Oecologia (2005) 144: 125–136
DOI 10.1007/s00442-005-0017-4

COMMUNITY ECOLOGY



Pedro A. Quijon Æ Paul V. R. Snelgrove
       ´

Predation regulation of sedimentary faunal structure:
potential effects of a fishery-induced switch in predators
in a Newfoundland sub-Arctic fjord

Received: 30 March 2004 / Accepted: 19 January 2005 / Published online: 11 May 2005
Ó Springer-Verlag 2005

                                 results suggest that crab predation is a significant
Abstract The collapse of the cod fishery in Newfound-
                                 structuring force in Newfoundland sedimentary com-
land has coincided with marked increases in abundances
                                 munities. Given the historical changes that have oc-
of snow crab, pandalid shrimp, and other crustaceans
                                 curred in predator composition as a result of cod over-
that prey on sedimentary infauna. A 3-year sampling
                                 fishing, we hypothesize that broad-scale community
program in Bonne Bay, Newfoundland indicates differ-
                                 changes may be taking place in North Atlantic benthic
ences in composition and number of these predators in
                                 ecosystems.
the two main arms of the fjord that coincide with strong
differences in benthic community structure. To test
whether predation pressure contributes to the observed      Keywords Predation Æ Fishery Æ Sedimentary benthos Æ
patterns in sedimentary fauna, exclusion field experi-       Sub-Arctic fjord
ments with full and partial cages were deployed in both
arms at 30-m depth and sampled along with ambient
sediments at 0-, 4-, and 8-week periods. Predation sig-
nificantly influenced species composition, abundance        Introduction
and, in some cases, diversity. The most striking changes
included increases in the polychaetes Pho¨loe tecta and      Among the most pervasive effects of fisheries is the
Ophelina cylindricaudata in exclusions relative to con-      alteration of food webs through removal or alteration of
trols, and concurrent declines in the polychaete Parad-      top predators (e.g. Botsford et al. 1997; Pauly et al. 1998;
oneis lyra and the cumacean Lamphros fuscata. In         Jackson et al. 2001), and the potential establishment of
laboratory experiments, fresh non-disturbed sediment       alternate states that favor different predator fields. In
cores from each experimental area were either protected      coastal Newfoundland, overfishing led to a complete
or exposed to snow crab, the most abundant predator in      collapse of all cod stocks (Hutchings 1996; Myers et al.
the bay. A snow crab inclusion experiment was also        1996), with an associated increase in primarily benthic
carried out in the field, using cages similar to those used    predators such as snow crab and shrimp (Koeller 2000;
for exclusions. Despite differences in sedimentary faunas     Worm and Myers 2003). The collapse of cod, a natural
in the two arms, both types of experiments detected a       predator of snow crab and shrimp, may represent a
predator effect that was very similar to that documented      predator release that has resulted in increased numbers
in exclusion experiments. Thus, despite differences in the     of both crustacean species (Lilly et al. 2000; Bundy
scales associated with each type of manipulation, our       2001). This switch in top predators is expected to have
                                 significant ramifications for benthic infauna, given that
                                 snow crab and shrimp, in contrast with adult cod, are
Communicated by Pete Peterson
                                 primarily benthic feeders (Brethes et al. 1984; Bergstrom
                                                ˆ             ¨
      ´
P. A. Quijon Æ P. V. R. Snelgrove                 2000). Few studies have examined cascading effects of
Ocean Sciences Centre and Biology Department,
                                 ecosystem alteration in the marine realm, but there is
Memorial University of Newfoundland,
                                 evidence that top-down effects may be more important
St. John’s, NL A1C 5S7, Canada
                                 than bottom-up effects (Jennings and Kaiser 1998;
              ´
Present address: P. A. Quijon (&)
                                 Micheli 1999). Thus, the rapid increase in shrimp and
Institute of Marine and Coastal Sciences, Rutgers University,
                                 crab in coastal Newfoundland over the last decade may
71 Dudley Road, New Brunswick, NJ 08901-8521, USA
E-mail: quijon@marine.rutgers.edu                 have cascading effects in sedimentary systems.
                                   Numerous benthic predators including blue crab
P. V. R. Snelgrove
                                 reach their northern distribution limit near Cape Cod
Canada Research Chair in Boreal and Cold Ocean Systems
126

                               summer season (Hooper 1996; Ennis et al. 1990).
(Williams 1984), and it has been suggested that preda-
                               Additional studies have focused on predator life histo-
tion plays a lesser role in benthic communities located
                               ries in the bay (snow crab: Comeau et al. 1998, 1999;
further north in the western Atlantic (Woodin 1976).
                               Conan et al. 1996), and in the Gulf of St. Lawrence
Nonetheless, increasingly large numbers of northern-
                               (snow crab: Brethes et al. 1987; Sainte-Marie and Gil-
                                       ˆ
native species (rock crab, Jonah crab, snow crab,
                               bert 1998; pandalid shrimp: Ouellet and Lefaibre 1994;
pandalid shrimp, mud shrimp) and invasive species (e.g.
                               Ouellet et al. 1995; Simard et al. 1990; Rock crab: Hu-
green crab in Nova Scotia) suggest otherwise (Hudon
                               don and Lamarche 1989). Based on these preliminary
and Lamarche 1989; Jamieson 2002). Predation is
                               observations, we hypothesize that there are strong epi-
thought to play a key role in marine sedimentary sys-
                               faunal predatory influences on infaunal abundance,
tems, in part, because of the lack of clear evidence for
                               diversity and dominance.
competitive exclusion (Peterson 1979; Wilson 1991;
Woodin 1999). Although some effects of predation have
               ´
been demonstrated (see Olafsson et al. 1994; Lenihan
and Micheli 2001), numerous experimental studies have    Materials and methods
found no consistent regulatory role (Thrush 1999).
Explanations for the absence of a clear effect include    Study area
prey mobility and exchange that mask predation losses
(Frid 1989; Englund 1997; Cooper et al. 1990), prey     Bonne Bay fjord is located in Western Newfoundland
recruitment outpacing post-settlement consumption      (Fig. 1) and is comprised of two main arms. East Arm
(Thrush 1999), and indirect interactions counterbalanc-   is a deep (up to 230 m) inner basin that is partly sep-
ing negative effects of epibenthic predators (Commito     arated from the outer bay by a shallow sill ($12-deep),
and Ambrose 1985; Kneib 1991). Variation in predator     whereas South Arm is a shallower basin (up to 55-m
density, mobility, and feeding rates also hinder our     deep) that is fully open to the adjacent Gulf of
capacity to detect predation effects (Clark et al. 1999;   St. Lawrence. Study sites for crab abundance estimates
Seitz et al. 2001).                     and experiments were established in each of these main
  Detection of predation is challenging. Field manipu-
lations have significant limitations (Hulberg and Oliver
1980; Peterson and Black 1994) but remain the best tool
for testing predator effects (Hall et al. 1990). Nonethe-
less, cage experiments alone may not suffice if they are
restricted to a single site (Fernandes et al. 1999) or are
not combined with surveys and/or other types of
manipulations (Thrush et al. 1997 and references there-
in). Combined field and laboratory experiments have
proven to be among the most informative experimental
approach because they examine different scales, have
different strengths, and may potentially complement
each other (Wiens 2001). We used this combined ap-
proach to study the role of predation in Bonne Bay, a
Newfoundland sub-Arctic fjord in the northwest
Atlantic. Preliminary observations from inner and outer
areas of the bay indicated strong differences in benthic
community structure, and in the number and composi-
tion of epibenthic predators (Wieczorek and Hooper
1995). The fact that infaunal organisms constitute the
main part of crab and shrimp diets (Squires and Dawe
2003; Scarrat and Lowe 1972; Bergstrom 2000) suggests
                    ¨
that increased predation pressure from these species may
play a key regulatory role for benthic community
structure.
  We tested this hypothesis by deploying cage exclusion
experiments and an inclusion experiment in the two
main arms of the fjord, and by using sediments (with
intact infauna) from these sites to carry out parallel
laboratory predation experiments. Bonne Bay also offers    Fig. 1 Map of Bonne Bay, with the location of a South and b East
                               Arms where predator sampling and manipulative experiments were
a unique opportunity to study these interactions because
                               conducted. Lower panels indicate mean summer abundances
an abundant guild of crabs and shrimps, which typically   (±95% confidence intervals) of the most abundant epibenthic
occur at greater depths, congregate in sedimentary      predators measured in baited traps during 1999–2001. SN Snow
habitats that are accessible by divers during the spring–  crab, SH Shrimp, TO Toad crab, RO Rock crab
                                                           127

                                sediment-related artifacts associated with caging treat-
arms. Currents and circulation in some areas of the bay
                                ments. Logistical constraints precluded sediment sam-
have been described by Gilbert and Pettigrew (1993).
                                pling after 4 weeks, although any artifact effects would
Detailed studies of benthic communities are lacking
                                be expected to be much stronger after 8 weeks than after
except for lists of invertebrates (Rivard and Bowen
                                4 weeks.
1971; Hooper 1975), and selected communities (Wiec-
sorek and Hooper 1995).

                                Laboratory experiments
Predator distribution
                                Two laboratory experiments were conducted in June
                                2000 to evaluate the potential impact of snow crab
Relative abundances of epibenthic predators were
estimated with traps ($40·30·60 cm, $5·15 cm open-       predation on benthic organisms under controlled con-
ing, $1 cm net) that were baited with mackerel and       ditions. A series of flow-through tanks (1–2°C) at the
                                Bonne Bay Field Station were supplied with cold water
deployed during the summer seasons of 1999–2001.
                                pumped from depths where cores were collected.
Traps were deployed at 35- to 50-m depth, separated
by $50–80 m and kept at the bottom for 1–2 days        Within these tanks, freshly collected sediment cores
(data standardized as crab trapÀ1 dayÀ1) every 2–       with intact infauna were exposed to snow crab feeding.
                                Sediment cores (7-cm diameter) were obtained from
3 weeks. Direct comparison of catch numbers and
                                each experimental site (South and East Arms) by divers
frequencies was not possible because deployments were
                                who gently pushed corers into the sediment to avoid
not simultaneous and catch rates were highly variable
                                physical disturbance, sealed them with rubber corks,
(within and among traps, sites, and summers). Instead,
                                and brought them to the surface where they were
summer averages were calculated by using average
                                transported to the laboratory in coolers to minimize
standardized daily catches per sampling period as rep-
                                disturbance of infauna. This protocol maximized the
licates. Baited traps do not provide absolute density
                                likelihood that initial core communities would be rep-
estimates, and this approach yields only relative density
                                resentative of nature. Six sediment cores were placed in
comparisons between the two sites.
                                each tank and a plastic plate was used to create a false
                                bottom so that the plastic core tube was flush with the
                                plate. Sediment inside the cores was gently extruded so
Field-exclusion experiments
                                that it was also flush with the acrylic plate, creating a
Two exclusion experiments were deployed at $30-m        smooth transition between sediments, core tube, and
                                plastic plate. Sediments (and infauna) were acclimated
depth in South and East Arms (Fig. 1). Each experiment
                                to these conditions for 24 h prior to initiation of
included three treatments and four replicates that were
                                experiments. Male snow crabs of 60–75 mm CW, a
haphazardly interspersed; treatments included full cages
                                range including immature, adolescent, and small adult
or ‘‘exclusions’’, partial cages or ‘‘artifact treatments’’,
                                snow crab (cf. Sainte-Marie et al. 1995) were consid-
and ambient undisturbed sediments or ‘‘controls’’. Ca-
ges (1-m diameter · 15 cm high, pushed 3 cm into sed-     ered representative of the size structure reported for the
                                                         ´
                                depth and location of the study area (P.A. Quijon and
iments) were circular in shape to minimize erosion/
                                P.V.R. Snelgrove, unpublished; Hooper 1996; Comeau
deposition of sediments in different areas of the cages.
Cages were anchored to the bottom by four $4-cm-long      et al. 1998). One of these snow crabs was added to each
                                tank and offered open access to three randomly se-
‘‘legs’’ extended from the main frame into the sediment.
                                lected cores (controls). The other three cores in the
Plastic 1 cm·1 cm mesh covered partial (50% of top and
                                tank were protected with horizontal plastic mesh, thus
side) and full cages. Infaunal organisms were sampled
                                excluding the predators. Experiments lasted for 96 h,
with tube cores (7-cm diameter; 10-cm deep; two cores
                                after which the snow crabs were removed, tanks were
per sample) that were collected by scuba divers. Initial
                                carefully drained, and sediment cores were collected
sampling (two groups of four samples at each of the two
                                and processed (see below).
study locales) took place on 25th June 1999, immediately
prior to deployment of full and partial cages. These
samples were used for comparison with ambient sedi-
                                Inclusion experiment
ments and cages sampled after 4 and 8 weeks (see BACI
design below). Sampling was never repeated within a
                                In order to provide a linkage between field-exclusion
given caged or ambient location, because cages were
                                experiments and laboratory manipulations, full cages
removed immediately after sampling. This approach
                                similar to those used for exclusion experiments were
minimized potential disturbance effects and created
                                used to confine snow crabs (one crab per cage; same
statistical independence in evaluating predation after 4
                                CW range reported above) for 96 h. Crabs were then
and 8 weeks. Coincident with the 8-week samples,
                                released and samples were collected from cages and
additional sediment cores were collected from all treat-
                                ambient sediments as described above for exclusion
ments in order to study grain size distribution and CHN
                                experiments. Inclusion experiments were initiated in
content. These analyses allowed us to evaluate potential
128

                                Four community response variables were calculated:
both arms of the bay during June 1999, but weather
                              total density and number of species P sample (77 cm2),
                                                  per
constraints made recovery of samples from South Arm
impossible. Thus, only results from East Arm are re-    Shannon–Wiener Diversity (H¢=À piloge(pi); with pi
ported.                           density of i species/total density), and Evenness (J¢=H¢/
                              H¢max; with H¢max=logeS). Selection of indices was
                              based on their widespread use in the literature (H¢),
Sample processing and analysis               sensitivity to rare species and independence from species
                              richness (J¢), and discriminant ability (H¢) (Magurran
Cores of sediments from field and laboratory experi-     1988; Smith and Wilson 1996). Statistical comparisons
ments were processed through a 500-lm sieve and pre-    were all carried out with ANOVAs in SPSS (version 10).
served in a 10% sea water–formalin solution, prior to    For the field-exclusion experiments, a ‘‘before–after,
transfer to 70% ethanol with Rose Bengal to facilitate   control-impact’’ (BACI) design was used. In this facto-
sorting and identification. Macrofaunal organisms were    rial design, the evidence for an impact (predation effect)
enumerated and identified to the lowest taxonomic level   appears as a significant time by treatment interaction
possible, which was usually species. Samples for grain   (Green 1979). The model for this ANOVA was
                              y = l + time + treatment + time · treatment + ,
size analysis were pre-treated with a 1:1 water:peroxide
solution and heated to 300°C to remove organic matter.   where y refers to each response variable, l is a mean
They were then disaggregated by re-suspension with     constant, time refers to the ‘‘before–after’’ comparison
0.1% Calgon solution, and passed through sieves to     (0–4 week or 0–8 week), treatment refers to the ‘‘impact’’
separate fractions of >350, >250, >177, >125, >88,     comparison (control versus predator exclusion), and 
and >62.5 lm by wet sieving. Finer fractions were sub-   refers to the error term. Because artifact treatments were
sampled (50 ml) and analyzed with a Sedigraph 5100     available only for the 8-week period, artifact data were
Particle Size Analyzer. Based on grain settling velocity,  analyzed separately using the model y = l +
                              site + treatment + site · treatment + . In this model,
the Sedigraph separated >53, >44, >37, >31, >15,
>7.8, >3.9, >2.0, >0.98, and >0.49 lm fractions.      site is South or East Arm, treatment is control or artifact,
Each fraction was then expressed as percentage of total   and  is the error term. The model for the laboratory
dry weight, and pooled into categories based on the     experiments was y = l + tank + treatment + ,
Wentworth scale (Folk 1980): fine + very fine sand      where tank refers to replicate tanks 1–3, and treatment
(>62.5 lm), silt (>3.9 lm), and clay (<3.9 lm).       refers to control (exposed to crab predation) versus
Additional sediment samples were processed with a      exclusion, with no interaction term. The model for the
CHN analyzer (Perkin Elmer Model 2400) to estimate C    inclusion experiment was y = l + treatment + ,
and N as a function of sediment dry weight. C:N ratios   where treatment refers to crab inclusion versus ambient
(an estimator of food quality for deposit feeders;     sediments. All variables, with the exception of ‘‘tank’’
Blackburn et al. 1996) were also calculated.        (laboratory experiments) were treated as fixed factors.
                              Assumptions of normality and heterogeneity were tested
                              in each analysis by plotting residual histograms and
Data analysis                        applying Levene’s test, respectively. Application of loge
                              transformation proved sufficient to homogenize vari-
Patterns in benthic community structure were studied    ances in those instances where data transformation was
using Chord Normalized Expected Species Shared       necessary (Sokal and Rohlf 1994).
(CNESS). This similarity index estimates the number of
species shared between two samples based on a random
                              Results
draw of m=10 individuals (cf. Trueblood et al. 1994)
that makes the index sensitive enough to detect the
                              Predator abundance
contribution of rare as well as abundant species (Grassle
and Smith 1976). The CNESS dissimilarity sam-
ple · species matrix was also used to cluster samples    Four species of decapods dominated average summer
                              abundances of epibenthic predators (Fig. 1). Snow crabs
based on un-weighted pair-group mean average sorting.
                              (Chionoecetes opilio, South Arm mean=0.96 crabs
The program COMPAH 90 (E.D. Gallagher, U. Mas-
                              trapÀ1 dayÀ1) and pandalid shrimp (Pandalus montagui,
sachusetts, Boston) was used for this analysis. The
                              East Arm mean=0.85 shrimp trapÀ1 dayÀ1) dominated
CNESS sample by species matrix was then transformed
                              the two study sites respectively. Snow crabs were almost
to a normalized hypergeometric probability matrix (H),
                              one-fifth as abundant in East Arm (0.21 crab
which was used in a principal components analysis of
                              trapÀ1 dayÀ1), whereas shrimp were absent from South
hypergeometric probabilities (hereafter called PCA-H)
                              Arm. Toad crabs (Hyas sp.) were less abundant but
to produce a two-dimensional metric scaling of CNESS
                              similar in density between sites (0.15 and 0.10 crabs
distances among samples. Gabriel Euclidean Distance
                              trapÀ1 dayÀ1). Rock crabs (Cancer irroratus) abun-
Biplots (Gabriel 1971) identified the species that were
                              dances were 0.08 and 0.30 crab trapÀ1 dayÀ1 at South
most important for among-sample variation, and thus,
                              and East Arms, respectively.
driving community composition differences.
                                                                  129

                                   treatments in East Arm. The polychaete Paradoneis lyra
Ambient communities and predator exclusion
                                   was important in describing ambient and control sedi-
experiments
                                   ments in South Arm, whereas the cumacean, Lamphros
                                   fuscata, and the amphipod Bathymedon obtusifrons were
Overall, abundances of benthic invertebrates in ambient
                                   important in ambient sediments and partial cages in East
sediments from South Arm were significantly higher
                                   Arm. Comparisons of species densities (Fig. 4) were
than in East Arm (P<0.05; Fig. 2). The three most
                                   consistent with the interpretation based on biplots
abundant species from South Arm (the clam Astarte sp.
                                   (Fig. 3). For example, P. tecta was abundant in exclu-
and the polychaetes Paradoneis lyra and Prionospio
                                   sion treatments, whereas L. fuscata was more abundant
steenstrupii) were all significantly more abundant than in
                                   in controls (P<0.001). Densities of O. cylindricaudata
East Arm (P<0.05) for each time period. The cumacean
                                   and P. lyra were also consistent with the biplots, though
Lamphros fuscata was consistently more abundant at
                                   differences were not significant.
East Arm than in South Arm (P<0.05); however, the
                                     Predation effects (i.e. significant time · treatment
two next most abundant species from East Arm (the
                                   interactions) on density and evenness were detected after
bivalve Thyasira flexuosa and the amphipod Bathyme-
                                   4 and 8 weeks in South Arm (Table 1). Similar effects
don obstusifrons) were generally not significantly differ-
                                   were detected on density, number of species, and diver-
ent from corresponding densities in South Arm (Fig. 2).
                                   sity after 4 week at East Arm, but these effects did not
  Exclusion experiments carried out in both arms of the
                                   persist to the 8th week (Table 1). A control–exclusion
bay are summarized in Fig. 3. Together, the first two
                                   comparison at each sampling date (Fig. 5) indicates that
principal components of the analysis explained 44% of
                                   the exclusion of predators increased the density and re-
the data variation. As was apparent in the clustering
                                   duced evenness (South Arm), whereas species richness
analysis, the PCA-H clearly separated South from East
                                   and Shannon diversity were not significantly affected. In
Arm communities (PCA1), and predator exclusions
                                   East Arm, exclusion of predators for 4 weeks signifi-
from ambient and partial cages treatments (PCA2). At
                                   cantly increased the density, species richness and Shan-
both sites, sampling period (4th vs. 8th week) had no
                                   non diversity but did not affect evenness (Fig. 5).
clear effect on patterns in the PCA-H plot. Gabriel bi-
                                   Sedimentary and faunal response variables were used to
plots identified two polychaetes, Pho¨loe tecta and Prio-
                                   test for potential artifacts (Table 2). In all cases, site was
nospio steenstrupi, as particularly important in exclusion
                                   the only significant factor, indicating no measurable
sediments in South Arm (Fig. 3). Three other poly-
                                   caging artifacts on sediment composition or community
chaetes, Ophelina cylindricaudata, Euchone papillosa,
                                   structure. These results coincide with diver observations
and Praxillella praetermissa, were important in exclusion




                                   Fig. 3 Cluster and metric scaling plot of treatments and ambient
Fig. 2 Mean total densities and most abundant infaunal taxa      samples using PCA-H of CNESS similarities (NESSm=10). South
(±95% confidence intervals) in ambient (control) sediments from    Arm (upper case) and East Arm (lower case) treatments are
South (open bars) and East Arms (shaded bars) at a 0-week, b 4-    indicated as follows: C, c control, E, e exclusion, A, a artifact.
week, and c 8-week periods in the field experiments. Asta Astarte   Numbers indicate sampling periods (0, 4, or 8 weeks) and subscript
sp., Para Paradoneis lyra, Prio Prionospio steenstrupi, Thya     numbers denote replicates (1–4). Vectors represent Gabriel biplots
                                   that identify species that explain for the most variability among
Thyasira flexuosa, Lamp Lamphros fuscata, Bath Bathymedon
obstusifrons. Asterisks indicate significant difference between South  samples. Dashed circles indicate samples forming subgroups into
and East Arms (*P<0.05, **P<0.01, ***P<0.001)             the groups represented by solid lines
130




Fig. 4 Mean densities (±95% confidence intervals) of four species
that explain most of the between-sample variation between controls
(ambient) (open bars) and exclusion treatments (shaded bars) in
Fig. 3. a P. tecta, b O. cylindricaudata, c P. lyra, and d L. fuscata.
Asterisks indicate significant differences between treatments
(***P<0.001)


at the study sites, which indicated that predators did
enter the partial cages.


Laboratory and inclusion experiments                   Fig. 5 Mean values (±95% confidence intervals) for density (a, b),
                                     species richness (c, d), diversity or H¢ (e, f), and evenness or J¢ (g, h)
                                     estimated from control (open bars) and exclusion (shaded bars)
The use of snow crab as a predator in laboratory
                                     treatments. Mean values are based on four replicates except at the
experiments yielded similar results to those observed in
                                     beginning of the experiments (week 0; n=8) when two groups of
the field experiments (Fig. 6). The first two principal          four samples were averaged and plotted as a single open bar.
components of the laboratory experiments explained            Asterisks indicate significant differences between treatments at each
                                     period. *P<0.05; **P<0.01
50% and 45% of the variation in South and East Arm,
respectively. Irrespective of the source of the sediments
(South or East Arms), cores exposed to predators were          chaetes O. cylindricaudata and E. papillosa were impor-
distinct from predator-exclusion treatments (Fig. 6, top         tant in exclusions for East Arm. Mediomastus ambiseta
and middle panels). The polychaete P. tecta and the           and E. papillosa (South Arm) and Aricidea nolani (East
bivalve Macoma calcarea were important in describing           Arm) were important to control treatments. In the field-
exclusion treatments for South Arm, whereas the poly-          inclusion experiment (Fig. 6, bottom panel), the first two

Table 1 Predation effects on community response variables

            Source         df       N            S            H¢           J¢

South Arm        Time          1       885.06**        2.25          0.0728         0.0133**
0–4 week        Treatment        1       95.06          0.25          0.0169         0.0026
            Interaction       1       1040.06**        2.25          0.0748         0.0048*
            Error          12       946.25         65.00          0.2842         0.0088
South Arm        Time          1       1139.06**        7.56          0.1561         0.0302**
0–8 week        Treatment        1       175.56         7.56          0.0184         0.0074*
            Interaction       1       1278.06**        0.56          0.0779         0.0108*
            Error          12       1147.75         86.25          0.4970         0.0165
East Arm        Time          1       10.56          203.06***        2.9451***        0.1036***
0–4 week        Treatment        1       126.56         14.06          0.0092         0.0008
            Interaction       1       351.56**        45.56**         0.2424**        0.0015
            Error          12       444.25         48.25          0.2659         0.0235
East Arm        Time          1       162.56         156.25***        2.3846***        0.0853***
0–8 week        Treatment        1       0.56          1.00          0.0305         0.0015
            Interaction       1       45.56          4.00          0.0491         0.0007
            Error          12       738.75         64.50          0.2093         0.0175

Values are sum of squares (SS) from two-way ANOVAs (BACI         *P<0.05, **P<0.01,
design, see text). Factors include time (before–after; 0–4 and 0–    ***P<0.001
8 week), treatment (control–exclusion) and their interaction.
Asterisks indicate significance associated with each SS.
                                                                  131

Table 2 Artifact effects on sedimentary and community response variables

Sedimentary variables        df        Fine sand         Silt          Clay          C:N

Site                 1         406.51***         350.43*         302.71*        84.08*
Treatment              1         2.73           116.97         9.89          0.84
Site · treatment           1         1.06           19.35          19.48         1.44
Error                12        142.3           471.24         99.83         18.32

Community variables          df         N            S           H           J¢

Site                 1         1444.0**         5.06          0.008         0.007**
Treatment               1         1.0           7.56          0.074         0.002
Site · treatment                                                        $0.000
                   1         4.0           10.56         0.014
Error                 12         1,376          158.75         0.445         0.005

Values are sums of squares (SS) from two-way ANOVAs. Factors include site (South vs. East), treatment (Control vs. Artifact), and their
interaction.
*P<0.05, **P<0.01; ***P<0.001

components explained 54% of the variation, and clearly        in our experiments (Thrush et al. 1997; Schneider
separated inclusion from ambient sediments. E. papill-        2001).
osa,Yoldia sp., and Tharyx acutus were important spe-
cies in the inclusion treatment, whereas Lamphros
fuscata was the most important species in ambient sed-        Predator abundance
iments. In general, densities of most of the representative
species identified in Fig. 6 were significantly different        Though noisy, our baited trap data suggest differences in
between treatments (Fig. 7).                     epifaunal predator abundance between East and South
  In terms of community variables, results from the         Arm. Predator numbers may differ because of produc-
laboratory and the inclusion experiments were similar to       tivity differences; South Arm is considered to be more
those in exclusion experiments. In general, site (South or      productive than East Arm (R. Hooper, Memorial Uni-
East Arm) explained most of the significant differences        versity Personal Communication) Recruitment may also
in variables (P<0.05 for all variables, Table 3); how-        play a role. The sill that limits exchange with East Arm
ever, treatment (predator exclusion vs. exposed) also had      also limits larval transport, which may contribute to
significant effects on density and evenness (P<0.05).         fewer snow crab and more shrimp recruiting in East
                                           ´
Because site effects were significant, data were re-ana-        Arm (P.A. Quijon, P.V.R. Snelgrove, submitted).
lyzed separately for each site. For South Arm, snow
crabs significantly reduced density (N), and increased
evenness (J¢) (P<0.05), but did not affect species rich-       Predation effects on composition
ness or Shannon diversity. For East Arm, snow crabs
reduced total density and increased species richness and       Two groups of species were expected to benefit most
Shannon diversity (H¢) (P<0.05), but had no effect on         from the exclusion of predators: sedentary polychaetes
evenness (P>0.05) (Fig. 8). The results of the inclusion       or clams unable to escape by emigration or burial
experiment were very similar to the laboratory experi-        (Roberts et al. 1989), and infaunal predatory species
ment: confined snow crabs reduced significantly the          (Commito and Ambrose 1985). In our experiments,
number of species and diversity (P<0.05), but did not        sedentary polychaetes such as the maldanid P. prae-
significantly reduce the total density, or modify evenness      termissa, the sabellid Euchone papillosa, and the amp-
(P>0.05).                              heretid Lyssipe labiata, were nearly twice as abundant in
                                   exclusion treatments than in ambient sediments in East
                                   Arm. Similarly, Mediomastus ambiseta was twice as
Discussion                              abundant in exclusion than in ambient sediments in
                                   South Arm. The clams Yoldia sp. and Macoma calcarea
                                   also benefited from the refuge created by exclusion
Overall, our results indicate that predation significantly
                                   treatments. Yoldia sp. was two times and M. calcarea
contributes to patterns of infaunal composition and
                                   five times more abundant in East and South Arm
abundance in Bonne Bay. This conclusion is based on
                                   exclusion treatments, respectively. These results are
laboratory and field experiments that were consistent in
                                   consistent with previous studies on predator diet.
their findings despite their obvious differences in scale
                                   Stomach content analyses have shown that clams and
(Kemp et al. 2001; Wiens 2001). Among-site differences
                                   sedentary polychaetes are important dietary components
reflect spatial variation that cannot be fully understood
                                   of snow crab populations from Bonne Bay (Wieczorek
with manipulative experiments that are limited to a
                                   and Hooper 1995), Gulf of St. Lawrence (Powles 1968),
single site (Fernandes et al. 1999) and exemplify the
                                   and Eastern Newfoundland (Squires and Dawe 2003).
need for including more than one spatial/temporal scale
132

  Pho¨loe tecta is a member of a predatory guild that is      cumacean Lamphros fuscata, the amphipod Bathymedon
believed to generate trophic complexity in soft-sediment      obtusifrons, and the polychaete Paradoneis lyra, are all
communities (Ambrose 1984; Commito and Ambrose           highly mobile species that were indeed more abundant in
1985; Posey and Hines 1991). Predatory infauna are         ambient sediments than in exclusion treatments. Two
expected to aggregate in exclusion treatments to take        notable exceptions were the clam Astarte sp., which is
advantage not only of the refuge from top predators but       characterized by a very robust shell, and the polychaete
also the enhanced infaunal prey beneath cages (Kneib        Ophelina cylindricaudata; neither species differed signif-
1988, 1991). In South Arm P. tecta was five times more        icantly between ambient and exclusion treatments.
abundant in exclusion treatments than in ambient sedi-       However, there is also no evidence to indicate that these
ments. Similarly, Phyllodoce mucosa, the only other         species are important in the diets of snow crab (Lefebvre
abundant predatory species (>1% of total) was $twice        and Brethes 1991), rock crab (Hudon and Lamarche
                                      ˆ
more abundant in exclusion treatments than in ambient        1989), pandalid shrimp (Bergstrom 2000), or toad crab
                                                    ¨
sediments. Species that were able to escape crab preda-       (Squires 1996).
tion were expected to dominate ambient sediments. The

                                  Predation effects on community variables

                                  The exclusion of predators produced an increase in
                                  total abundance in both sites over 4 weeks but the in-
                                  crease persisted through 8 weeks only in South Arm.
                                  Predation effects are ‘‘strong’’ when a 100% density
                                  increase is detected in exclusion versus ambient sedi-
                                       ´
                                  ments (Olafsson et al. 1994). This strong an effect is
                                  clearly not the case in Bonne Bay, where field and
                                  laboratory experiments show that the predation influ-
                                  ence is moderate and varies among sites. Spatial dif-
                                  ferences in predation influence and persistence may be
                                  related to predator foraging rates (Micheli 1997; Seitz
                                                          ´
                                  et al. 2001) and predator composition (Quijon and
                                  Snelgrove 2005). On the one hand, snow crabs were
                                  nearly five times more abundant in South Arm, sug-
                                  gesting that their foraging in this area may be much
                                  more frequent than in East Arm. On the other hand,
                                  predation effects on species richness that were detected
                                  only in East Arm may be related to higher density of
                                  rock crabs relative to South Arm. In laboratory con-




Fig. 6 Cluster and metric scaling plot of samples collected in
laboratory snow crab feeding experiments carried out with
sediments (communities) from a South and b East Arms, and from
a c field-inclusion experiment carried out in East Arm (see text).
NESSm=10 except in the B (NESSm=5). Vectors represent        Fig. 7 Mean densities (±95% confidence intervals) of a P. tecta,
Gabriel biplots that identify species that explain the most     b O. cylindricaudata, c L. fuscata, d E. papillosa, e A. nolani, and
variability among samples. Treatments are represented by letters  f E. papillosa. These species are among the ones that explained most
(see Fig. 3; with the addition of i=crab inclusion), whereas    of the between-sample variation between treatments with
numbers refer to tanks (1–3) and subscript numbers to replicates  (‘‘pres’’=present) and without (‘‘abs’’=absent) crabs in laboratory
(1–3)                                and field-inclusion experiments (see Fig. 6)
                                                                  133

ditions, rock crabs were at least four times more
effective than snow crabs in reducing species richness
    ´
(Quijon and Snelgrove 2005). These differences are
consistent with feeding rates reported for both species
(Himmelman and Steele 1971; Drummond-Davis et al.
1982; Thompson and Hawryluk 1989) and with labo-
ratory observations that suggest higher rates of sedi-
ment alteration by rock crab. Thus, although snow
crab had significant effects in both sites, the effects on
individual species reflected differences between com-
munities, which in turn may reflect the complex influ-
ence of multiple epifaunal predators that vary between
sites.
  The influence of rock crab on species richness also
explains differences in diversity (H¢), but not necessarily
in evenness and dominance. Predation may indirectly
increase the evenness when predators are non-selective
foragers, i.e., when they primarily target the most
abundant prey (Schneider 1978). This seems to be the
case in South Arm, where the reduction in density by
predation tends to equalize numbers per species (both in
the field and in the laboratory). Moreover, disturbance
per se, in addition to predation, can have significant
consequences for sedimentary infauna (Virnstein 1977)
through non-selective mortality. Most of the literature
suggests that our four predators are primarily generalists
(Squires and Dawe 2003; Bergstrom 2000; Scarrat and
                  ¨
Lowe 1972), despite some degree of prey selectivity by
snow crab (Wiecsorek and Hooper 1995). In East Arm,
                                   Fig. 8 Mean values (±95% confidence intervals) for density,
the reduction of density by predation (in field and lab-
                                   species richness, diversity (H¢), and evenness (J¢) for treatments
oratory experiments) resulted in the loss of species         with (‘‘pres’’=present) and without (‘‘abs’’=absent) crabs in
without changes in evenness. This pattern may suggest        laboratory and field-inclusion experiments (see Fig. 6). Asterisks
that equalization of individuals among species (Schnei-       indicate significant differences between treatments in two way
                                   ANOVAs (*P<0.05)
der 1978) is more likely in communities where abun-
dance and species richness are comparatively high as is
                                   evaluate and minimize caging effects. The round shape
the case in South Arm.
                                   of the cages effectively eliminated variable deposition
                                   within the cage interior because no visual evidence of
                                   sediment erosion or deposition was detected, nor were
Artifact effects
                                   significant changes in sediment parameters observed.
                                   Although separate analysis of East Arm data indicated
Cage artifacts are a recurrent concern in predation
     ´                             an increase in silt content in the cages, this effect was
studies (Olafsson et al. 1994; McGuinness 1997). It is
                                   probably not meaningful for overall sediment quality;
impossible to completely eliminate cage influences on
                                   no other grain size fraction changed significantly, nor
sediments, prey, or predators, but it is possible to


Table 3 Snow crab predation effects on community response variables in the laboratory and in the inclusion experiment

             Source        df      N          S           H          J

Laboratory        Site         1       41,877***      1080.21***       3.218***      0.192***
             Tank         2       399         1.47          0.255        0.021
             Treatment       1       1,039*        15.05         0.004        0.031**
             Error         30      4,577        124.61         2.065        0.099
                                                               $0.000
Field inclusion      Treatment       1       84.5         24.5*         0.141*
             Error         6       147.5        15.5          0.146        0.008

Values are sums of squares (SS) from three-way and one-way ANOVAs, respectively. In the laboratory experiments, factors include site
(South and East), tank, and treatment (exposed to crab vs. exclusion). In the inclusion experiment treatment refers to inclusion (crab)
versus ambient sediments.
Asterisks indicate significant effects associated with each SS.
*P<0.05; **P<0.01; ***P<0.001
134

                               (fishery) that created them. There are exceptions (Barkai
did the C:N ratio, our closest surrogate of food quality
                               and Branch 1988), however, that may include scenarios
for deposit feeders (Blackburn et al. 1996) in the ab-
                               such as the Newfoundland ecosystem where cod have
sence of chlorophyll data. More importantly, no com-
                               failed to recover even 10 years after a fishing morato-
munity responses to partial cages were detected. We
                               rium was declared. Irrespective of whether or not an
were unable to test for artifact effects during the first
                               ‘alternate state’ applies to the Newfoundland ecosystem,
half of the experiment (0–4 weeks), but caging effects
                               it is clear that the consequences of cod collapse have
tend to be cumulative over time (Hindell et al. 2001)
                               been far more severe than anticipated and, as our results
and, if present, should therefore have been apparent in
                               indicate, may have been paralleled by a fundamental
partial cages after 8 weeks.
                               change in the structure of benthic communities.
                                 Ironically, fishing pressure now focuses on three of
                               the four crab predators studied here. The exploitation of
Implications for marine conservation
                               rock crab (Mallet and Landsburg 1996), and at a much
                               larger scale, snow crab (Paul et al. 2002), and pandalid
The collapse of cod and other major predators on large
                               shrimp (Bergstrom 2000), grew partly as a consequence
decapods, that were once extraordinarily abundant in              ¨
                               of the cod collapse and subsequent moratorium on cod
coastal Newfoundland, has contributed to an explosion
                               fishing (Bundy 2001; Schiermeier 2002). Our results
in shrimp, snow crab (Worm and Myers 2003) and
                               indicate a clear influence of these predators on key as-
presumably, other crustaceans. Previous studies have
                               pects of the structure of benthic communities. It follows
documented spatial differences in predation pressure on
                               that the decimation of these predators will have indirect
epifaunal taxa in the Gulf of Maine, and historical
                               consequences on the bottom component of the ecosys-
declines in cod abundance were hypothesized to have
                               tems that they currently structure. Cascading effects, as a
resulted in long-term changes in predation impact
                               result of fishery exerted at the top of the trophic web
(Witman and Sebens 1992). Given that adult cod are
                               (Agardi 2000), have been proposed for systems domi-
not primarily an infaunal predator, and instead, ado-
                               nated by fish predators. Similar cascading effects may be
lescent and adult snow crabs display clear effects on
                               playing a role in benthic communities of the North
benthic infauna, it is reasonable to expect that the
                               Atlantic, although this remains largely unknown to date.
structure of Newfoundland infaunal communities may
                               If overfishing leads to the collapse of crab stocks, as
have changed in the last few decades with the
                               some data are beginning to suggest (Bundy 2001),
replacement of cod by a trophic guild that feeds pri-
                               additional shifts in sedimentary communities may be
marily on infauna. Admittedly, in the absence of his-
                               expected.
torical data on infaunal composition and structure, we
must infer that the predator-mediated changes we
                               Acknowledgements We thank D. Schneider, R. Haedrich, and three
observe in short-term experiments are reflective of
                               anonymous reviewers for comments that significantly improved
long-term changes related to increased crab effects. We
                               earlier versions of this manuscript. R. Hooper, M. Kelly,
believe this inference is reasonable given that the recent  M. Norris, M. Parsons, K. Carter, J. Barry, and multiple students
increases in snow crab abundance in Newfoundland       at the Bonne Bay Field Station provided field assistance. K. Gil-
waters are well documented (Worm and Myers 2000)       kinson, D. Steele, and P. Ramey assisted in the identification of
                               bivalves, amphipods, and polychaetes, respectively. Funding was
and the results of our different experiments are con-
                               provided by a Discovery Grant from the Natural Sciences and
sistent and unambiguous. Small crabs manipulated here    Engineering Council of Canada to PVRS, and fellowships from the
are representative of the study area and the region for   Fisheries Conservation Chair, Memorial University’s School of
depths and the spring–summer season (Comeau et al.      Graduate Studies, and the Canadian International Development
                               Agency (CIDA) to PAQ.
1998) but may not necessarily represent those popula-
tions living in deeper waters. Therefore, the extrapo-
lation of our experiments should be done with caution
until similar manipulations with larger crabs at deeper   References
bottoms can be done.
  Our results indicate that crabs modify benthic com-   Agardi T (2000) Effects of fisheries on marine ecosystems: a con-
position and abundance, and in more depauperate sed-      servationist’s perspective ICES. J Mar Sci 57:761–765
                               Ambrose WG Jr (1984) Role of predatory infauna in structuring
imentary habitats, also reduce diversity. As in other
                                marine soft-bottom communities. Mar Ecol Progr Ser 17:109–
systems affected by long-term overfishing (Goni 1998;
                                115
Jackson et al. 2001), the elimination of cod may have    Barkai A, Branch GM (1988) Contrast between the benthic com-
established an alternate stable state. The existence of     munities of subtidal hard substrata at Marcus and Malgas Is-
these states has been debated for decades (Connell and     lands: a case of alternative stable states. S Afr J Mar Sci 7:117–137
                               Bergstrom BI (2000) Biology of Pandalus. Adv Mar Biol 38:57–256
                                   ¨
Sousa 1983; Peterson 1984; Petraitis and Latham 1999;
                                                       ´
                               Blackburn TH, Hall POJ, Hulth S, Landen A (1996) Organic-N
van de Koppel et al. 2001), though rarely have they been    loss by efflux and burial associated with a low efflux of inor-
described to persist once overfishing has halted. Alter-     ganic N and with nitrate assimilation in Arctic sediments
nate systems in fisheries ecosystems have often been       (Svalbard, Norway). Mar Ecol Progr Ser 141:283–293
                               Botsford LW, Castilla JC, Peterson CH (1997) The management of
assumed to be unstable, in that they revert back to the
                                fisheries and marine ecosystems. Science 277:509–515
previous state with the cessation of the disturbance
                                                                     135

Brethes J-C, Desrosiers G, Coulombe F (1984) Aspects de l’ ali-
   ˆ                                   Himmelman JH, Steele DH (1971) Foods and predators of the
    mentation et du comportement alimentaire du crabe-des-nei-       green sea urchin Strongylocentrotus droebachiensis in New-
    ges, Chionoecetes opilio (O. Fabr.) dans le sud-ouest du Golfe     foundland waters. Mar Biol 9:315–322
    de St-Laurent (Decapoda, Brachyura). Crustaceana 47:235–      Hindell SJ, Jenkins GP, Keough MJ (2001) Spatial and temporal
    244                                   variability in the effects of fish predation on macrofauna in
Brethes J-C, Coulombe F, Lafleur F-E, Bouchard R (1987) Habitat
  ˆ                                      relation to habitat complexity and cage effects. Mar Ecol Progr
    and spatial distribution of early benthic stages of the snow crab    Ser 224:231–250
    C. opilio O. Fabricius off the north shore of the Gulf of St.    Hooper RG (1975) Bonne Bay marine resources. An ecological and
    Lawrence. J Crust Biol 7:667–681                    biological assessment. Parks Canada report. Contract No.
Bundy A (2001) Fishing on ecosystems: the interplay of fishing and       ARO 74-83, p 295
    predation in Newfoundland-Labrador. Can J Fish Aquat Sci      Hooper RG (1996) A spring breeding migration of the snow crab,
    58:1153–1167                              Chionoecetes opilio (O Fabr.) into shallow water in New-
Clark ME, Wolcott TG, Wolcott DL, Hines AH (1999) Intraspe-          foundland. Crustaceana 50:257–264
    cific interference among foraging blue crabs Callinectes sapidus:  Hudon C, Lamarche G (1989) Niche segregation between Ameri-
    interactive effects of predator density and prey patch distribu-     can lobster Homarus americanus and rock crab Cancer irrora-
    tion. Mar Ecol Progr Ser 178:69–78                   tus. Mar Ecol Progr Ser 52:155–168
Comeau M, Conan GY, Maynou F, Robichaud G, Therriault J-          Hulberg LW, Oliver JS (1980) Caging manipulations in marine
    C, Starr M (1998) Growth, spatial distribution, and abun-        soft-bottom communities: importance of animal interactions or
    dance of benthic stages of the snow crab (Chionoecetes opilio)     sedimentary habitat modifications. Can J Fish Aquat Sci
    in Bonne Bay, Newfoundland, Canada. Can J Fish Aquat Sci        37:1130–1139
    55:262–279                             Hutchings JA (1996) Spatial and temporal variation in the density
Comeau M, Starr M, Conan GY, Robichaud G, Therriault J-C            of northern cod and a review of hypotheses for the stock’s
    (1999) Fecundity and duration of egg incubation for multipa-      collapse. Can J Fish Aquat Sci 53:943–962
    rous female snow crabs (C. opilio) in the fjord of Bonne Bay,    Jackson JBC et al (2001) Historical overfishing and the recent
    Newfoundland. Can J Fish Aquat Sci 56:1088–1095             collapse of coastal ecosystems. Science 293:629–637
Commito JA, Ambrose WG Jr (1985) Multiple trophic levels in        Jamieson GS (2002) In: Claudi R, Nantel P, Muckle-Jeffs E (eds)
    soft-bottom communities. Mar Ecol Progr Ser 26:289–293         Alien invaders in Canada’s waters, wetlands and forests. Nat-
Conan GY, Starr M, Comeau M, Therriault J-C, Hernandez             ural Resources Canada, Ottawa, pp 179–186
    FXMI, Robichaud G (1996) Life history strategies, recruitment    Jennings S, Kaiser MJ (1998) The effects of fishing on marine
    fluctuations, and management of the Bonne Bay fjord Atlantic       ecosystems. Adv Mar Biol 34:201–352
    snow crab (Chionoecetes opilio). In: High latitude crabs: biol-   Kemp WM, Petersen JE, Gardner RH (2001) Scale-dependence
    ogy, management, and economics. Alaska Sea Grant Program.        and the problem of extrapolation: implications for experi-
    AK-SG-96-02, pp 59–97                          mental and natural coastal ecosystems. In: Gardner RH,
Connell JH, Sousa WP (1983) On the evidence needed to judge          Kemp WM, Kennedy VS, Petersen JE (eds) Scaling relations
    ecological stability or persistence. Am Nat 121:789–824         in experimental ecology. Columbia University Press, New
Cooper SD, Walde SJ, Pecharsky BL (1990) Prey exchange rates          York, pp 3–57
    and the impact of predators on prey populations in streams.     Kneib RT (1988) Testing for indirect effects of predation in an
    Ecology 71:1503–1514                          intertidal soft-bottom community. Ecology 69:1795–1805
Drummond-Davis NC, Mann KH, Pottle RA (1982) Some esti-          Kneib RT (1991) Indirect effects in experimental studies of marine
    mates of population density and feeding habits of the rock crab     soft-sediment communities. Am Zool 31:874–885
    Cancer irroratus, in a kelp bed in Nova Scotia. Can J Fish     Koeller P (2000) Relative importance of abiotic and biotic fac-
    Aquat Sci 39:636–639                          tors to the management of the Northern shrimp (Pandalus
Englund G (1997) Importance of spatial scale and prey movements        borealis) fishery on the Scotian shelf. J Northwest Atl Fish
    in predator caging experiments. Ecology 78:2316–2325          Sci 27:37–50
Ennis PG, Hooper RG,Taylor DM (1990) Changes in the com-          van de Koppel J, Herman PMJ, Thoolen P, Heip CHR (2001) Do
    position of snow crab, Chionoecetes opilio, participating in the    alternate stable states occur in natural ecosystems? Evidence
    annual breeding migration in Bonne Bay, Newfoundland. Can        from a tidal flat. Ecology 82:3449–3461
    Atl Fish Adv Comm Res Doc 90(5):1–18                Lefebvre L, Brethes J-C (1991) Orientation of movements of tag-
                                              ˆ
Fernandes TF, Huxman M, Piper SR (1999) Predator caging            ged, male, snow crabs (Chionoecetes opilio) in the southwestern
    experiments: a test of the importance of scale. J Exp Mar Biol     Gulf of St. Lawrence. Can J Fish Aquat Sci 48:1167–1175
    Ecol 241:137–154                          Lenihan HS, Micheli F (2001) Soft-sediment communities. In:
Folk RL (1980) Petrology of sedimentary rocks, 2nd edn. Hemphill        Bertness MD, Gaines SD, Hay ME (eds) Marine community
    Publishing                               ecology. Sinauer Associates, Sunderland, pp 253–287
Frid CLJ (1989) The role of recolonization processes in benthic      Lilly GR, Parsons DG, Kulka DW (2000) Was the increase in
    communities, with special reference to the interpretation of      shrimp biomass on the Northeast Newfoundland Shelf a
    predator-induced effects. J Exp Mar Biol Ecol 126:163–171        consequence of a release in predation pressure from cod?
Gabriel KR (1971) The biplot graphic display of matrices with         J Northwest Atl Fish Sci 27:45–61
    application to principal component analysis. Biometrika       Magurran AE (1988) Ecological diversity and its measurement.
    58:453–467                               Princeton University Press, Princeton
Gilbert D, Pettigrew G (1993) Current-meter data from Bonne Bay,      Mallet P, Landsburg M (1996) Gulf rock crab (Cancer irroratus).
    NFLD, during the summer of 1991. Can Rep Hydr Ocean Sci         DFO Atlantic fisheries stock status report 96/98 E p 4
    122:63                               McGuinness KA (1997) Tests for artifacts in some methods used to
Goni R (1998) Ecosystem effects of marine fisheries: an overview.        study herbivory and predation in mangrove forests. Mar Ecol
    Ocean Coast Manag 40:37–64                       Progr Ser 153:37–44
Grassle JF, Smith W (1976) A similarity measure sensitive to the      Micheli F (1997) Effects of predator foraging behavior on patterns
    contribution of rare species and its use in investigation of var-    of prey mortality in marine soft bottoms. Ecol Monogr 67:203–
    iation in marine benthic communities. Oecologia 25:13–22        224
Green RH (1979) Sampling design and statistical methods for        Micheli F (1999) Eutrophcation, fisheries, and consumer-resource
    environmental biologists. Wiley, New York                dynamics in marine pelagic ecosystems. Science 285:1396–1398
Hall SJ, Raffaelli D, Turrell WR (1990) Predator-caging experi-       Myers RA, Hutchings JA, Barrowman NJ (1996) Hypothesis for
    ments in marine systems: a re-examination of their value. Am      the decline of cod in the north Atlantic. Mar Ecol Progr Ser
    Nat 136:657–672                             138:293–308
136

´
Olafsson EB, Peterson CH, Ambrose WG Jr (1994) Does recruit-     Schneider DC (2001) The rise of the concept of scale in ecology.
  ment limitation structure populations and communities of       Bioscience 51:545–553
  macroinvertebrates in marine soft sediments: the relative sig-  Seitz RD, Lipcius RN, Hines AH, Eggleston DB (2001) Density-
  nificance of pre- and post-settlement processes. Oceanogr Mar     dependent predation, habitat variation, and the persistence of
  Biol Annu Rev 32:65–109                        marine bivalve prey. Ecology 82:2435–2451
Ouellet P, Lefaibre D (1994). Vertical distribution of Northern   Simard Y, Brunelle P, Lacele J (1990) Distribution and growth of
  Shrimp (Pandalus borealis) larvae in the Gulf of St. Lawrence;    pre-recruits cohorts of the Northern shrimp (Pandalus borealis)
  implications for trophic interactions and transport. Can J Fish    in the Western Gulf of St. Lawrence as related to hydrographic
  Aquat Sci 51:123–132                         conditions. Can J Fish Aquat Sci 47:1526–1533
Ouellet P, Taggart CT, Frank KT (1995) Early growth, lipid      Smith B, Wilson JB (1996) A consumer’s guide to evenness indices.
  composition, and survival expectations of shrimp Pandalus       Oikos 76:70–82
  borealis larvae in the Northern Gulf of St. Lawrence. Mar Ecol  Sokal RR, Rohlf FJ (1994) Biometry. The principles and practice
  Progr Ser 126:163–175                         of statistics in biological research. W H Freeman, San Francisco
Paul AJ, Dawe EG, Elner R, Jamieson GS, Kruse GH, Otto RS,      Squires HJ (1996) Decapod crustaceans of Newfoundland, Lab-
  Sainte-Marie B, Shirley TC, Woodby D (2002) Crabs in cold       rador and the Canadian Eastern Arctic. Can Man Rep Fish
  water regions: biology, management, and economics. In: Alaska     Aquat Sci 2359:235
  Sea Grant, Proceedings of the Crab 2001 Symposium.        Squires HJ, Dawe EG (2003) Stomach contents of snow crab
  Anchorage, Alaska p 876                        (Chionoecetes opilio, Decapoda, Brachyura) from the Northeast
Pauly D, Christensen V, Dalsgaar J, Froese R, Torres F Jr (1998)     Newfoundland Shelf. J Northwest Atl Fish Sci 32:27–38
  Fishing down marine food webs. Science 279:860–863        Thompson RJ, Hawryluk M (1989) Physiological energetics of the
Peterson CH (1979) Predation, competitive exclusion, and diversity    snow crab, Chionoecetes opilio. In: Proceedings of the interna-
  in the soft-sediment benthic communities of estuaries and la-     tional symposium on King and Tanner crabs. University of
  goons. In: Livingston RJ (ed) Ecological processes in coastal     Alaska Sea Grant Report. 90–04. Fairbanks pp 283–291
  and marine systems. Plenum, New York, pp 233–264         Thrush SF (1999) Complex role of predators in structuring soft-
Peterson CH (1984) Does a rigorous criterion for environmental      sediment macrobenthic communities: implications of changes
  identity precludes the existence of multiple stable points? Am    in spatial scale for experimental studies. Aust J Ecol 24:344–
  Nat 124:127–133                            354
Peterson CH, Black R (1994) An experimentalist’s challenge: when   Thrush SF et al (1997) The sandflat habitat: scaling from experi-
  artifacts of intervention interact with treatments. Mar Ecol     ments to conclusions. J Exp Mar Biol Ecol 216:1–9
  Progr Ser 111:289–297                       Trueblood DD, Gallagher ED, Gould DM (1994) Three stages of
Petraitis PS, Latham RE (1999) The importance of scale in testing    seasonal succession on the Savin Hill Cove mudflat, Boston
  the origin of alternative community states. Ecology 80:429–442    Harbor. Limnol Oceanogr 39:1440–1454
Posey MH, Hines AH (1991) Complex predator–prey interactions     Virnstein RW (1977) The importance of predation by crabs and
  within an estuarine benthic community. Ecology 72:2155–2169      fishes on benthic infauna in Chesapeake Bay. Ecology 58:1199–
Powles H (1968) Distribution and biology of the spider crab       1217
  Chionoecetes opilio, in the Magdalen Shallows, Gulf St. Law-   Wieczorek SK, Hooper RG (1995) Relationship between diet and
  rence. Fish Res Board Can Ms Rep Ser 997:1–106            food availability in the snow crab Chionoecetes opilio (O.
   ´
Quijon PA, Snelgrove PVR (2005) Differential regulatory roles of     Fabricius) in Bonne Bay, Newfoundland. J Crust Biol 15:236–
  crustacean predators in a sub-Arctic, soft-sediment system. Mar    247
  Ecol Progr Ser 285:137–149                    Wiens JA (2001) Understanding the problem of scale in experi-
Rivard DH, Bowen BE (1971) A survey of marine fauna of Bonne       mental ecology. In: Gardner RH, Kemp WM, Kennedy VS,
  Bay, Newfoundland. Transcript Manuscript p 18             Petersen JE (eds) Scaling relations in experimental ecology.
Roberts D, Rittschof D, Gerhart DJ, Schmidt AR, Hill LG (1989)      Columbia University Press, New York, pp 61–88
  Vertical migration of Mercenaria mercenaria (L.) (Mollusca:    Williams AB (1984) Shrimps, lobsters, and crabs of the Atlantic
  Bivalvia). Environmental correlates and ecological significance.    Coast of the Eastern United States, Maine to Florida/Austin B.
  J Exp Mar Biol Ecol 126:271–280                    Williams Smithsonian Institution Press, Washington
Sainte-Marie B, Gilbert D (1998) Possible effects of changes in CIL  Wilson WH Jr (1991) Competition and predation in marine soft-
  temperature and thickness in population dynamics of snow crab     sediment communities. Annu Rev Ecol Syst 21:221–241
  (C. opilio) in the Gulf of Saint Lawrence. Can Atl Fish Sci Adv  Witman JD, Sebens KP (1992) Regional variation in fish predation
  Comm Res Doc 98/38 p 19                        intensity: a historical perspective in the Gulf of Maine. Oeco-
Sainte-Marie B, Raymond S, Brethes J-C (1995) Growth and
                   ˆ                  logia 90:305–315
  maturation of the benthic stages of male snow crab, Chion-    Woodin SA (1976) Adult-larval interactions in dense infaunal
  oecetes opilio (Brachyura: Majidae). Can J Fish Aquat Sci       assemblages: patterns of abundance. J Mar Res 34:25–41
  52:903–924                            Woodin SA (1999) Shallow water benthic ecology: a North
Scarrat DJ, Lowe R (1972) Biology of the rock crab (Cancer ir-      American perspective of sedimentary habitats. Aust J Ecol
  roratus) in the Northumberland Strait. J Fish Res Board Can      24:291–301
  29:161–166                            Worm B, Myers RA (2003) Meta-analysis of cod-shrimp interac-
Schiermeier Q (2002) How many more fish in the sea? Nature        tions reveals top-down control in oceanic food webs. Ecology
  419:662–665                              84:162–173
Schneider DC (1978) Equalization of prey numbers by migratory
  shorebirds. Nature 271:353–354