Faunal assemblages of benthic megainvertebrates inhabiting sea scallop grounds from eastern Georges Bank, in relation to environmental factors.

: Faunal assemblages of benthic megainvertebrates from Georges Bank (NW Atlantic) commercial scallop beds were defined from video-monitored sled-dredge samples. The distribution, diversity, abundance, and biomass of the megafauna were studied in relation to water depth and sediment types. A total of 140 species of megainvertebrates representing 10 phyla were idenhfied, from which epibenthic taxa accounted for ca 76 %. Molluscs, crustaceans, annelids, and echinoderms were best represented, with bivalves ranking first in abundance (55 %) and biomass (86 %). Significant changes of species diversity, total abundance, and total biomass were found according to sediment type, but not according to depth (except for total biomass). No clear bathymetric pattern was observed, although the mean number of species, mean total density (below 80 m), and mean total biomass (below 60 m) decreased with increasing depth in the deepest zone. Maximum megafaunal richness was found in biogenic bottoms while minimum value occurred in sand dunes. Total biomass and total density were consistently dominated by a small number of taxa. Three bivalve species (Spisula solidissima, Arctica islandica, Placopecten rnagellanicus) made up to 7 l % of total biomass overall, while 14 species (predominantly Anornia spp., Arctica islandica, and Ophjopholis aculeata) accounted for 70 % of total density. Six megafaunal associations related to 2 major assemblages (biogenic sand-gravel and sand-shell fauna) were defined from multivariate analyses. Sediment type, tidal current speed, turbulent mixing, and food availability appear to be the major distribution-regulating factors of the megabenthos in the 55 to 105 m depth range. Density-dependent predator-prey relationshps were the main biological associahons shown by megainvertebrates


INTRODUCTION
Extensive benthic fauna1 studies have been carried out in recent years on the northeastern continental shelf of North America (Wigley & Haynes 1958, Wigley  1960, 1961b, 1965, 1968, Wigley & Theroux 1965, 1981,  Williams & Wigley 1977, Maurer & Leathem 1980,  1981a, b, Dickinson & Wigley 1981, Franz et al. 1981,  Maurer 1983, Michael et al. 1983, Theroux & Wigley  1983, Maurer & Wigley 1984, Maciolek & Grassle 1987,  Michael 1987, Steimle 1987, Theroux & Grosslein 1987,  Wildish et al. 1989), because of the interest in present or potential effects of expanded fisheries, oil and gas drilling, mineral recovery, and ocean dumping (see Grosslein et al. 1979) on benthic organisms.In particular, the benthos is highly sensitive to bioaccumulation of contaminants (trace metals, hydrocarbons) discharged in the course of oil and gas drilling (see Neff O Inter-Research/Printed in Germany 1987 for review), and its diversity and abundance constitute a measure of the health of the ecosystem.
Benthic invertebrates also play a critical role in trophic relationships on the shelf, providing a major source of energy to economically and ecologically important groundfish (Cohen et al. 1982).On Georges Bank, the benthos contributes directly and indirectly to highly productive fisheries of shellfish (scallop) and bottom-feeding fishes such as haddock, cod, and various flounders (Wigley 1965, Edwards & Bowman 1979, Grosslein et al. 1979, Steimle 1987).In addition, benthic organisms play a supporting role in nutrient exchange, providing a mechanism for flux of nutrients initially trapped in sediments to the water column (Zeitzschel 1980).
Despite the number of comprehensive studies dealing with benthos from Georges Bank, no study has specifically featured the megabenthos (organisms retained on a 10 mm mesh sieve; Hily 1989).Moreover, most benthic fauna1 surveys were conducted with grab samplers (0.1 m2 Smith/McIntyre grab; 0.56 m2 Campbell grab; 0.04 and 0.1 m2 modified Van Veen grab) providing small sample sizes, even with replicates at each station.Grab samplers do not give quantitative estimates of the abundance of low-density species, particularly of epibenthic megafauna (Holme 1971, McIntyre 1971, Ogden 1980, Frontier 1983, Caddy & Carter 1984, Watson et al. 1984).Moreover, most of the mobile species, such as pectinids and decapod crustaceans, escape the grabs (Holme 1971, Maurer & Wigley 1984, Steimle 1987, Thouzeau 1989), because such bottom-samplers trigger an avoidance response when approaching the seabed.
Camera and video transects have been used to estimate the abundance of megabenthic epifauna on the Northwest Atlantic continental shelf, slope, and rise (Caddy 1970, Cummins 1971, Maciolek et al. 1987, Langton & Robinson 1990), and in the Bay of Fundy (Caddy 1976, Caddy & Carter 1984).None of these studies were carried out on the Canadian side of Georges Bank.Such techniques preclude information on most infaunal taxa and only allow identification and counting of epifaunal individuals over 20 to 50 mm in size on photographs (Dare 1987, Langton & Robinson 1990) and 40 mm on video pictures (Franklin et al. 1980).
The present study investigates the distribution of benthic megainvertebrates on Canadian Georges Bank scallop (Placopecten rnagellanicus Gmelin) grounds, and patterns of megafaunal zonation with depth and sediment types.This investigation is part of a program dealing with benthic pre-recruitment patterns in P. magellanicus from the Bank (Thouzeau et al. in press).Environmental factors generate strong variability in sea scallop year-class strength (Mohn et al. 1988), and biotic interactions (spatial and trophic competition, predation) are likely to regulate the survival or growth of benthic organisms, especially pre-recruits (post-larvae and juveniles).This study focuses on large-size (> 10 mm) organisms, some of which may present biotic interactions with sea scallops, in order to define the spatial variability of the trophic niche occupied by scallops.
The distance travelled on the bottom was measured by an odometric wheel.The steel collecting box (1.0 m width X 0.4 m high X 1.8 m depth) was drilled with regularly spaced holes of 10 mm in diameter.A box closing device adapted from Aldred et al. (1976) was triggered when the sled left the bottom.Once on board, samples were washed, sorted (mechanical sorting table; grid with 10 mm diameter holes), and preserved in 5 O/O formaldehyde solution.
The targeted fauna was megabenthic animals retained on a l 0 mm sieve (Hily 1989), including age-l Placopecten magellanicus.A 2-dimensional systematic lattice sampling scheme was used because little was known about the distribution and behaviour of sea scallop post-larvae and juveniles.Thirteen latitudinal transects spaced 7.4 km apart (and stations at 8.3 km intervals) were defined within the 110 m isobath (Fig. 1).Stations were positioned in alternate rows to maximize the distance between them; this design was selected because it avoids redundancy and information gaps, as may occur with a random sampling design (Frontier 1983).Of 99 potential stations east of the International Court of Justice line (maritime boundary separating the U.S. and Canadian parts of the Bank), operational constraints reduced the coverage to 65 stations (Fig. 1).A total area of 1776 m2 was sampled during the survey.The mean sample size was 25.1 m2 (SD,,-I, = 9.4 m2), with a range of 11 to 50 m2 according to sediment types.Data analyses.Organisms were identified to species level for most taxonomic groups (except for hydroids and bryozoans), counted (except for cirripeds), and weighed (wet weight; mollusc shells included).
Samples were standardized to numbers of individuals and g m-2.Density and biomass isopleths were plotted with the ACON software using Delaunay triangles (see Robert & Black 1990), and tested for homogeneity of variances using the F,,, test (Sokal & Rohlf 1981).Since the variances were nonhomogeneous for both density and biomass, over the bathymetric and sedimentary gradients studied, a logarithmic transformation (loglo(x+ l ) ; Sokal & Rohlf 1981) was applied to the data before further statistical analysis.One-way ANOVA was performed to detect significant changes in species diversity, density and biomass of megainvertebrates, according to depth and sediment types.
Multivariate analyses, specific diversity indices and a dominance index were used to determine megafaunal assemblages and conlmunity structure.Hierarchical ascending classification (mean distance criteria; Legendre & Legendre 1984b) and correspondence analysis (Hill 1974) were performed on transformed density data to determine similarity among samples and spatial patterns of species associations.The Basic programs were adapted from Lebart et al. (1982) and Legendre & Legendre (1984b), and data processed on an IBM PC computer.Correspondence analysis was preferred over principal component analysis, because of the great number of zeros in the data matrix (Hill 1973, Daget 1976, Gauch et al. 1977).The occurrence and density criteria (Lebart et al. 1982, Legendre & Legendre 1984b) were used to remove rare (occurrence < 5 ) , low-density species, and undersampled ones (mainly polychaetes), from the initial contingency table (65 stations; 140 species).Stations with very low total abundance (40,51,59) were also removed.Sixtytwo stations and 58 species (see Table 2) were retained.
Shannon-Wiener diversity H' and Evenness J' (Pielou 1966) were used to define community structure and spatial variability.The Evenness was calculated because the Shannon index varies with sample size (total number of species), prohibiting comparisons between samples (Legendre & Legendre 1984a).Dominance ranking of the leading species in each faunal assemblage was calculated using the Le Bris index D',, (Le Bris 1988): where P, = number of samples including the species i in the assemblage j; P, = total number of samples in the assemblage j ; N,, = density or biomass of the species i in the kth sample of the assemblage j ; Nk = total density or total biomass of the kth sample.
When defining a faunal assemblage, this index is weighted towards species found in high densities or high biomass but only at a few stations (Le Bris 1988).

Relationship to bottom sediment type
Significant variations in the mean number of species, mean total abundance and mean total biomass (SAB) were found according to sediment types (l-way ANOVA; p < 0.01).The biogenic bottoms (admixture of sand, gravel, cobble and tubes of the polychaete Filograna implexa) from the northeast showed highest number of species per sample and highest total density (Fig. 5).Intermediate values were found on gravel (particle size 2 to 15 mm, similar to the scale of Buchanan & Kain 1971) and medium-to-coarse sands (except sand dunes); lowest values occurred on sand dunes (sand waves and megaripples) and pebbles (15 to 64 mm particle size) mixed with cobbles (64 to 256 mm) and boulders (> 256 mm).A different pattern was observed   for biomass; coarse sand (0.5 to 2.0 mm) sustained the highest mean biomass (733 g m-2; Fig. 5)., which showed decreasing trends with particle size (260 g m-2 in biogenic bottoms; 103 g m-2 in gravel; 15 g m-2 in pebbles).Medium sand (0.25 to 0.50 mm) of the southern half of the Bank showed high biomass variations (13 to 130 g m-2), depending on the presence of sand dunes and the amount of shell debris in surficial sediments.
Depth-and sediment-related distribution of dominant taxa

Bathymetric distribution
Molluscs prevailed in terms of density in all depth ranges except 100-109 m (polychaetes; Fig. 6).and showed decreased relative abundance with increas-Highest values of density were found on biogenic bottoms and lowest values in pebbles and sand dunes (extremes of sediment texture), for all taxa (Fig. 7).Molluscs prevailed in all sediment types (43 to 78 O/O of total numbers); mean density ranged from between 1 and 2 to 38 ind.m-' (Astarte spp.dominant on biogenic bottoms).Crustaceans (5 to 14 O/O of numbers; 0.3 to 9 ind.m-2) showed lowest abundances on the finest sediments.Echinoderms accounted for 35 O/O of total density on biogenic bottoms (30 ind.m-2; mainly Ophiopholis aculeata), 17 % on sand-shell (2.2 ind.m-2; mainly Echinarachniusparma), and 7 % on gravel (1.4 ind.m-2; mainly Strongylocentrotus droebachiensis and Asterias vulgaris).Polychaetes (8 to 24 % of total density; 0.3 to 9 ind.m-2), although 3 times more numerous on biogenic bottoms, showed higher relative abundance on sand-shell.
All taxa, except molluscs, showed maximum biomass on biogenic bottoms (Fig. 7).Molluscs dominated biomass in all sediment types (70 to 99.5 O/O of total biomass), despite sharp decreases in the finest and coarsest sediments.Echinoderms (0.9 to 47.4 g mP2) showed minimum biomass in pebbles, while molluscs (10 to 729 g m-2), crustaceans (0.2 to 7.4 g m-2), polychaetes (0.3 to 3.2 g m-2), poriferans (0.3 to 17.6 g m-2), and ascidians (0.1 to 1.8 g m-2) exhibited lowest values in sand dunes.8).The analysis opposes the finest sediments against the coarsest ones, along a sedimentary gradient from medium sand to pebbles and boulders.Group I contains central-western stations (including sand dune locations) mainly defined by medium sand with little shell debris.Low total abundance (1.2 to 3.8 ind.m-') and biomass (4 to 28 g m-') were found in these samples.Group I1 includes southernmost and eastern (deep water) stations (sand-shell bottoms).The southwestern area (Stns 81,82,88,94), which showed high abundance and biomass, is isolated within the group.All the coarse sediment stations of the northern half define Group 111.Group IIIA (northern points) includes most of the gravel and biogenic bottom stations, while Group IIIB refers to heterogeneous sediments (sand mixed with pebbles, cobbles, or boulders).According to another HAC, including all the stations and 87 species, Stns 40, 51 and 59 belong to Groups MA, I1 and IIIB, respectively.

Hierarchical ascending classification on species
Four major clusters (species code in Table 2) and 3 isolated points (PAG, THU, SIC) were separated by the classificahon analysis (Fig. g), along the sedimentary gradient.
Group I contains epifaunal species (filter feeders or predators) mainly sampled on gravel and biogenic bottoms from the northeastern area (species defining Group IIIA of HAC on stations).Group I1 (largest unit) entale, Phascolion strombi, polychaetes), mainly located at the deepest southern stations.Group IV includes common species of the southern half of the study area (Arctica islandica, Echinarachnius parma, Actinothoe gracillima, Epizoanthus incrustatus, Pagurus arcuatus etc.).Siliqua costata (segregated from all the other species) was mainly distributed in the southsouthwest area of the study.Undersampling may have caused its position on the dendrogram.Transition to sand-shell bottoms of the southern half is made with Crepidula plana and Astarte castanea, the latter being mainly distributed in sand dunes.Group I11 contains filter-and deposit-feeding species (Dentalium The first 2 axes are the main contributors to total inertia (Table 3), since percentage relative difference is maximum between Axes 2 and 3 (Benzecri & Benzecri 1984).
Axis 1 : the 2 groups of variables isolated by Axis 1 account for 78 O/O of total variance on the axis for stations, and 86 % of total variance for species.The first axis sets the southern half of the Bank (Group B) against the northeastern area (A), along increasingly coarser sediments (Fig. 10).Stns 30 and 41 (biogenic bottoms) strongly influence Axls 1 definition (29 % of total variance).Ophiopholis aculeata, Astarte undata, A. elliptica, A, borealis, A. crenata subequilatera, and Modiolus modiolus are major contributors to total variance for species (29 %).The negative part of the axis is mostly defined by Stns 88, 94, 100, 74, and 81 (20 % of total variance), and Echinarachnius parma and Arctica islandica (28 % of total variance).
Axis 2: the 2 groups of variables (C and D) account for 76 % of total variance on the axis for stations, and 91 O/O of total variance for species (Fig. 10).The second axis sets the deepest southern (except 61) points (C) against the shallowest stations of the northwest (D), with a decreasing depth gradient along the axis.Arctica islandica, Euchone rubrocincta and Potamilla reniformis prevail in the deepest waters (29 % of total variance), while Spisula solidissirna, Anornia spp.(coarse and mixed bottoms), and Astarte castanea (sand dunes) define the axis in shallow waters (34 O/O of total variance).Placopecten magellanicus (Group D ) contributes to the definition of Axis 2, showing preferential distributions for water depths less than 85 m.Previous analysis showed that Leptasterias tenera and Table 3 Axes contributions to total inertia of correspondence analysis performed on log-transformed (loglo(x+l)) density data.
Sixty-two stations and 58 species were retained for the analysis (see parma ( 1 1.5 O / O of total variance on the axis) are the main Axis 3: the 2 groups of variables (E and F) account for species associated with sand dunes.87 % of total variance on the axis for stations and 90 % of total variance for species.A strong 'Guttman' distribution (Guttman 1941, Benzecri 1973) 2 and Fig. 9 for species codes).Eigen value (V) and percentage of total variance (T) are given for each axis   6) shows that each assemblage is dominated by a small number of species, the top 10 species accounting for 73 to 86 % of mean total density and 84 to 98 % of mean total biomass.Moreover, 1 or 2 species (particularly Arctica islandica, Potamilla reniformis, Astarte castanea or Echinarachniusparma) make up 52 to 60 O/O of mean total density in the southern half.One or two species also dominate the biomass, except for the northeastern area (A). A. islandica predominates in sand-shell (62 to 91 O/U of total biomass); E. parma and A. castanea are the leading species in sand dunes (48 and 31 ' 10 of total biomass, respectively).Placopecten magellan~cus (56 % of total biomass in sand-gravel) and Spisula solidissima (84 % of total biomass in coarse sand) are the domlnant taxa in the northern half.P. nlagellanicus is a major component of each assemblage (except for the sand dunes), being ranked in the top 15 of all density indices, and first (Group E) or second in biomass.  of epifaunal taxa among the 783 species collected with a Van Veen grab.The specificity of the AQUAREVE as a better collector of epibenthic macrofauna compared to grabs (Thouzeau 1989) also applies to the megabenthos; it may partly reside in the greater areas of sampled bottom relative to studies using grabs.The inverse relationship existing between species richness and sieve mesh size on Georges Bank (1019 and 783, 700, 259, and 160 to 165 species retained on 0.3, 0.5, 1.0, and 10.0 mm, respectively) precludes any quantitative comparisons referring to different sampling methods, as pointed out by Maciolek & Grassle (1987) and Steimle (1987) Different sampler types and sieve mesh sizes may explain most of the differences among studies for benthos species richness.However, the absence of several megabenthic taxa in this study may be related to environmental factors such as water depths, bottomwater (b-W) temperatures, and sediment types.This study was limited to depths in the range 55 to 110 m (commercial scallop grounds), while greater bathymetric ranges (between 60 to 200 and 25 to 250 m) were reported in Theroux & Grosslein (1987), Michael  (19871, and Maciolek & Grassle (1987).Maciolek et al. (1987) sampled a deeper area of the Bank between 250 and 2155 m.The annual range and extremes of b-W temperatures (linked to depth) also present smaller variations in the study area.The annual range of b-W temperature varies between 0 to 4 "C (100 to 200 m depth) and 12 to 16 "C on top of the Bank (Dickinson & Wigley 1981, Michael et al. 1983).The magnitude of temperature variations in the study area ranges between 4 to 8 "C in the mid-east (80 to 110 m depth) and 8 to 12 "C in the mid-west (60 to 80 m).Northwest Atlantic outer shelf boreal species such as Astropecfen an~ericanus, Sclerasterias tanneri, Porania insignis (Franz et al. 1981), Catapagurus sharren, and Munida i n s (Theroux & Grosslein 1987) were not sampled because they occur in deeper waters where winter b-W temperatures remain above 5 to 6 "C.Summer b-W temperatures explain the absence of cold-water species such as Hippasteria phrygiana, Ceramaster granularis, Pteraster militaris (found in areas where b-W temperatures do not exceed 7 to 8 "C; Franz et al. 1981), Pandalus borealis, Lebbeus polaris (Williams & Wigley 1977), and Cyclopecten pustulosus (Theroux & Wigley 1983), usually found in waters deeper than 100 to 120 m.On the other hand, the absence or scarcity of major epibenthic species such as Brisaster fragilis, Ophiura sarsi, and Geryon quinquedens (species defining the Western Basin assemblage of Theroux & Grosslein 1987) may be related to the absence (except for small patches) of fine-grain sediments (silt-clay and muddy sand) in the study area.On Georges Bank, these sediments mainly occur in the Western Basin (northwestern U.S. zone), between 150 and 200 m.

Depth-and sediment-related distribution of taxonomic groups
Though most of the differences were not significant, the mean number of species, mean total density (below 80 m ) , and mean total biomass (below 60 m) decreased with increasing depth in this study (Fig. 4).All taxa showed sharp decreases in density and biomass below 100 m (Fig. 6), in agreement with Theroux & Grosslein (1987).The same bathymetric patterns (but high spatial variations), with abrupt faunal discontinuities below 100 m, were found in different locations of the Middle Atlantic Bight region (Wigley & Theroux 1976, Boesch et al. 1977, Wigley & Theroux 1981, Maurer & Wigley 1984) and elsewhere (see Parsons et al. 1977).Recent studies on the Bank (Maciolek & Grassle 1987, Michael 1987) did not show reduced macrofaunal richness and density in deeper waters (25 to 200 m depth range).It may result from polychaete density increasing with depth on Georges Bank (Maurer & Leathem 1980).
Bathymetric distribution of the megabenthos varied depending on taxa in this study (Fig. 6).Mollusc density decreased with depth below 80 m on the northern edge of the Bank, which agrees with Theroux & Grosslein (1987).Sediment instability (in sand dunes) or inappropriate substrata (pebbles and cobbles) might account for the low bivalve densities in the 60-69 m depth range (Maurer & Wigley 1984).The inverse relationship between mollusc biomass and depth (except at 90-99 m) agrees with both studies cited above.Patchy distributions of heavy species (Arctica islandica a n d Spisula solidissima) created spatial variability along the depth gradient however, as in Theroux & Grosslein (1987).Decapod crustaceans and echinoderms showed the same bathymetric patterns overall, with increasing densities from the shallowest waters to mid-depths (70-79 m for decapods; 80-89 m for echinoderms), and decreasing values in deeper waters.The same trend was observed for echinoderm biomass, while there was no clear pattern for decapods except decreased biomass below 100 m.Theroux & Grosslein (1987) did not find a significant relationship between depth and density or biomass of these taxa.Maurer (1983) found significant decreasing echinoderm biomass with increasing depths, but no relationship for density.Wigley & Theroux (1981) and Maurer & Wigley (1984) showed a negative correlation between depth (0 to 200 m depth range) and decapod density or biomass, off southern New England, but the opposite for echinoderms.
Discrepancies between studies may be explained by variations in faunal composition (and dominant species presenting different bathymetric preferences), and patchy distributions of major species.Nonsignificant differences (except for total biomass in shallow water) and great variations of SAB values with depth (Fig. 4) in this study would indicate that depth is not the main factor regulating megabenthos distribution in the 55 to 105 m depth range.Significant differences and smaller variations with sediment type (Fig. 5), as well as the prevalence of this factor in defining the faunal assemblages, would suggest that sediment type (depending on bottom currents and turbulent mixing) is the main distribution factor of the megabenthos in the study area.
The lowest values of SAB were found at the extremes of sediment texture (Fig. 5).Except for maximum values, similar patterns in abundance and biomass were observed, i.e. decreased values with increased particle size in the coarsest sediments but positive relationship with grain size in sands.Wigley & Theroux (1981) and Theroux & Grosslein (1987) also found increased total biomass with grain size in sands.The latter study reported average total biomasses of 371 g m-2 in coarse sand, 230 g m-' in medium sand (but less than 50 g m-' on most of the Canadian side of the Bank), and 85 g m-2 in gravel of the Northeast Peak (neither biogenic bottoms nor pebbles were separated from gravels in their study).Total biomass in AQUAREVE samples was consistently higher than in grab samples, which may be explained by the efficiency of the collecting device, preservation techniques, contagious distributions of the species and year-to-year differences.
The escapement of mobile organisms with grab sampling (see 'Introduction') leads to underestimated biomass of large-size taxa (pectinids, decapods) which accounted for 15 '10 of total biomass in thls study (Table 2).Moreover, these taxa could represent up to 61 O/O of total biomass on coarse sediments (Table 6).The sled-dredge towed at 1 to 1.5 knots is likely to catch greater numbers of these slow-moving individuals, compared to grabs.In addition, the higher biomasses found in the coarse sediments of the Northeast Peak may reflect better sampling with the AQUAREVE, since grabs do not dig deeply in hard bottoms (Holme 1971, McIntyre 1971, Wildish et al. 1989).Preservation techniques may account for some discrepancy between studies since Mills et al. (1982) showed that methods of preservation and storage may alter biomass estimates by as much as 15 % .The absence of Spisula solidissima from the dominant taxa of Theroux & Grosslein (1987), and the scarcity of Echinarachnius parma in this study, also point out to the impact of distribution patterns of dominant species on the results.
The present study emphasizes 2 sediment types (sand dunes and biogenic bottoms) which were not segregated by Theroux & Grosslein (1987).SAB values of sand dune samples were markedly lower than those found in other medium sand samples (Fig. 5).A 10-fold reduction of total megabenthos density and biomass was found in sand dunes, compared to sand-shell.This result indicates that, despite its size, the megabenthos (both epifauna and infauna) does not appear to be better adapted to sediment instability than the macrobenthos, which shows decreased diversity with increased sediment mobility (Maurer & Wigley 1984).One should remember however, that factors affecting the settling and survival of juveniles cvlll also determine to a large extent the distributional patterns of the adults, since most of the sampled megafaunal species are poorly motile.Biogenlc hottoms were identified as a specific biosedimentary category, because of their richness and particular megafauna (several species were exclusive to this unit).Hulsemann (1967) also distinguished the northeastern corner from the rest of the Bank, because of higher level of calcium carbonate in surface sediments (> 20 " n , versus < 5 " 0 ) .SAB values In biogenic bottoms were markedly higher than those recorded in gravel (Fig. 5).Moreover, each taxonomic group showed highest density and biomass (except for mollusc biomass) for this bottom type (Fig. 7).Sediment heterogeneity and porosity, as well as the functional importance of polychaete tubes and cobbles (in provid-ing spatial refuges from predators and suitable microhabitats for invertebrates, especially the juveniles), may partly explain biogenic bottom richness.The abundance of infaunal taxa can be ascribed to sediment porosity (polychaetes) and the occurrence of a sandy fraction between biogenic fragments (allowng bivalves to inhabit the sediment).Maurer & Leathem  (1981b) showed that the abundance of motile carnivorous polychaetes on Georges Bank was related to coarse-grained sediments.Enhanced movement and feeding processes within the pore spaces existing between grains would promote their proliferation.This relationship applies to the biogenic bottoms compared to other coarse sediments: motile polychaetes were 3 times more numerous in biogenic bottoms than in coarse sand.The abundance of sessile epibenthic taxa results from the occurrence of numerous substrata (biogenic fragments, gravels, and cobbles) on the surface of the sediment, allowing individuals to settle.
Distribution patterns of the megabenthos were expected to be less obvious than those of macrobenthos, because the former comprises a proportionately higher abundance of vagile epifaunal taxa, which are supposedly less sensitive to sediment texture.Welldefined patterns of megafaunal zonation were found however, pointing out habitat preference and likely functional relationships.

Dominant taxa and functional relationships between species
Dominant megabenthic taxa of Georges Bank belong to the same major taxonomic groups as the macrobenthos (see Theroux & Grosslein 1987), i.e. molluscs, echinoderms, crustaceans, and annelids (large-size tubiferous polychaetes).Despite the number of epi-fauna1 species among the megabenthos, low abundances and biomasses of epibenthic sessile taxa were found (Tables l & 2).The sparsity of epifauna on exploited scallop grounds has been pointed out by Caddy (1973), Caddy & Carter (1984), and Lough et al.  (1989).Sediment furrowing, broken epifaunal organisms (worm tubes, poriferans, ascidians, anthozoans), and the absence of erect ectoprocts and hydrozoans result from trawl and scallop dredge activities.This study emphasizes the sparsity of epibenthic sessile taxa on coarse sediments of the northern half of the Bank (heavily fished area), despite numerous substrata available for settlement.
The present study establishes bivalve prevalence (55 and 86 O/O of total density and biomass, respectively) among megainvertebrates, and emphasizes the leading role of filter-feeding bivalves in the area under investigation (Table 2).Bivalve dominance (57 to 80 % of total biomass) is a recurrent trend in the whole Middle Atlantic Bight region (Wigley & Theroux 1981, Maurer & Wigley 1984).The U.S. side of Georges Bank exhibits different patterns with echinoderms and molluscs being CO-dominant in biomass (Wigley 1961b, 1968, Theroux & Grosslein 1987).Echinarachnius parma (40.9 % of total biomass), Arctica islandica (25.7 %), and Modiolus modiolus (8.4 %) were reported as major contributors to biomass while Placopecten magellanicus (0.1 %), Spisula solidissima, and Astarte spp.were not listed among the dominant taxa (Theroux & Grosslein 1987).The low abundances of Spisula may indicate undersampling by grabs (as for scallops), this species being commercially exploited on the Bank.Surf clams were mostly sampled in coarse sand in depths less than 70 m in this study, which agrees with its shallow-water preference (Theroux & Wigley 1983).The scarcity of Echinarachnius parma (sand dollar) on the Canadian side of the Bank (Table 2) is related to its sedimentary affinity for fine-to-medium, often rippled, sands (Maurer & Wigley 1984, Lawrence et al. 1985  Maciolek & Grassle 1987, Theroux & Grosslein 1987), and its bathymetric preference for shallow waters.Sand dollars are mainly located in the western half of Georges Bank ( U S .side), a large high-biomass area (100 to 4450 g m-') extending from 40 to 60 m (Theroux & Grosslein 1987).
The largest filter-feeding bivalves (Spisula solidissima, Arctica islandica, Placopecten magellanicus, and Modiolus modiolus) showed discrete distributions in the study area, suggesting limited trophic competition between species, and possibly adaptative strategies.Caddy & Carter (1984) stated that the distribution of most epifaunal species in relation to other taxa was random; the common presence of most species within a habitat would be determined by physical characteristics of the habitat.However, numerous functional relationships between epifaunal species have been shown in benthic communities (see Ojeda & Dearborn 1989 for review).In addition to the barnacle-sea scallop association reported by Caddy & Carter (1984), this study mainly pointed out predator-prey relationships (Fig. 9) such as the close distributions of P. magellanicus and Asterias vulgaris (one of the main sea scallop predators on the Bank); Anomia spp.(prey) and Buccinum undatum, Colus stimpsoni and Pagurus acadianus; A. islandica (prey) and Pagurus arcuatus and Colus pygmaeus; A. vulgaris (prey) and Crossaster papposus.The latter, which mainly feeds on asteroids and especially on Asterias (see Franz et al. 1981), showed maximum abundance at the same stations as Asterias (northern half).Bivalve predators (decapods, boring gastropods, echinoderms) exhibited density-dependent relationships with potential preys (Figs. 6 & 7), suggesting that distribution patterns of motile carnivorous n~egabenthos depend on trophic relationships.Maximum abundance of sea scallops (northern half) was associated with maximum density of spatio-trophic competitors and potential predators; the areas of high scallop occurrence are also likely to have the greatest mortality rates resulting from biotic interactions.

Fauna1 assemblages in relation to environmental factors
The first attempt to identify faunal associations on Georges Bank and in the Gulf of Maine was made by Wigley & Haynes (1958) who defined 4 communities (I to IV) relating to sand fauna, silty-sand fauna, gravel fauna, and muddy basin fauna.'Community 11' (Northern edge and Northeast Peak) and 'Community 111' (extending on most of the bank in depths less than 100 m) correspond to the present study area.Theroux & Grosslein (1987) also recognized 4 major faunal zones, the 'Northeast Peak assemblage' and 'Central Georges assemblage' relating to Communities I1 and 111, respectively.Macrofaunal assemblages emphasize taxa (amphipods, polychaetes) which were missed or undersampled in this study.Based on sediment types and bathymetry however, the megafaunal assemblages (Fig. 12A) clearly match the faunal zones of Theroux & Grosslein (1987).In addition, the present study allows the definition of small-scale variations (facies) in megafaunal zonation, which are likely to b e related to sediment patchiness, bottom currents, turbulent mixing, food availability and depth.
The sand-gravel facies (E) extends on each side of the biogenic facies, on gravels (10 to 69 O/O of sediment weight), pebbles and cobbles interspersed with sand.The sand-gravel fauna differs from the biogenic gravel fauna by decreased abundance and biomass of suspension feeders (epi-and infauna), the absence of several epibenthic taxa and the advent or increase of mediumsand species common in the southern half (Arctica islandica, Clymenella torquata, Ensis directus, Echinarachnius parma, Astarte castanea, Pagurus arcuatus).Ranlung of Dichelopandalus leptocerus in the top 10 of the density index indicates silty-sand or mud patches; this species is characteristic of the 'silty-sand and mud patch fauna' (Wigley 1968, Maurer & Wigley  1984).
The mixed sand facies (D) occupies a large area extending from the northwest (55 to 60 m) to the central part (75 to 82 m) of the study area.Sediments reflect spatial variations of tldal current direction and speed (Fig. 12B) and contain an admixture of coarse sand (dominant fraction) interspersed with boulders and pebbles of different sizes.Percent gravel by weight ranges from less than 10 % in the northwestern area up to 49 % in the central zone (Lough et al. 1989).An increase of sand particles allows the inclusion of new sand-bottom species among the dominant taxa (Spisula solidissima, Astarte castanea, A. quadrans, Aphrodite hastata) and surf clams predominate.Typical medium- sand species also show up in this unit (Thyone spp., Lunatia heros, Sillqua costata, Cerebratulus sp.), or s ~~s t a i n higher abundance than in sand-gravel (Echinarachnlus parma, Pandora inornata, Dendrodoa carnea).Hard-bottom epibenthic taxa and infaunal bivalves common to the biogenic gravel and sandgravel facies either show decreased density and biomass (Table 7) or are absent (Polynoidae, Subertechinus hispidus, Boltenia spp.).
The 3 northern facies exhibit marked differences in fauna1 composition or dominance, which justify their separate identity within the Northeast Peak assemblage of Theroux & Grosslein (1987).The gravel fauna on the northern edge is one of the nchest and most complex benthic communities on Georges Bank (Wig- (++++) 5.0 to 9.9 ind m-'    7) may be related to higher food availability due to physical processes, compared to the southern half.In addition to the clockwise flow around Georges Bank (Greenberg 1983, Butman et al. 1987), a complex frontal system exists along the northern edge of the Bank separating Georges Bank water from Gulf of Maine water (Flagg 1987).The extensive gravel deposit is in a transition area where the subsurface Gulf of Maine-Georges Bank water front moves northward and southward across the Bank edge with the rotary tidal currents (Lough & Trites 1989).The summer intensification of the front-jet system and the intrusion of cold, nutrient-rich Gulf of Maine Intermediate Water along the northern edge (Lough et al. 1989) lead to maximum abundance of phytoplankton centered near the northern flank during summer (Cura 1987).Distribution of large-size suspension feeders, such as Spisula solidissima, in northwestern shallow waters is associated with a primary production which may be 4 to 5 times greater than at the Bank 100 m isobath (O'Reilly et al. 1987).In depths shallower than about 60 m, averaged water-column concentration of chlorophyll a and annual total primary production are 2.58 mg m-3 and 455 g C m-2 respectively, compared to 1.18 mg m-3 and 310 g C m-2 between 60 and 100 m (O' Reilly et al. 1987).In addition, tidal currents are strong enough in shallow waters to cause sediment resuspension (Twichell et al. 1987) and complete vertical mixing of the water column throughout the year (Butman & Beardsley 1987, Flagg 1987), providing live phytoplankton cells and resuspended particulate organic matter (250 to 500 bg 1-' In summer; Twichell et al. 1987) to suspension feeders.Decreased primary production from shallow to deep waters does not explain the abundance of suspension feeders in the biogenic gravel facies.This may result from higher sedimentation of organic matter (providing food to benthic organisms) because of the low rotary tidal currents in this area, compared to the sandgravel facies.

Sand-shell assemblage
The 3 southern facies (B, C, F in Fig. 12A) belong to the Central Georges assemblage of Theroux & Grosslein (1987), and exhibit a sand-shell fauna.Medium sands predominate in the southern half but numerous valves of ocean quahogs, surf clams and sea scallops, near and on top of the sediment, provide a substratum to sessile polychaetes and Zoantharia.
The typical fauna of the sand-shell assemblage (B) extends on most of the southern half of the study area, from 75 to 90 m.Arctica islandica and Echinarachnius parma are the leading species.Actinothoe gracillima, Epizoanthus incrustatus, Pagurus arcuatus, Spisula solidissima, Clymenella torquata, Aphrodite hastata, Dentalium entale, Phascolion strombi, and Cerebratulus sp. are other typical components.Free-living forms include most of the carnivorous decapods, gastropods and asteroids found in the northern assemblage (ubiquitous), but in lower densities.Motile carnivorous polychaetes are reduced to Nephtyidae and Nereidae; sessile epibenthic taxa also show lower diversity and abundance than in the northern assemblage (Table 7).The sand-shell fauna presents similarities with the sand fauna off southern New England (Maurer & Wigley 1984); it contains more deposit feeders (feeding on the benthic boundary layer) than the coarse sediment assemblage.It also includes fewer suspension-feeding infaunal bivalves than coarse sands of the northwest.Differences may result from lower food availability (suspended particles) in southern waters (Cura 1987), which do not benefit from the intrusion of nutrient-rich subsurface Gulf of Maine water.Water stratification (preventing particulate organic matter resuspension) and lower sedimentation of organic matter resulting from lower initial phytoplankton biomass and decreased productivity with depth (O'Reilly et al. 1987) may explain the scarcity of large-size suspension feeders in the eastern half of the sand-shell assemblage.Higher abundances of A, islandica and S. solidissima in the western half may be related to higher dinoflagellate abundance (see Cura 1987) and increased tidal currents (Fig. l2B).Surf clams and ocean quahogs were both sampled in areas (Stns 11,12,81,88,94,95 & 100) presenting some of the highest bottom currents within the study area.
The deep-water sand-shell facies (C) is located in the southernmost deepest zone (90 to 105 m) of the study area, but it probably extends northwards on sand-shell bottoms in waters deeper than 90 to 95 m (e.g.Stn 61).Tubiferous and motile polychaetes predominate, which agrees with the depth-related distribution of annelids on the Bank (Maurer & Leathem 1980).Heterogeneous megafauna includes a n admixture of coarse-sediment and sand-shell species (Table 7).Increased abundances of small-size suspension-feeding polychaetes (in contrast to large-size bivalves) in this deep facies may result from the smaller size of primary producers in deep waters.O'Reilly et al. (1987) 1987).The southern limit of the study area corresponds to the southward extension limit of the sand-shell assemblage: A. circinata is a common member of the Southern Georges assemblage (fine sand bottoms) extending to the southern and southwestern flanks of the Bank at depths ranging from 80 to 200 m (Theroux & Grosslein 1987).
The sand dunes facies (F) is located in the centralwest between 60 and 70 to 75 m.Sand waves (up to 8 m high; Fader, in Lough et al. 1989) and megaripples are formed of well-sorted medium-to-coarse sand with few shells (98 to 99 O/O sand in dune crests; Lough et al. 1989).The megafauna, dominated by Astarte castanea and Echinarachnius parma (Table 7), is an impoverished one with low species diversity and abundance caused by limiting current-sediment erosion conditions (Wildish et al. 1989).Sand waves and megaripples are present where surface tidal currents are more than 40 cm S-' on the Bank (Twichell et al. 1987), which agrees with Greenberg's (1983) model predicting increased tidal currents in the facies area (Fig. 12B).Epibenthic sessile taxa are restricted to few Field work was carried out from the RV 'E.E. Prince' between August m LAG PEBBLE GRAVEL DEPOSIT m COARSE SAND COBBLES AND BOULDERS I I Ill SAND DUNES (sand waves or megaripplesl + + + + + BlOGENlC BOlTOM (sands, gravels, cobbles, and bryozoan debris)

Fig. 4 .
Fig. 4. Mean number of species, mean density (ind.m-'), and mean biomass (wet weight; g m-2) of all taxonomic groups of megainvertebrates combined, in relation to depth (95 % confidence limits are represented)

Fig. 8 .
Fig. 8. Stations clustered by hierarchical ascending classification (HAC) using the mean distance criteria.Dendrogram of similarity based on log-transformed (loglo(x+ l ) ) density values of the 58 most common megabenthic taxa at 62 stations.Three major clusters of stations (I to 111) are separated by HAC, with 2 groups (A & B) isolated within Cluster 111 Fig. 9. Species clustered by hierarchical ascending classification (HAC) using the mean distance criteria.Dendrogram of similarity based on log-transformed (loglo(x+ 1)) density values of the 58 most common taxa at 62 stations.Species codes in Table 2 except for CUF = Cucumaria frondosa; G A C = Gattyana cirrosa; GLC = Glycera capitata; MUD = Musculus discors, PAG = Pandora gouldiana; PHS = PhascoLion strornb~; PHU = Pholis gunnellus; and PRG = Propebela gouldii.Four major clusters of species (I to IV) are separated by HAC

Fig. 10 .
Fig. 10.Plane 1-2 of the correspondence analysis performed on log-transformed (loglo(x+ 1)) density values of the 58 most common megabenthic taxa at 62 stations (see Table2and Fig.9tor species codes).Eigen value (V) and percentage of total variance (T) are given for each axis

A
moderate number of species was collected with the AQUAREVE sampler compared to previous studies on Georges Bank.The NorthEast Fisheries Center (NEFC) reference list comprises 259 benthic species (Theroux & Table 6 Dominance ranlung in biomass (Le Bris Index D',,, Le Bris 1988) o f the top 5 species o f megalnvertebrates in each fauna1 group defined by correspondence analysls ( A = biogenic gravel, B = sand-shell, C = deep sand-shell, D = mixed sediments, E = sand-gravel, F = sand dunes) Mean biomass (g m ' + standard error), percentage o f total biomass, and frequency of occurrence o f the species are given in each group Gioups Fig.12(A) Geographic location of megafaunal assemblages as outlined by multivariate analyses.Facies A, D and E define the biogenic sand-gravel assemblage while facles B, C and F refer to the sand-shell assemblage.(B) Depth-averaged mean Eulerian current driven by rectificahon of the M? tidal current In the Georges Bank region as predicted by Greenberg's 7 km grid, nonlinear numerical model of the Gulf of Maine(Greenberg 1983, modified)

Table 7 .
Distribution by density intervals (mean values) of the main megabenthic taxa sampled with the AQUAREVE 111, within the 6 fauna1 groups defined by multivariate analyses.( + ) 1 0 1 ind m-'; ( + + ) 0.1 to 0.9 ind.m-*; ( -C + + ) 1.0 to 4 .9 ind.m-'; primary production (265 to 455 g carbon m-2; in O'Reilly et al. 1987) of the Northwest Atlantic shelf, except for the Middle Atlantic coastal band (Cohen & Grossleln 1987).Higher abundance of suspension feeders in the northern half of the Bank (Table individuals of Actinothoe gracillima, Anomia spp., Crenella glandula a n d Potarnilla reniformis.Several major c o m p o n e n t s of t h e sand-shell a s s e m b l a g e a r e missing, s u c h as Arctica islandica, Epizoan thus incrustatus, Clymenella torquata, Placopecten magellanicus, Den talium entale a n d s o m e predators (Lunatia heros, Hyas coarctatus, Pagurus pubescens, Colus pygmaeus).In s u m m a r y , well-defined zonation patterns oi b e n t h i c m e g a i n v e r t e b r a t e s occur o n G e o r g e s Bank scallop g r o u n d s , i n relation to environmental conditions.Results a g r e e with previous s t u d i e s d e a l i n g with G e o r g e s B a n k macrofauna a n d allow t h e inclusion of m e g a i n v e r t e b r a t e s within t h e faunal a s s e m b l a g e s of Wigley & H a y n e s (1958) a n d T h e r o u x & Grosslein (1987).T h i s s t u d y s h o w s small-scale variations i n m e g a f a u n a l zonation, a p p a r e n t l y l i n k e d t o spatial variations of environmental factors.T h e s e results provide a basis for further investigation i n t o t h e roles of s e d i m e n t texture as modified b y physical disturbances, a n d ecological factors s u c h a s food availability, competition a n d predation, on distribution p a t t e r n s of commercially exploited s p e c i e s l i k e s e a scallops a n d surf clams.

Table 1 .
Total biomass (wet weight; g m-"), total density (numbers n1C2), number of species, percentage composition and frequency of occurrence of taxonomic groups of megabenthic invertebrates (except hydroids, bryozoans, and cirripeds) and vertebrates, sampled with the AQUAREVE 111 on Georges Bank

Table 2
Rank order o f some of the more important species in the taxonomic groups sampled with the AQUAREVE I11 on Georges Bank, by occurrence, percentage of total biomass, and percentage of total density.Anomia spp.= A. squamula L. and A. simplex Orbigny

Table 2 )
Cancer borealis were exclusive to Group C. Several rnagellanicus, Strongylocentrotus droebachiensis, Divariables (32 %) defining Axis 2 were not characteristic chelopandalus leptocerus, Cancer irroratus, and Colus of the axis, suggesting that water depth was not the stirnpsoni are characteristic of the sand-gravel statlons main distribution factor of the megabenthos; the bathy- (maximum relative contributions).Astarte castanea, Silimetric gradient would b e superposed on a sedimentary qua costata, Thyone unisernita, and Echinarachnius gradient.

Table 4
Synthetic parameters of the megafaunal groups defined by correspondence analysis A = biogenlc gravel, B = sandshell, C = deep sand-shell D = m ~x ~d wriiments E = sand-gravel F = sand dunes a Except hydroids and bryozoans

Table 5 .
Dominance rankinLe Bris 1988)(LeBris index D',,;Le Bris 1988)of the top 5 species of megainvertebrates in each fauna1 group defined by correspondence analysis ( A = biogenic gravel; B = sand-shell; C = deep sand-shell; D = mixed sediments; E = sand-gravel; F = sand dunes).Mean density (ind.m-' f standard error), percentage of total density, and frequency of occurrence of the species are given in each group.Groups A. D, and E define the biogenic sand-gravel assemblage; Groups B. C , and F define the sand-shell assemblage A, D, and E define the biogenic sand-gravel assemblage, Groups B, C , and F deflne the sand-shell assemblage Michael (1987)), butMichael (1987),Maciolek &  Grassle (1987), and Maciolek et al. (1987)reported greater numbers (700, 783, and 1019 specles, respectively).However, this study reports the greatest number of epifaunal taxa despite the largest sieve mesh size.The prevalence of epifaunal taxa among megainvertebrates collected with the AQUAREVE 111 (76.4 'lo) constitutes a major difference to macrofaunal studies.Maciolek & Grassle (1987) reported only 9.4 O / O