Densities of Wintering Scoters in Relation to Benthic Prey Assemblages in a North Atlantic Estuary Author(S): Pamela H

Home , Anomia simplex, Mergini

Densities of Wintering Scoters in Relation to Benthic Prey Assemblages in a North Atlantic Estuary Author(s): Pamela H. Loring , Peter W. C. Paton , Scott R. McWilliams , Richard A. McKinney and Candace A. Oviatt Source: Waterbirds, 36(2):144-155. 2013. Published By: The Waterbird Society URL: http://www.bioone.org/doi/full/10.1675/063.036.0204

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/ page/terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non- commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research. Densities of Wintering Scoters in Relation to Benthic Prey Assemblages in a North Atlantic Estuary

Pamela H. Loring1,4,*, Peter W. C. Paton1, Scott R. McWilliams1, Richard A. McKinney2 and Candace A. Oviatt3 1Department of Natural Resources Science, University of Rhode Island, 1 Greenhouse Road, Kingston, RI, 02881, USA

2U.S. Environmental Protection Agency, Atlantic Ecology Division, 27 Tarzwell Drive, Narragansett, RI, 02882, USA

3Graduate School of Oceanography, University of Rhode Island, 215 South Ferry Road, Narragansett, RI, 02882, USA

4Current address: Department of Environmental Conservation, 160 Holdsworth Way, University of Massachusetts, Amherst, MA, 01003, USA

*Corresponding author; E-mail: [email protected] Abstract.—During winter, molluscivorous sea ducks often form high density feeding flocks in association with patchily distributed prey, although few studies have documented the substrate and prey characteristics where sea ducks (tribe: Mergini) aggregate and thus what constitutes high-value sea duck foraging habitat. Sea duck surveys were conducted and benthic grab samples collected at sites with different benthic substrate characteristics to compare macroinvertebrate community assemblages in relation to densities of three species of sea ducks: (Surf (Mela- nitta perspicillata), Black (M. americana), and White-winged (M. fusca) scoters (hereafter “scoters” when combined) during winter (2010-2011) in Narragansett Bay, Rhode Island, an urbanized estuary in the Northwest Atlantic. Overall, the highest densities of scoters (104 ± 17 per 0.5 km2) were found over sand substrates with homogeneous assemblages of infaunal prey. Significantly lower densities of scoters (21 ± 4 per 0.5 km2) were associated with mixed sand-gravel-mud substrates that supported epifaunal assemblages and patchily distributed infauna. Mean energy densities (kcal/g) were higher among infaunal (0.64-4.49) vs. epifaunal (0.17-0.53) prey. Overall biomass (g/m2) of polychaetes was higher in mixed substrates, and biomass of infaunal bivalves did not significantly differ among habitat type. However, infaunal prey may have been less accessible to scoters at sites with mixed substrates due to a barrier effect created by high-densities of epifauna with low energetic value. Thus, sand-substrate sites supporting infaunal benthic communities may provide high-quality feeding habitat for scoters wintering in the Northwest Atlantic. Understanding the influences of benthic habitat characteristics and macroinvertebrate prey assemblages on distribution of feeding scoters is particularly important for managing sea ducks in areas with increasing anthropogenic development in the coastal zone. Received 16 November 2012, accepted 13 February 2013. Key words.—benthic community, habitat use, Melanitta, Mergini, Narragansett Bay, scoters, winter surveys. Waterbirds 36(2): 144-155, 2013

Prey availability and quality are among high densities and biomass of prey, includ- the most important habitat characteristics ing bivalves and other benthic macroinverte- affecting the distribution and abundance brate taxa (Guillemette et al. 1993). Despite of many animals (Pyke et al. 1977), includ- the assumed importance of benthic prey to ing sea ducks (tribe: Mergini) during winter many sea ducks, few studies have explicitly (Guillemette et al. 1992; Lewis et al. 2008). examined how local-scale spatial variation Winter is energetically demanding for sea in benthic community composition relates ducks, in part due to increased thermoreg- to sea duck abundance (Degraer et al. 1999; ulatory costs (McKinney and McWilliams Larsen and Guillemette 2000; Kirk et al. 2005), so quality and availability of prey may 2008). be particularly important for these species. The distribution of benthic macroinverte- Decreases in winter food availability have brates is spatially heterogeneous and driven been linked to mass sea duck mortality and by dynamic biophysical interactions between decreased reproductive output the following the water column and seafloor (Zajac 2008). breeding season (Camphuysen et al. 2002; In shallow, subtidal habitats of western Atlan- Oosterhuis and van Dijk 2002). Sea ducks tic estuaries and coastal areas, latitudinal re- are characteristically aggregated during gion, salinity, and sediment type have strong nonbreeding periods and may form high- influences on the distribution and abundance density feeding flocks at sites with relatively patterns of benthic macroinvertebrate assem-

144 Scoter Densities and Benthic Prey 145 blages. A distinct benthic macroinvertebrate sub-region has been identified from New York Bight to Cape Cod (Hale 2010). Within Narra- gansett Bay and adjacent Rhode Island Sound, sediment grain size is an important factor affecting variation in benthic species composition, diversity, and abundance (Frithsen 1989; Theroux and Wigley 1998). Individual benthic macroinvertebrates also differ widely in value as potential sea duck prey because of variation in shell mass, size, energy and nutrient content, burial depth, and accessibility (Hamilton et al. 1999; Richman and Lovvorn 2003). Thus, information on the distribution, abundance, and relative energetic value of potential prey species is important for understanding the ecology of wintering sea ducks. On the Atlantic coast of North America, sea duck managers have identified a need to better understand sea duck ecology in relation to benthic systems (Sea Duck Joint Venture 2008). Identifying characteristics associated with high-value sea duck forag- Figure 1. Bathymetry of six land-based survey sites (numbered points), six 0.5 km2 study plots (boxes), ing habitat is particularly important for and 36 benthic grab sampling stations (triangles) in the monitoring and managing sea duck popu- Sakonnet Passage in Narragansett Bay, Rhode Island, lations in areas subjected to increasing lev- sampled during winter of 2010-2011. els of anthropogenic impacts in the coastal zone (Sea Duck Joint Venture 2008). Within Island (Fig. 1). Narragansett Bay is a large estuary (342 Rhode Island’s Narragansett Bay, an urban- km2), with semi-diurnal tides (mean range = 1.1 m) and ized estuary in the Northwest Atlantic, we a mean depth of 7.8 m. One of three major north-south passages that divide the bay is the Sakonnet Passage, investigated associations between benthic which covers 61 km2 and has a mean depth of 6.5 m habitat characteristics, macroinvertebrate (Chinman and Nixon 1985). The Sakonnet Passage community assemblages, and densities of is mixed primarily by tidal currents averaging 0.5-1.0 three species of scoters (Surf (Melanitta knots and has a mean salinity (31‰) similar to the ad- perspicillata), Black (M. americana), and jacent Rhode Island Sound (Hicks 1959). Sediment classification in Narragansett Bay is pre- White-winged (M. fusca) scoters; hereafter dominantly clayey silt and sand-silt-clay, although the “scoters”) during winter. Our specific ob- Sakonnet Passage has a unique and abrupt substrate jectives were to: 1) assess scoter densities grain size transition zone that occurs approximately during winter among sites that differed by mid-way between its northern headwaters and south- substrate type; 2) compare scoter densities ern entrance to Rhode Island Sound. The northern (upper) reach is mixed silt-clay-gravel (< 25% sand), among sites as a function of benthic macro- whereas the southern (lower) reach is predominately invertebrate community composition; and (> 75%) well-sorted sand (McMaster 1960). This tran- 3) quantify interspecific variation in en- sition zone provided us a unique opportunity to com- ergetic value of various species of benthic pare benthic prey communities in relation to sea duck macroinvertebrates as prey for scoters. densities among selected sites with different substrate types. Baseline data from the 2004-2010 Narragansett Bay Methods Winter Waterfowl Survey indicated that a large proportion of scoters in Narragansett Bay used the Sakonnet Passage during winter (U.S. Environmental Protection Study Area and Site Delineation Agency 2010). To better understand the relationship We conducted fieldwork in the Sakonnet Passage between scoter densities, benthic substrate classifica- within the eastern portion of Narragansett Bay, Rhode tion, and community assemblages of potential benthic 146 Waterbirds macroinvertebrate prey, we selected six 0.5 km2 study replicate grabs (0.04 m2 each) per sampling station sites in the Sakonnet Passage that had similar mean wa- for a representative area of 0.12 m2. We multiplied ter depths (4.8 m; range 3.5-6.6 m) and that were strati- overall abundance of each genera per station by 8.33 fied by benthic substrate classificationn ( = 3 sites from to estimate the mean number of individuals per m2. the lower and upper reach, respectively; Fig. 1). We Statistical Analyses used ArcGIS (Environmental Systems Research Insti- tute 2009) to delineate site boundaries as 0.5 km2 poly- Sea duck densities. We retained, for subsequent gons extending seaward from each land-based survey analyses, scoter surveys with observation conditions point and generated six random points (stratified by we classified as excellent, good, or fair and with dis- distance to shore) within each site for benthic macro- turbance events classified as none. We calculated invertebrate sampling. monthly averages of scoters (all species combined) from weekly survey counts conducted at each site Sea Duck Surveys from December 2010 to March 2011. The following From 15 December 2010 to 31 March 2011, an ob- statistical analyses were conducted using SAS (SAS In- server conducted weekly sea duck counts from fixed, stitute, Inc. 2008). To determine if scoter abundance land-based survey points adjacent to each 0.5 km2 significantly differed between upper and lower strata, study site (n = 6; Fig. 1). Surveys lasted 30 min and we used a mixed effects model (Proc GLIMMIX) to were completed within 4 hr following local morning fit models with total scoter abundance (monthly aver- civil twilight (U.S. Naval Observatory 2010). Before ages of all species combined) as the response vari- each survey was initiated, observers classified overall able, month and stratum (upper or lower Sakonnet observation conditions as poor, fair, good, or excel- Passage) and the interaction between month and lent based on wind speed, visibility, wave height, and stratum as fixed effects, and site as a random effect. precipitation. Observers then used a 20-60 power We used an Autoregressive (AR-1) covariance struc- spotting scope (Swarovski HD-ATS 80) and binocu- ture to account for correlation of survey data col- lars to count, by species, all scoters on the water with- lected at fixed sites over time. We assessed the sig- in site plot boundaries, which were visually estimated nificance of each fixed effect using Wald Statistics for using established landmarks. Disturbance events Type III contrasts. were classified as: none, hunting, boat, or other. To Benthic biomass. We used allometric modeling pro- quantify foraging behavior, the estimated percentage cedures outlined in McKinney et al. (2004) to estimate of scoters observed diving during the entire 30-min genus-level dry biomass for genera that comprised > survey period was scored as: 1 (0-25%), 2 (26-50%), 3 5% of overall or within-stratum benthic macroinver- (51-75%), or 4 (76-100%). tebrate abundance. For each grab sample, we measured length and dry mass of shell and tissue from at Benthic Macroinvertebrate Sampling least 10 randomly selected individuals representing We sampled benthic macroinvertebrates at six each genus. We collected length measurements with randomly located stations within each site from 14 digital calipers to nearest 0.01 mm along the longest October 2010 to 8 April 2011. At each station, we body axis. After drying at 60 °C for 48 hr, we mea- used a 0.04 m2 Young-modified Van Veen grab af- sured mass using a Mettler scale with an accuracy of fixed with a planar-view camera to collect sediment 0.0001 g. For molluscs, we removed soft tissue from samples and underwater video of the benthos. Grabs shell parts and weighed separately. We measured with level sediment to a depth of 7 to 10 cm were whole animal mass for amphipods and polychaetes. considered successful. We collected three replicate, We fit linear and non-linear regression models successful grabs per station. We discarded unaccept- to explain the relationship between dry biomass and able grabs (e.g., overfilled or washed out). After col- length. For all taxa, power models (y = axb) provided lection, we immediately rinsed contents with seawater the best fit (highest 2r coefficient) for explaining the over a stainless steel No. 12 (1.7 mm mesh size) sieve relationship between length (x) and whole animal and packed the samples on ice. This relatively coarse (shell + tissue for molluscs) dry mass (y). We used mesh size may have under-sampled the density and these models to estimate biomass at each sampling biomass of small macrofauna such as amphipods, station using the average shell length of 10 randomly polychaetes, and isopods (Schlacher and Wooldridge selected individuals per genus. For amphipods, we 1996). However, diet studies indicate that while in estimated dry biomass using an overall average dry marine habitats, scoters primarily consume macro- mass value of individuals (n = 100) collected from invertebrates exceeding 5 mm in size (Savard et al. Narragansett Bay (U.S. Environmental Protection 1998), so we assumed that the 1.7 mm mesh was ad- Agency, unpubl. data; Appendix). We combined ma- equate for characterizing assemblages of potential rine worms of the genera Nereis, Nephtys, and Glycera prey available to scoters within our study sites. We at the suborder level Phyllodocida. Because biomass either stored samples at 4 °C and processed them data were not normally distributed, we compared bio- within 48 hrs, or froze samples at -15 °C for later pro- mass of benthic prey between upper and lower strata cessing. In the laboratory, we sorted samples by taxa using Wilcoxon rank-sum tests. Specifically, we tested with all individuals identified to species or lowest pos- for differences in biomass of infaunal bivalves and sible taxon, up to the class level. We pooled the three polychaetes as previous research has shown that these Scoter Densities and Benthic Prey 147 prey categories comprise relatively high proportions of scoter diets (e.g., Stott and Olsen 1973; Fox 2003; Anderson et al. 2008). We estimated percent shell mass for each molluscan genus by calculating: mass of shell / (mass of tissue + shell). We used the following published soft- tissue energy equivalents to estimate overall energy density of biomass-modeled benthic macroinvertebrates in kilocalories (kcal) per unit overall dry mass: McKinney et al. (2004) for bivalves and gastropods, and McKinney et al. (2004) or Steimle and Terranova (1985) for polychaetes and molluscs. We compared energy densities across molluscan genera with varying proportions of shell biomass using paired Wilcox- on tests with Bonferroni-adjusted significance values. Figure 2. Seasonal variation in scoter densities (mean ± Benthic community multivariate analyses. We used SE birds 0.5 km-2) at upper and lower sites in the Sakon- Plymouth Routines in Multivariate Research software to net Passage in Narragansett Bay, Rhode Island, based analyze benthic community structure between strata on on surveys conducted weekly at each site from 15 De- 4th-root transformed biomass data (PRIMER; Clarke and cember 2010 to 31 March 2011. Counts of Surf, Black, Gorley 2006). This transformation is conventional for and White-winged scoters are pooled for this analysis. analysis of such benthic data to reduce the influence of the most dominant taxa (Clarke and Gorley 2006). We calculated resemblance between sampling stations using Densities and Biomass of Benthic Macroin- the Bray-Curtis similarity coefficient. From the Bray-Cur- vertebrates tis similarity matrix, we created a dendogram using the group-average linkage method. Non-metric multi-dimen- Benthic macrofaunal abundance (indi- sional scaling (MDS) ordination was used to graphically viduals m-2) at upper sites was dominated display, in two-dimensional space, relative Bray-Curtis dis- similarity distances between sampling stations, with con- by epifaunal common slipper snails (Crep- tour lines delineating sampling stations with Bray-Curtis idula fornicata) and common jingleshells similarity values ≥ 0.5. We used the ANOSIM (analysis of (Anomia simplex) (Table 1). These high similarities) procedure to test the a priori null hypothesis density Crepidula-Anomia beds covered of no assemblage differences between sampling stations 66% (n = 12) of the sampling stations at stratified by position. We then used the SIMPER (similarity percentages) sub-routine to quantify the contribution upper sites. The remaining 34% of upper- of each genus to the within-group similarity of upper vs. site sampling stations contained relatively lower strata (Clarke and Gorley 2006). high densities of infaunal Tellinid bivalves. Benthic macrofaunal abundance (individuals m-2) at lower sites was dominated Results by relatively homogeneously distributed, tube-dwelling Ampeliscid amphipods, gas- Densities of Scoters tropods of the genus Ilyanassa, and Tellinid We observed significantly higher average bivalves (Table 1). Overall, benthic bio- densities of scoters within lower vs. upper Sa- mass (g m-2) of Phyllodocid polychaetes was higher in upper vs. lower strata (Z = 2.073, konnet Passage sites (F1,16 = 5.47; P = 0.03) with no evident monthly variation during P = 0.038; Fig. 3). Biomass of infaunal bivalves did not differ between strata (Z = winter (F3,16 = 1.31; P = 0.30; Fig. 2). Scoter density (mean [± SE] individuals 0.5 km-2) -1.191, P = 0.234). within upper sites was 20.64 ± 4.04 (range 2-115; 84% Surf, 10% White-winged, 6% Food Value of Benthic Macroinvertebrates Black scoters). Scoter density within lower sites averaged 104.38 ± 17.42 (range 8 - 500; Energy density was inversely related 69% Surf, 0% White-winged, and 31% Black to proportion of shell biomass (Table 2). scoters). Over 50% of scoters observed were Shell-less marine worms of the order Phyl- diving during each survey across strata, sug- lodocida had the highest overall energy den- gesting that all sites were used as feeding sity (4.49 kcal/g), followed by Ampeliscid habitat. amphipods (2.51 kcal/g) and the relatively 148 Waterbirds

Table 1. Mean (± SE) dry biomass (g m-2), density (individuals m-2), frequency of occurrence (percent of grabs pres- ent), and overall range of lengths (mm) of benthic macroinvertebrate taxa comprising > 5% of overall abundance of benthic macroinvertebrates collected from upper and lower sites in the Sakonnet Passage, Narragansett Bay, Rhode Island, during the winter of 2010-2011. Taxaa Biomass Density Frequency Length Upper Sites

Crepidula 863.83 ± 267.48 2,071 ± 539 94 10-23 Anomia 127.13 ± 37.33 186 ± 50 83 11-27 Ilyanassa 5.97 ± 1.29 39 ± 11 72 6-26 Pagurus 5.68 ± 1.45 11 ± 3 67 5-17 Phyllodocidab 3.13 ± 1.02 34 ± 8 94 27-207 Tagelus 1.86 ± 1.50 30 ± 23 17 9-23 Tellina 1.41 ± 0.80 51 ± 24 61 5-12 Ensis 0.81 ± 0.43 2 ± 1 22 27-45 Ampelisca 0.01 ± 0.01 44 ± 30 22 2-6c Lower Sites Pagurus 8.79 ± 3.36 10 ± 3 50 8-16 Ilyanassa 6.72 ± 2.43 45 ± 16 89 7-15 Phyllodocidab 0.91 ± 0.28 26 ± 5 89 16-63 Ensis 0.81 ± 0.19 15 ± 3 72 12-34 Mulinia 0.76 ± 0.39 7 ± 3 39 6-17 Tellina 0.57 ± 0.12 41 ± 8 83 5-10 Ampelisca 0.09 ± 0.02 495 ± 110 83 2-6c

aThe remaining 5% of overall biomass of benthic macroinvertebrates was composed of the following taxa: Acmaea, Anachis, Anadara, Asterias, Busycon, Cancer, Carcinus, Diopatra, Epitonium, Gemma, Haminoea, Isopoda, Littorina, Mercenaria, Mitrella, Nucula, Ovalipes, Palaemonetes, Pandalus, Panopeus, Pectinaria, Sabellaria, Spiochaetopterus, Turbonilla, and Yoldia. bIncludes polychaetes of the genera Nereis, Nephtys, and Glycera. cLength range derived from Ampelisca sampled throughout Narragansett Bay, RI (U.S. Environmental Protection Agency, unpubl. data). low shell mass (< 75%) infaunal bivalves of Multivariate Benthic Community Analyses the genera Tagelus and Ensis (1.20 and 1.12 kcal/g, respectively). The genera with the An Analysis of Similarities Test (ANO- highest shell mass, Anomia (96%) and Crep- SIM) indicated a significant difference in idula (93%), had the lowest overall energy benthic macroinvertebrate community com- densities (0.17 and 0.26 kcal/g, respective- position between upper vs. lower sites (glob- ly). al R = 0.406, P < 0.01). The upper stations

Figure 3. Fourth-root transformed mean (± SE) biomass (g m-2) of benthic macroinvertebrate taxa by stratum (upper vs. lower sites), arranged along a continuum of low to high energy densities (including shell). Scoter Densities and Benthic Prey 149

Table 2. Food value of benthic macroinvertebrate genera collected in Narragansett Bay, Rhode Island. Food value relates to ease of capture and nutritional quality which here includes: prey mobility, exposure (infauna or epifauna), percent shell mass (dry mass of shell/dry mass of shell and tissue), and overall mean (± SE) energy density (kcal/g dry mass of whole animal). Taxa are arranged on a continuum of high to low energy densities. Molluscan genera connected by a vertical line did not significantly differ in energy density (P < 0.002, Wilcoxon, Bonferroni corrected). Genera Mobility and Exposure Shell Mass (%) Energy Density Phyllodocida Highly mobile marine worm — 4.49 Ampelisca Tube-dwelling amphipod — 2.51 Tagelus Sessile infaunal bivalve 72.38 ± 2.26 1.20 ± 0.10 Ensis Sessile infaunal bivalve 74.36 ± 2.69 1.12 ± 0.12 Tellina Sessile infaunal bivalve 81.61 ± 0.76 0.80 ± 0.03 Mulinia Sessile infaunal bivalve 84.80 ± 0.89 0.64 ± 0.04 Ilyanassa Epifaunal gastropod 84.42 ± 0.74 0.53 ± 0.03 Pagurus Highly mobile epifaunal crustacean 90.16 ± 2.15 0.30 ± 0.06 Crepidula Sessile epifaunal gastropod 92.54 ± 0.45 0.26 ± 0.02 Anomia Sessile epifaunal gastropod 96.05 ± 0.52 0.17 ± 0.02 show relatively scattered (variable) benthic Benthic macroinvertebrates comprising macroinvertebrate composition compared this cluster were largely represented by the to the more clustered (homogeneous) ben- taxa Ilyanassa, Phyllodocida, Tellina, and En- thic macroinvertebrate composition among sis. The other cluster contained the major- the lower sampling stations (Fig. 4). ity (n = 11) of upper sampling stations and The MDS ordination delineated two ma- was dominated by large proportions of low jor clusters of sampling stations. Sampling energy density genera such as Crepidula and stations comprising each cluster shared at Anomia. least 50% Bray-Curtis similarity in benthic The SIMPER sub-routine indicated that macroinvertebrate biomass composition. overall similarity in benthic macroinver- One cluster contained the majority (n = tebrate taxa was greater among the lower 16) of lower sampling stations and a lesser (61%) vs. upper (49%) sampling stations, number (n = 7) of upper sampling stations. indicating that benthic communities at upper sites were more variable in composition of macroinvertebrate taxa. Benthic macroinvertebrates that were consistently found within the lower sites included relatively high-energy potential prey items such as Ilya- nassa, Phyllodocida, Tellina, and Ensis (Fig. 5). Low-energy epifauna of the genera Crepidu- la, Anomia, and Pagurus contributed to over 50% of the benthic community composition within the upper sites.

Discussion

Overall, we found the highest mean den- -2 Figure 4. Non-metric multidimensional scaling ordina- sities of scoters (approximately 200 km ) in tion plot of fourth-root transformed benthic macroin- the lower vs. upper Sakonnet Passage. Mc vertebrate biomass data, showing relative Bray-Curtis Kinney (2004) estimated that the mean den- similarities between upper and lower sampling stations sity of winter waterfowl (all species combined) (light gray upward and black downward facing triangles, -2 respectively; stress = 0.14) in the Sakonnet Passage in in Narragansett Bay was 39 individuals km , Narragansett Bay, Rhode Island. Contour lines delin- which was similar to the mean density of sco- eate stations with Bray-Curtis similarity values ≥ 0.50. ters (41 km-2) that we detected in the upper 150 Waterbirds nearshore Rhode Island Sound (McMaster 1960). Conversely, benthic habitats in the upper Sakonnet are a mosaic of silt, clay, and gravel sediments. These mixed sediments are more typical of sediment composition found throughout Narragansett Bay (Mc- Master 1960). We found in this study that differences in sediment characteristics in the lower vs. upper Sakonnet Passage were associated with distinct assemblages of benthic macroinvertebrate taxa. Sandy substrates in the lower Sakonnet supported dense beds of Ampeliscid amphipods, as well as relatively high biomass of Ilyanassa whelks, and the infaunal bivalves Tellina and Ensis. At upper Sakonnet Passage sites over mixed gravel, sand, or mud substrates, epifaunal beds of Crepidula fornicata dominated the biomass and we found significantly higher biomass of Phyllodocid worms relative to lower sites. Stickney and Stringer (1957) also found increased densities and larger sizes of Nereis Figure 5. Contribution (%) of benthic taxa to the av- polychaetes occurring within Crepidula beds erage Bray-Curtis similarity between sampling stations in Narragansett Bay. stratified by position (upper vs. lower) in the Sakonnet High-density populations of Ampeliscid Passage in Narragansett Bay, Rhode Island. Taxa are amphipods or Crepidula create distinct habi- shaded along a gradient of lowest energy density (light) to highest energy density (dark) and are denoted within tats by directly modifying the sediment and each slice by the following letters: Ph (Phyllodocida), Am distribution of benthic fauna (French et (Ampelisca), E (Ensis), T (Tellina), I (Ilyanassa), Pg (Pagu- al. 1992). High densities of Ampeliscid am- rus), C (Crepidula), An (Anomia). phipods alter the surficial sediment layer through tube-building, which may enhance Sakonnet Passage. Scoters are likely attract- growing conditions for infaunal bivalves ed to the lower Sakonnet Passage by favor- (MacKenzie et al. 2006). Dense Crepidula able abiotic and biotic habitat conditions. A beds form biogenetic reefs that increase previous study found that in Narragansett habitat complexity and provide refuge for Bay, the abundance of open-water habitat other species of benthic invertebrates (Lind- species (including Surf, Black, and White- sey et al. 2006). Thus, benthic community winged scoters) decreased with increasing structure within estuarine systems appears residential development along adjacent to be largely influenced by interactions be- shorelines and that sites used by wintering tween benthic substrate type and dominant waterfowl in the Sakonnet had less shoreline macroinvertebrate species. development than sites located elsewhere in Energy density of potential prey items the bay (McKinney et al. 2006). within the upper and lower Sakonnet ben- Past research has shown that substrate thic communities differed among taxa and, type influences benthic community struc- for hard-bodied organisms, was inversely re- ture in marine environments (e.g., Zajac lated to mean amount of shell mass. Anomia et al. 2000). The lower Sakonnet Passage is and Crepidula had the highest overall shell unusual, compared to most of Narragan- mass and lowest overall energy densities of sett Bay, in that the dominant benthic sedi- all taxa we sampled. Among molluscs, razor ment type was well-sorted sand that is more clams (i.e., Tagelus, Ensis) and Tellins had the characteristic of benthic habitats within lowest shell mass and highest energy density. Scoter Densities and Benthic Prey 151 Sea ducks select prey items with the highest with spatially variable prey supported lower tissue to shell ratio when offered bivalves densities of birds. varying by length and shell morphology Our findings indicate that nearshore, (Bustnes 1998), and this prey selection strat- sandy habitats support relatively high-value egy would provide sea ducks in our study benthic prey communities that likely pro- area with more energy-dense prey. Mini- vide high-quality foraging habitat for sco- mizing amounts of ingested shell may also ters on southern New England wintering reduce the energy required to crush and grounds. Stott and Olsen (1973) found that process indigestible materials (Richman and in northern New England, Surf Scoters also Lovvorn 2003). selected sites with sandy substrates, where Soft-bodied organisms such as Ampelisca they fed primarily on Arctic wedge clams and Phyllodocid marine worms provided the (Mesodesma arctatum) and Atlantic razor highest energy per unit mass of all the taxa clams (Siliqua costata). The razor clam (En- that we sampled, although their contribution sis directus) is invasive to sand-flat habitats to diets of sea ducks depends in part on body in Western Europe, where it has become size. For example, small, mobile prey, such an important prey item for scoters (Tulp et as amphipods, may not provide a sufficient al. 2010). Scoters in Western Europe have amount of predictable biomass for large- also been associated with high densities of bodied sea ducks during winter (Goudie and Tellinid bivalves in shallow (< 10 m), sandy Ankney 1986). Larger-bodied White-winged habitats (Degraer et al. 1999). Overall bio- Scoters tend to consume more hard-shelled mass of infaunal bivalves was not signifi- prey, whereas smaller-bodied Surf Scoters cantly different between upper vs. lower usually consume a mixture of bivalves and Sakonnet Passage sites. However, Ensis and soft-bodied prey (Anderson et al. 2008). Tellins were evenly distributed at sites in the Black Scoters, the smallest of the three spe- Sakonnet Passage with relatively high densi- cies, consume bivalves, polychaetes, and am- ties of scoters. Conversely, we found highly phipods in marine environments (Bordage heterogeneous distributions of Tellins and and Savard 1995). The range of size and en- razor clams of the genus Tagelus at sites with ergy densities of prey available in the Sakon- relatively low densities of scoters. The major- net Passage may allow for resource partition- ity of sampling stations at sites with low sco- ing among the different-sized scoter species ter use were dominated by high densities of feeding within mixed flocks (Richman and Crepidula, which formed pavement-like beds Lovvorn 2009). over the substrate that may have prevented Flocking behavior of wintering sea ducks scoters from accessing infaunal prey in these helps them to locate high-quality feeding areas (Witman 1985). Diet analyses from sea sites (Guillemette et al. 1993) that are tempo- ducks collected throughout the U.S. Atlan- rally and spatially variable (Larsen and Guil- tic coast indicate that scoters do not typically lemette 2000). The highest densities of sco- consume Crepidula fornicata (P. Osenton, un- ters in our study were associated with sand publ. data). Thus, in the Sakonnet Passage, substrates that supported homogeneous dis- more scoters are attracted to sand-substrate tributions of benthic macroinvertebrate taxa habitats that support homogeneous distribu- with relatively high energy per unit mass. tions of infaunal bivalves relative to mixed- Smaller flocks of scoters were associated substrate habitats with variably distributed with mixed-substrate habitats that support- infaunal and epifaunal prey. It may also be ed benthic communities composed of taxa easier for scoters to detect and sift prey items with heterogeneous energy densities (i.e., from clean, sandy substrates relative to finer 66% low-value epifaunal Crepidula beds, 33% substrates with higher organic content (Fox high-density, energetically valuable infauna) 2003). and accessibility. Our survey data showed During the non-breeding season, scoters relatively consistent numbers of birds within forage over both hard and soft substrate each stratum during winter, although sites benthic habitats (Perry et al. 2007). More 152 Waterbirds information on associations between scoters cornell.edu.bnaproxy.birds.cornell.edu/bna/spe- and hard substrate prey communities, such cies/177, accessed 10 June 2012. as blue mussels ( s), is needed Bustnes, J. O. 1998. Selection of blue mussels, Mytilus Mytilus eduli edulis, by Common Eiders, Somateria mollissima, by to better understand foraging dynamics of size in relation to shell content. Canadian Journal scoters in southern New England. Although of Zoology 76: 1787-1790. blue mussels are a predominant prey spe- Camphuysen, C. J., C. M. Berrevoets, H. J. W. M. Cre- cies for scoters wintering on the Canadian mers, A. Dekinga, R. Dekker, B. J. Ens, T. M. Van Atlantic coast (Goudie and Ankney 1986), der Have, R. K. H. Kats, T. Kuiken, M. F. Leopold, J. van der Meer and T. Piersma. 2002. Mass mortal- the abundance and stability of blue mussel ity of Common Eiders (Somateria mollissima) in the populations in southern New England may Dutch Wadden Sea, winter 1999/2000: starvation be constrained by lack of suitable hard sub- in a commercially exploited wetland of interna- strate and increased heat stress at this south- tional importance. Biological Conservation 106: ern limit of the species range (Tam and 303-317. Chinman, R. A. and S. W. Nixon. 1985. Depth-area- Scrosati 2011). In Narragansett Bay, blue volume relationships in Narragansett Bay. NOAA/ mussels form temporally variable and patch- Sea Grant Marine Technical Report 87, University ily distributed epifaunal reefs (French et al. of Rhode Island, Graduate School of Oceanogra- 1992). Therefore, infaunal bivalves occupy- phy, Narragansett, Rhode Island. ing sandy substrates may provide a more Clarke, K. R. and R. N. Gorley. 2006. PRIMER v.6. User Manual/Tutorial. PRIMER-E Ltd., Plym- stable source of prey capable of supporting outh, U.K. a local population of scoters throughout Degraer, S., M. Vincx, P. Meire and H. Offringa. 1999. winter, whereas mussel-dominated habitats The macrozoobenthos of an important wintering over hard substrate may be more susceptible area of the Common Scoter (Melanitta nigra). to depletion by foraging scoters (Kirk . Journal of the Marine Biological Association of et al the UK 79: 243-251. 2008). In summary, certain types of benthic Environmental Systems Research Institute (ESRI). sediments are associated with different mac- 2009. ArcInfo and ArcMap GIS software, 9. 3.1. roinvertebrate prey communities in south- Environmental Systems Research Institute, Red- ern New England that in turn determine the lands, California. distribution of wintering scoters, and this Fox, A. D. 2003. Diet and habitat use of scoters Mela- nitta in the western Palearctic - a brief overview. information is useful in predicting how off- Wildfowl 54: 163-182. shore development may affect these compo- French, D., H. Rines, J. Boothroyd, C. Galagan, M. nents of the ecosystem. Harlin, A. Keller, G. Klein-McPhee, S. Pratt, M. Gould and M. Villalard-Bohnsack. 1992. Atlas and habitat inventory/resource mapping for Nar- Acknowledgments ragansett Bay and associated coastlines. Final Re- port to the Narragansett Bay Project, Applied Sci- We thank Don Cobb and the U.S. Environmental ence Associates, Providence, Rhode Island. Protection Agency, Atlantic Ecology Division, for pro- Frithsen, J. 1989. The benthic communities within viding extensive support for benthic sampling. We also Narragansett Bay. Final Report to the Narragan- thank Joe Attwater for collecting survey data, and Pe- sett Bay Project, Rhode Island Department of nelope Pooler for providing statistical advice. Dan Esler Environmental Management, Providence, Rhode and John Takekawa provided helpful reviews that im- Island. Graduate School of Oceanography, Uni- proved the manuscript. Funding for this study was pro- versity of Rhode Island, Narragansett, Rhode Is- vided by the Rhode Island Department of Environmen- land. tal Management and the University of Rhode Island. Goudie, R. I. and C. D. Ankney. 1986. Body size, ac- tivity budgets, and diets of sea ducks wintering in Literature Cited Newfoundland. Ecology 67: 1475-1482. Guillemette, M., R. C. Ydenberg and J. H. Himmel- Anderson, E. M., J. R. Lovvorn and M. T. Wilson. 2008. man. 1992. The role of energy intake rate in prey Reevaluating marine diets of Surf and White-winged and habitat selection of Common Eiders Somateria Scoters: interspecific differences and the impor- mollissima in winter: a risk-sensitive interpretation. tance of soft-bodied prey. Condor 110: 285-295. Journal of Animal Ecology 61: 599-610. Bordage, D. and J.-P. L. Savard. 1995. Black Scoter Guillemette, M., J. H. Himmelman, C. Barette and A. (Melanitta americana). No. 177 in The Birds of North Reed. 1993. Habitat selection by Common Eiders America Online (A. Poole, Ed.). Cornell Laboratory in winter and its interaction with flock size. Cana- of Ornithology, Ithaca, New York. http://bna.birds. dian Journal of Zoology 71: 1259-1266. Scoter Densities and Benthic Prey 153

Hale, S. S. 2010. Biogeographical patterns of marine ben- tion by Spectacled Eiders in the Bering Sea. Marine thic macroinvertebrates along the Atlantic coast of the Ecology Progress Series 261: 283-297. northeastern USA. Estuaries and Coasts 33: 1039-1053. Richman, S. E. and J. R. Lovvorn. 2009. Predator size, Hamilton, D. J., T. D. Nudds and J. Neate. 1999. Size- prey size and threshold food densities of diving selective predation of blue mussels (Mytilus edulis) ducks: does a common prey base support fewer large by Common Eiders (Somateria mollissima) under con- animals? Journal of Animal Ecology 78: 1033-1042. trolled field conditions. Auk 116: 403-416. SAS Institute, Inc. 2008. SAS Statistical Software, v. 9.2 Hicks, S. D. 1959. The physical oceanography of Nar- Cary, North Carolina. ragansett Bay. Limnology and Oceanography 4: 316- Savard, J.-P., L. D. Bordage and A. Reed. 1998. Surf Sco- 327. ter (Melanitta perspicillata). No. 363 in The Birds of Kirk, M., D. Esler, S. A. Iverson and W. S. Boyd. 2008. North America Online (A. Poole, Ed.). Cornell Lab- Movements of wintering Surf Scoters: predator re- oratory of Ornithology, Ithaca, New York. http:// sponses to different prey landscapes. Oecologia 155: bna.birds.cornell.edu.bnaproxy.birds.cornell.edu/ 859-867. bna/species/363, accessed 23 Feb 2013. Larsen, J. K. and M. Guillemette. 2000. Influence of an- Schlacher, T. A. and T. H. Wooldridge. 1996. How sieve nual variation in food supply on abundance of win- mesh size affects sample estimates of estuarine ben- tering Common Eiders Somateria mollissima. Marine thic macrofauna. Journal of Experimental Marine Ecology Progress Series 201: 301-309. Biology and Ecology 201: 159-171. Lewis, T. L., D. Esler and W. S. Boyd. 2008. Foraging Sea Duck Joint Venture. 2008. Sea Duck Joint Venture behaviors of Surf and White-winged Scoters in rela- strategic plan 2008-2012. U.S. Fish and Wildlife Ser- tion to clam density: inferring food availability and vice, Anchorage, Alaska and Canadian Wildlife Ser- habitat quality. Auk 125:149-157. vice, Sackville, New Brunswick. Lindsey, E. L., A. H. Altieri and J. D. Witman. 2006. In- Steimle, F. W. and R. J. Terranova. 1985. Energy fluence of biogenic habitat on the recruitment and equivalents of marine organisms from the conti- distribution of a subtidal xanthid crab. Marine Ecol- nental shelf of the temperate northwest Atlantic. ogy Progress Series 306: 223-231. Journal of Northwest Atlantic Fisheries Science 6: MacKenzie, C. L., R. Pikanowski and D. G. McMillan. 117-124. 2006. Ampelisca amphipod tube mats may enhance Stickney, A. P. and L. D. Stringer. 1957. A study of the abundance of northern quahogs Mercenaria merce- invertebrate bottom fauna of Greenwich Bay, Rhode naria in muddy sediments. Journal of Shellfish Re- Island. Ecology 38: 111-122. search 25: 841-847. Stott, R. and D. Olson. 1973. Food-habitat relationship McKinney, R. A. 2004. Habitat relationships of water- of sea ducks on the New Hampshire coastline. Ecol- fowl wintering in Narragansett Bay. Rhode Island ogy 54: 996-1007. Naturalist 11: 3-6. Tam, J. C. and R. A. Scrosati. 2011. Mussel and dogwhelk McKinney, R. A. and S. R. McWilliams. 2005. A new distribution along the north-west Atlantic coast: test- model to estimate daily energy expenditure for win- ing predictions derived from the abundant-centre tering waterfowl. Wilson Bulletin 117:44-55. model. Journal of Biogeography 38: 1536-1545. McKinney, R. A., S. M. Glatt and S. R. McWilliams. 2004. Theroux, R. B. and R. L. Wigley. 1998. Quantitative Allometric length-weight relationships for benthic composition and distribution of the macrobenthic prey of aquatic wildlife in coastal marine habitats. invertebrate fauna of the continental shelf ecosys- Wildlife Biology 10: 241-249. tems of the northeastern United States. Volume 140. McKinney, R. A., S. R. McWilliams and M. A. Charpen- U.S. Dept. of Commerce, National Oceanic and At- tier. 2006. Waterfowl-habitat associations during mospheric Administration, National Marine Fisher- winter in an urban North Atlantic estuary. Biologi- ies Service, Scientific Publications Office. Washing- cal Conservation 132: 239-249. ton, D.C. McMaster, R. L. 1960. Sediments of Narragansett Bay Tulp, I., J. Craeymeersch, M. Leopold, C. van Damme, system and Rhode Island Sound, Rhode Island. F. Fey and H. Verdaat. 2010. The role of the invasive Journal of Sedimentary Research 30: 249-274. bivalve Ensis directus as food source for fish and birds Oosterhuis, R. and K. van Dijk. 2002. Effect of food in the Dutch coastal zone. Estuarine, Coastal and shortage on the reproductive output of Common Shelf Science 90: 116-128. Eiders Somateria mollissima breeding at Griend (Wad- U.S. Environmental Protection Agency. 2010. Narragan- den Sea). Atlantic Seabirds 4: 29-38. sett Bay winter waterfowl survey data and annual re- Perry, M. C., A. M. Wells-Berlin, D. M. Kidwell and P. ports. U.S. Environmental Protection Agency, Wash- C. Osenton. 2007. Temporal changes of populations ington, D.C. http://www.nbnerr.org/Content/ and trophic relationships of wintering diving ducks WaterfowlSurvey/data.htm, accessed 13 April 2013. in Chesapeake Bay. Waterbirds 30: 4-16. U.S. Naval Observatory. 2010. Astronomical applica- Pyke, G. H., H. R. Pulliam and E. L. Charnov. 1977. Op- tions civil twilight data. Astronomical Applications timal foraging: a selective review of theory and tests. Department, U.S. Naval Observatory, Washington, Quarterly Review of Biology 52: 137-154. D.C. http://www.usno.navy.mil/USNO/astronom- Richman, S. E. and J. R. Lovvorn. 2003. Effects of clam ical-applications/data-services, accessed 11 Novem- species dominance on nutrient and energy acquisi- ber 2012. 154 Waterbirds

Witman, J. D. 1985. Refuges, biological disturbance, lationships among sea-floor structure and benthic and rocky subtidal community structure in New communities in Long Island Sound at regional and England. Ecological Monographs 55: 421-445. benthoscape scales. Journal of Coastal Research 16: Zajac, R. N. 2008. Macrobenthic biodiversity and sea 627-640. floor landscape structure. Journal of Experimental Marine Biology and Ecology 366: 198-203. Zajac, R. N., R. S. Lewis, L. J. Poppe, D. C. Twichell, J. Vozarik and M. L. DiGiacomo-Cohen. 2000. Re- Scoter Densities and Benthic Prey 155

Appendix. Length range, total mean dry animal mass (g), and allometric power model parameters for estimating total dry mass (g) from length for the most abundant benthic macroinvertebrate genera collected in the Sakonnet Passage in Narragansett Bay, Rhode Island, during the winter of 2010-2011. Equations are in the form of y = axb where x is the shell length (total length for polychaetes) in mm, a is the length coefficient and b is the shell length exponent, and y is the predicted whole animal dry mass (g). Shell Length Whole Dry SE for Predicted Taxon (mm) Mass (g) n A b Mass (g) r2 Pagurus 13-28 1.8018 ± 0.31 10 0.00326 2.1242 0.31 0.94 Mulinia 4-19 0.1551 ± 0.05 10 0.00011 2.8244 0.05 0.99 Tellina 9-15 0.0453 ± 0.02 10 7.2 E-05 2.6078 0.01 0.84 Ensis 20-39 0.1307 ± 0.03 8 7.7 E-08 4.1392 0.03 0.99 Tagelus 14-20 0.0962 ± 0.01 9 4.7 E-06 3.4392 0.01 0.81 Anomia 15-32 0.7509 ± 0.12 10 0.00127 1.9944 0.12 0.76 Ilyanassa 12-18 0.4526 ± 0.03 11 0.00095 2.2298 0.04 0.92 Crepidula 9-28 0.5475 ± 0.20 10 5.8 E-07 4.5071 0.20 0.97 Phyllodocida 22-184 0.1604 ± 0.06 13 1.3 E-05 2.1098 0.06 0.93 Ampelisca 2-6 0.0002 ± 1.0 E-4a 100 — — — —

aDerived from dry-mass average of individuals collected within Narragansett Bay, RI.