Caloric content of Chukchi Sea benthic invertebrates: Modeling spatial and environmental variation

Deep-Sea Research II 102 (2014) 97–106

Contents lists available at ScienceDirect

Deep-Sea Research II journal homepage: www.elsevier.com/locate/dsr2

Caloric content of Chukchi Sea benthic invertebrates: Modeling spatial and environmental variation Lisa M. Wilt n, Jacqueline M. Grebmeier, Thomas J. Miller, Lee W. Cooper Chesapeake Biological Laboratory, University of Maryland Center for Environmental Science, Solomons, MD, USA

ar t ic l e i nf o

a b s t r a c t

Available online 2 October 2013

The Chukchi Sea shelf off the northern coast of Alaska is rich with infauna and epibenthic macroinvertebrates used by foraging Pacific walrus (Odobenus rosmarus divergens) and other benthic-feeding consumers. Recent seasonal sea-ice retreat on the Chukchi Sea shelf has resulted in walrus hauling out in late summer on beaches of the Chukchi Sea in Russia and Alaska rather than on sea ice. Additional energetic costs may be imposed upon walruses traveling from these haul-outs to more productive foraging areas. Here, we provide an energetic assessment of prey items that could be relevant to the foraging energetics of walruses in haul-out locations. Caloric values for 171 potential walrus prey items (comprising 11 classes of benthic fauna) were obtained over 15 southeastern Chukchi Sea stations in 2010. There were statistically significant relationships between caloric content and increasing latitude (R¼ 0.661) and bottom temperature (R¼  0.560). Linear modeling indicated that taxon and latitude were the most important explanatory variables for caloric content in the study area, whereas a second model with taxon dependencies removed returned significant coefficients for the explanatory variables of latitude, depth, bottom water temperature, and sediment total organic carbon and nitrogen. K-means cluster analysis identified 6 clusters based mainly upon environmental variables such as bottom temperature, bottom salinity, and other water-column and sediment parameters that explained 86% of the variation in the data. The finding that caloric content varies strongly with latitude, which is a proxy for both water-mass type and associated water-mass productivity in the study area, may have implications for Pacific walrus, whose historical foraging patterns well offshore on sea ice have been disrupted by sea-ice decline. & 2013 Published by Elsevier Ltd.

Keywords: Chukchi Sea Benthic invertebrates Caloric content Walrus prey

1. Introduction Biological processes in the Chukchi Sea exhibit a strong seasonality, similar to other regions of the Arctic Ocean (Grebmeier et al., 1995). The Chukchi Sea supports one of the highest levels of marine productivity in the world (Bluhm and Gradinger, 2008; Gradinger, 2009; Grebmeier et al., 2006; Hill and Cota, 2005) particularly during ice melt, the movement of nutrient rich water masses north through the Bering Strait (Coachman, 1987; Weingartner et al., 2005), and tight benthic–pelagic coupling of upper water column organic carbon production settling to the underlying shallow continental shelf (Campbell et al., 2009; Grebmeier et al., 1988; Iken et al., 2010). Estimates of primary production in localized regions of the Chukchi Sea have surpassed 250 g C m  2 yr  1 (Walsh et al., 1989a). On a daily basis, typical production rates are o0.3 g C m  2 d  1 during the ice covered period, but can reach 8 g C m  2 d  1 during the ice break up period (Hill and Cota, 2005). n Correspondence to: 8539 Hayshed Lane, Columbia, MD 21045, USA. Tel.: þ 1 610 442 2025. E-mail address: [email protected] (L.M. Wilt).

0967-0645/$ - see front matter & 2013 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.dsr2.2013.09.025

Within the Chukchi Sea, numerous water masses have been identified and studied with relevance to benthic communities, including the high nutrient Anadyr Water (AW) entering the southern Chukchi Sea from the western side of Bering Strait, the low nutrient Alaska Coastal Water (ACW) entering the southern Chukchi Sea from the eastern side of Bering Strait, and the mixed Bering Shelf Water (BSW) between these two water masses (Fig. 1; Coachman, 1987; McRoy, 1993; Weingartner et al., 2005). Although the ACW water mass remains distinct from the other two water masses as it moves northward along the Alaska Coast into the Chukchi Sea, portions of the AW and BSW water masses mix as they move north and westward. This merged water mass has been designated as Bering Shelf–Anadyr Water (BSAW), or also Bering Sea Water (with winter and summer variants) in the central and northern Chukchi Sea (Weingartner et al., 2005), which is known to provide a much higher quality organic carbon supply to the benthos than the ACW in summer (Grebmeier et al., 1988). Higher nutrient supply in BSAW supports greater overall annual primary and secondary production offshore than in the ACW water mass nearshore (Stoker, 1978; Walsh et al., 1989b). Annual primary production in the ACW water mass is characteristically

98

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

Fig. 1. Map of stations analyzed for caloric content from the Chukchi Sea Offshore Monitoring in Drilling Area (COMIDA) Chemical and Benthos (CAB) project study area with schematic of water mass and current structure (modified from Grebmeier, 2012). Key: BSW ¼Bering Sea Water that is a combined water mass north of Bering Strait of both Anadyr Water and Bering Shelf Water; ACW ¼Alaska Coastal Water.

low (20–70 g C m  2 yr  1), whereas the annual primary production rate in AW tends to be high (470 g C m  2 yr  1) (Grebmeier et al., 2006; Sakshaug, 2004; Springer et al., 1996). In the northern Bering and Chukchi Seas, benthic infaunal biomass is estimated to reach nearly 150 g C m  2, with the highest biomasses found in the Gulf of Anadyr, southwest of St. Lawrence Island, in the southern Chukchi Sea, and at the head of Barrow Canyon, where there is organic carbon delivery from the southern Chukchi Sea and high local primary production following ice melt in the spring (Grebmeier, 2012). Specifically for BSAW in the Chukchi Sea, a high benthic faunal abundance of 13,554 ind m  2 has been observed, with carbon biomass ranging from 0.3– 56.2 g C m  2 (Feder et al., 2007). In the last decade, bivalves (Tellinidae and Nuculanidae), sipunculids (Golfingiidae), amphipods (Ampeliscidae and Lysianassidae), isopods (Idoteidae) and polychaetes (Maldanidae and Nephtyidae) have dominated the biomass in the Chukchi Sea, although assemblages of other organisms such as sea anemones, gastropods and sand dollars have also been observed (Grebmeier, 2012). More recently, during the summers of 2009 and 2010, infaunal abundance in the northeastern Chukchi Sea was dominated first by annelids, followed by molluscs and arthropods (see Schonberg et al., 2014). Epibenthic invertebrate communities in the Chukchi Sea have an estimated gross abundance range of 229–70,879 individuals per meter squared and a biomass estimate range of 1628–21,7023 g wet weight per meter squared (Bluhm et al., 2009). These communities are dominated by echinoderms, crustaceans and molluscs (Bluhm et al., 2009). Molluscs in the southeastern Chukchi Sea are highly diverse, but echinoderms dominate by biomass, representing 59.7% of epifaunal biomass (Feder et al., 2005). During the summers of 2009 and 2010, echinoderms were among the most dominant epibenthic invertebrates in the northeastern Chukchi Sea, but crustaceans had significantly higher Simpson's dominance and Pielou's evenness values in the area (see Ravelo et al., 2014). Feder et al. (1994) found that the abundance of infaunal molluscs was related to the percentage of sand and bottom salinity, whereas the abundance of epifaunal molluscs was related to the percent gravel and bottom temperature. Food availability in the form of entrained suspended particulate organic carbon (POC) was also

noted as a key driver of molluscan epifaunal abundance (Feder et al., 1994). Food chains in the Chukchi Sea do not statistically differ in length between the higher nutrient AW water mass and the lower nutrient ACW water mass (Iken et al., 2010), but higher proportions of consumers in the first trophic level in AW indicate a more direct coupling of benthic macroinvertebrates to pelagic primary producers than in the ACW water mass (Iken et al., 2010). In keeping with these short food chain lengths, benthic macroinvertebrates are key sources of food for predators such as bottom feeding fish, whales, seals, diving birds, and pinnipeds like Pacific walrus (Cui et al., 2009; Fay, 1982; Hazard and Lowry, 1984; Highsmith and Coyle, 1992; Iken et al., 2010; Lovvorn et al., 2003; Lowry et al., 1980). Pacific walrus forage in shallow ( o100 m) water, with nearby ice or land to haul-out on close to their feeding grounds (Fay, 1982). While they feed upon a wide variety of benthic invertebrates, Pacific walrus prefer softer bodied organisms that are high in fat content (Fay et al., 1977; Fay, 1982; Sheffield and Grebmeier, 2009). In particular, analyses of walrus stomach contents indicate that bivalves, gastropods and polychaete worms are the most frequently consumed prey (Sheffield and Grebmeier, 2009). Because they utilize ice in the marginal ice zone as transport and resting platforms for feeding grounds that are far from shore (Kovacs et al., 2010), receding sea ice has recently become problematic for Pacific walrus (Jay et al., 2011, 2012; Rausch et al., 2007). Declining Arctic sea ice led to unusual and massive haul-outs in northwestern Alaska and along the Russian Arctic coast in the summer and fall of 2007 and 2009 (Jay et al., 2011). Some walrus have been observed via satellite telemetry to travel from these onshore haul-outs to the northeastern Chukchi Sea to reach preferred feeding grounds (Jay et al., 2012). Travel from these haulout locations to better foraging areas may be imposing additional energetic requirements on walruses. Estimates of energetic costs walruses expend while foraging are dependent on accurate estimates of the caloric values of prey consumed (Noren et al., 2012). While numerous caloric surveys have been conducted throughout the European and Canadian Arctic (e.g. Percy and Fife, 1980; Szaniawska and Wolowicz, 1986; Tyler, 1973; Wacasey and Atkinson, 1987), few have been conducted in the Chukchi Sea. Stoker (1978) conducted the most comprehensive caloric surveys in the Bering and Chukchi Seas in the 1970s, and reported formalin-preserved caloric values for 52 species of benthic infauna encompassing 5 taxonomic classes. In that survey, formalinpreserved caloric contents averaged 4.85 70.13 (mean7s.d.) kcal g  1 for bivalves, 3.60 7 0.76 kcal g  1 for polychaetes, and 5.227 0.24 kcal g  1 for amphipods. Significant correlations were found to exist between organic carbon and caloric content, but analysis for the influence of other spatial and environmental variation on caloric content were not undertaken (Stoker, 1978). A more recent survey by Hondolero et al. (2011) in the Bering and Chukchi Seas evaluated the caloric content of 18 epifaunal and 6 infaunal taxa, including bivalves, polychaetes and crustaceans. Reported caloric densities for formalin-preserved benthic invertebrates from that study ranged from 2.45 to 5.00 kcal g  1, and for frozen benthic invertebrates values ranged from 2.45 to 4.77 kcal g  1 (Hondolero et al., 2011). In 3 out of the 7 invertebrate taxa surveyed, significant differences in caloric content were observed between formalin-preserved specimens and frozen specimens, including a decapod (Argis lar) (p ¼0.013), and two anthozoans (p ¼0.046 and 0.050). However, general conclusions regarding preservation techniques were limited by the small sample size (Hondolero et al., 2011). In light of limited sources of current caloric data for benthic fauna in the Pacific Arctic and the need to analyze foraging

Table 1 Sampling, caloric, and environmental data for all stations surveyed for caloric content during the July–August 2010 COMIDA CAB cruise. All environmental data were obtained through the core COMIDA CAB program (COMIDA CAB 2012 final report available at http://www.comidacab.org/). Station name

Collection date (mm/dd/yyyy)

Latitude (1N)

Longitude (1W)

4 6 8 10 16 18 20 21 24 27 30 32 34 36 38

RDM CBL1 107 CBL5 CBL4 UTX16 1014 CBL16 CBL15 HSH1 UTX11 UTX5 1030 UTX3 CBL8

7/27/2010 7/28/2010 7/29/2010 7/29/2010 7/31/2010 8/1/2010 8/1/2010 8/3/2010 8/4/2010 8/5/2010 8/5/2010 8/6/2010 8/6/2010 8/7/2010 8/7/2010

67.562 69.04 70.086 70.023 70.831 71.249 70.84 71.414 71.727 72.101 71.453 71.702 72.103 71.93 71.485

 164.178  166.594  166.455  163.761  167.787  165.448  163.291  157.491  160.718  162.975  162.611  164.515  165.456  167.389  167.782

Station number 4 6 8 10 16 18 20 21 24 27 30 32 34 36 38

Station name RDM CBL1 107 CBL5 CBL4 UTX16 1014 CBL16 CBL15 HSH1 UTX11 UTX5 1030 UTX3 CBL8

Number of caloric observations 6 7 8 6 14 11 13 6 16 10 18 12 18 13 13

Number of classes found at station 4 5 6 5 6 4 8 4 6 5 7 6 6 5 6

Depth (m)

Station number

Station name

Sediment grain size 2Φ (%)

Sediment grain size 3Φ (%)

4 6 8 10 16 18 20 21 24 27 30 32 34 36 38

RDM CBL1 107 CBL5 CBL4 UTX16 1014 CBL16 CBL15 HSH1 UTX11 UTX5 1030 UTX3 CBL8

0.99 0.57 6.15 22.2 0.14 0.75 3.47 – 0.05 7.82 6.89 1.52 0.1 0.05 0.1

3.67 2.71 36.43 57 2.32 18.17 11.82 – 0.84 41.63 12.61 53.25 1.06 0.72 1.01

Sediment grain size 4Φ (%) 52.56 31.4 20.83 7.12 22.66 16.53 37.29 – 4.5 12.31 10.07 16.44 8.06 5.27 8.52

18 35 47 27 55 43 45 126 45 36 44 38 45 48 48

Minimum caloric Mean caloric content (MJ kg  1) observation (MJ kg  1)

Maximum caloric observation (MJ kg  1)

Variance

Standard deviation

14.95 15.07 15.17 15.13 16.06 15.15 16.00 17.49 17.07 13.70 15.79 17.41 14.20 18.41 15.82

21.29 22.38 23.23 20.77 21.77 22.06 22.25 21.46 21.88 23.24 21.32 22.56 22.29 22.67 23.49

6.746 9.793 7.954 3.376 3.471 5.43 5.372 1.839 2.649 7.748 2.29 2.274 3.822 1.409 4.584

2.597 3.129 2.82 1.837 1.863 2.33 2.318 1.356 1.627 2.783 1.513 1.508 1.955 1.187 2.141

Bottom salinity (psu) 30.58 31.25 31.94 32.15 32.46 32.46 32.16 32.77 32.86 32.67 32.84 32.66 32.49 32.84 32.7

Sediment chlorophyll a (mg/m²) 31.88 9.63 6.24 9.31 30.45 44.68 16.95 – 17.01 6.92 9.19 7.56 59.87 42.24 41.59

Sedimet grain size o 0Φ (%) 0 0.14 0.58 0.09 0.05 0.5 0.14 – 0 0.78 3.19 1.34 0.65 0 0

Sediment grain size 1Φ (%) 0.19 0.29 0.53 0.6 0.05 0.15 0.24 – 0.05 1.01 2.39 0.23 0.1 0.05 0.05

Sand (grain size 1–4Φ) (%)

Silt grain size Z 5Φ (%)

Sediment modal size

Sediment TOC (%)

Sediment TON (%)

Sediment C:N ratio

57.4 34.96 63.95 86.92 25.16 35.59 52.82 – 5.45 62.76 31.96 71.44 9.32 6.09 9.69

42.6 64.89 35.47 12.98 74.79 63.91 47.03 – 94.55 36.46 64.85 27.22 90.03 93.91 90.31

4 5 3 3 5 5 5 – 5 3 5 3 5 5 5

0.41 1.1 0.46 0.13 0.97 0.88 0.58 – 1.35 0.41 0.88 0.36 1.48 1.47 1.24

0.06 0.12 0.06 0.02 0.13 0.13 0.07 – 0.21 0.06 0.12 0.05 0.19 0.2 0.18

6.83 9.17 7.67 6.5 7.46 6.77 8.29 – 6.43 6.83 7.33 7.2 7.79 7.35 6.89

18.31 18.77 19.73 17.48 19.47 20.34 19.22 19.44 20.28 20.18 19.58 20.07 19.45 20.90 20.12 Bottom temperature (1C) 7.38 6.08 0.14 0.86  1.47  1.38 0.22  0.87  1.63  1.63  1.67  1.54  1.36  1.76  1.69

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

Station number

99

100

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

energetics for predators reliant upon them, the primary objective of this study was to determine the current caloric energy values of benthic and epibenthic macroinvertebrates in the Chukchi Sea. A second objective was to analyze relationships between caloric content and other spatial and environmental variables, with a particular focus upon latitude, which varies in part as distance from shore and thus is a proxy for overlying water mass and biological productivity.

2. Material and methods 2.1. Sample collection and preparation Samples were collected at 15 randomly selected stations for this caloric study within the 45 core stations sampled between July 25th and August 16th, 2010 from the RV Moana Wave in the Chukchi Sea as part of the COMIDA CAB project (Fig. 1 and Table 1). Stations for the core COMIDA CAB project were selected in 2009, using both a general randomized tessellation stratified design (GRTS) in the core study area and a spatially-oriented, nearshore to offshore, south to north grid overlaying the stratified design. In addition, during 2010, stations were also sampled in Bering Strait and other regions in the Chukchi Sea to increase spatial coverage. Epifauna were collected at each station with a 3 m beam trawl, lasting approximately 3 min or less at a towing speed of 1–1.5 kt speed over ground (SOG) (for further information see Ravelo et al. (2014)). Infauna for caloric analyses were collected at each station with one deployment of a weighted 0.1 m2 van Veen grab. Sediment from the grab was sieved through 1 mm screen mesh, and the all of the retained infauna were subsequently sorted shipboard to family or lowest taxon possible. In the case of the epibenthic trawls, only 3 organisms (selected randomly) from each family collected were available to be kept for caloric analyses. These faunal types included sea stars (Asteroidea), tunicates (Pyuridae), decapods (Oregoniidae and Paguridae), gastropods (Buccinidae) and some bivalves (Astartidae and Carditidae). All animals were sorted onboard to the lowest taxon possible (typically family, but only class for sea stars), and frozen. Samples were transported to the Chesapeake Biological Laboratory (CBL) in Solomons, Maryland, where they were stored in a  20 1C freezer prior to processing for caloric analyses. Animals were prepared for caloric analyses by first removing all non-organic materials, including the calcium carbonate shells of all bivalves and gastropods, and polychaete sediment tubes, following the methodologies of Wacasey and Atkinson (1987) since only the soft portion of the animal is typically consumed by a feeding walrus (Fay, 1982). Crustaceans and echinoderms were processed whole, although members of the subphylum Asterozoa (sea stars and brittle stars) were omitted because the high amounts of dry skeletal material prevented complete combustion with the instrument available, a problem that has also been noted by other investigators (Stoker, 1978). Other echinoderms, such as sand dollars (Echinoidea) and sea cucumbers (Holothuroidea) were included, as these organisms had smaller proportions of dry skeletal material than the Asterozoans and it was possible to obtain reproducible caloric values (within 2% error). All processed samples were weighed before placement in preweighed aluminum trays for desiccation in an oven at 80 1C. Samples were reweighed periodically until constant weight was achieved (typically in 5–6 days). For each station, all individuals within each family were combined and ground to form homogeneous powders, which were stored in glass desiccators containing DRIERITEs until subsequent analyses. A pellet press (Parr Instruments, Moline IL) was used to create either 1–3 g pellets or

0.1–0.5 g pellets from each station's homogenized family samples, depending on the amount of sample available. For samples too dry for pelletization, gel capsules were used. Pellet weights and gel capsule weights were recorded for each sample prior to calorimetry. 2.2. Calorimetry procedures All pellets were combusted in a Bomb Calorimeter (Model 6200, Parr Instruments, Moline, Illinois, USA). A large bomb (Model 1108, Parr Instruments) was used to analyze 1–3 g pellets, and a semi-micro bomb (Model 1109, Parr Instruments) was used for 0.1–0.5 g pellets. Caloric density estimates for both bombs were calibrated using a 1 g or 0.1 g benzoic acid (C6H5COOH) pellet. Caloric densities are reported in megajoules per kilogram (MJ kg  1), and were corrected for the amount of fuse wire consumed in combustion and the remaining sample weight. Samples combusted in gel capsules were further corrected according to the formula: Ec ¼ ððEs wp Þ–ðEgc wgc ÞÞ=ðwp  wgc Þ where Ec is the corrected energy density, Es is the energy density of pellet containing sample, wp is the weight of the pellet containing sample, Egc is the energy density of the empty gel capsule, and wgc is the weight of the empty gel capsule. For this capsule correction, five gel capsules were weighed to determine an average gel capsule weight, and combusted for the calculation of an average gel capsule caloric density. The average weight was found to be 0.115 7 o 0.01 g and the average caloric density was 19.51 70.09 MJ kg  1. These values were used in the formula to calculate the corrected caloric densities. Replicate tissue samples were combusted until less than 2% difference was achieved between replicate samples (typically within 2–5 replicates), at which point the replicates were averaged and recorded as that family's caloric measurement at that station. 2.3. Statistical analysis All statistical analyses were undertaken using R (version 2.15.0, see http://www.r-project.org). Basic statistics for all stations and classes were calculated, including (a) the mean caloric content over the study area (the average of all recorded caloric measurements for all stations), (b) mean caloric content, variance and standard deviation for each station (the average of all recorded family caloric measurements at the station), and (c) mean caloric content, variance and standard deviation for each class that was observed in the study. Because normality testing could not be carried out for the full dataset (2 classes had only 1 caloric observation), a non-parametric Kruskal–Wallis rank sum test was applied to evaluate if significant differences in caloric content exist by class. Relationships between average measured caloric densities at each station and the following parameters were evaluated using a Spearman correlation analysis: latitude, longitude, depth, bottom temperature, bottom salinity, sediment chlorophyll a, grain size, and modal size, sediment total organic carbon (TOC), sediment total organic nitrogen (TON), and carbon to nitrogen ratio (C:N) (Table 1). All environmental data were obtained through the core COMIDA CAB program (COMIDA CAB 2012 final report available at http://www.comidacab.org/). Two different analyses of variance (ANOVA) models were developed to explain the distribution of caloric content: one that included the influence of taxon, and one without the effects of taxon. The first model was created on a dataset including the influence of taxon (class) on caloric content. Because the variable “lowest taxon identified” proportionately was responsible for

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

many degrees of freedom, it was dropped for the first linear model. The second linear model was created in two stages. In the first stage, the dependencies on the two taxonomic variables (class and lowest taxon identified) were modeled. Then, these dependencies were regressed out, leaving the residuals in caloric content to comprise the new response variable for the second stage of the model. Since the remaining variables after regressing caloric content on class and “lowest taxon identified” were all associated with the station where the data was taken, the residuals were averaged by station before performing further regression analysis on the residuals. For both ANOVAs, zone, quadrant, sediment grain size, and sediment organic carbon to nitrogen ratio (C:N) were not used because of strong correlations with other variables. For example, because sediment chlorophyll a was so highly correlated with grain size, grain sizes were not used in the analysis to avoid overfitting the model through autocorrelation. Sediment TOC and TON were used in favor of the organic C:N ratio. Diagnostic plots were generated to evaluate the assumptions associated with ANOVA for both models. Adjusted r2 and Mallow's Cp were used to calculate and plot the number and combination of variables in the best fit for two different linear models. For both linear models, Tukey Honest Significant Differences (HSD) tests were applied to investigate differences between factor levels. Following these ANOVA analyses, a nested mixed effects analysis was also conducted for this dataset. Class, and lowest taxon identified within class were identified as random effects, and latitude, longitude, sediment chlorophyll a, bottom salinity, and bottom temperature were identified as fixed effects. Two plots of spatially interpolated caloric content were generated for the study area: one with the influence of taxon, and the other without (residual caloric content). The R packages akima, lattice, colorspace, and mapdata were used to interpolate the data to a grid of points for both energy and residual energy and plot them within the latitude and longitude boundaries of the Chukchi Sea study area (Akima, 1978, 1996). Sampling stations were clustered using the K-means (Hartigan and Wong, 1979) method by the average of the following numerical variables: mean station caloric content, depth, bottom temperature, bottom salinity, sediment chlorophyll a, sediment grain size, sediment modal size, sediment TOC, sediment TON, and carbon to nitrogen ratio (C:N) for each station. Since the variables have different means and standard deviations, prior to cluster analysis, all variables were Z-transformed. Because a number of environmental observations were not available for station 21 (CBL6), station 21 was excluded from cluster analysis. The recommended number of cluster groups was identified using a plot of the within groups sum of squares against clusters extracted.

101

Using the matrix of averaged environmental data and caloric content scaled to zero mean and unit standard deviation generated for the cluster analysis, a principal components analysis (PCA) was also conducted with ten principal components, though only the first two were used for plotting.

3. Results 3.1. Energy content A total of 171 caloric values were determined for 11 classes of infaunal and epifaunal macroinvertebrates across all 15 stations throughout the Chukchi Sea study area (see supplement – Appendix A and Wilt (2012) for full appendix of caloric values, http://drum.lib.umd.edu/handle/1903/13271). The number of energy observations (recorded caloric content for a lowest taxon identified, typically family, at each station; Wilt, 2012), ranged from 6 to 18, and the number of classes represented at each station ranged from 4 (station 4, RDM) to 8 (station 20, 1014). The mean energy density for the entire Chukchi Sea study area was 19.71 72.08 MJ kg  1, with individual measurements ranging from a low of 13.70 MJ kg  1 for a tunicate (Styelidae) to a high of 23.49 MJ kg  1 for a bivalve (Nuculanidae). Mean station caloric densities ranged from 17.48 71.84 MJ kg  1 (station 10, CBL5) to 20.907 1.19 MJ kg  1 (station 36, UTX3), station variances ranged from 1.41 to 9.79, and standard deviations ranged from 1.19 to 3.13 (Table 1). The Kruskal–Wallis rank sum test identified a significant difference (Kruskal–Wallis χ2 ¼ 88.16, df ¼10, po 0.001) among the 11 classes of benthic and epibenthic macroinvertebrates analyzed in this study. Echinoidea (n ¼1), Holothuoridea (n ¼ 2), and Ascidiacea (n ¼6) were the classes with the lowest mean caloric content (15.13, 16.09 70.42, and 16.11 71.48 MJ kg  1, respectively – Table 2), and Polychaeta (n ¼15), Gastropoda (n ¼41), and Bivalvia (n ¼48) were the classes with the highest mean caloric content (20.49 70.80, 20.857 0.73, and 20.98 71.31 MJ kg  1, respectively – Table 2).

3.2. Correlations The Spearman's rank test determined that latitude (r ¼0.661; p¼ 0.009) and bottom temperature (r ¼  0.560; p ¼0.033) are significantly correlated with mean station caloric content throughout the Chukchi Sea study area. Increasing latitude was positively correlated with mean station caloric content, while bottom temperature was negatively correlated.

Table 2 Caloric observations for infaunal and epifaunal animals collected during the July–August 2010 COMIDA CAB cruise. Class

Number of caloric observations

Mean caloric content (MJ kg  1)

Minimum caloric observation (MJ kg  1)

Maximum caloric observation (MJ kg  1)

Variance Standard deviation

Amphipoda Anthozoa Ascidiacea Asteroidea Bivalvia Echinoidea Gastropoda Holothuroidea Malacostraca Polychaeta Sipunculidea

1 5 6 11 48 1 41 2 35 15 6

19.48 18.52 16.11 16.73 20.98 15.13 20.85 16.09 18.50 20.49 18.42

– 16.68 13.67 14.24 17.60 – 18.93 15.79 14.95 19.00 16.25

– 20.06 17.49 19.93 23.49 – 22.38 16.39 23.24 21.86 21.47

– 1.56 2.19 2.67 1.72 – 0.54 0.18 3.59 0.64 4.22

– 1.25 1.48 1.64 1.31 – 0.73 0.42 1.90 0.80 2.05

102

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

3.3. Linear models and nested mixed effects model Two linear models were created: one with the influence of taxon included, and the other with the influence of taxon screened. The Adjusted r2 and Mallow's Cp analysis indicated that the best-fit model including the influence of taxon incorporates 2 variables: 7 levels of class (Ascidiacea, Asteroidea, Bivalvia, Echinoidea, Gastropoda, Holothuroidea, and Polychaeta) (p ¼ o0.001), and latitude (p ¼0.03) (n¼ 171, df ¼159). A Tukey HSD test on the class variable returned 22 significant (po0.05) differences out of 55 possible pairings of class levels. Gastropoda and Asteroidea had the most significant difference in caloric content (p o0.001). For the second linear model, the effects of spatial and environmental variables on caloric energy were evaluated separately from taxonomic variables. In the first stage, a linear model was constructed to model caloric content as a function of class and the lowest taxon identified (n ¼171, df ¼136). Taxonomic class dependency was again found to be a significant explanatory variable for caloric content (p o0.001). However, at finer taxonomic levels (i.e. the lowest identifiable taxon, Wilt, 2012), dependency was found to be non-significant (p ¼0.055). Tukey HSD analysis revealed significant differences in 22 out of 55 possible pairings at class levels. Classes Bivalvia and Ascidiacea had the most significant difference in caloric content in this model (p o0.001). After these taxon dependencies were regressed out, the residuals in caloric content from the first stage model ranged from  3.81 MJ kg  1 to 3.20 MJ kg  1 and averaged 9.77  10  18 MJ kg  1. In the second stage of the second linear model, ANOVA tested the effect of environmental variables on the caloric content residuals. The adjusted r2 and the Mallow's Cp analysis indicated that the best-fit model contains 5 variables (latitude, depth, bottom water temperature, sediment TOC and TON) and explained 87% of variation in residual caloric content. Five explanatory variables returned as significant: latitude (p o0.001), sediment TOC (p ¼0.002), bottom water temperature (p¼ 0.003), depth (p ¼0.004), and sediment TON (p ¼ 0.005) (n ¼15, df ¼8) (Table 3). Spatially interpolated plots of caloric content and residual caloric content (no influence of taxon) (Figs. 2 and 3) indicate a northward increase in caloric content in the Chukchi Sea study area. In the mixed effects model, the environmental variables of longitude, bottom water salinity and bottom water temperature were non-significant, so they were dropped from the model. In the new model, the fixed effects of latitude and sediment chlorophyll a

Table 3 ANOVA output for a model explaining residual caloric content (without taxonomic influence). Animals were collected during the July–August 2010 COMIDA CAB cruise. Variable

Coefficient estimate

Standard error

t-Value p-Value

Intercept Latitude Sediment TOC Bottom temperature Depth Sediment TON

 62.446 0.857  5.273 0.395 0.065 29.368

12.147 0.169 1.195 0.096 0.017 7.630

 5.141 o 0.001 5.077 o 0.001  4.412 0.002 4.099 0.003 3.954 0.004 3.849 0.005

Residual standard error Degrees of freedom Multiple r2 Adjusted r2 F-statistic p-Value n

were found to be significant explanatory variables for caloric content, and the random effect of class was found to be significant, as the high posterior density (HPD) 95% confidence interval (0.742–1.432) did not include the origin. The random effect of lowest taxon nested within class was found to be non-significant at the α ¼0.05 level. 3.4. Cluster analysis and ordination The K-means analysis identified 6 cluster groups (Fig. 4) that reinforce the spatial gradients identified in the ANOVA modeling. The clustering explained 86% variance among the clusters, and 14% variance within clusters. Cluster groups K1, K3 and K5, located in the northern region of the study area, had the highest caloric densities, while cluster groups K2, K4 and K6, located further to the south and along the Alaskan coast, had the lowest caloric densities (Table 4). The higher caloric density northern clusters (K1, K3, and K5) had lower bottom water temperatures and higher salinities than the averages for all 14 stations. The lower caloric density southern clusters along the Alaskan coast (K2, K4, and K6) were uniformly shallow, warm with respect to bottom seawater temperatures, and less saline than the averages for all 14 stations. Notably, cluster group K5, mostly located at the northwestern portion of the study area, had the highest caloric density, the greatest depth, the highest sediment chlorophyll a, the highest percentage of silt and clay, the highest sediment TOC and the highest sediment TON of all cluster groups (Table 4). Conversely, cluster group K4, located in the central portion of the study area near the Alaskan coast, had the lowest caloric density, the lowest percentage of silt and clay, the lowest sediment TOC, and the lowest sediment TON of all cluster groups. PCA produced an acceptable ordination of environmental variables. The first principal component of the PCA explained 46.7% (s ¼2.73) of the variance among stations, and the second principal component explained 23.8% (s ¼1.94). The variables that most strongly influenced the first principal component (PC1) of the PCA were percent sand (grain size 1–4Φ), percent silt and clay (grain size Z5Φ), and sediment TON and TOC. Of all the 16 variables

included in the analysis, percent coarse sand (grain size r0Φ) and carbon to nitrogen ratio were the only two that did not influence PC1 at all (Table 5). Bottom temperature and bottom salinity most strongly influenced the second principal component (PC2), with a positive correlation for bottom temperature and a negative correlation for bottom salinity. Percent sand, percent silt and clay, sediment TOC and sediment TON were the only variables that did not influence PC2 (Table 5). When scored by station, station 10 (CBL5) had the largest component of PC1, followed by stations 36 (UTX3) and 34 (1030). Station 4 (RDM) had the largest component of PC2, followed by station 6 (CBL1). Plotting the stations by component PC1 and component PC2 yielded a strong grouping of stations 16 (CBL4), 18 (UTX16), 24 (CBL15), 34 (1030), 36 (UTX3) and 38 (CBL8). Though they tended to have small components of PC2, they contained a large component of PC1. Stations 4 (RDM) and 10 (CBL5) were not tightly grouped with any other stations, but had strong components of PC2 and PC1 respectively (Fig. 4).

0.252 8.000 0.870 0.788 10.670 0.002 15

4. Discussion Analyses presented here indicate a strong latitudinal gradient in environmental variables that may drive a parallel gradient in the caloric content of epifaunal and infaunal taxa. The latitudinal associations with caloric content identified in this survey may

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

103

Fig. 2. Spatially interpolated plots of (A) caloric content including the influence of taxonomic variables and (B) residual caloric content (influence of taxonomic variables regressed out) for animals collected during the July–August 2010 COMIDA CAB cruise. Black dots are stations surveyed, with station number.

Fig. 4. Plot of PC1 score by PC2 score for each of the 15 stations sampled during the July–August 2010 COMIDA CAB cruise.

Fig. 3. Map of the six cluster station groups produced by K-means cluster analysis. Animals were collected during the July–August 2010 COMIDA CAB cruise.

have important implications for higher trophic level predators, particularly Pacific walrus. Walruses rely upon seasonal ice floes in the marginal ice zone for transport to preferred feeding grounds and for resting platforms over the continental shelf during foraging (Fay, 1982). Hauling out on Alaskan and Russian shores as a response to decreasing amounts of seasonal sea ice (Jay et al., 2011) separates walrus geographically from the highest quality benthic prey. Geographical distributions of highest caloric density are consistent with historic feeding grounds where they have been previously observed. This study indicates that caloric densities of benthic prey are highest offshore and to the northwest

in the COMIDA CAB study area. This finding is also consistent with satellite telemetry data showing walruses making energetically costly efforts to reach these feeding grounds from land (Jay et al., 2012). The caloric gradient identified in our analyses is likely reflective of strong gradients in environmental variables. Indeed, results of our PCA indicated that stations in this sampling area are distinguished from each other mostly by environmental variables rather than by energy content. The first principal component was largely driven by sediment characteristics (substrate type and sediment TOC and TON). Sediment type has also been identified as a key variable in cluster analysis for both infauna (Feder et al., 1994, 2007; Grebmeier et al., 1989; see Grebmeier and Cooper section in final COMIDA CAB report, www.comidacab.org; Schonberg et al., 2014) in the southern and northern Chukchi Sea, and epifaunal taxa in the southern Chukchi Sea (Feder et al., 1994, 2005) and northern Chukchi Sea (Konar et al., 2014; Ravelo et al., 2014). Of the biological variables, sediment TOC, TON, and chlorophyll a were found to have significant relationships to caloric content. The best-fit model for residual energy (with the effects of taxon removed) included both sediment TOC and TON as significant explanatory variables for caloric content, suggesting that the importance of sediment TOC and TON may have been obscured

104

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

Table 4 Summary of the 6 cluster groups produced by K-means cluster analysis, with 86% variance between clusters. All variables normalized in 0 mean and 1 standard deviation. Animals were collected during the July–August 2010 COMIDA CAB cruise. Cluster identifier

Station numbers

Average caloric content

Depth

Bottom water temperature

Bottom water salinity

Sediment chlorophyll a

Sediment grain size r0Φ

Sediment grain size 1Φ

Sediment grain size 2Φ

K1 K2 K3 K4 K5

30 4 8, 27, 32 10 16, 18, 24, 34, 36, 38 6, 20

0.02  1.39 0.48  2.31 0.59

0.314  2.408  0.070  1.466 0.663

 0.648 2.443  0.422 0.216  0.457

0.936  2.595 0.285  0.142 0.436

 0.826 0.455  0.955  0.819 0.874

3.082  0.618 0.426  0.514  0.386

3.129  0.372 0.265 0.281  0.555

0.542  0.438 0.255 3.085  0.570

 0.63

 0.105

Station numbers 30 4 8, 27, 32 10 16, 18, 24, 34, 36, 38 6, 20

Sediment grain size 3Φ  0.23  0.66 1.27 1.91  0.65

Sediment grain size 4Φ  0.577 2.471  0.114  0.788  0.516

K6 Cluster identifier K1 K2 K3 K4 K5 K6

 0.49

0.998

1.164

Sand (grain size 1–4Φ)  0.284 0.669 0.992 1.774  0.910 0.163

 0.837

 0.595

 0.456

 0.252

 0.267

Sediment grain size Z 5Φ 0.183  0.646  1.002  1.750 0.919

Sediment modal size 0.692  0.385  1.461  1.461 0.692

Sediment TOC 0.095  0.950  0.950  1.572 0.877

Sediment TON 0.091  0.867  0.920  1.506 0.943

Sediment C:N ratio 0.011  0.663  0.120  1.107  0.279

 0.148

0.692

 0.308

1.895

PC8

PC9

PC10

0.342 0.492 0.120  0.428 0.238  0.203 0.370 0.356  0.145

 0.518 0.324

0.006

Table 5 A matrix of the relative contributions of each variable to the axes of the reduced space in the principal components analysis (PCA). Variable

PC1

PC2

Average caloric content Depth Bottom temperature Bottom salinity Sediment chlorophyll a % Grain size o 0Φ % Grain size 1Φ % Grain size 2Φ % Grain size 3Φ % Grain size 4Φ % Sand (grain size 1–4Φ) % Silt (grain size 45Φ) Sediment modal size Sediment TOC Sediment TON C:N ratio (weight/weight)

 0.232  0.254 0.115  0.128  0.250

 0.221  0.217 0.452  0.460 0.124  0.293  0.253  0.173  0.244 0.404

0.114 0.267 0.295 0.129 0.359  0.359  0.303  0.351  0.354

PC3

0.117  0.244 0.549 0.556  0.176 0.168

0.129

0.225

0.225

0.431

PC4

PC5

PC6

 0.369  0.413 0.149

 0.462 0.267

0.240 0.118 0.296 0.356  0.147  0.215  0.139 0.134 0.139 0.111 0.165  0.467

 0.126  0.300

0.302  0.226 0.359  0.157  0.573  0.131

in the correlation by the large dependency upon taxon. The K-means cluster analysis further supports the importance of sediment TOC and TON as driving variables in the pattern reported here. The cluster with the highest caloric density also had the highest sediment TOC and TON content, and the cluster with the lowest caloric density also had the lowest TOC and TON of all 6 clusters. Sediment TOC and TON can be interpreted as representative of food availability in the Chukchi Sea study area, thus supporting their importance in driving prey caloric content. When food availability is high, the lipid content of invertebrates increases (Luis and Passos, 1995), and organisms with higher lipid content have higher caloric densities than animals with low lipid content (Falk-Petersen et al., 1990; Weslawski et al., 2010). In addition to sediment TOC and TON, sediment chlorophyll a was identified as a significant explanatory variable for caloric content in the mixed effects model, which also reflects the importance of food availability to macroinvertebrates when evaluating caloric content as it does with benthic standing stock (Grebmeier et al., 2006). Bottom temperature was resolved to be a significant explanatory variable for caloric content, which is related to latitude (due to the spatial arrangement of water masses on a north to south basis) in the Chukchi Sea study area. Latitude was consistently the strongest non-taxonomic predictor of caloric content in this investigation, which coincides with the observed higher benthic

PC7

0.178 0.146  0.595  0.299

0.485  0.285

0.138 0.141 0.480

 0.352  0.122 0.126  0.294 0.230 0.235

 0.303 0.305

0.287  0.182  0.414

 0.193

 0.121 0.202  0.248  0.147

0.394 0.293  0.311  0.112 0.388  0.344

 0.150 0.143  0.628

0.280  0.197 0.102  0.114 0.393

 0.139

 0.279

infaunal carbon biomass in the region as compared with other parts of the Chukchi Sea (Grebmeier et al., 2006). High concentrations of lipid rich bivalves have also been observed in the northern part of the Chukchi Sea (Grebmeier, 2012). To the south of the COMIDA CAB study area and along the Alaskan coast, the low nutrient ACW water mass flows northward from the Bering Strait, and in the northwest portion of the COMIDA CAB study area, the higher nutrient BSAW (Bering Sea water in the Chukchi Sea; Weingartner et al., 2005) transits first west and north across the shelf, and then flows northeast along the shelf break of the Chukchi Sea (Weingartner et al., 2005). The BSAW water mass is known to support a higher water column production and subsequent export of carbon to the benthos, resulting in higher benthic productivity to the north and west of the ACW water mass (Grebmeier et al., 2006). While there is a clear latitudinal gradient of increasing caloric density moving from south to north in the study area, it is important to note that there were fewer sites observed in the south than in the north, and that the southern sites had fewer caloric observations than those in the north. As caloric density increases from south to north, the number of total caloric observations increases, which likely reflects a previously described increase in diversity that can be found moving from south to north in the Chukchi Sea (Grebmeier et al., 1989).

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

Cluster analysis also highlights the importance of water mass type to determining caloric content in the Chukchi Sea study area. Nearly all of the stations in the two K-means cluster groups with the highest caloric densities were located within the BSAW water mass, while the two K-means cluster groups with the lowest caloric densities were both located to the south near the Alaskan coast. The PCA echoed this finding, as the plot of stations by component PC1 and component PC2 also grouped the high caloric density northern stations strongly, suggesting that these stations are significantly distinguished from the others by similar water mass driven characteristics. Our caloric analyses clearly indicate the presence of taxon specific differences in energy content. In the Canadian Arctic, a caloric survey including 10 of the same classes included in this investigation indicated similar caloric values to those measured here. We also note the important influence of preservation technique when comparing estimates of caloric content in individual taxa (Hondolero et al., 2011; Wilt, 2012). With that caveat, the results of this and previous studies clearly indicate spatial and seasonal variability in the caloric content of benthic organisms. For example, Stoker (1978) conducted a caloric survey on frozen specimens from 9 bivalve taxa and 6 other miscellaneous taxa, including two polychaete families (Maldanidae and Nephtyidae). The average caloric value for frozen bivalves was 4.42 kcal g  1 (Stoker, 1978), while the average caloric value for frozen bivalves surveyed in this current study was much higher (5.014 kcal g  1). The average caloric value between the two polychaete taxa in Stoker's survey (3.64 kcal g  1) (Stoker, 1978) is also less than determined during the current survey (4.90 kcal g  1). These measurement differences are possibly due to instrumentation, geography (Gallagher et al., 1998), species analyzed, or sampling season (Mann, 1978; Okumus and Stirling, 1998). By way of illustration, Stoker's field sampling was conducted over a 4-year period in both summer and winter. Because the lipid content of polychaete worms is known to depend largely upon diet (Luis and Passos, 1995), strong seasonal changes in Chukchi Sea annual primary production would likely result in seasonal differences in polychaete lipid content, and therefore caloric content. Different polychaete species may also feed at different trophic levels, which could also lead to differences in caloric content.

Key findings In this investigation, taxon was the most significant explanatory variable for caloric content, likely due to the higher lipid levels that are found in softer bodied macroinvertebrate organisms. After the influence of taxon is ruled out, however, a number of other variables may explain the variability in caloric content. Geographically, zones with high sediment TOC and TON (signifying high food availability) tend to have higher caloric densities. Caloric content was found in multiple statistical analyses to increase with increasing latitude in the COMIDA CAB area, an effect that was strong enough to be observed even with the influence of taxon still included in the analysis. As indicated by cluster analysis, the connection between caloric content and latitude in this study area is likely related to water mass type (Anadyr Water mixed with Bering Sea water); this water mass carries higher water column nutrients and primary productivity than in ACW, along with colder bottom water temperatures, all conducive to increasing carbon export to maintain higher benthic biomass and more caloric-rich benthic taxa. This caloric data will be helpful for modeling foraging energetics for walrus and other higher trophic level predators such as bottom feeding fish, whales, seals, and diving birds that are known to consume infaunal and epifaunal macroinvertebrates (Cui et al.,

105

2009; Fay, 1982; Hazard and Lowry, 1984; Highsmith and Coyle, 1992; Iken et al., 2010; Lovvorn et al., 2003; Lowry et al., 1980; Noren et al., 2012). It is important to note however that the caloric density of feeding grounds is not the only criteria of feeding ground quality. For example, biomass and accessibility are other factors that should be taken into account. Understanding foraging energetics for Pacific walrus in particular will enhance our understanding of cost and benefit tradeoffs associated with walrus traveling from haul-out sites to preferred feeding grounds, which may be a powerful tool to evaluate prospects for Pacific walrus if the extent of seasonal sea ice continues to decline in the Arctic Ocean.

Acknowledgments We thank Dr. Brenda Konar (University of Alaska, Fairbanks) for access to epifauna, and Monika Kedra, Adam Peer, Regan Simpson, Linton Beaven, Christian Johnson, Arvind Shantharam, Stephanie Soques, and Mike Studivan for their assistance in data collection, faunal identification, and statistical analyses during the study. Comments by two anonymous reviewers helped improve a prior version of the manuscript. Funding for this project was provided by U.S. Department of the Interior, Bureau of Ocean Energy Management (BOEM), Alaska Outer Continental Shelf Region, Anchorage, Alaska under Contract number M08PC20056 to PIs Grebmeier and Cooper, CBL/UMCES.

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.dsr2.2013.09.025. References Akima, H., 1978. A method of bivariate interpolation and smooth surface fitting for irregularly distributed data points. ACM Transactions on Mathematical Software 4, 148–164. Akima, H., 1996. Algorithm 761: scattered-data surface fitting that has the accuracy of a cubic polynomial. ACM Transactions on Mathematical Software 22, 362–371. Bluhm, B., Gradinger, R., 2008. Regional variability in food availability for Arctic marine mammals. Ecological Applications 18, S77–S96. Bluhm, B.A., Iken, K., Mincks Hardy, S., Sirenko, B.I., Holladay, B.A., 2009. Community structure of epibenthic megafauna in the Chukchi Sea. Aquatic Biology 7, 269–293. Campbell, R.G., Sherr, E.B., Ashjian, C.J., Plourde, S., Sherr, B.F., Hill, V., Stockwell, D.A., 2009. Mesozooplankton prey preference and grazing impact in the western Arctic Ocean. Deep Sea Research II 56, 1274–1289. Coachman, L.K., 1987. Advection and Mixing on the Bering-Chukchi Shelves. Component A. Advection and Mixing of Coastal Water on High Latitude Shelves. ISHTAR 1986 Progress Report, vol. 1. Institute of Marine Science, University of Alaska Fairbanks, pp. 1–42. Cui, X., Grebmeier, J.M., Cooper, L.W., Lovvorn, J.R., North, C.A., Kolts, J.M., 2009. Spatial distributions of groundfish in the northern Bering Sea in relation to environmental variation. Marine Ecology Progress Series 393, 147–160. Falk-Petersen, S., Hopkins, C.C.E., Sargent, J.R., 1990. Trophic relationships in the pelagic, Arctic food web. In: Barnes, M., Gibson, R.N. (Eds.), Trophic Relationships in the Marine Environment. Aberdeen University Press, Aberdeen, pp. 315–333. Fay, F.H., 1982. Ecology and Biology of the Pacific Walrus, Odobenus rosmarus divergens Illiger. US Fish and Wildlife Service. North American Fauna 74, 279 p. Fay, F.H., Feder, H.M., Stoker, S.W., 1977. An Estimation of the Impact of the Pacific Walrus Population on its Food Resources in the Bering Sea. Final report, US Marine Mammal Commission, Contract nos. MM4AC-006, MM5AC-024. Feder, H.M., Foster, N.R., Jewett, S.C., Weingartner, T.J., Baxter, R., 1994. Mollusks in the northeastern Chukchi Sea. Arctic 47, 145–163. Feder, H.M., Jewett, S.C., Blanchard, A., 2005. Southeastern Chukchi Sea (Alaska) epibenthos. Polar Biology 28, 402–421, http://dx.doi.org/10.1007/s00300-0040683-4. Feder, H.M., Jewett, S.C., Blachard, A.L., 2007. Southeastern Chukchi Sea (Alaska) macrobenthos. Polar Biology 30, 261–275, http://dx.doi.org/10.1007/s00300006-0180-z.

106

L.M. Wilt et al. / Deep-Sea Research II 102 (2014) 97–106

Gallagher, M.J., Ambrose Jr., W.G., Renaud, P.E., 1998. Comparative studies in biochemical composition of benthic invertebrates (bivalves, ophiuroids) from the Northeast Water (NEW) Polynya. Polar Biology 19, 167–171. Gradinger, R., 2009. Sea-ice algae: major contributors to primary production and algal biomass in the Chukchi and Beaufort Seas during May/June 2002. DeepSea Research II 56, 1201–1212. Grebmeier, J.M., 2012. Shifting patterns of life in the Pacific Arctic and sub-Arctic seas. Annual Review of Marine Science 4, 63–78. Grebmeier, J.M., McRoy, C.P., Feder, H.M., 1988. Pelagic–benthic coupling on the shelf of the northern Bering and Chukchi Seas. I. Food supply source and benthic biomass. Marine Ecology Progress Series 48, 57–67. Grebmeier, J.M., Feder, H.M., McRoy, C.P., 1989. Pelagic–benthic coupling on the shelf of the northern Bering and Chukchi Seas. II. Benthic community structure. Marine Ecology Progress Series 51, 253–268.Grebmeier, J.M., Smith, W.O., Conover, R.J., 1995. Biological processes on Arctic continental shelves: iceocean-biotic interactions. In: Smith, W.O., Grebmeier, J.M. (Eds.), Arctic Oceanography: Marginal Ice Zones and Continental Shelves. American Geophysical Union, Washington, D.C., pp. 231–261. Grebmeier, J.M., Cooper, L.W., Feder, H.M., Sirenko, B.I., 2006. Ecosystem dynamics of the Pacific-influenced northern Bering and Chukchi Seas. Progress in Oceanography 71, 331–361. Hartigan, J.A., Wong, M.A., 1979. A K-means clustering algorithm. Applied Statistics 28, 128–137. Hazard, K.W., Lowry, L.F., 1984. Benthic prey in a bowhead whale from the northern Bering Sea. Arctic 37, 166–168. Highsmith, R.C., Coyle, K.O., 1992. High productivity of arctic amphipods relative to gray whale energy requirements. Marine Ecology Progress Series 83, 141–150. Hill, V., Cota, G., 2005. Spatial patterns of primary production on the shelf, slope and basin of the western Arctic in 2002. Deep-Sea Research II 52, 3344–3354. Hondolero, D., Bluhm, B.A., Iken, K., 2011. Caloric content of dominant benthic species from the northern Bering and Chukchi Seas: historical comparisons and the effects of preservation. Polar Biology 35, 637–644. Iken, K., Bluhm, B.A., Dunton, K.H., 2010. Benthic food web structure under differing water 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 Sea during sparse sea ice cover. Marine Ecology Progress Series 468, 1–13. Jay, C.V., Marcot, B.G., Douglas, D.C., 2011. Projected status of the Pacific Walrus (Odobenus rosmarus divergens) in the twenty-first century. Polar Biology 34, 1065–1084. Konar, B., Ravelo, A., Grebmeier, J.M., Trefry, J.H., 2014. Influence of environmental parameters on the size frequencies of key epibenthic organisms in the eastern Chukchi Sea. Deep Sea Research II 102, 107–118. Kovacs, K.M., Lydersen, K., Overland, J.E., Moore, S.E., 2010. Impacts of changing seaice conditions on Arctic marine mammals. Marine Biodiversity 41, 181–194. Lovvorn, J.R., Richman, S.E., Grebmeier, J.M., Cooper, L.W., 2003. Diet and body condition of spectacled eiders wintering in pack ice of the Bering Sea. Polar Biology 26, 259–267. Lowry, L.F., Frost, K.J., Burns, J.J., 1980. Feeding of bearded seals in the Bering and Chukchi Seas and trophic interaction with Pacific walruses. Arctic 33, 330–342. Luis, O.J., Passos, A.M., 1995. Seasonal changes in lipid content and composition of the polychaete Nereis (Hediste) diversicolor. Comparative Biochemistry and Physiology 111B, 579–586. Mann, R., 1978. A comparison of morphometric, biochemical, and physiological indexes of condition in marine bivalve mollusks. In: Throp, J.H., Gibbons, J.W. (Eds.), Energy and Environmental Stress in Aquatic Systems. US Department of Energy, Washington, D.C., pp. 484–497.

McRoy, C.P., 1993. ISHTAR, the project: an overview of Inner Shelf Transfer and Recycling in the Bering and Chukchi Seas. Continental Shelf Research 13, 473–479. Noren, S., Udevitz, M., Jay, C., 2012. Bioenergetics model for estimating food requirements of female Pacific walruses Odobenus rosmarus divergens. Marine Ecology Progress Series 460, 261–275. Okumus, I., Stirling, H.P., 1998. Seasonal variations in the meat weight, condition index and biochemical composition of mussels (Mytilus edulis L.) in suspended culture in two Scottish sea lochs. Aquaculture 159, 249–261. Percy, J.A., Fife, F.J., 1980. The proximate composition and caloric content of Arctic marine invertebrates from Frobisher Bay. Canadian Data Reports of Fisheries and Aquatic Sciences 214, 1–35. Rausch, R.L., George, J.C., Brower, H.K., 2007. Effect of climatic warming on the Pacific walrus, and potential modification of its helminth fauna. Journal of Parasitology 93, 1247–1251. Ravelo, A.M., Konar, B., Trefry, J.H., Grebmeier, J.M., 2014. Epibenthic community variability in the northeastern Chukchi Sea. Deep Sea Research II 102, 119–131. Sakshaug, E., 2004. Primary and secondary production in the Arctic Seas. In: Stein, R., MacDonald, R.W. (Eds.), The Organic Carbon Cycle in the Arctic Ocean. Springer, Berlin, pp. 57–81. Schonberg, S.V., Clarke, J.T., Dunton, K.H., 2014. The distribution, abundance, biomass and diversity of benthic infauna in the northeast Chukchi Sea, Alaska in relation to walrus and gray whale sightings. Deep Sea Research II 102, 144–163. Sheffield, G., Grebmeier, J.M., 2009. Pacific walrus (Odobenus rosmarus divergens): differential prey digestion and diet. Marine Mammal Science 25, 761–777. Springer, A.M., McRoy, C.P., Flint, M.V., 1996. The Bering Sea green belt: shelf-edge processes and ecosystem production. Fisheries Oceanography 5, 205–223. Stoker, S.W., 1978. Benthic invertebrate macrofauna of the eastern continental shelf of the Bering/Chukchi Seas. University of Alaska, Fairbanks, Alaska. (Ph.D. thesis). Szaniawska, A., Wolowicz, M., 1986. Changes in energy content of common species from Hornsund, southwest Spitsbergen. Polar Research 4, 85–90. Tyler, A.V., 1973. Caloric values of some North Atlantic invertebrates. Marine Biology 19, 258–261. Wacasey, J.W., Atkinson, E.G., 1987. Energy values of marine benthic invertebrates from the Canadian Arctic. Marine Ecology Progress Series 39, 243–250. Walsh, J.J., McRoy, C.P., Coachman, L.K., Goering, J.J., Nihoul, J.J., Whitledge, T.E., Blackburn, T.H., Parker, P.L., Wirick, C.D., Shuert, P.G., Grebmeier, J.M., Springer, A.M., Tripp, R.D., Hansell, D.A., Djenidi, S., Deleersnijder, E., Henriksen, K., Lund, B.A., Andersen, P., Muller-Karger, F.E., Dean, K., 1989a. Carbon and nitrogen cycling within the Bering/Chukchi Seas: source regions for organic matter effecting AOU demands of the Arctic Ocean. Progress in Oceanography 22, 277–359. Walsh, J.J., Mcroy, C.P., Blackburn, T.H., Coachman, L.K., Goering, J.J., Henriksen, K., Andersen, P., Nihoul, J.J., Parker, P.L., Springer, A.M., Tripp, R.B., Whitledge, T.E., Wirick, C.D., 1989b. The role of Bering Strait in the carbon/nitrogen fluxes of polar marine ecosystems. In: Rey, L., Alexander, V. (Eds.), Marine Living Systems of the Far North. E.J. Brill, Leiden, pp. 90–120. Weingartner, T., Aagaard, K., Woodgate, R., Danielson, S., Sasaki, Y., Cavalieri, D., 2005. Circulation on the north central Chukchi Sea Shelf. Deep-Sea Research II 52, 3150–3174. Weslawski, J.M., Wiktor Jr., J., Kotwicki, L., 2010. Increase in biodiversity in the arctic rocky littoral, Sorkappland, Svalbard, after 20 years of climate warming. Marine Biodiversity 40, 123–130. Wilt, L.M., 2012. Caloric Content of Bering and Chukchi Sea Benthic Invertebrates (MS thesis). University of Maryland. Available from: 〈http://drum.lib.umd.edu/ handle/1903/13271〉.