooper, L.W., Bump, J.K., Richman, S.E., 2009.Modeling marine protected areas for threatened eiders in a climaticallychanging Bering Sea. Ecological Applications 19, 1596–1613.

174 J.R. Lovvorn et al. / Progress in Oceanography 136 (2015) 162–174

Lovvorn, J.R., Mossotti, R.H., Wilson, J.J., McKay, D., 2012. Eiders in offshore pack iceshow previously unknown courtship behavior: acceleration of readiness for aconstrained breeding period? Polar Biology 35, 1087–1095.

Lovvorn, J.R., De La Cruz, S.E.W., Takekawa, J.Y., Shaskey, L.E., Richman, S.E., 2013.Niche overlap, threshold food densities, and limits to prey depletion for a divingduck assemblage in an estuarine bay. Marine Ecology Progress Series 476, 251–268.

Lovvorn, J.R., Anderson, E.M., Rocha, A.R., Larned, W.W., Grebmeier, J.M., Cooper,L.W., Kolts, J.M., North, C.A., 2014. Variable wind, pack ice, and prey dispersionaffect the long-term adequacy of protected areas for an Arctic sea duck.Ecological Applications 24, 396–412.

Lovvorn, J.R., 2007. Thermal substitution and aerobic efficiency: measuring andpredicting effects of heat balance on endotherm diving energetics. PhilosophicalTransactions of the Royal Society of London B 362, 2079–2093.

Mahoney, A.R., Eicken, H., Shapiro, L.H., Gens, R., Heinrichs, T., Meyer, F.J., Gaylord,A.G., 2012. Mapping and Characterization of Recurring Spring Leads andLandfast Ice in the Beaufort and Chukchi Seas. U.S. Department of the Interior,Bureau of Ocean Energy Management, Alaska Region, Anchorage, AK. OuterContinental Shelf Study BOEM 2012-067, 179 pp.

Mahoney, A.R., Eicken, H., Gaylord, A.G., Gens, R., 2014. Landfast sea ice extent in theChukchi and Beaufort Seas: the annual cycle and decadal variability. ColdRegions Science and Technology 103, 41–56.

Massom, R., Comiso, J.C., 1994. The classification of Arctic sea ice types and thedetermination of surface temperature using Advanced Very High ResolutionRadiometer data. Journal of Geophysical Research 99, 5201–5218.

Merkel, F.R., Mosbech, A., Jamieson, S.E., Falk, K., 2007. The diet of king eiderswintering in Nuuk, Southwest Greenland, with reference to sympatricwintering common eiders. Polar Biology 30, 1593–1597.

Mizutani, H., Kabaya, Y., Wada, E., 1991. Nitrogen and carbon isotope compositionsrelate linearly in cormorant tissues and its diet. Isotopenpraxis 27, 166–168.

Nagy, K.A., 1987. Field metabolic rate and food requirement scaling in mammalsand birds. Ecological Monographs 57, 111–128.

Nelson, C.H., Johnson, K.R., Barber, J.H., 1987. Gray whale and walrus feedingexcavation on the Bering shelf, Alaska. Journal of Sedimentary Petrology 57,419–430.

North, C.A., Lovvorn, J.R., Kolts, J.M., Brooks, M.L., Cooper, L.W., Grebmeier, J.M.,2014. Deposit-feeder diets in the Bering Sea: potential effects of climatic loss ofsea ice-related microalgal blooms. Ecological Applications 24, 1525–1542.

NRC (National Research Council), 2014. Responding to Oil Spills in the U.S. ArcticMarine Environment. National Academy Press, Washington, DC, USA.

Oliver, J.S., Slattery, P.N., O’Connor, E.F., Lowry, L.F., 1983. Walrus, Odobenusrosmarus, feeding in the Bering Sea: a benthic perspective. Fishery Bulletin 81,501–512.

Oliver, J.S., Kvitek, R.G., Slattery, P.N., 1985. Walrus feeding disturbance: scavenginghabits and recolonization of the Bering Sea benthos. Journal of ExperimentalMarine Biology and Ecology 91, 233–246.

Oosterhuis, R., Van Dijk, K., 2002. Effect of food shortage on the reproductive outputof common eiders Somateria mollissima breeding at Griend (Wadden Sea).Atlantic Seabirds 4, 29–38.

Oppel, S., Powell, A.N., 2009. Does winter region affect spring arrival time and bodymass of king eiders in northern Alaska? Polar Biology 32, 1203–1209.

Oppel, S., Powell, A.N., 2010. Age-specific survival estimates of king eiders derivedfrom satellite telemetry. Condor 112, 323–330.

Oppel, S., Dickson, D.L., Powell, A.N., 2009. International importance of the easternChukchi Sea as a staging area for migrating king eiders. Polar Biology 32,775–783.

Oppel, S., Powell, A.N., O’Brien, D.M., 2010. King eiders use an income strategy foregg production: a case study for incorporating individual dietary variation intonutrient allocation research. Oecologia 164, 1–12.

Overeem, I., Anderson, R.S., Wobus, C.W., Clow, G.D., Urban, F.E., Matell, N., 2011.Sea ice loss enhances wave action at the Arctic coast. Geophysical ResearchLetters 38, L17503. http://dx.doi.org/10.1029/2011GL048681.

Paszkowski, C.A., Gingras, B.A., Wilcox, K., Klatt, P.H., Tonn, W.M., 2004. Trophicrelations of the red-necked grebe on lakes in the western boreal forest: a stableisotope analysis. Condor 106, 638–651.

Petersen, M.R., Grand, J.B., Dau, C.P., 2000. Spectacled Eider (Somateria fischeri). TheBirds of North America, No. 547, Philadelphia, PA, USA.

Petersen, M.R., 2009. Multiple spring migration strategies in a population of Pacificcommon eiders. Condor 111, 59–70.

Phillips, L.M., Powell, A.N., Taylor, E.J., Rexstad, E.A., 2007. Use of the Beaufort Sea byking eiders breeding on the North Slope of Alaska. Journal of WildlifeManagement 71, 1892–1898.

Phillips, D.L., Inger, R., Bearhop, S., Jackson, A.L., Moore, J.W., Parnell, A.C., Semmens,B.X., Ward, E.J., 2014. Best practices for use of stable isotope mixing models infood web studies. Canadian Journal of Zoology 92, 823–835.

Podlesak, D.W., McWilliams, S.R., 2007. Metabolic routing of dietary nutrients inbirds: effects of dietary lipid concentration on d13C of depot fat and itsecological implications. Auk 124, 916–925.

Post, D.M., Layman, C.A., Arrington, D.A., Takimoto, G., Quattrochi, J., Montaña, C.G.,2007. Getting to the fat of the matter: models, methods and assumptions fordealing with lipids in stable isotope analyses. Oecologia 152, 179–189.

Quakenbush, L.T., Suydam, R.S., Acker, R., Knoche, M., Citta, J., 2009. Migration ofKing and Common Eiders Past Point Barrow, Alaska, during Summer/Fall 2002through Spring 2004: Population Trends and Effects of Wind. Final Rep, OuterContinental Shelf Study MMS 2009-036, Fairbanks, AK.

Reeves, R., Rosa, C., George, J.C., Sheffield, G., Moore, M., 2012. Implications of Arcticindustrial growth and strategies to mitigate future vessel and fishing gearimpacts on bowhead whales. Marine Policy 36, 454–462.

Richman, S.E., Lovvorn, J.R., 2003. Effects of clam species dominance on nutrient andenergy acquisition by spectacled eiders in the Bering Sea. Marine EcologyProgress Series 261, 283–297.

Richman, S.E., Lovvorn, J.R., 2004. Relative foraging value to lesser scaup ducks ofnative and exotic clams from San Francisco Bay. Ecological Applications 14,1217–1231.

Richman, S.E., Lovvorn, J.R., 2009. Predator size, prey size, and threshold fooddensities of diving ducks: does a common prey base support fewer largeanimals? Journal of Animal Ecology 78, 1033–1042.

Richman, S.E., Lovvorn, J.R., 2011. Effects of air and water temperatures on restingmetabolism of auklets and other diving birds. Physiological and BiochemicalZoology 84, 316–332.

Sexson, M.G., Pearce, J.M., Peterson, M.R., 2014. Spatiotemporal Distribution andMigratory Patterns of Spectacled Eiders. BOEM 2014-665, Bureau of OceanEnergy Management, Alaska Outer Continental Shelf Region, Anchorage, Alaska.

Smale, D.A., Brown, K.M., Barnes, D.K.A., Fraser, K.P.P., Clarke, A., 2008. Ice scourdisturbance in Antarctic waters. Science 321, 371.

Smith, L.C., Stephenson, S.R., 2013. New trans-Arctic shipping routes navigable bymidcentury. Proceedings of the National Academy of Sciences of the UnitedStates 110, E1191–E1195.

Speer, L., Loughlin, T.L., 2011. Workshop Report: IUCN/NRDC Workshop toIdentify Areas of Ecological and Biological Significance or Vulnerabilityin the Arctic Marine Environment. <http://www.iucn.org/about/work/programmes/marine/marine_resources/?7470/IUCNNRDC-Workshop-to-Identify-Areas-of-Ecological-and-Biological-Significance-or-Vulnerability-in-the-Arctic-Marine-Environment>.

Stehn, R.A., Larned, W.W., Platte, R.M., 2013. Analysis of Aerial Survey IndicesMonitoring the Waterbird Populations of the Arctic Coastal Plain, Alaska, 1986–2012. U.S. Fish and Wildlife Service, Anchorage, AK.

Stoker, S.W., 1978. Benthic Invertebrate Macrofauna of the Eastern ContinentalShelf of the Bering and Chukchi Sea. Ph.D. Thesis, Universityof Alaska, Fairbanks.

Suydam, R.S., Dickson, D.L., Fadely, J.B., Quakenbush, L.T., 2000. Population declinesof king and common eiders of the Beaufort Sea. Condor 102, 219–222.

Systad, G.H., Bustnes, J.O., Erikstad, K.E., 2000. Behavioral responses to decreasingday length in wintering sea ducks. Auk 117, 33–40.

Toimil, L.J., 1978. Ice-gouged Microrelief onthe Floor of the Eastern Chukchi Sea, Alaska:A Reconnaissance Survey. U.S. Geological Survey Open-File Report 78-693.

Weingartner, T.J., Winsor, P., Potter, R.A., Statscewich, H., Dobbins, E.L., 2013.Application of high frequency radar to potential hydrocarbon developmentareas in the Northeast Chukchi Sea. Final Report, Outer Continental Shelf StudyBOEM 2012-079, Fairbanks, AK. 162 pp.

Weingartner, T.J., 1997. A review of the physical oceanography of the northeasternChukchi Sea. American Fisheries Society Symposium 19, 40–59.

Wenzel, G.W., 2009. Canadian Inuit subsistence and ecological instability – if theclimate changes, must the Inuit? Polar Research 28, 89–99.

Winsor, P., Chapman, D.C., 2004. Pathways of Pacific water across the Chukchi Sea: anumerical model study. Journal of Geophysical Research 109, C03002. http://dx.doi.org/10.1029/2003JC001962.

Young, E.F., Bigg, G.R., Grant, A., Walker, P., Brown, J., 1998. A modelling study ofenvironmental influences on bivalve settlement in The Wash, England. MarineEcology Progress Series 172, 197–214.

Zhang, J., Lindsay, R., Schweiger, A., Rigor, I., 2012. Recent changes in the dynamicproperties of declining Arctic sea ice: a model study. Geophysical ResearchLetters 39, L20503. http://dx.doi.org/10.1029/2012GL053545.