Contents lists available at ScienceDirect

Continental Shelf Research

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

The death assemblage as a marker for habitat and an indicator of climatechange: Georges Bank, surfclams and ocean quahogs

Eric N. Powell⁎, Kelsey M. Kuykendall, Paula Moreno

Gulf Coast Research Laboratory, University of Southern Mississippi, 703 E, Beach Dr. Ocean Springs, MS 39564, United States

A R T I C L E I N F O

Keywords:SpisulaArcticaSurfclamOcean quahogDeath assemblageClimate changeHabitatGeorges BankContinental shelf

A B S T R A C T

A comprehensive dataset for the Georges Bank region is used to directly compare the distribution of the deathassemblage and the living community at large spatial scales and to assess the application of the deathassemblage in tracking changes in species’ distributional pattern as a consequence of climate change. Focus isplaced on the biomass-dominant clam species of the northwest Atlantic continental shelf: the surfclam Spisulasolidissima and the ocean quahog Arctica islandica, for which extensive datasets exist on the distributions ofthe living population and the death assemblage. For both surfclams and ocean quahogs, the distribution of deadshells, in the main, tracked the distribution of live animals relatively closely. Thus, for both species, the presenceof dead shells was a positive indicator of present, recent, or past occupation by live animals. Shell dispersionwithin habitat was greater for surfclams than for ocean quahogs either due to spatial time averaging, animalsnot living in all habitable areas all of the time, or parautochthonous redistribution of shell. The regionaldistribution of dead shell differed from the distribution of live animals, for both species, in a systematic wayindicative of range shifts due to climate change. In each case the differential distribution was consistent withwarming of the northwest Atlantic. Present-day overlap of live surfclams with live ocean quahogs was consistentwith the expectation that the surfclam's range is shifting into deeper water in response to the recent warmingtrend. The presence of locations devoid of dead shells where live surfclams nevertheless were collected measuresthe recentness of this event. The presence of dead ocean quahog shells at shallower depths than live oceanquahogs offers good evidence that a range shift has occurred in the past, but prior to the initiation of routinesurveys in 1980. Possibly, this range shift tracks initial colonization at the end of the Little Ice Age.

1. Introduction

Death assemblages have received much attention by taphonomistsin investigations relevant to the process of preservation and ultimatelythe interpretation of the fossil record. Applications of the deathassemblage in investigations of ecological change have been manyfewer, but these investigations demonstrate the potential of the deathassemblage as a long-term record of change in community structureand function. Kidwell (2007, 2008) considered community change inresponse to anthropogenic activities such as fishing and documentedthe record of such in the death assemblage. Aller (1995), Poirier et al.(2009), and Tomašových and Kidwell (2009), among others, consid-ered the record of spatial and environmental gradients as recorded inthe death assemblage. Stratigraphic variation in death assemblagecomposition records temporal changes in community structure wheresedimentation rate is sufficient to overcome time averaging (Gasseet al., 1987; Alin and Cohen, 2004; Powell et al., 1992). Warwick andLight (2002) and Tomašových and Kidwell (2010) considered applica-

tion of the death assemblage in estimating regional biodiversity.Extracting such information from the death assemblage is compro-

mised by a range of processes among the most important being spatialand temporal time averaging (Powell et al., 1989; Kidwell and Holland,2002; Kidwell et al., 2005; Dexter et al., 2014), taphonomic degrada-tion (Smith and Nelson, 2003; Kosnik et al., 2009; Powell et al., 2011),and resuspension and transport (Parsons and Brett, 1991; Zenatas,1990; Callender et al., 1992). Nonetheless, evidence of the potential ofthe death assemblage in the study of recent changes in communitycomposition over space and time continues to accumulate.

Few studies have considered the death assemblage of continentalshelves, particularly over large spatial scales (e.g., MacIntyre et al.,1978; Frey and Dörjes, 1988; Powell et al., 1990; Staff and Powell,1990a; Powell et al., 1998; Staff and Powell, 1999; Grill and Zuschin,2001) and fewer still have addressed use of the death assemblage intracking changes in community composition as a product of anthro-pogenic impact or climate change (e.g., Kidwell, 2007, 2008; Albanoet al., 2016; Tomašových et al., 2017). Climate change is substantively

http://dx.doi.org/10.1016/j.csr.2017.05.008Received 17 January 2017; Received in revised form 11 May 2017; Accepted 14 May 2017

⁎ Corresponding author.E-mail address: eric.n.powell@usm.edu (E.N. Powell).

Continental Shelf Research 142 (2017) 14–31

Available online 19 May 20170278-4343/ © 2017 Elsevier Ltd. All rights reserved.

MARK

affecting community structure over large spatial scales on continentalshelves today (Rose, 2005; Lucey and Nye, 2010; Perry et al., 2010).Most studies emphasize commercial species because long-term timeseries over large spatial scales are primarily contributed by routinestock surveys (Kerr et al., 2009; Brander, 2010; Perry et al., 2010).Typically, the benthos of the continental shelf are poorly surveyed (e.g.,van der Meer, 1997; Morehead et al., 2008; Powell and Mann, 2016)and few time series exist. Potentially, comparison between the deathassemblage and the living community might provide evidence of shiftsin distributional patterns in response to climate change when long-term time series do not exist. Unfortunately, evaluation of thispotential remains unaddressed, as surveys of adequate geographicscale are extremely rare. Of the few that exist, the most important maybe the survey datasets for the commercial clam species of the northwestAtlantic continental shelf: the surfclam Spisula solidissima and theocean quahog Arctica islandica.

Both surfclams and ocean quahogs are biomass dominants and longlived and thus provide a potentially rich opportunity to study theinfluence of climate change on the continental shelf benthos. The oceanquahog is a pan-boreal species with a western Atlantic range extendingsouth to near Chesapeake Bay (Merrill and Ropes, 1969; Dahlgrenet al., 2000). The population distribution has been stable in thenorthwest Atlantic region, probably due to the ability of the speciesto burrow and estivate during summer months at the southern andinshore extent of the range and thereby escape high summer bottomwater temperatures (Taylor, 1976a; Ridgway and Richardson, 2011).Recent documentation of population age frequencies from sites on theU.S. continental shelf, however, show that ocean quahogs occupiedtheir present range beginning near the end of the Little Ice Age, duringthe late 1700s to early 1800s (Pace et al., 2017; for more on the LittleIce Age, see Cronin et al., 2010; Mann et al., 2009; for timing of initialwarming in the northwest Atlantic, see Moore et al., 2017). That is, thespecies has shifted its range over historical time beginning more or lesscoincident with a period of rapid warming that began in the early 1800swith the present range established between 1860 and 1910. Presentbottom water temperatures impinge on this species’ thermal limitsalong the inshore boundary of the range, however, so that the species isbecoming increasingly sensitive to ongoing climate change (see alsoDahlgren et al., 2000 for a historical perspective).

In contrast, surfclams have proven to be much more sensitive toclimate change over the last two decades. Surfclams are generally notfound in areas where bottom temperature exceeds 25 °C (Cargnelliet al., 1999a); scope for growth becomes negative at temperaturesabove about 21 °C and animals starve to death if such high tempera-tures are experienced for extended periods (Kim and Powell, 2004;Narváez et al., 2015). Thus, the southern and inshore range boundariesare controlled primarily by bottom water temperatures in late summer-early fall (August–October). The sensitivity of surfclams to warm watertemperatures (Munroe et al., 2013; Narváez et al., 2015) positions thespecies to be sensitive to warming of the Mid-Atlantic Bight, the mostrecent phase of which began circa 1970 and accelerated circa 1990(Nixon et al., 2004; Friedland and Hare, 2007). In the 1960s and1970s, and probably at earlier times, the range of the Atlantic surfclamas documented by stock surveys extended from Georges Bank almost toCape Hatteras (e.g., Ropes, 1980, 1982) and encompassed the innerhalf of the continental shelf from the Chesapeake Bay mouth to HudsonCanyon at depths of 10–50 m, with nearshore populations along LongIsland and Southern New England (Goldberg and Walker, 1990;Weinberg, 1998; Jacobson and Weinberg, 2006), extending onto theshallower portion of Georges Bank (NEFSC, 2013). As a consequence ofrising bottom water temperatures, the southern and inshore rangeboundary of the Atlantic surfclam has shifted north and into deeperwaters (Cargnelli et al., 1999a; Weinberg, 2005; Munroe et al., 2013).Early evidence of this trend is the disappearance of surfclams fromVirginia and Maryland state waters between the 1970s and the 1990s(Loesch and Ropes, 1977; Powell, 2003) and the development of the

New Jersey state fishery in the 1990s. During the 1997–1999 period,the surfclam population was judged to be near carrying capacitythroughout most of its range (NEFSC, 2013). However, surveys in2002 revealed a large mortality event after 1999 that eliminatedsurfclams from the southern inshore region off the DelmarvaPeninsula, an event followed soon thereafter by stock declines in bothstate and federal waters off New Jersey (Kim and Powell, 2004). Theresults of an additional survey conducted in 2004 (Weinberg et al.,2005) confirmed the northward and offshore range shift.

The contraction of the southern boundary of the surfclam's rangehas not been compensated by an expansion northward at the northernboundary of the range. The stock appears to have extended into federaloffshore waters off eastern Long Island, but the degree to which thiswill continue and be substantive for the stock remains unclear. Thenorthern limit of the surfclam stock is primarily a result of limitedsandy habitat north of Georges Bank where only a limited extensioninto the inshore region of the Gulf of Maine is documented (Palmer,1991). Thus, the trends in aggregate have resulted in a contraction ofthe surfclam's range. Munroe et al. (2016) showed that maximum sizealso declined over much of the stock since 1980. Simulation modelingof surfclam population dynamics demonstrates that this outcome canbe derived solely from rising temperatures, as temperature influencesscope for growth primarily through its effect on filtration rate (Munroeet al., 2013, 2016), although a change in food supply would provide thesame outcome.

Here, we utilize a comprehensive long-term dataset for the region ofGeorges Bank, a database that includes documentation of the abun-dance of live surfclams and ocean quahogs and also their dead shells, todirectly compare the distribution of these species in the deathassemblage and the living community over large spatial scales and toassess the application of the death assemblage in tracking changes inspecies distributional pattern as a consequence of climate change.Georges Bank approaches the northern boundary of the surfclam'srange while being situated well within the center of the ocean quahog'srange (for additional documentation of the North Atlantic range ofocean quahogs, see Brey et al., 1990; Rowell et al., 1990; Ragnarssonand Thórarinsdóttir, 2002; Butler et al., 2009). Georges Bank providesa unique opportunity because clam fishing has been limited in thisregion by a multidecadal fishery closure (Jacobson and Weinberg,2006), the bank is primarily self-recruiting (Zhang et al., 2015, 2016),habitat complexity is pronounced (Collie et al., 1997), surfclams andocean quahogs are both benthic biomass dominants on the bank, andtheir survey data are extensive (e.g., Lewis et al., 2001; NEFSC, 2009,2013). In addition, an extensive dataset documents the geographicdistribution of dead surfclam and ocean quahog shells on the bank, aswell as a selection of substrate types that permit consideration ofhabitat as a modulator of clam distribution.

2. Methods

2.1. Data resources

Surfclam and ocean quahog survey data from 1980 to 2011 wereobtained from the NMFS-NEFSC (National Marine Fisheries Service –Northeast Fisheries Science Center) survey database. These datainclude standardized catch of live surfclams and ocean quahogs,information on substrate and in particular the presence of cobbles,rocks, and boulders, and documentation of the occurrence of oceanquahog and surfclam dead shells from each survey tow. The data wereobtained, in most cases, from triennial surveys that invoked a stratifiedrandom design. The gear was a hydraulic dredge, with well-knownselectivity and efficiency characteristics (NEFSC, 2009, 2013; see alsoMeyer et al., 1981; Smolowitz and Nulk, 1982; Hennen et al., 2012).Selectivity was good for live clams down to a size of approximately 50–55 mm. As the dredge liner controlled selectivity and no post-catchmechanical sorting occurred on the boat prior to on-deck sorting of

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

15

constituents by personnel, selectivity for dead clam shells and sedi-mentary constituents at least this large would have been about thesame.1 Tows were typically of 5-min duration at 1.5 knots.

2.2. Analytical approach

2.2.1. General considerationsSurvey tow locations were specified by the recorded position at the

initiation of the tow. Tows with similar initial positions were consid-ered replicates, a replicate being defined relative to a distanceapproximately twice the distance traveled by a typical survey tow:approximately 0.29 min of latitude or 0.39 min of longitude. Surveytows initiated within this distance apart were considered replicateseven if taken in different years. In general, the highest value amongstreplicates was taken for further analysis. This emphasized the presenceof indicators of complex habitat and also retained information on liveanimals that might not be stable constituents over the 30+ years of thesurvey time series.

Dead shells of ocean quahogs and surfclams can be considered to bestable constituents of the death assemblage over much, if not all, of theentirety of the survey time series. For ocean quahogs, for example,experimental deployment of ocean quahog shells in a variety of habitatsfor up to 13 years suggests that taphonomic loss rates are low (Powellet al., 2002, 2008, 2011) and this can be anticipated also to be the casefor surfclams based on the deployment of a range of clam species (seethese same studies), although in neither species’ case does directevidence exist for the Georges Bank region. Substantive anecdotalevidence does exist, however. Observation of many tows by the authorsshows that epibionts occur very infrequently on surfclam and oceanquahog dead shells and this is confirmed by documentation fromsurvey data where epibionts such as barnacles, anemones, tunicates,and other fauna were routinely recorded when present on dead shells.Such recordings are rare over the entirety of the 1980–2011 dataset.The absence of epibionts is consistent with rapid burial of these shells(Parsons-Hubbard et al., 1999; Hageman, 2001; Rodland et al., 2006;Powell et al., 2008) and, by inference, low rates of taphonomic loss(Walker and Goldstein, 1999; Tomašových et al., 2006; Powell et al.,2012).

Stability of the living population over time might not be questionedfor surfclams as life span is less than the duration of the time series andrecruitment, although relatively continuous (Weinberg et al., 1999;Lewis et al., 2001; Powell et al., 2016), is patchy. Thus, surfclams mightvary in their presence and extent of occupation of any particularlocation throughout the time series. In contrast, ocean quahog lifespans vastly exceed the duration of the time series (Ropes et al., 1984;Begum et al., 2010; Ridgway and Richardson, 2011; Pace et al., 2017).Thus, for surfclams, presence should be interpreted to indicate thepotential for occupation of a site, not the expectation of continualoccupancy over the 30+ year time series, whereas for ocean quahogs,continuous habitation is a defensible inference.

2.2.2. Data preparation – substrateThe dredge efficiently captured large sedimentary particles such as

rocks, cobbles, and boulders. Cobbles encompassed anything smallerthan six inches but larger than gravel. Rocks were defined to bebetween six and twelve inches and boulders were defined as anythinglarger than twelve inches. Over the history of the survey, the annota-tions regarding substrate varied. From 1978 to 1980, substrate datawere recorded in either liters or bushels. The survey dredge usedduring this time period was considerably smaller than the dredge usedfrom 1982 to 2011. Due to the extreme variability of recorded data

from 1978 to 1980, presence and predominance values were assignedto these data. A value of 0 indicated an absence of a particular substrate(e.g., cobbles). A value of 1 was given to volumes ≤1 bushel or wherepresence was indicated without a volume given (e.g., “trace” wasrecorded). A value of 2 was assigned to any volume > 1 bushel. Datafor Georges Bank in this portion of the dataset are limited to a fewstations taken in 1980.

From 1982 to 2011, substrate data were recorded at sea in terms ofpresent or predominant. In addition, the total volume caught wasroutinely recorded, as was the percent of total volume present assubstrate. To provide more quantitative and consistent values forsubstrate, the total volume of substrate in bushels was calculated foreach tow from the percent of total volume. The total substrate volumewas then divided proportionally by the sum of presence and predomi-nance values, given values of 1 and 2 respectively, in order to estimatethe number of bushels of cobbles, rocks, and boulders. This providedan estimate of the volume of each constituent in the dredge haul. Thus,a substrate type listed as ‘present’ in a tow with a high total substratevolume would have contributed a higher portion of the total catchrelative to a substrate type listed as ‘predominant’, but with a lowertotal substrate volume. Using these volumetric estimates, the data wererecoded as 0 for absence or < 1 bushel, 1 where the volume of aparticular category was ≥ 1 but < 30 bushels, and 2 where the volumewas ≥30 bushels.

Data prior to 2002 were recorded on hand-written data sheets. For2002–2011, data were entered directly into an electronic database.Subsequent statistical analysis showed that the substrate volumesrecorded in the electronic database were consistently lower per towthan those values on the pre-2002 data sheets, by a factor of 10.Further investigation, including interviews with crew who participatedin the survey across the 1999–2002 transition, did not elucidate anexplanation for the differential, but evaluation across a series of surveysshowed that the differential coincided with the transition from datasheet to electronic file and that the differential was consistent forwardsand backwards in time from that point. To standardize the data, the2002–2011 values were increased by a factor of 10.

The divisions between zero and one bushel and between 29 and 30bushels used to distinguish absent, present, and predominant wereobtained by examining the corrected electronic data from 2002 to 2011where the tows for the entire survey domain, not just Georges Bank,could be analyzed as they were already in electronic format. Themedian and 75th percentile for all tows was 0 (no substrate larger thangravel collected) for these tows, except for cobbles where the 75thpercentile fell near the tail of the distribution. That is, cobbles, rocks,and boulders were rarely encountered by the survey. The value of 30fell between the 90th and 99th percentiles of all tows for thesesubstrate types except cobbles where it fell close to the 90th percentile(Table 1). The value 1 fell at or above the 90th percentile of all tows forthese substrate types except cobbles where it fell near the 75thpercentile (Table 1). Thus, we defined as present all tows for whichat least one bushel of material was obtained and as predominant therare tows in which 30 or more bushels were obtained.

In addition, substrate information was gleaned from haul and gearcodes associated with each tow. These encompass a range of incidents,mechanical problems, and miscellaneous misfortunes that might havecompromised the tow. We extracted the haul and gear codes thatindicated problems stemming from bottom contact. These included: (1)locations where the bottom was too rough for dredge deployment; (2)locations where dredge damage occurred, including broken nipples,broken or bent knife blades, torn hoses, or damage to the dredge frame;and (3) locations where rocks were caught in sufficient number to bejudged to have compromised the tow, but which did not causesignificant/any damage to the dredge. Each of these tows was assignedto the set that contained tows that returned ≥30 bu of cobbles, rocksand/or boulders. In nearly all cases, these same tows were identified tothis substrate class based on documentation obtained from the catch.

1 Commercial dredges use shakers or roller tables to remove dead shell, which beingsingle valves, fall between the shaker or roller bars. The survey protocol did not includethe use of such post-catch mechanical sorting gear.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

16

In only 37 cases (3.7% of the total dataset) was the tow not so identifiedand 31 of these were cases where the tow was abandoned due to bottombeing too rough for dredge deployment.

2.2.3. Data preparation – dead shellFor dead shell, abundance data were entered as absence, presence,

and predominance values (0, 1, 2). Generally, shell volume as apercentage of total catch was recorded for each tow, as it was forsubstrate. Thus, the afore-described analysis for substrate could berecapitulated for shell. However, our approach was to focus on therelative importance of shell types at each location rather than compar-ing the absolute quantity across all tows because this provided arelative ranking of the habitat as a function of species preference.Determining whether total quantity was interpretable with respect tothe time-averaged intensity of occupation of the site would requirefurther analyses, although the present database may support such ananalysis. Thus, we assigned values of 0, 1, and 2 for absent, present,and predominant for each tow in the time series.

2.3. Data analysis

We defined the domain for analysis as Georges Bank and thewestward extension across the Great South Channel to approximately69°40′W (Fig. 1). In total, the dataset encompassed approximately1000 tows, not counting replicates (996–1005 depending on thenumber of missing data for each datum type). Each of the datum typeswere patchily distributed across Georges Bank, as can be gleanedreadily from the Figures referred to later in the Results section.Characterizing patchiness, however, would not directly address thequestions raised in this contribution; namely, the degree to which thedispersion of dead shell and live animals overlapped spatially and thedegree to which recorded differences in the distributional pattern oflive animals and dead shells occurred coherently over large spatialscales within and between species. Consequently, analysis focused onthe characteristics of neighboring tows in order to assess the degree towhich different tow characteristics such as the presence of livesurfclams and dead surfclam shells agreed with the expectation that,in this case, the distribution would be similar. We thus adopted amodified nearest neighbor analysis without invoking the statisticalevaluation of patchiness as is often done (e.g., Clark and Evans, 1954;Solow, 1989; Brown, 2003; Leighton and Schneider, 2004).

We adopted a presence-absence approach, although information onquantity is present for live animals and for dead shells, for severalreasons. First, although the abundance of ocean quahogs at anylocation is likely a conservative feature over the 30+ year survey timeseries, this time span being not much longer than the time to maturityfor this animal (Thórarinsdóttir and Jacobson, 2005; Thórarinsdóttirand Steingrímsson, 2000), the same cannot be said for surfclams, giventhe higher mortality rate and shorter, albeit still long, life span(Weinberg, 1999; Munroe et al., 2016). Thus, for surfclams, abundanceat any time is not necessarily an indicator of integrated abundance overa tricennial time frame. Second, the amount of dead shell retained inthe tow is not necessarily a function of shell production over time, assome shell may be transported from the site of death and, at least aslikely, buried to a depth beneath that accessed by the hydraulic dredge(dredge exhumation depths ≤20 cm, see Smolowitz and Nulk, 1982;NEFSC, 2013; exhumation depths are less than the height of sand

waves and megaripples, see Twichell, 1983).For each tow, we assigned a pair-wise set of characteristics. For

example, the tow might be characterized by the presence of livesurfclams and the presence of dead surfclam shells. Four possiblestates exist; in the above example, they are: (1) neither live surfclamsnor dead surfclam shells present; (2) live surfclams but no deadsurfclam shells present; (3) no live surfclams but dead surfclam shellspresent; and (4) both live surfclams and dead surfclam shells present.We then identified the same characteristics in the four nearestneighbors, using a modified bishops moves approach. We identifiedthe nearest neighbor in the northeast (NE), southeast (SE), southwest(SW), and northwest (NW) quadrant, with one exception. Tows nearboundaries typically were missing one of the four neighbors. Thesetows were excluded as parent tows from further analysis, althoughretained as neighboring tows. For each of the four possible character-istics of the parent tow, we tallied the characteristics of these fournearest neighbors and evaluated the degree to which the tallies wererandomly distributed using a chi-square test for goodness of fit (Daniel,1978). A coherent set of neighbors, for example, might suggest thateither the environment encompassed by the 5 sites was uniformlyconducive to the species in question, that the history of the five siteswas similar with respect to species shifts in range, or that shelltransport processes were similar. For the latter, for example, coherencybetween live animals and dead shells might suggest autochthonousaccumulation if both live animals and dead shells were present,whereas the presence of dead shells only might suggest a change inspecies range or regionally coherent shell transport.

The previous analysis targets the degree to which a given char-acteristic pair resides in a region surrounded by that or anothercharacteristic pair. To further examine the complexity of the surround-ing region, we tallied the number of times a given characteristic pairwas surrounded by a set of specified characteristic pairs. For example,if the parent tow was characterized by characteristic 4 listed in thepreceding paragraph, both live surfclams and dead surfclam shellsbeing present, the four nearest neighbors might be characterized by4,3,1,1; that is, by one neighbor characterized by both live surfclamsand dead surfclam shell present (4), one by no live surfclams present,but dead surfclam shell present (3), and two by the absence of livesurfclams and dead surfclam shell (1). Discounting the order ofappearance, a total of 36 unique tetrads exist. Certain of these tetradsbeing common suggests that certain environmental and/or biologicalprocesses are important. For example, if 4,3,3,3 occurs commonly asnearest neighbors of a site with both live surfclams and dead surfclamshells present as in the above example, the suggestion would be thatdead shell has been transported away from a production site as onlydead shell was observed in 3 of 4 quadrants surrounding a locationwhere both live animals and dead shell were found. We compared thedistribution of the number of times a particular neighborhood char-acteristic occurred, namely the sequence of the four neighboringvalues, against the expectation of homogeneity, using a two-sidedone-sample exact Kolmogorov-Smirnov test (Conover, 1972) to deter-mine if certain tetrads were unusually common. The comparison wasdone using all possible combinations of the four neighborhoodcharacteristics, of which there are 36. However, certain of thesecombinations do not occur in the dataset and, arguably, are combina-tions that are implausible given the distribution of habitat on GeorgesBank. As a consequence, we also conducted the test using only the

Table 1Percentiles of bushel catch per tow for all survey tows taken in 2002–2011.

Substrate constituent 25th percentile 50th percentile 75th percentile 90th percentile 95th percentile 99th percentile

Cobble 0.0 0.0 3.3 42.7 71.3 120.Rock 0.0 0.0 0.0 8.8 41.6 100.Boulder 0.0 0.0 0.0 0.0 0.0 47.5

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

17

observed combinations. Test results depend to some extent on theorder of the tetrads. For consistency, the same order was used for alltests. Reference to that order can be found in the accompanying figures.

3. Results

3.1. Live animal comparison

Fig. 1a depicts the distribution of stations on Georges Bank and theeasternmost portion of southern New England encompassing the GreatSouth Channel characterized by each of the four possible pairwisecombinations defined by the presence or absence of living oceanquahogs and living surfclams. Fig. 1b and c distinguish more clearlythe locations where the two species occur together and the locationswhere they occur separately. The distribution of stations characterizedby any one of the four possible combinations is obviously non-random.Cases where only live ocean quahogs were found are dominantlylocated along the southern margin of Georges Bank and parts west(Fig. 1a, c). Stations where only live surfclams were found aredominantly located in the central shallower region and along the

northern edge of the bank (Fig. 1a, c). A relatively clearly defined lineexists, particularly along the southern portion of the bank, where thetwo species comingle (Fig. 1a, b).

Stations where neither live surfclams nor live ocean quahogs werecollected were neighbored primarily by stations with the same character-istic or by stations where only live surfclams were found (Table 2). Suchstations occurred disproportionately in shallower water distant from thedeeper-water habitats occupied by ocean quahogs. Stations where livesurfclams were found, sans live ocean quahogs, dominantly neighboredstations of the same kind and rarely neighbored stations where only liveocean quahogs were found. This distribution was highly non-random (chi-square, Table 2, Fig. 1c). The same pattern was true for live ocean quahogs(Table 2, Fig. 1c). Stations where only live ocean quahogs were foundneighboring stations where only live surfclams were present occurred withlow frequency. The ecotone between stations with only live surfclams oronly live ocean quahogs present was characterized by the presence of bothspecies (Fig. 1b). These stations, interestingly, were neighbored moreevenly by other station types, as would be expected by the ecotonal natureof this station type. In particular, neighboring stations often had only liveocean quahogs and relatively often had only live surfclams.

Fig. 1. The distribution of live Atlantic surfclams, Spisula solidissima (AS), and live ocean quahogs, Arctica islandica (OQ) in the region of Georges Bank as assessed by surveys over the1980–2011 time frame. Upper plot (A), all four station types. Lower left (B), stations with both surfclams and ocean quahogs. Lower right (C), stations with either surfclams only orocean quahogs only.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

18

A closer look at stations where both species were present shows thatnearest neighbors were not distributed randomly among the tetradicpossibilities. Cases where neighbors were either stations with bothspecies present or live surfclams only were most common; however,cases where neighbors were stations with both species present or liveocean quahogs only also occurred with some frequency (Fig. 2). Thedistribution of stations among tetradic combinations was highly non-random (K-S exact test; P = 0.018). In contrast, stations where livesurfclams were present without ocean quahogs, were neighboredprimarily by stations where either both species were found or livesurfclams only (Fig. 2). The distribution among observed tetrads wasalso inhomogeneous (K-S exact test; P = 0.013); thus locations withboth species present clearly fell into an ecotone between habitatsoccupied solely by one species or the other. In contrast, stations with

only live ocean quahogs present were surrounded primarily by stationsof the same type or by stations where neither species was found(Fig. 2). The occurrence rate of tetrads among those observed wasrandom (K-S exact test; P = 0.059, but the occurrence rate was non-random if all possible tetradic combinations were included (K-S exacttest; P = 0.003. That is, in this case, many possible tetradic combina-tions were not observed. These missing tetrads were cases wherestations with only live ocean quahogs were neighbored by stations withonly live surfclams and stations where neighbors were locations whereneither species was found. The absence of the former defines theextensiveness of the two-species ecotone in the Georges Bank region.The absence of the latter confirms that stations where neither species isfound tend to be on the shallower parts of Georges Bank well separatedfrom the deeper-water ocean quahogs.

Table 2Summary of characteristics of the four nearest neighbors to each parent site, chosen using a bishops moves strategy. Combination indicates the characteristic of the parent site, with thesequence of “no” and “yes” indicating the attribute of the first and second characteristic in the header. Numbers 1–4 are consistent with the tetradic identification scheme used in thefigures. Neighbor combinations refer to the characteristics of the neighboring sites. Thus, in the first dataset the combination “no-yes” indicates parent sites where live surfclams werenot present, but live ocean quahogs were found. The neighbor combination “yes-no” indicates neighbors where live surfclams were found sans live ocean quahogs. The occurrence rateindicates that such neighbors were found 19 times.

Combination no-no (1) no-yes (2) yes-no (3) yes-yes (4)

Neighborcombination

no-no (1) no-yes(2)

yes-no(3)

yes-yes(4)

no-no(1)

no-yes(2)

yes-no(3)

yes-yes(4)

no-no(1)

no-yes(2)

yes-no(3)

yes-yes(4)

no-no(1)

no-yes(2)

yes-no(3)

yes-yes(4)

Live surfclams-live ocean quahogsOccurrence 301 50 280 93 64 571 19 214 283 19 493 77 109 220 71 272Chi-square 271.3: P < 0.0001 866.1: P < 0.0001 639.1: P < 0.0001 157.3: P < 0.0001

Dead surfclam shell-dead ocean quahog shellOccurrence 250 91 234 193 91 250 36 339 222 36 432 210 212 337 208 691Chi-square 79.8: P < 0.0001 328.7: P < 0.0001 350.2: P < 0.0001 428.4: P < 0.0001

Live surfclams-dead surfclam shellOccurrence 408 226 86 304 218 259 33 218 88 46 19 135 317 224 137 626Chi-square 215.7: P < 0.0001 167.7: P < 0.0001 107.1: P < 0.0001 417.8: P < 0.0001

Live ocean quahogs-dead ocean quahog shellOccurrence 1002 200 52 166 210 57 12 49 59 12 16 93 174 53 95 1306Chi-square 1606.1: P < 0.0001 280.5: P < 0.0001 98.4: P < 0.0001 2666.2: P < 0.0001

Live surfclams-dead ocean quahog shellOccurrence 285 98 223 162 102 626 44 284 216 47 343 146 185 227 156 272Chi-square 100.8: P < 0.0001 851.2: P < 0.0001 247.1: P < 0.0001 36.5: P < 0.0001

Live ocean quahogs-dead surfclam shellOccurrence 238 282 53 83 291 658 28 115 63 33 262 334 87 115 338 576Chi-square 233.4: P < 0.0001 855.5: P < 0.0001 378.9: P < 0.0001 557.2: P < 0.0001

Fig. 2. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of live surfclams and live ocean quahogs. The bar fills identify thecharacteristic of the parent site. Thus, black fill indicates parent sites where both live surfclams and live ocean quahogs were collected. The tetradic groups represent the characteristics ofthe four nearest neighbors using a bishops moves identification scheme. For example, 4311 indicates that one neighbor had both live surfclams and live ocean quahogs (4), one neighborhad live surfclams, but no live ocean quahogs (3), and two neighbors yielded neither species (1). A (2) would indicate live ocean quahogs, but no live surfclams. The y-axis records thenumber of times each of the tetrads occurred.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

19

3.2. Dead shell comparison

The distribution of surfclam and ocean quahog dead shell (Fig. 3a–c) is reminiscent of the distribution of live animals (Fig. 1a–c), but withless coherency. Generally, stations where both shell types were foundmore commonly had neighbors where only one shell type was found, incomparison to live animals (Table 2). The distribution of neighboringtypes remained highly non-random, however (chi-square – Table 2).

Perusal of the characteristics of neighboring stations shows dis-tributional characteristics similar to that observed for live animals,though more dispersed. The occurrence rate for neighboring tetrads ofparent stations containing both dead shell types was not randomlydistributed (K-S exact test; P = 0.00031, Fig. 3b); rather, neighborswere of the same type, having both dead ocean quahog and deadsurfclam shells, or were characterized by stations where only deadocean quahog shells were found (Fig. 4). The occurrence rate ofneighboring tetrads of parent stations containing only dead oceanquahog shells also was inhomogeneous (K-S exact test; P = 0.026,Fig. 3c). In contrast parent stations where only dead surfclam shellswere found were much richer in tetradic complement and distributed

more evenly among the various tetradic combinations than dead oceanquahog shells whether only the observed tetrads were considered (K-Sexact test; P = 0.396) or all possible tetradic combinations (K-S exacttest; P = 0.11). The increased dispersion of dead surfclam shells(Fig. 3c) was often illustrated by the presence of one or more neighborswhere neither shell type was found (Fig. 4). So, for example, in the liveanimal comparison, parent stations characterized by the presence oflive surfclams only were surrounded primarily by stations of the sametype or stations also containing live ocean quahogs (e.g., 4222 inFig. 2). In the dead shell case, the additional tetrads containing onestation with neither shell type present (e.g., 4221 – Fig. 4) alsocommonly occurred. This increased dispersion suggests that spatialtime averaging (see Powell et al., 1989) or local transport (para-utochthony) contributes importantly to the distribution of deadsurfclam shells to a larger extent than dead ocean quahog shells.

3.3. Live surfclams and dead surfclam shell

Live surfclams and dead surfclam shell were routinely collectedtogether (Fig. 5a, b), but stations where dead shell was collected

Fig. 3. The distribution of Atlantic surfclam, Spisula solidissima (AS), dead shell and ocean quahog, Arctica islandica (OQ), dead shell in the region of Georges Bank as assessed bysurveys over the 1980–2011 time frame. Upper plot (A), all four station types. Lower left (B), stations with both dead surfclams and dead ocean quahog shell. Lower right (C), stationswith either dead surfclam shell only or dead ocean quahog shell only.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

20

Fig. 4. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of dead surfclam shell and dead ocean quahog shell. The bar fills identifythe characteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using a bishops moves identification scheme. The y-axis records thenumber of times each of the tetrads occurred. For additional explanation, see Fig. 2.

Fig. 5. The distribution of live Atlantic surfclams, Spisula solidissima, and Atlantic surfclam dead shell in the region of Georges Bank as assessed by surveys over the 1980–2011 timeframe. Upper plot (A), all four station types. Lower left (B), stations with both dead surfclams shells and live surfclams. Lower right (C) , stations with either dead surfclam shells only orlive surfclams only.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

21

without living surfclams were also common, particularly along thesouthern deeper portion of Georges Bank and around the Great SouthChannel (Fig. 5a, c). Both being collected together occurred commonlyon the shallower central portion of Georges Bank and along thenorthern edge (Fig. 5b). Cases where only living surfclams werecollected occurred much less often, but were most common along thenorthern edge of the bank (Fig. 5c).

Not surprisingly, stations where live surfclams and dead surfclamshells were collected frequently were associated with stations of thesame type (Table 2). Moreover, most neighboring stations of stationswhere live surfclams were collected, but where no dead shells werefound, were characterized by the presence of both. However, manystations where live surfclams were found sans dead shells also wereneighbored by stations with neither live surfclams nor dead surfclamshells. Stations where only dead surfclam shells were collected werealso commonly neighbored by stations of the same type or by stationswhere neither live surfclams nor dead shells were collected (Table 2).In very few cases were the neighbors of these stations characterized bythe presence only of live surfclams.

The distribution of tetradic sets associated with stations where bothlive surfclams and dead surfclam shells were found was non-random(K-S exact test; P = 0.0028) whereas the distribution of tetradic setsassociated with stations where only live surfclams were found wasrandom (K-S exact test; P = 0.89) (Fig. 6). The tendency towardsrandomness in the distribution of stations where live surfclams werefound without dead shells is due to the frequency of tetrads in whichneighbors had neither live surfclams nor dead surfclam shell. Thisneighbor type was proportionally more common for parent stationswhere live surfclams were found sans dead shells in comparison toparent stations with both live surfclams and dead surfclam shell(Fig. 6). Many such sites were distributed in the region where livesurfclams and live ocean quahogs were jointly collected, the absence ofdead shell indicating the newness of occupation by live surfclams inlocations inhabited by ocean quahogs. The occurrence rate of tetradictypes associated with stations where only dead surfclam shells werecaught was random if only the observed tetrads were considered (K-Sexact test; P = 0.425), and remained barely so if all possible tetradswere considered (K-S exact test; P < 0.07). That is, many possibletetradic combinations were not observed. Of note is the number ofstations where dead surfclam shells were found for which none or onlyone of the four neighbors contained live surfclams (Figs. 5a, c and 6).Many of these locations are associated with sites where live oceanquahogs were collected: the absence of live surfclams may indicate off-bank transport of dead surfclam shells, a possibility considered furtherin the Discussion section.

3.4. Live ocean quahogs and dead ocean quahog shell

Generally, stations were characterized by the joint collection of liveocean quahogs and dead ocean quahog shells (Fig. 7a, b) or thecollection of dead ocean quahog shells only (Fig. 7a, c). The formeroccurred along the southern deeper water portion of Georges Bank,along the southern terminus of the Great South Channel, and along anarrow northern rim of the bank. The latter were distributed inshallower portions of Georges Bank, particularly towards the east,and on the northwestern side of the Great South Channel (Fig. 7c).

In comparison to the surfclam case, the neighbors of stationscontaining both live ocean quahogs and dead ocean quahog shellswere highly likely to be of the same type (Table 2). Cases where oceanquahogs were collected without dead shells were uncommon andusually were near stations where both dead shells and live animalswere collected. Cases where only dead ocean quahog shells were foundwere also uncommon, but disproportionately neighbored by stationswhere neither live animals nor dead shells were found. Once again, thedistribution of neighbors among neighbor types was significantlyinhomogeneous in every case (chi square – Table 2).

The tetradic complement for parent stations where both live oceanquahogs and dead ocean quahog shells were present was highly non-random (K-S exact test; P < 0.000001). Such stations typically weresurrounded by at least 3 of 4 neighbors of the same type (Fig. 8). Thedistribution of tetrads surrounding parent stations with live oceanquahogs, sans dead shells, was also non-random (K-S exact test; P =0.061), but most neighboring stations remain stations where both liveanimals and dead shells were collected (Fig. 8). The same is not true forparent stations where only dead ocean quahog shells were collected.The occurrence rate for the observed tetradic combinations was alsohighly non-random (K-S exact test; P = 0.001), but most parentstations were neighbored by stations where only dead ocean quahogshells were found or where neither live animals nor dead shells werefound (Fig. 8). As neighboring stations only yielding dead oceanquahog shells rarely neighbored parent stations yielding both liveocean quahogs and dead ocean quahog shells, the influence oftransportation as an explanation for the observed distributional patternseems unlikely. Rather, the distribution bespeaks a possible shift in thespecies’ range.

3.5. Cross-species comparison: live surfclams and dead ocean quahogshells

Confirmation of the differential in distribution of dead shells andlive animals comes from cross-species comparisons. Stations where live

Fig. 6. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of live surfclams and dead surfclam shell. The bar fills identify thecharacteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using a bishops moves identification scheme. The y-axis records thenumber of times each of the tetrads occurred. For additional explanation, see Fig. 2.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

22

Fig. 7. The distribution of live ocean quahogs, Arctica islandica, and ocean quahog dead shell in the region of Georges Bank as assessed by surveys over the 1980–2011 time frame.Upper plot (A), all four station types. Lower left (B), stations with both dead ocean quahog shells and live ocean quahogs. Lower right (C), stations with either dead ocean quahog shellsonly or live ocean quahogs only.

Fig. 8. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of live ocean quahogs and ocean quahog dead shell. The bar fills identifythe characteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using a bishops moves identification scheme. The y-axis records thenumber of times each of the tetrads occurred. For additional explanation, see Fig. 2.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

23

surfclams were found with dead ocean quahog shells were character-ized by a diversity of neighbors (Table 2, Fig. 9). The occurrence rateamong the observed tetrads was random (K-S exact test; P = 0.315)and remained so if all possible tetrads were included (K-S exact test; P= 0.178). These stations were in the zone where live surfclams haveinvaded ocean quahog territory, so that neighbors were of manydifferent types. Similarly, cases where live surfclams were foundwithout dead ocean quahog shells were neighbored primarily bystations of the same type with a lesser number of stations where bothlive surfclams and dead ocean quahog shells were found (Table 2). Thefew neighbors characterized by the absence of live surfclams and thepresence of dead ocean quahog shells did not lessen the randomdistribution of the occurrence rate of neighboring tetrads about thesesites (K-S exact test; P = 0.353). The opposite was true for stationswhere live surfclams were not found, but dead ocean quahog shellswere present. Most neighbors were of the same type and the distribu-tion of tetradic types was significantly non-random (K-S exact test; P =0.0063) (Fig. 9). Thus, live surfclams were often found with or hadneighbors that contained dead ocean quahog shells, but the stationswithout live surfclams but with dead ocean quahog shells were muchmore restricted in their distribution. This was true because deeperwaters inhabited by live ocean quahogs often were not inhabited bysurfclams (Fig. 1).

3.6. Cross-species comparison: live ocean quahogs and dead surfclamshells

A different picture emerges from the distribution of live oceanquahogs and dead surfclam shells. Most stations where live oceanquahogs were found coincidentally with dead surfclam shells were ofthe same type or were cases where live ocean quahogs were foundwithout dead surfclam shells being present (Table 2, Fig. 10). Thedistribution of occurrences for neighboring tetrads was highly non-random (K-S exact test; P = 0.00000059). A similar distributionalpattern was found for stations where live ocean quahogs were foundsans dead surfclam shell (Fig. 10). The occurrence pattern forneighboring tetrads was also highly non-random (K-S exact test; P =0.00295). The same outcome is true for stations characterized by deadsurfclam shells sans live ocean quahogs (K-S exact test; P =0.00000173). Thus, parent stations with live ocean quahogs wereprimarily neighbored by a limited number of station types with orwithout dead surfclam shells being present whereas neighboringstations characterized by the absence of live ocean quahogs and thepresence of dead surfclam shell were extremely rare. The mirrordistribution occurred for stations without live ocean quahogs, but withdead surfclam shell, where neighbors with live ocean quahogs were alsorare (Fig. 10). The differential distribution of live ocean quahogs and

Fig. 9. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of live surfclams and dead ocean quahog shell. The bar fills identify thecharacteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using a bishops moves identification scheme. The y-axis records thenumber of times each of the tetrads occurred. For additional explanation, see Fig. 2.

Fig. 10. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of live ocean quahogs and dead surfclam shell. The bar fills identify thecharacteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using a bishops moves identification scheme. The y-axis records thenumber of times each of the tetrads occurred. For additional explanation, see Fig. 2.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

24

dead surfclam shell is clearly recorded in these comparisons, suggest-ing that large-scale transport of dead surfclam shell is not a ubiquitousoccurrence on Georges Bank.

3.7. Substrate relationships

Relatively few stations were characterized by the presence of livesurfclams and also evidence of complex habitat as defined by thepresence of any combination of cobbles, rocks, and boulders (Table 3).The occurrence rate for the four nearest neighbors was not randomlydistributed among the observed tetrads (P = 0.027). Nearest neighborsoften were locations where live surfclams were caught without complexhabitat being present or where no surfclams were caught, but complexhabitat was present (Fig. 11). The occurrence of tetradic combinationsof nearest neighbors for locations where live surfclams were foundwithout complex habitat was non-randomly distributed among theobserved tetrads (P = 0.022); combinations in which one or more of thefour contained complex habitat occurred relatively infrequently.

The distribution of dead surfclam shells relative to complex habitatwas almost identical. Relatively few stations were characterized by thepresence of surfclam shells and also evidence of complex habitat

(Table 3). The occurrence rate for the combinations of the four nearestneighbors was not randomly distributed among the observed tetrads (P= 0.059). Nearest neighbors often were locations where dead surfclamshells were caught without complex habitat being present or where nodead surfclam shells were caught, but where complex habitat waspresent (Fig. 12). The occurrence rate for tetradic combinations ofnearest neighbors for parent locations where dead surfclam shells werefound without complex habitat was non-randomly distributed amongthe observed tetrads, however (P = 0.0076). Combinations in whichone or more of the four contained complex substrate occurred relativelyinfrequently (Fig. 12). Overall, surfclam shells were not obviously moreassociated with complex habitat than were live surfclams, as one mightexpect would be the case if local transport of shells was important indetermining the distribution of shells.

The distribution of ocean quahogs and ocean quahog shells relativeto complex habitat is similar in that the trends are nearly identicalbetween live ocean quahogs and dead ocean quahog shells (Table 3,Figs. 13 and 14). Noticeable, however, is the fact that cases wherestations with dead ocean quahog shells were neighbored by stationswith complex habitat are more frequent than cases where live oceanquahogs were so neighbored (Table 3). This increased frequency can be

Table 3Summary of characteristics of the four nearest neighbors to each parent site, chosen using a bishops moves strategy. Combination indicates the characteristic of the parent site, with thesequence of “no” and “yes” indicating the attribute of the first and second characteristic in the header. Numbers 1–4 are consistent with the tetradic identification scheme used in thefigures. Neighbor combinations refer to the characteristics of the neighboring sites. Thus, in the first dataset the combination “no-yes” indicates parent sites where live surfclams werenot present, but complex habitat were found. The neighbor combination “yes-no” indicates neighbors where live surfclams were found without complex habitat. The occurrence rateindicates that such neighbors were found 181 times.

Combination no-no (1) no-yes (2) yes-no (3) yes-yes (4)

Neighborcombination

no-no(1)

no-yes(2)

yes-no(3)

yes -yes(4)

no-no(1)

no-yes(2)

yes-no(3)

yes-yes(4)

no-no(1)

no-yes(2)

yes-no(3)

yes-yes(4)

no-no(1)

no-yes(2)

yes-no(3)

yes-yes(4)

Live surfclams-complex habitatOccurrence 676 130 337 77 128 177 181 118 374 100 468 142 75 126 137 170Chi-square 725.5: P < 0.0001 24.4: P < 0.0001 351.7: P < 0.0001 36.6: P < 0.0001

Dead surfclam shell-complex habitatOccurrence 369 66 465 100 74 193 115 122 471 124 937 200 88 124 194 210Chi-square 458.3: P < 0.0001 58.2: P < 0.0001 935.9: P < 0.0001 64.9: P < 0.0001

Live ocean quahogs-complex habitatOccurrence 465 265 133 37 263 476 50 59 147 56 1235 122 39 56 104 49Chi-square 457.8: P < 0.0001 575.2: P < 0.0001 2452.4: P < 0.0001 40.2: P < 0.0001

Dead ocean quahog shells-complex habitatOccurrence 383 198 278 77 191 366 76 99 286 82 1275 133 81 98 123 86Chi-square 214.0: P < 0.0001 284.5: P < 0.0001 2124.5: P < 0.0001 10.9: P=0.012

Fig. 11. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of live surfclams and complex habitat as defined by the presence ofcobbles, rocks, and/or boulders. The bar fills identify the characteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using a bishopsmoves identification scheme. The y-axis records the number of times each of the tetrads occurred. For additional explanation, see Fig. 2.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

25

mapped to the shallower water stations where dead ocean quahogshells were found without live ocean quahogs (Figs. 7 and 8).

4. Discussion

4.1. Perspective

Application of the death assemblage for tracking changes in thedistribution of the living community over time offers an importantopportunity in that many species’ shifts in range as a consequence of,for example, warming temperatures, cannot be assessed because timeseries of the living population are not present. Thus, even if thepresent-day distribution is suspected of being a product of, forexample, global warming, information on the past distributionalpattern, being poorly known or absent, does not permit empiricalvalidation. This challenge is particularly severe for continental shelfbenthos, for which we know little of distributional patterns beyond afew commercial species (for exceptions, see for example, Parker, 1960;Cerame-Vivas and Gray, 1966; Davis and VanBlaricom, 1978; Zuschinand Piller, 1997; Staff and Powell, 1988; Buhl-Mortensen et al., 2012),particularly at large geographic scales and for which sampling isinsufficient to be confident in patterns for which we might have some

distributional data (Powell and Mann, 2016). The opportunity alsopresents itself to better evaluate a species’ habitat, as living individualsmay often not be found in otherwise habitable areas, either due to thevagaries of recruitment or the inadequacies of sampling design, but thedeath assemblage reliably records occupation by most molluscanspecies at present or past times as the species’ ensemble is relativelyfaithfully preserved (Kidwell and Flessa, 1995; Kidwell, 2001; seeCallender and Powell, 2000; Powell et al., 2011 for example excep-tions). Powell et al. (1989) termed this process spatial time averaging.

Unfortunately, the assemblage of dead shells may represent a range oftime spans depending on local biological and environmental processes.Time averaging is a well-documented phenomenon (Kowalewski andLaBarbera, 2004; Kosnik et al., 2009; Tomašových and Kidwell, 2010).Often, the suite of times-since-death of shells includes a disproportionatenumber of relatively recent deaths and a long tail of increasingly rarershells that died at increasingly earlier times. Vicennial to centurial orlonger time scales are not uncommon (e.g., Powell and Davies, 1990;Kidwell et al., 2005; Dexter et al., 2014; Tomašových et al., 2014; Albanoet al., 2016). The time scales for the assemblages described herein are notknown, although the simple recognition of the maximum age of the livinganimals and thus the age at death for the dead shells strongly implies half-century to bicentennial time scales. As discussed subsequently, the record

Fig. 12. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of dead surfclam shell and complex habitat as defined by the presenceof cobbles, rocks, and/or boulders. The bar fills identify the characteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using abishops moves identification scheme. The y-axis records the number of times each of the tetrads occurred. For additional explanation, see Fig. 2.

Fig. 13. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of live ocean quahogs and complex habitat as defined by the presenceof cobbles, rocks, and/or boulders. The bar fills identify the characteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors using abishops moves identification scheme. The y-axis records the number of times each of the tetrads occurred. For additional explanation, see Fig. 2.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

26

of climate change in the northwest Atlantic likely limits the time ofaccumulation of dead shells from these long-lived species in any onelocation proffering an opportunity to evaluate shifts in species’ range thatmight not be present elsewhere.

As unfortunately, the death assemblage also is impacted by spatialtime averaging caused by local to regional transport of shells (Miller,1988; Zenetas, 1990; Callender et al., 1992) and the tendency of speciesnot to live in all habitable areas all of the time [the case of Mulinialateralis being a particularly useful example (Levinton, 1970; Powell et al.,1986)]. Thus, the distribution of dead shells may not always be adependable indicator of present or past habitat, just as it may ofteninform on the same. This ambiguity places a restriction on interpretation.In this study, we used an extensive dataset for the Georges Bank region ofthe northwestern Atlantic continental shelf to evaluate the distribution oflive animals and dead shell for two clam species, Spisula solidissima andArctica islandica. For surfclams, this region has been sampled repeatedlyover more than 30 years, so that variations in the distribution of livingindividuals within habitable area very likely have occurred and thecumulative distribution of locations where live surfclams have beenobserved very likely is an effective descriptor of the habitat. For oceanquahogs, life spans are so long that the distribution is also likely aneffective descriptor of the habitat. In each case, we compared thedistribution of dead shells with that of the living individuals.

4.2. Review and interpretation – range shifts

Warming of the northwest Atlantic has resulted in a documentedshift in the distribution of the surfclam south of Hudson Canyon in theMid-Atlantic Bight, with the southern range boundary receding northand the offshore boundary extending farther offshore (Weinberg, 2005;Weinberg et al., 2005; Hofmann et al., in press). These shifts permit thespecies to remain within the relatively narrow temperature rangeconducive to survival. For ocean quahogs, the species’ range has beenrelatively stable over the survey time span that began, for the GeorgesBank region, in 1980. However, ocean quahogs invaded the region atsome prior time (Dahlgren et al., 2000) and recent evidence suggeststhat the most recent occupation of the Georges Bank region began inthe late 1700s to early 1800s (Pace et al., 2017), some earliest colonistsof which are still to be found in the living population. Thus, weanticipate a complex comparison between the distribution of livesurfclams and dead surfclam shell, with shell distributions being bothmore diffuse than the live animals and possibly biased in space as aresult of a range shift. For ocean quahogs, we anticipate a less complexcomparison, with dead shell distributions being more or less mirroredby the living community.

Consideration can be given to the reason why increasing tempera-

tures have resulted in a northward and offshore shift in the southernand inshore boundary of the surfclam, but no apparent shift in thesouthern and inshore boundary for the ocean quahog. For surfclams,temperatures resulting in substantial loss in scope for growth are only afew degrees Celsius above optimal (Munroe et al., 2013). Animalsentering what Woodin et al. (2013) termed the transient event marginwhere, in this case, scope for growth begins to decline with increasingtemperature, are rapidly compromised as ingestion drops relative tometabolic needs (Munroe et al., 2013; Narváez et al., 2015). Kim andPowell (2004) document the reduction in condition as such animal'sscope for growth becomes negative, whereupon, if persistent, theyultimately die, a process termed deficit stress mortality by Narváezet al. (2015). Marzec et al. (2010) showed that condition is also reducedoffshore probably due to lower temperature leading to lower ingestionrates or to lower food availability, so that the leading range edge verylikely is also directly temperature controlled. In contrast ocean quahogshave the ability to bury and estivate for extended time periods (seeTaylor, 1976a, 1976b; Oeschger, 1990). As a consequence, theseanimals do not experience the highest summer temperatures alongtheir southern and inshore boundary. This insulation buffers thisspecies against rising temperatures and stabilizes the range boundary.Whether ocean quahogs are expanding their range northward oroffshore into deeper water remains unknown.

The two species have temperature tolerances that would normallylimit distributional overlap. The summer temperature optimum forsurfclams is in the range of 18–21 °C, whereas the optimal summertemperature for ocean quahogs is below 15 °C (Mann, 1982; Cargnelliet al., 1999a, 1999b; Munroe et al., 2013). But, during times of climatechange, the insulation of the ocean quahog and expansion of thesurfclam can result in a range overlap. This overlap is well documentedand has become increasingly apparent in the southern portion of therange over the > 30-year history of the survey (NEFSC, 2013). Asimilar overlap in species’ range is readily observed in the GeorgesBank dataset (Figs. 1 and 2). Nearest neighbors to stations having livesurfclams typically also have live surfclams. The same is true for oceanquahogs. However, a significant minority of stations have both speciesand, in these cases, typically one or more of the four nearest neighborsis a station where only ocean quahogs were caught or where only livesurfclams were caught. Although the dynamics of range shift onGeorges Bank are less clear than farther south, the strong tendencyof the sites where both species were found to have nearest neighborswhere live ocean quahogs were found sans live surfclams stronglyindicates advancement of the offshore boundary of the surfclam intohistorically ocean quahog habitat. The fact that this pattern isrecapitulated in the distribution of dead shell (Fig. 4) adds substan-tively to the belief that this distributional pattern is not simply

Fig. 14. The characteristics of the four neighboring sites for parent sites characterized by the presence or absence of ocean quahog dead shell and complex habitat as defined by thepresence of cobbles, rocks, and/or boulders. The bar fills identify the characteristic of the parent site. The tetradic groups represent the characteristics of the four nearest neighbors usinga bishops moves identification scheme. The y-axis records the number of times each of the tetrads occurred. For additional explanation, see Fig. 2.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

27

produced by the occasional settlement of surfclams in suboptimalhabitat.

For ocean quahogs, the distribution of live animals and dead shellwere similar with two noticeable exceptions. First, in a few cases,certain nearest neighbors were characterized by the presence of liveocean quahogs, but no dead shell. These locales were almost alwaysneighbored by at least three of the tetradic neighboring ensemblehaving both live ocean quahogs and dead ocean quahog shell (Fig. 8).Most were distributed near the edge of the region occupied by thisspecies (Fig. 7). Likely, these sites represent recruitment in marginalhabitat where total shell production is low.

Second, although overall, the distribution of dead ocean quahogshell tracked the distribution of live ocean quahogs very closely (Fig. 8),in an important minority of sites, dead ocean quahog shell was foundshallower than live animals. This region was discrete spatially (Figs. 7cand 8). Neighbors were of the same type, being characterized by deadocean quahog shell but no live animals; also, however, sites whereneither live ocean quahogs nor dead ocean quahog shell were collectedoccurred as neighbors with some frequency (Fig. 8). The suggestionfrom the regionalism and neighboring site complement is that thesedead shells document the occupation by ocean quahogs at an earliertime when ocean quahogs occupied depths on Georges Bank shallowerthan today. Over time, with warming of the northwest Atlantic thatlikely started at the end of the Little Ice Age (Schöne et al., 2005;Cronin et al., 2010), these depths became uninhabitable. When thisshift in range occurred is unclear, save that the event surely precededby many years the first surveys which began in 1980, as no live oceanquahogs have ever been collected in this region and the shells almostcertainly represent animals that were at least 60 years old at the time ofdeath, given the normal growth rates of the species (Murawski et al.,1982; Harding et al., 2008).

4.3. Review and interpretation – shell distributional patterns

Ocean quahogs today are rarely associated with sites where complexsubstrate such as cobbles, rock, or boulders are found and no evidenceexists that they have been so in the past (Figs. 13 and 14). For liveanimals, the few such sites in the dataset have neighboring sites wherethree or more of the tetrad do not have complex substrate associatedwith them (Fig. 13). The pattern is recapitulated with dead oceanquahog shell (Fig. 14), and the somewhat increased frequency isexplained by the set of sites where dead ocean quahog shell occurredat sites shallower than the depth range of live ocean quahogs today,giving additional credence to the conclusion that dead ocean quahogshell is rarely transported any distance beyond its original restingplace.

The distribution of dead surfclam shells is more complex. Siteswhere live surfclams and dead surfclam shell were collected have aneclectic set of neighbors, including sites of the same kind, sites wherelive surfclams were collected without dead shell, and sites where deadshell was collected without live animals. The sites with live animalsonly tend to be in the region where live surfclams and live oceanquahogs overlap; that is, these sites provide additional evidence of thenewness of occupation as surfclams colonize deeper water.

Sites where only dead surfclam shells were collected are of twotypes. One type consists of locations intermingled with sites having livesurfclams. Possible explanations for this distributional pattern are thatthese are sites capable of supporting surfclams, but none were presentwhen the sampling events occurred, or that regional transport of shellhas dispersed shell beyond surfclam habitat. Surfclams live in a regionof Georges Bank that abuts complex habitat where cobbles, rocks andboulders prevail. Sites where complex habitat only was found oftenneighbor sites where both complex habitat and live surfclams werecollected and sites where only live surfclams were collected, stressingthe local heterogeneity of the region (Table 3, Fig. 11). The distributionof dead surfclam shells is very similar suggesting that transport of dead

surfclam shells substantial distances is not an important process;otherwise the number of sites in which dead shells were associatedwith complex habitat would be clearly higher than the number of siteswhere live animals were associated with complex habitat. The distribu-tional pattern appears more diffuse than that of live ocean quahogs anddead ocean quahog shells as one might expect from the shallowerwater, higher energy habitat, but allochthonous shell beds do notappear to be an important contributor to the distribution of deadsurfclam shell.

Georges Bank is a relatively high-energy environment (Grant et al.,1997), leading to the expectation that dead surfclam shell might betransported to some degree during storms. Storm waves in shallowwater are easily capable of moving molluscan shells (Allen, 1984; Frey,1987; Chattopadhyay et al., 2013) and other objects much heavier thansurfclam shells (e.g., Papili et al., 2014). Storms resuspend sedimentacross much of the continental shelf in the Mid-Atlantic region andelsewhere (Hill and Nadeau, 1989; Gagan et al., 1990; Glenn et al.,2008) and can be expected to do so on Georges Bank. Although longdistance shell transport might occur, the more likely outcome is localreworking and burial (Staff and Powell, 1990b; Best and Kidwell, 2000;Papili et al., 2014). Of note, only rarely are dead surfclam shellsretrieved with even minor epibiont coverage (our unpubl. data). Theabsence of epibionts is a good indicator of nearly continuous burial(Parsons-Hubbard et al., 1999; Rodland et al., 2006; Powell et al.,2008), which leads credence to the belief that although dead surfclamshells may be exhumed briefly during storms, only limited directionaltransport is likely to occur.

The issue becomes murky, however, when the second type of sitewhere only dead surfclam shells were collected is evaluated. These aresites where surfclam shell is found without neighboring sites with livesurfclams (Figs. 5 and 6). Many of these sites or nearest neighbors aresites that also contain dead ocean quahog shell (Fig. 7). Many of thesesites are in deeper water where live surfclams are rarely collected(Fig. 1). A question arises as to the explanation for the distribution ofdeep-water surfclam shells. One possibility is that these shells are of asimilar species, Spisula polynyma. Spisula polynyma is the cold-waterequivalent of S. solidissima with a present range that extends down toGeorges Bank, but the species is primarily an Arctic denizen (Cassitaand Hart, 2007; Kilada et al., 2009; Selin, 2010). At one time, however,the Georges Bank climate may have been cold enough (see Wanamakeret al., 2011) to support significant numbers of this species, as exist nowon the Grand Banks and Scotian Shelf. More likely, these shells aresubfossil S. solidissima shells representing an earlier occupation of thebank during a previous warm period (e.g., Medieval Climate Anomaly –Wanamaker et al., 2011), but off-bank transport cannot be excluded;the latter, however, receives little support from the shallow waterdistributional pattern earlier discussed. Dating time-since-death ofthese shells would resolve this uncertainty.

5. Conclusions

Here, we utilize a comprehensive dataset for the region of GeorgesBank to compare the distribution of species in the death assemblageand the living population at large spatial scales and evaluate the recordof changes in species range over time over large spatial scales. Thesurfclam and the ocean quahog are large long-lived species with robustshells that can be expected to be retained in the death assemblage forlong periods of time: preservation potential is high. Thus, the deathassemblage should retain a long-term record. This presents bothopportunities and constraints in that the death assemblage may recordrecent occupation or occupation in the more distant past, the latter ofwhich no longer maps the extent of the present-day living population.Conflation of historical and recent distributional patterns challengesinterpretation.

For both surfclams and ocean quahogs, the distribution of deadshells, in the main, tracked the distribution of live animals relatively

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

28

closely. That is, for both species, the presence of dead shells was apositive indicator of present, recent, or past occupation by live animals.In both cases, the discrepancies in the distributional pattern wereunlikely to be due to regional transport of shell, although localtransport, particularly for surfclams, is likely and longer distancetransport of surfclam shell off bank cannot be ruled out. Theassemblage was primarily autochthonous, likely for ocean quahogs,or parautochthonous, more likely for surfclams. Dispersion of deadsurfclam shell relative to live surfclams within habitat originated eitherfrom spatial time averaging, animals not living in all habitable areas allof the time, or within habitat redistribution of dead shell. Evidence ofdispersion was limited for ocean quahogs, likely due to the deeper-water habitat, the extended life span, and the recent colonization (inocean quahog terms - past 200+ years) which would promote a closemapping between dead shell and live animal distributions.

The regional distribution of dead shell differed from the distribu-tion of live animals, for both species, in a systematic way. In each casethe differential distribution was consistent with the anticipated shift inrange during a period of warming of the northwest Atlantic. It isunlikely that shifts in range are a product of the North AtlanticOscillation (NAO) as the life spans of both species far exceed theNAO periodicities (Hurrell and Loon, 1997; Soniat et al., 2009; Xuet al., 2015). For surfclams, shifts in range might be produced by theAtlantic Multidecadal Oscillation (AMO) as the life span is half or lessof the cycle's period, but the same is unlikely for ocean quahogs, as thatlife span far exceeds the AMO periodicity (Baines and Folland, 2007;Lucey and Nye, 2010; Lanci and Hirt, 2015). For ocean quahogs,response to minimally a century long warming trend would be required(see Cronin et al., 2010; Shen et al., 2012). This is possibly anotherreason why range shifts have been observed in surfclams and not oceanquahogs, as the latter responds to climate change on a much differenttemporal scale.

Regardless, the distribution for surfclams was recapitulated by thepresent-day overlap of live animals with live ocean quahogs and wasconsistent with the expectation that surfclams are moving into deeperwater in response to global warming as has been observed elsewherethroughout the species’ range. The presence of locations where liveanimals were collected without dead shells measures the recentness ofthis event. The tendency for sites neighboring other sites having bothspecies or both shell types to be more often sites with live oceanquahogs and/or dead ocean quahog shell only re-enforces this conclu-sion.

The same distributional dynamics for ocean quahogs has not beenotherwise observed. That is, present-day survey data do not record arange shift for this species. The death assemblage, however, offers goodevidence that such a range shift has occurred in the past, but muchlonger ago. That record is provided by the presence of sites in shallowerwater than inhabited today where dead ocean quahog shells arepresent, sites well separated from the present-day habitat of thespecies.

For both species, the death assemblage provides evidence of shiftsin range corroborating and extending the limited time series record ofmodern survey technology. For most species occupying the continentalshelf in the northwest Atlantic, data on the distribution of livingindividuals is at best rudimentary. The very common sand dollarEchinarachnius parma is a good example (e.g., Sisson et al., 2002;Murillo et al., 2016). The analysis presented here suggests that thedeath assemblage can be used to evaluate the present and pastdistribution of such species, and may be particularly useful if thepresent-day distribution is at least rudimentarily known, thus permit-ting assessment of shifts in range over recent historical time. The oceanquahog, in this study, is a particularly good example. The analysispresented here also suggests that the death assemblage can be used toweigh the adequacy of sampling design for long-lived species, as deadshells are often much more common than live animals, so that thepresence of dead shells without live animals becomes a useful indicator

that sample intensity may be too low. Finally, the analysis presentedhere suggests that the distribution of dead shells may be a goodindicator of habitat in defining survey geographic boundaries, providedthat shifts in range can be identified. An interesting question thatremains is the degree to which dead shell content is an indicator ofhabitat quality. Given the relatively good agreement often observed inrank-order abundance between the living community and the deathassemblage (Kidwell, 2002; for limitations, see Tomašových andKidwell, 2010), it would be useful to determine if shell contentindicates the degree to which an area has supported a species longterm. This could be an indicator of the stability of patches in the livingcommunity.

Acknowledgements

This research was supported by the National Science FoundationIndustry/University Cooperative Research Center SCeMFiS (ScienceCenter for Marine Fisheries) through membership fees under thedirection of the Industry Advisory Board (IAB). SCeMFiS adminis-trative support is provided by National Science Foundation award#1266057. Conclusions and opinions expressed herein are solely thoseof the authors. The authors thank the NMFS Northeast FisheriesScience Center for making available the datasets used in this studyand we acknowledge the many individuals of the NEFSC survey branchthat participated in survey data collection over the last tricennial.

References

Allen, J.R.L., 1984. Experiments on the settling, overturning and entrainment of bivalveshells and related models. Sedimentology 31, 227–250.

Aller, J.Y., 1995. Molluscan death assemblages on the Amazon shelf: implication forphysical and biological controls on benthic populations. Palaeogeogr. Palaeoclimatol.Palaeoecol. 118, 181–212.

Alin, S.R., Cohen, A.S., 2004. The live, the dead, and the very dead: taphonomiccalibration of the recent record of paleoecological change in Lake Tanganyika, EastAfrica. Paleobiology 30, 44–81.

Albano, P.G., Filippova, N., Steger, J., Kaufman, D.S., Tomašových, A., Stachowitsch, M.,Zuschin, M., 2016. Oil platforms in the Persian (Arabian) Gulf: living and deathassemblages reveal no effects. Cont. Shelf Res. 121, 21–34.

Baines, P.G., Folland, C.K., 2007. Evidence for a rapid global climate shift across the late1960s. J. Clim. 20, 2721–2744.

Begum, S., Basova, L., Heilmayer, O., Philipp, E.E.R., Abele, D., Brey, T., 2010. Growthand energy budget models of the bivalve Arctica islandica at six different sites in thenortheast Atlantic realm. J. Shellfish Res. 29, 107–115.

Best, M.M.R., Kidwell, S.M., 2000. Bivalve taphonomy in tropical mixed siliciclastic-carbonate settings. I. Environmental variation in shell condition. Paleobiology 26,80–102.

Brander, K., 2010. Impacts of climate change on fisheries. J. Mar. Syst. 79, 389–402.Brey, T., Arntz, W.E., Pauly, D., Rumohr, H., 1990. Arctica (Cyprina) islandica in Kiel

Bay (Western Baltic): growth, production and ecological significance. J. Exp. Mar.Biol. Ecol. 136, 217–235.

Brown, J.A., 2003. Designing an efficient adaptive cluster sample. Environ. Ecol. Stat. 10,95–105.

Buhl-Mortensen, L., Buhl-Mortensen, F., Dolan, M.F.J., Dannheim, J., Bellec, V., Holte,B., 2012. Habitat complexity and bottom fauna composition of different scales on thecontinental shelf and slope of northern Norway. Hydrobiologia 685, 191–219.

Butler, P.G., Richardson, C.A., Scourse, J.D., Witbaard, R., Schöne, B.R., Fraser, N.M.,Wanamaker, A.D., Jr., Bryant, C.L., Harris, I., Robertson, I., 2009. Accurateincrements identification and the spatial extent of the common signal in five Arcticaislandica chronologies from the Fladen Ground, northern Baltic Sea.Paleoceanography 24, 1–18.

Callender, R., Powell, E.N., 2000. Long-term history of chemoautotrophic clam-dominated faunas of petroleum seeps in the northwestern Gulf of Mexico. Facies 43,177–204.

Callender, W.R., Powell, E.N., Staff, G.M., Davies, D.J., 1992. Distinguishingautochthony, parautochthony and allochthony using taphofacies analysis: can coldseep assemblages be discriminated from assemblages of the nearshore andcontinental shelf? Palaios 7, 409–421.

Cargnelli, L.M., Griesbach, S.J., Packer, D.B., Weissberger, E., 1999a. Atlantic surfclam,Spisula solidissima, life history and habitat characteristics. NOAA Tech. Mem.NMFS-NE-142, 13 pp.

Cargnelli, L.M., Griesbach, S.J., Packer, D.B., Weissberger, E., 1999b. Ocean quahog,Arctica islandica, life history and habitat characteristics. NOAA Tech. Mem. NMFS-NE-148, 12 pp.

Cassita, M.C., Hart, M.W., 2007. Spatial and temporal genetic homogeneity in the Arcticsurfclam (Mactromeris polynyma). Mar. Biol. 152, 569–579.

Cerame-Vivas, M.J., Gray, I.E., 1966. The distribution pattern of benthic invertebrates of

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

29

the continental shelf off North Carolina. Ecology 47, 260–270.Chattopadhyay, D., Rathie, A., Das, A., 2013. The effect of morphology on postmortem

transportation of bivalves and its taphonomic implications. Palaios 28, 203–209.Clark, P.J., Evans, F.C., 1954. Distance to nearest neighbor as a measure of spatial

relationships in populations. Ecology 35, 445–453.Collie, J.S., Escanero, G.A., Valentine, P.C., 1997. Effects of bottom fishing on the benthic

megafauna of Georges Bank. Mar. Ecol. Prog. Ser. 155, 159–172.Conover, W.J., 1972. A Kolmogorov goodness-of-fit test for discontinuous distributions.

J. Am. Stat. Assoc. 67, 591–595.Cronin, T.M., Hayo, K., Thunell, R.C., Dwyer, G.S., Saenger, C., Willard, D.A., 2010. The

Medieval Climate Anomaly and Little Ice Age in Chesapeake Bay and the NorthAtlantic Ocean. Palaeogeogr. Palaeoclimatol. Palaeoecol. 297, 299–310.

Dahlgren, T.G., Weinberg, J.R., Halanych, K.M., 2000. Phylogeography of the oceanquahog (Arctica islandica): influences of paleoclimate on genetic diversity andspecies range. Mar. Biol. 137, 487–495.

Daniel, W.W., 1978. Applied Nonparametric Statistics. Houghton Mifflin Company,Boston, 503 pp.

Davis, N., VanBlaricom, G.R., 1978. Spatial and temporal heterogeneity in a sand bottomepifaunal community of invertebrates in shallow water. Limnol. Oceanogr. 23,417–427.

Dexter, T.A., Kaufman, D.S., Krause, R.A., Jr., Barbour Wood, S.I., Simões, M.G.,Huntley, J.W., Yanes, Y., Romanek, C.S., Kowalewski, M., 2014. A continuous multi-millennial record of surficial bivalve mollusk shells from the São Paulo Bight,Brazilian shelf. Quat. Res. 81, 274–283.

Frey, R.W., 1987. Distribution of ark shells (Bivalvia: Anadara), Cabretta Island beach,Georgia. Geology 27, 155–163.

Frey, R.W., Dörjes, J., 1988. Carbonate skeletal remains in beach-to-offshore sediments,Pensacola, Florida. Senck. Marit. 20, 31–57.

Friedland, K.D., Hare, J.A., 2007. Long-term trends and regime shifts in sea surfacetemperature on the continental shelf of the northeast United States. Cont. Shelf Res.27, 2313–2328.

Gagan, M.K., Chivas, A.R., Herczeg, A.L., 1990. Shelf-wide erosion, deposition, andsuspended sediment transport during cyclone Winifred, central Great Barrier Reef,Australia. J. Sediment. Pet. 60, 456–470.

Gasse, F., Fontes, J.C., Plaziat, J.-C., Carbonel, P., Kaczmarska, I., de Dockker, P., Soulié-Marsche, I., Callot, Y., Dupeuble, P.A., 1987. Biological remains, geochemistry andstable isotopes for the reconstruction of environmental and hydrological changes inthe Holocene lake from north Sahara. Palaeogeogr. Palaeoclimatol. Palaeoecol. 60,1–46.

Glenn, S., Jones, C., Twardowski, M., Bowers, L., Kerfoot, J., Kohut, J., Webb, D.,Schofeld, O., 2008. Glider observations of sediment resuspension in a MiddleAtlantic Bight fall transition storm. Limnol. Oceanogr. 53, 2180–2196.

Goldberg, R., Walker, R.L., 1990. Cage culture of yearling surfclams, Spisula solidissima(Dillwyn, 1817), in coastal Georgia. J. Shellfish Res. 9, 187–193.

Grant, J., Cranford, P., Emerson, C., 1997. Sediment resuspension rates, organic matterquality and food utilization by sea scallops (Placopecten magellanicus) on GeorgesBank. J. Mar. Res. 55, 965–994.

Grill, B., Zuschin, M., 2001. Modern shallow- to deep-water bivalve death assemblages inthe Red Sea – ecology and biogeography. Palaeogeogr. Palaeoclimatol. Palaeoecol.168, 75–96.

Hageman, S.J., 2001. Encrusting invertebrate faunas and shipwreck histories: ballaststones from North Carolina shipwreck 03BUI (Queen Anne's Revenge). Southeast.Geol. 40, 65–73.

Harding, J.M., King, S.E., Powell, E.N., Mann, R., 2008. Decadal trends in age structureand recruitment patterns of ocean quahogs Arctica islandica from the Mid-AtlanticBight in relation to water temperatures. J. Shellfish Res. 27, 667–690.

Hennen, D.R., Jacobson, L.D., Tang, J., 2012. Accuracy of the Patch model used toestimate density and capture efficiency in depletion experiments for sessileinvertebrates and fish. ICES J. Mar. Sci. 69, 240–249.

Hill, P.R., Nadeau, O.C., 1989. Storm-dominated sedimentation on the inner shelf of theCanadian Beaufort Sea. J. Sediment. Petrol. 59, 455–468.

Hofmann, E.E., Powell , E.N., Klinck, J.M., Mann, D.M., Hadivogel, R., Narváez, D.B.,Zhang, X., Kuykendall, K.M., in press. Factors affecting distribution of the Atlanticsurfclam (Spisula solidissima), a continental shelf biomass dominant, during aperiod of climate change J. Shellfish Res.

Hurrell, J.W., Loon, H. van, 1997. Decadal variations in climate associated with theNorth Atlantic Oscillation. Clim. Change 36, 301–326.

Jacobson, L., Weinberg, J., 2006. Atlantic surfclam (Spisula solidissima). Status offishery resources of the northeastern US. NOAA NEFSC Resour. Eval. Assess. Div.,8 pp.

Kerr, L.A., Connelly, W.J., Martino, E.J., Peer, A.C., Woodland, R.J., Secor, D.H., 2009.Climate change in the U.S. Atlantic affecting recreational fisheries. Rev. Fish. Sci. 17,267–289.

Kidwell, S.M., 2001. Preservation of species abundance in marine death assemblages.Science 294, 1091–1094.

Kidwell, S.M., 2007. Discordance between living and death assemblages as evidence foranthropogenic ecological change. Proc. Natl. Acad. Sci. USA 104, 17701–17706.

Kidwell, S.M., 2008. Ecological fidelity of open marine molluscan death assemblages:effects of post-mortem transportation, shelf health, and taphonomic inertia. Lethaia41, 199–217.

Kidwell, S.M., Best, M.M.R., Kaufman, D.S., 2005. Taphonomic trade-offs in tropicalmarine death assemblages: differential time averaging, shell loss, and probable biasin siliciclastic vs. carbonate facies. Geology 33, 729–732.

Kidwell, S.M., Flessa, K.W., 1995. The quality of the fossil record: populations, speciesand communities. Annu. Rev. Ecol. Syst. 26, 269–299.

Kidwell, S.M., Holland, S.M., 2002. The quality of the fossil record: implications for

evolutionary analyses. Annu. Rev. Ecol. Syst. 33, 561–588.Kilada, R.W., Campana, S.E., Roddick, D., 2009. Growth and sexual maturity of the

northern propellerclam (Cyrtodaria siliqua) in eastern Canada, with bombradiocarbon age validation. Mar. Biol. 156, 1029–1037.

Kim, Y., Powell, E.N., 2004. Surfclam histopathology survey along the Delmarvamortality line. J. Shellfish Res. 23, 429–441.

Kosnik, M.A., Hua, Q., Kaufman, D.S., Wüst, R.A., 2009. Taphonomic bias and time-averaging in tropical molluscan death assemblages: differential shell half-lives inGreat Barrier Reef sediment. Paleobiology 35, 565–586.

Kowalewski, M., LaBarbera, M., 2004. Actualistic taphonomy: death, decay, anddisintegration in contemporary settings. Palaios 19, 423–427.

Lanci, L., Hirt, A.M., 2015. Evidence of Atlantic Multidecadal Oscillation in the magneticproperties of Alpine lakes during the last 2500 years. Palaeogeogr. Palaeoclimatol.Palaeoecol. 440, 47–52.

Leighton, L.R., Schneider, C.L., 2004. Neighbor proximity analysis, a technique forassessing spatial patterns in the fossil record. Palaios 19, 396–407.

Levinton, J., 1970. The paleoecological significance of opportunistic species. Lethaia 3,69–79.

Lewis, C.V.W., Weinberg, J.R., Davis, C.S., 2001. Population structure and recruitment ofthe bivalve Arctica islandica (Linnaeus, 1767) on Georges Bank from 1980–1999. J.Shellfish Res. 20, 1135–1144.

Loesch, J.G., Ropes, J.W., 1977. Assessments of surf clam stocks in nearshore watersalong the Delmarva Peninsula and in the fishery south of Cape Henry. Proc. Natl.Shellfish Assoc. 67, 29–34.

Lucey, S.M., Nye, J.A., 2010. Shifting species assemblages in the northeast UScontinental shelf large marine ecosystem. Mar. Ecol. Prog. Ser. 415, 23–33.

MacIntyre, I.G., Pilkey, O.H., Stuckenrath, R., 1978. Relict oysters on the United StatesAtlantic continental shelf. Geol. Soc. Am. Bull. 89, 277–282.

Mann, M.E., Zhang, Z., Rutherford, S., Bradley, R.S., Hughes, M.K., Shindell, D.,Ammann, C., Foluvegi, G., Ni, F., 2009. Global signatures and dynamical origins ofthe Little Ice Age and Medieval Climate Anomaly. Science 326, 1256–1260.

Mann, R., 1982. The seasonal cycle of gonadal development in Arctica islandica from thesouthern New England shelf. Fish. Bull. 80, 315–326.

Marzec, R.J., Kim, Y., Powell, E.N., 2010. Geographic trends in weight and conditionindex of surfclams (Spisula solidissima) in the Mid-Atlantic Bight. J. Shellfish Res.29, 117–128.

Merrill, A.S., Ropes, J.W., 1969. The general distribution of the surf clam and oceanquahog. Proc. Natl. Shellfish Assoc. 59, 40–45.

Meyer, T.L., Cooper, R.A., Pecci, K.J., 1981. The performance and environmental effectsof a hydraulic clam dredge. Mar. Fish. Rev. 43 (9), 14–22.

Miller, A.I., 1988. Spatial resolution in subfossil molluscan remains: implications forpaleobiological analysis. Paleobiology 14, 91–103.

Moore, G.W.K., Halfar, J., Majeed, H., Adey, W., Kronz, A., 2017. Amplification of theAtlantic Multidecadal Oscillation associated with the onset of the industrial-erawarming. Sci. Rep. 7 (#40861), 9. http://dx.doi.org/10.1038/srep40861.

Morehead, S., Montagna, P., Kennicutt II, M.C., 2008. Comparing fixed-point andprobabilistic sampling designs for monitoring the marine ecosystem near McMurdoStation, Ross Sea, Antarctica. Ant. Sci. 20, 471–484.

Munroe, D.M., Narváez, D.A., Hennen, D., Jacobson, L., Mann, R., Hofmann, E.E.,Powell, E.N., Klinck, J.M., 2016. Fishing and bottom water temperature as drivers ofchange in maximum shell length in Atlantic surfclams (Spisula solidissima). Estuar.Coast. Shelf Sci. 170, 112–122.

Munroe, D.M., Powell, E.N., Mann, R., Klinck, J.M., Hofmann, E.E., 2013.Underestimation of primary productivity on continental shelves: evidence frommaximum size of extant surfclam (Spisula solidissima) populations. Fish. Oceanogr.22, 220–233.

Murawski, S.A., Ropes, J.W., Serchuk, F.M., 1982. Growth of the ocean quahog, Arcticaislandica, in the Middle Atlantic Bight. Fish. Bull. 80, 21–34.

Murillo, F.J., Serrano, A., Kenchington, E., Mora, J., 2016. Epibenthic assemblages of thetail of the Grand Bank and Flemish Cap (northwest Atlantic) in relation toenvironmental parameters and trawling intensity. Deep-Sea Res. Part I 109, 99–122.

Narváez, D.A., Munroe, D.M., Hofmann, E.E., Klinck, J.M., Powell, E.N., Mann, R.,Curchitser, E., 2015. Long-term dynamics in Atlantic surfclam (Spisula solidissima)populations: the role of bottom water temperature. J. Mar. Syst. 141, 136–148.

NEFSC, 2009. 48th Northeast Regional Stock Assessment Workshop (48th SAW)Assessment Report. NEFSC Ref. Doc. 09-15, 834 pp.

NEFSC, 2013. 56th Northeast Regional Stock Assessment Workshop (56th SAW)Assessment Report. NEFSC Ref. Doc. 13-10, 868 pp.

Nixon, S.W., Granger, S., Buckley, B.A., Lamont, M., Rowell, B., 2004. A one hundredand seventeen year coastal water temperature record from Woods Hole,Massachusetts. Estuaries 27, 397–404.

Oeschger, R., 1990. Long-term anaerobiosis in sublittoral marine invertebrates from thewestern Baltic Sea: Halicryptus spinulosus (Priapulida), Astarte borealis andArctica islandica (Bivalvia). Mar. Ecol. Prog. Ser. 59, 133–143.

Pace, S.M., Powell, E.N., Mann, R., Long, M.C., Klinck, J.M., 2017. Development of anage-frequency distribution for ocean quahogs (Arctica islandica) on Georges Bank.J. Shellfish Res. 36, 41–53.

Palmer, C.T., 1991. Life and death of a small-scale fishery surf clam dredging in southernMaine. MAST - Marit. Anthropol. Stud. 4, 56–72.

Papili, S., Wever, T., Dupont, Y., Lancker, V. van, 2014. Storm influence on the burial ofobjects in a shallow sandy shelf environment. Mar. Geol. 349, 61–72.

Parker, R.H., 1960. Ecology and distribution patterns of marine macro-invertebrates,northern Gulf of Mexico. In: Shepard, F.P., Phleger, F.B., Andel, T.H. van (Eds.),Recent Sediments, Northwest Gulf of Mexico. American Association of PetroleumGeologists, Tulsa, OK, 302–331.

Parsons, K.M., Brett, C.E., 1991. Taphonomic processes and biases in modern marine

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

30

environments: an actualistic perspective on fossil assemblage preservation. In:Donovan, S.K. (Ed.), The Processes of Fossilization. Belhaven Press, London, 22–65.

Parsons-Hubbard, K.M., Callender, W.R., Powell, E.N., Brett, C.E., Walker, S.E.,Raymond, A.L., Staff, G.M., 1999. Rates of burial and disturbance of experimentally-deployed molluscs: implications for preservation potential. Palaios 14, 337–351.

Perry, R.I., Cury, P., Brander, K., Jennings, S., Möllmann, C., Planque, B., 2010.Sensitivity of marine systems to climate and fishing: concepts, issues andmanagement responses. J. Mar. Syst. 79, 427–435.

Poirier, C., Sauriau, P.-G., Chaumillon, E., Allard, J., 2009. Can molluscan assemblagesgive insights into Holocene environmental changes other than sea level rise? A casestudy from a macrotidal bay (Morennes-Oléron, France). Palaeogeogr.Palaeoclimatol. Palaeoecol. 280, 105–118.

Powell, E.N., 2003. Maryland inshore surf clam, Spisula solidissima, survey August 2003cruise report. Final report to J.H. Miles & Co., Inc. Haskin Shellfish ResearchLaboratory, Port Norris, NJ. 19 pp.

Powell, E.N., Callender, W.R., Staff, G.M., Parsons-Hubbard, K.M., Brett, C.E., Walker,S.E., Raymond, A., Ashton-Alcox, K.A., 2008. Mollusc shell condition after eightyears on the sea floor – taphonomy in the Gulf of Mexico and Bahamas. J. ShellfishRes. 27, 191–225.

Powell, E.N., Callender, W.R., Stanton, R.J., Jr., 1998. Can shallow- and deep-waterchemoautotrophic and heterotrophic communities be discriminated in the fossilrecord? Palaeogeogr. Palaeoclimatol. Palaeoecol. 144, 85–114.

Powell, E.N., Davies, D.J., 1990. When is an "old" shell really old? J. Geol. 98, 823–844.Powell, E.N., Hu, X.P., Cai, W.J., Ashton-Alcox, K.A., Parsons-Hubbard, K.M., Walker,

S.E., 2012. Geochemical controls on carbonate shell taphonomy in northern Gulf ofMexico continental shelf and slope sediments. Palaios 27, 571–584.

Powell, E.N., Mann, R., 2016. How well do we know the infaunal biomass of thecontinental shelf? Cont. Shelf Res. 115, 27–32.

Powell, E.N., Parsons-Hubbard, K.M., Callender, W.R., Staff, G.M., Rowe, G.T., Brett,C.E., Walker, S.E., Raymond, A., Carlson, D.D., White, S., Heise, E.A., 2002.Taphonomy on the continental shelf and slope: two-year trends – Gulf of Mexico andBahamas. Palaeogeogr. Palaeoclimatol. Palaeoecol. 184, 1–35.

Powell, E.N., Staff, G.M., Callender, W.R., Ashton-Alcox, K.A., Brett, C.E., Parsons-Hubbard, K.M., Walker, S.E., Raymond, A., 2011. Taphonomic degradation ofmolluscan remains during thirteen years on the continental shelf and slope of thenorthwestern Gulf of Mexico. Palaeogeogr. Palaeoclimatol. Palaeoecol. 312,209–232.

Powell, E.N., Staff, G.M., Davies, D.J., Callender, W.R., 1989. Macrobenthic deathassemblages in modern marine environments: formation, interpretation andapplication. Crit. Rev. Aquat. Sci. 1, 555–589.

Powell, E.N., Stanton, R.J., Jr., Davies, D., Logan, A., 1986. Effect of a large larvalsettlement and catastrophic mortality on the ecologic record of the community in thedeath assemblage. Estuar. Coast. Shelf Sci. 23, 513–525.

Powell, E.N., Stanton, R.J., Jr., Logan, A., Craig, M.A., 1992. Preservation of Mollusca inCopano Bay, Texas. The long-term record. Palaeogeogr. Palaeoclimatol. Palaeoecol.95, 209–228.

Ragnarsson, S.A., Thórarinsdóttir, G.G., 2002. Abundance of ocean quahog, Arcticaislandica, assessed by underwater photography and a hydraulic dredge. J. ShellfishRes. 21, 673–676.

Ridgway, I.D., Richardson, C.A., 2011. Arctica islandica: the longest lived non colonialanimal known to science. Rev. Fish Biol. Fish. 21, 297–310.

Rodland, D.L., Kowalewski, M., Carroll, M., Simões, M.G., 2006. The temporal resolutionof epibiont assemblages: are they ecological snapshots or overexposures? J. Geol.114, 313–324.

Ropes, J.W., 1980. Biological and fisheries data on surf clam, Spisula solidissima(Dillwyn). NEFSC Tech. Rep. Ser. No. 24. 88 pp.

Ropes, J.W., 1982. The Atlantic coast surf clam fishery, 1965–1974. Mar. Fish. Rev. 44(8), 1–14.

Ropes, J.W., Murawski, S.A., Serchuk, F.M., 1984. Size, age, sexual maturity, and sexratio in ocean quahog, Arctica islandica Linné, off Long Island, New York. Fish. Bull.82, 253–266.

Rose, G.A., 2005. On the distributional responses of North Atlantic fish to climatechange. ICES J. Mar. Sci. 62, 1360–1374.

Rowell, T.W., Chaisson, D.R., McLane, J.T., 1990. Size and age of sexual maturity andannual gametogenic cycle in the ocean quahog, Arctica islandica (Linnaeus, 1767),from coastal waters in Nova Scotia, Canada. J. Shellfish Res. 9, 195–203.

Schöne, B.R., Fiebig, J., Pfeiffer, M., Gleß, R., Hickson, J., Johnson, A.L.A., Dreyer, W.,Oschmann, W., 2005. Climate records from bivalved Methuselah (Arctica islandica,Mollusca; Iceland). Palaeogeogr. Palaeoclimatol. Palaeoecol. 228, 130–148.

Selin, N.I., 2010. The growth and life span of bivalve mollusks at the northeastern coastof Sakhalin Island. Russ. J. Mar. Biol. 36, 258–269.

Shen, S.S.P., Lee, C.K., Lawrimore, J., 2012. Uncertainties, trends, and hottest andcoldest years of U.S. surface air temperature since 1895: an update on the USHCNV2 TOB data. J. Clim. 25, 4185–4203.

Sisson, J.D., Shimeta, J., Zimmer, C.A., Traykovski, P., 2002. Mapping epibenthicassemblages and their relations to sedimentary features in shallow-water, high-energy environments. Cont. Shelf Res. 22, 565–583.

Smith, A.M., Nelson, C.S., 2003. Effects of early sea-floor processes on the taphonomy oftemperate shelf skeletal carbonate deposits. Earth Sci. Rev. 63, 1–31.

Smolowitz, R.J., Nulk, V.E., 1982. The design of an electrohydraulic dredge for clamsurveys. Mar. Fish. Rev. 44 (4), 1–18.

Solow, A.R., 1989. Bootstrapping sparsely sampled spatial point patterns. Ecology 70,379–382.

Soniat, T.M., Hofmann, E.E., Klinck, J.M., Powell, E.N., 2009. Differential modulation ofeastern oyster (Crassostrea virginica) disease parasites by the El Niño-SouthernOscillation and the North Atlantic Oscillation. Int. J. Earth Sci. 98, 99–114.

Staff, G.M., Powell, E.N., 1988. The paleoecological significance of diversity: the effect oftime averaging and differential preservation on macroinvertebrate species richnessin death assemblages. Palaeogeogr. Palaeoclimatol. Palaeoecol. 63, 73–89.

Staff, G.M., Powell, E.N., 1990a. Taphonomic signature and the imprint of taphonomichistory: discriminating between taphofacies of the inner continental shelf and amicrotidal inlet. In: Paleocommunity Temporal Dynamics: The Long-termDevelopment of Multispecies Assemblages 5. Spec. Publ. (Paleontol. Soc.), Knoxville,TN, 370–390.

Staff, G.M., Powell, E.N., 1990b. Local variability of taphonomic attributes in aparautochthonous assemblage: can taphonomic signature distinguish aheterogeneous environment? J. Paleontol. 64, 648–658.

Staff, G.M., Powell, E.N., 1999. Onshore-offshore trends in community structuralattributes: death assemblages from the shallow continental shelf of Texas. Cont.Shelf Res. 19, 717–756.

Taylor, A.C., 1976a. Burrowing behaviour and anaerobiosis in the bivalve Arcticaislandica (L.). J. Mar. Biol. Assoc. U. K. 56, 95–109.

Taylor, A.C., 1976b. The cardiac responses to shell opening and closure in the bivalveArctica islandica (L.). J. Exp. Biol. 64, 751–759.

Thórarinsdóttir, G.G., Jacobson, L.D., 2005. Fishery biology and biological referencepoints for management of ocean quahogs (Arctica islandica) off Iceland. Fish. Res.75, 97–106.

Thórarinsdóttir, G.G., Steingrímsson, S.A., 2000. Size and age at sexual maturity and sexratio in ocean quahog, Arctica islandica (Linnaeus, 1767), off northwest Iceland. J.Shellfish Res. 19, 943–947.

Tomašových, A., Fürsich, F.T., Olszewski, T.D., 2006. Modeling shelliness and alterationin shell beds: variation in hardpart input and burial rates lead to opposingpredictions. Paleobiol 32, 278–298.

Tomašových, A., Gallmetzer, I., Haselmair, A., Kaufman, D.S., Vidović, J., Zuschin, M.,2017. Stratigraphic unmixing reveals repeated hypoxia events over the past 500 yr inthe northern Adriatic Sea. Geology 45, 363–366.

Tomašových, A., Kidwell, S.M., 2009. Preservation of spatial and environmentalgradients by death assemblages. Paleobiology 35, 119–145.

Tomašových, A., Kidwell, S.M., 2010. Predicting the effects of increasing temporal scaleon species composition, diversity, and rank-abundance distributions. Paleobiology36, 672–695.

Tomašových, A., Kidwell, S.M., Barber, R.F., Kaufman, D.S., 2014. Long-termaccumulation of carbonate shells reflects a 100-fold drop in loss rate. Geology 42,819–822.

Twichell, D.C., 1983. Bedform distribution and inferred sand transport on Georges Bank,United States Atlantic continental shelf. Sedimentology 30, 695–710.

van der Meer, J., 1997. Sampling design of monitoring programmes for marine benthos:a comparison between the use of fixed versus randomly selected stations. J. Sea Res.37, 167–179.

Walker, S.E., Goldstein, S.T., 1999. Taphonomic tiering: experimental field taphonomy ofmolluscs and foraminifera above and below the sediment water interface.Palaeogeogr. Palaeoclimatol. Palaeoecol. 149, 227–244.

Wanamaker, A.D., Kreutz, K.J., Schöne, B.R., Introne, D.S., 2011. Gulf of Maine shellsreveal changes in seawater temperature seasonality during the Medieval ClimateAnomaly and the Little Ice Age. Palaeogeogr. Palaeoclimatol. Palaeoecol. 302,43–51.

Warwick, R.M., Light, J., 2002. Death assemblages of molluscs on St. Martin's Flats, Isleof Scilly: a surrogate for regional biodiversity? Biodivers. Conserv 11, 99–112.

Weinberg, J.R., 1998. Density-dependent growth in the Atlantic surfclam, Spisulasolidissima, off the coast of the Delmarva Peninsula, USA. Mar. Biol. 130, 621–630.

Weinberg, J.R., 1999. Age-structure, recruitment, and adult mortality in populations ofthe Atlantic surfclam, Spisula solidissima, from 1978–1997. Mar. Biol. 134,113–125.

Weinberg, J.R., 2005. Bathymetric shift in the distribution of Atlantic surfclams:response to warmer ocean. ICES J. Mar. Sci. 62, 1444–1453.

Weinberg, J.R., Powell, E.N., Pickett, C., Nordahl, Jr,, V.A., Jacobson, L.D., 2005. Resultsfrom the 2004 Cooperative Survey of Atlantic Surfclams. NEFSC Ref. Doc. 05-01, 41pp.

Woodin, S.A., Hilbish, T.J., Helmuth, B., Jones, S.J., Wethey, D.S., 2013. Climate change,species distribution models, and physiological performance metrics: predicting whenbiogeographic models are likely to fail. Ecol. Evol. 3, 3334–3346.

Xu, H., Kim, H.-M., Nye, J.A., Hameed, S., 2015. Impacts of the North AtlanticOscillation on sea surface temperature on the northeast US continental shelf. Cont.Shelf Res. 105, 60–66.

Zenetas, A., 1990. Discrimination of autochthonous vs. allochthonous assemblages in theEden Estuary, Fife, Scotland, U.K. Estuar. Coast. Shelf Sci. 30, 525–540.

Zhang, X., Haidvogel, D., Munroe, D., Powell, E.N., Klinck, J., Mann, R., Castruccio, F.S.,2015. Modeling larval connectivity of the Atlantic surfclams within the MiddleAtlantic Bight: model development, larval dispersal and metapopulationconnectivity. Estuar. Coast. Shelf Sci. 153, 38–53.

Zhang, X., Munroe, D., Haidvogel, D., Powell, E.N., 2016. Atlantic surfclam connectivitywithin the Middle Atlantic Bight: mechanisms underlying variation in larvaltransport and settlement. Estuar. Coast. Shelf Sci. 173, 65–78.

Zuschin, M., Piller, W.E., 1997. Bivalve distribution on coral carpets in the Northern Bayof Safaga (Red Sea, Egypt) and its relation to environmental parameters. Facies 37,183–194.

E.N. Powell et al. Continental Shelf Research 142 (2017) 14–31

31