You are on page 1of 14

ICES Journal of Marine Science Advance Access published June 6, 2014

ICES Journal of

Marine Science
ICES Journal of Marine Science; doi:10.1093/icesjms/fsu065

Original Article

Kevin A. Sorochan1,2* and Pedro A. Quijon 1


1

University of Prince Edward Island, Department of Biology, 550 University Ave, Charlottetown, PE CIA 4P3, Canada
Dalhousie University, Department of Oceanography, 1355 Oxford St. Halifax, NS B3H 4R2, Canada

*Corresponding author: tel: +1 902 494 2830; fax: +1 902 494 3877; e-mail: kevin.sorochan@dal.ca
Sorochan, K. A., and Quijon, P. A. Horizontal distributions of Dungeness crab (Cancer magister) and red rock crab (Cancer
productus) larvae in the Strait of Georgia, British Columbia. ICES Journal of Marine Science, doi: 10.1093/icesjms/fsu065.
Received 17 July 2013; revised 16 March 2014; accepted 18 March 2014.
The supply of planktonic larvae to adult populations is an important contributor to the spatial and temporal variability of benthic marine organisms. The ability to predict spatial patterns of larvae and recruits from the physical and biological processes that facilitate dispersal is required in
order to advise and evaluate conservation and sheries management decisions. In the present study, the horizontal distribution of Dungeness crab
(Cancer magister) and red rock crab (Cancer productus) zoeae was described from surveys conducted in the Strait of Georgia in the spring of 2009
and 2010. Processes that may be responsible for generating spatial variability of larvae were evaluated based on (i) horizontal overlap between larvae
and water properties, (ii) spatial dependence of larvae and water properties, and (iii) changes in the dispersion of stage-specic distributions.
Interspecic variability between horizontal patterns of the rst and second larval stages was primarily attributed to differences in the distribution
of larval release locations, which appeared to be restricted to the southern and central strait for C. magister. Potential effects of physical processes on
larval distributions are also discussed.
Keywords: Cancer magister, Cancer productus, decapod larvae, Dungeness crab, larval distribution, Metacarcinus magister, population connectivity,
red rock crab, Salish Sea, Strait of Georgia.

Introduction
The dispersal of planktonic larvae facilitates exchange of individuals
between populations and is a fundamental determinant of the
dynamics of benthic marine organisms (Cowen and Sponaugle,
2009). Consequently, larval dispersal is an important management
consideration, especially in the strategic design of marine reserves
and their networks (Kritzer and Sale, 2004). Larval distributions
can be utilized to develop hypotheses about the physical and
biological processes responsible for generating spatial patterns or
to test predictions of larval distributions when information on dispersal processes is available (Sale and Kritzer, 2003). For example,
horizontal distributions are sometimes used to validate biophysical
models (Peliz et al., 2007; Snauffer, 2013) that are ultimately used
to predict the probability of larval exchange between locations
(Metaxas and Saunders, 2009).
# International

The Dungeness crab (Cancer magister) and the red rock crab
(Cancer productus) are benthic mobile predators in coastal waters
of the Northeast Pacific (Stevens et al., 1982; Yamada and
Boulding, 1996). Both species are permitted to be commercially
fished in British Columbia; however, C. magister is the primary
target for exploitation, and landings support an economically important fishery (DFO, 2013a). For these reasons, a substantial
volume of research has been conducted on the biology of the early
life history stages of C. magister and other cancrid species from
waters adjacent to Vancouver Island (Orensanz and Galluchi,
1988; DeBrosse et al., 1990; Jamieson and Phillips, 1993; Sulkin
and McKeen, 1994; Sulkin et al., 1996). In Puget Sound, monitoring
of ovigerous females has indicated that larvae of C. magister hatch
primarily from February through April (Mayer, 1973; Armstrong
et al., 1987) and that the majority of larvae of C. productus hatch

Council for the Exploration of the Sea 2014. All rights reserved.
For Permissions, please email: journals.permissions@oup.com

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Horizontal distributions of Dungeness crab (Cancer magister) and


red rock crab (Cancer productus) larvae in the Strait of Georgia,
British Columbia

Page 2 of 14

Methods
Study area
The SoG is a semi-enclosed ocean basin situated between Vancouver
Island and mainland British Columbia (Figure 1). Circulation and
water properties in the SoG are strongly influenced by seasonal
river runoff and are further modulated by tidal currents, wind,
and basin topography (Masson and Cummins, 2004). Although numerous rivers empty into the SoG, 75% of the freshwater input is
from the Fraser River (Thomson, 1981). Observations from drifters
and current meters have suggested residual northward surface flow
along the SoGs eastern margin, and southward flow along the
western margin. Based on these results it has been suggested that
counterclockwise circulation prevails in the central SoG during
the spring and summer months (reviewed by Thomson, 1981).
Ocean modelling has simulated two additional counter-rotating
gyres embedded within the straits south-central region (Masson
and Cummins, 2004) near the location of an empirically measured
rotary current (Stacey et al., 1987).

open during deployment and recovery. Since larvae cannot be


assumed to be distributed evenly in the vertical dimension, differences in sampling methodology between surveys prevented a
direct comparison between S1 and S2.
The carapace length of the first zoeae of C. magister and
C. productus has been demonstrated to be greater than the mesh
size used in the present study (Lough, 1975; Shirley et al., 1987;
Rice and Tsukimura, 2007). However, illustrations of zoeae of
C. magister by Poole (1966) and C. productus by Trask (1970) indicate that if the abdomen were to become compressed against the
ventral side of the carapace, the first two larval stages of C. magister
and first four larval stages of C. productus could potentially pass
through the net.
Plankton samples were preserved in 95% ethanol and subsampled using a Folsom splitter. A minimum of 200 reptantian
decapod larvae were identified from each sample that required splitting. Approximately 80% of the larvae identified from subsamples
consisted of cancrid zoeae (Sorochan, 2011). Larvae of C. magister
and C. productus were identified to species and stage, and counts
were standardized to concentration per 100 m3 of filtered seawater.
Although C. magister and C. productus were the only species considered in the present study, at least two other cancrid species
(Cancer oregonensis and Cancer gracilis) were present in the
samples. Identification was facilitated by guides (Lough, 1975;
Puls, 2001; Rice and Tsukimura, 2007), larval descriptions (Poole,
1966; Trask, 1970; Ally, 1975), and inspection of laboratory-reared
C. oregonensis zoeae (Sorochan, 2011). The morphological features

Sampling and larval identication


Larvae were collected from zooplankton surveys conducted in the
SoG during two consecutive spring seasons: 25 29 April 2009
and 24 27 April 2010 (hereafter referred to as S1 and S2, respectively). Samples were collected throughout the day and night from a grid
of stations spaced 9 km apart that covered the majority of the SoG
(Figure 2). In S1, 57 stations were visited. In S2, these stations were
revisted plus two additional stations.
At each station vertical profiles of salinity, temperature, oxygen
and fluorescence were logged from a Conductivity-TemperatureDepth profiler (Figure 3). Larvae were sampled with a Tucker
Trawl (1.5 m2 frame, 1-mm mesh size) equipped with a flowmeter
and depth-logging device. The duration and speed of each plankton
tow was 15 min and 1 m s21, respectively. In S1, samples were collected from a constant depth with a mean + SD of 34.3 + 6.2 m. In
S2, samples were collected by oblique tows from a depth with
mean + SD of 47.5 + 8.9 m (Figure 2). The Tucker Trawl remained

Figure 1. The Strait of Georgia and adjacent water bodies, collectively


referred to as the Salish Sea. Stippled line indicates the Canada USA
boarder. JS, Johnstone Strait; Te, Texada Island; SoG, Strait of Georgia;
Fr, Fraser River; JdF, Strait of Juan de Fuca.

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

by late March or early April (Knudsen, 1964). During larval development, C. magister and C. productus progress through five zoeal
stages and one megalopal stage (Poole, 1966; Trask, 1970).
Estimates of the natural planktonic larval duration of C. magister
from Oregon to Alaska range between 120 and 160 days (Fisher,
2006 and references therein); over this period there is potential for
the larvae of this species to disperse large distances (Shanks,
2009). Relationships between dispersal patterns and processes
have been studied extensively in coastal waters of the Northeast
Pacific; however, far less research has been conducted within semienclosed inlets and basins (reviewed by Rasmuson, 2012).
In the present study, the mesoscale distribution of the early larval
stages of C. magister and C. productus was described from the Strait
of Georgia (SoG hereafter), and relationships between larval distributions and their governing processes were investigated. The shoreline of the SoG is highly populated by humans, and although the SoG
ecosystem provides several services, it is subjected to numerous
stressors (Johannessen and McCarter, 2010). This has provided incentive for the consideration and implementation of marine
reserves (Parks Canada, 2012; DFO, 2013b) and ecosystem-based
management initiatives (DFO, 2013c). This research provides information relevant to the future management decisions on populations
of C. magister and C. productus in the SoG.

K. A. Sorochan and P. A. Quijon

Page 3 of 14

Larval distributions of the Dungeness crab and red rock crab

[a] = [a + b + c] [b + c].
[b] = [a + b] + [b + c] [a + b + c]
[c] = [a + b + c] [a + b].
Figure 2. Stations sampled in (a) survey 1 and (b) survey 2. Filled and
empty dots indicate stations sampled during darkness/twilight and
daylight, respectively. Insets within each map illustrate representative
net depth proles from each survey.

necessary to separate the zoeal stages of C. productus from


C. oregonensis have not been described, and inconsistencies were
found between the morphology of C. productus zoeae identified in
the present study and zoeae of the same species described by Trask
(1970). Consequently, morphological characteristics that were
used for separation of C. productus from C. oregonensis have been
outlined and illustrated (see Supplementary data).

Data analysis
All statistical analyses were conducted in the R programming environment version 2.12.2 using log(x+1)-transformed larval abundance. Variation partitioning (Legendre and Legendre, 2012) and
hierarchical partitioning (Chevan and Sutherland, 1991) were
used to evaluate relationships between explanatory data and larval
distributions. Both methods employ multiple linear regression
(MLR) but are not hampered by problems associated with multicolinearity because they utilize goodness of fit statistics, such as
the coefficient of determination (R-squared).
Relationships between explanatory variables and abundance of
each larval stage could not be evaluated by MLR due to the presence
of zeros in the larval abundance data. Therefore, these relationships
were evaluated for combined distributions of adjacent stages 1 2
and 3 4. This was not possible for C. productus in S1, in which all
stages were combined. Since the vertical distribution of the larvae
was unknown, the values of hydrographic variables (except

Hierarchical partitioning was used to determine the independent


contribution of each hydrographic variable towards accounting for
variation in larval abundance. The method was carried out by
obtaining R-squared values from 2k MLR models containing all possible combinations of k explanatory variables. The independent
effect (I ) for each hydrographic variable was then quantified by averaging the increase in R-squared that occurred from the addition of
the hydrographic variable of interest to each model (Chevan and
Sutherland, 1991; Christiansen, 1992). The results are presented as
the proportion of I (computed for each explanatory variable) of
total I (integrated across explanatory variables). The significance
of I was evaluated by a randomization test (n 1000 randomizations) described by Mac Nally (2002). Hierarchical partitioning
was carried out only after a significant relationship had been identified between larval abundance and hydrographic variables from
MLRs. Pearson correlations between hydrographic variables and
larval abundance were used to determine the sign of significant relationships.
Morans I correlograms were used to explore the relationship
between spatial dependence and distance between stations for
both larval abundance and hydrographic variables. Correlograms
have been used in larval ecology to determine if spatial dependence
of both larvae and hydrographic variables occur on similar length
scales (Maynou et al., 2006). The Morans I coefficient is a
measure of autocorrelation of the response variable from all pairs
of sites separated from each other by a distance specified by a predefined distance class:
n
I = n

i=1
i=j

1
n

j=1
j=i

i=1
i=j

wij (d)

n

wij (d)(Xi X)(Xj X)


,

1 n
2
(X

X)
i
i=1
n
j=1
j=i

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

fluorescence) were considered from 2 m in both surveys


(Figure 4) as well as 34 m in S1 and 48 m in S2 (Figure 5). These
values represent the depth range sampled in each survey and are indicative of temperature, salinity and oxygen conditions both above
(referred to hereafter as Ts, Ss and Os) and below (referred to hereafter as Td, Sd and Od) the thermocline (Figure 3). Since fluorescence
values did not continually increase or decrease with depth, they were
integrated over the upper 34 m in S1 and 48 m in S2 (hereafter referred to as Fint). Prior to MLR analyses, transformations were
applied to hydrographic variables to improve linearity between
explanatory and response data (Table 1).
Variation partitioning was conducted to isolate variation in
larval abundance accounted for by spatial variables independent
of hydrographic variables [a], the intersection between spatial and
hydrographic variables [b], and hydrographic variables independent of spatial variables [c]. Spatial variables were produced from
mononomial functions (x, y, x 2, y 2, xy, x 2y, xy 2, x 3, y 3) of centred
and standardized coordinates (UTM) of the sample sites. Multiple
linear regressions were used to determine the total variation in
larval abundance (R-squared) accounted for by spatial variables
[a + b], hydrographic variables [b + c], and both spatial and hydrographic variables combined [a + b + c]. Fractions [a], [b], and [c]
were then partitioned as follows:

Page 4 of 14

K. A. Sorochan and P. A. Quijon

where n is the number of sample pairs at locations i and j, X is the


variable of interest, and w(d) is a spatial weight for distance class d
(Fortin and Dale, 2005). In the present study, if i and j were separated
by a distance within or outside d, w(d) was set to 1 or 0, respectively.
A spatial lag of 10.6 km was selected so that the number of sample
pairs was spread across distance classes as evenly as possible.
Euclidian connections between samples on either side of Texada
Island were removed to avoid the inclusion of their spatial covariance within inappropriate distance classes.
Prior to calculation of Morans I co-efficients, larval abundance
and hydrographic variables were detrended. This step is necessary
because of the assumption of intrinsic stationarity (e.g. that the variance is constant and the mean difference in X between i and j is zero
within d) (Legendre and Fortin, 1989). Detrended data were obtained
by collecting residuals from polynomial regressions between hydrographic and larval abundance response data and a trend surface
generated from a first, second or third order polynomial of each stations centred and standardized UTM coordinates. Selection of the
polynomial model used for detrending was based on corrected
Akaike Information Criterion values (not shown) (Burnham and
Anderson, 2002). Morans I coefficients were evaluated for significance
by employing a permutation test and sequential Bonferroni correction. In a separate analysis, the coefficient of variation (CV) was calculated to evaluate the dispersion of larval abundance across samples for
each stage of each species within each survey.

A three factor ANOVA was carried out to test for the effects of
species, survey and daylight on total larval abundance (all stages
combined). Because sample size was unbalanced, Type III sums
of squares were calculated. Data did not exhibit substantial
deviations from normality, and homogeneity of variance was
confirmed using a Levenes test. In addition, a single factor analysis of similarity (ANOSIM) was used to compare Bray Curtis
similarity of larval stage composition for each species between
daylight and darkness within each survey. Only eight samples in
S1, and ten samples in S2, were collected during nautical twilight.
Abundances from these samples did not appear to be systematically
high or low, and were grouped within the darkness category for each
analysis. Tests comparing larval abundance and stage composition
between darkness/twilight and daylight conditions were conducted because previous evidence has suggested cancrid zoeae
exhibit diel or crepuscular vertical migratory behaviour (Reilly,
1983; Shanks, 1986; Hobbs and Botsford, 1992; Park and Shirley,
2005).

Results
Results from variation partitioning between spatial variables [a], the
intersection between spatial and hydrographic variables [b], hydrographic variables [c], and unexplained variation [d] are outlined in
Table 2. Explained variation in larval abundance [a+b+c] was
greater for C. magister (0.418 R 2 0.687) than C. productus

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 3. Vertical proles of median temperature, salinity, oxygen and uorescence for survey 1 (upper panels) and survey 2 (lower panels).
Horizontal grey lines represent the interquartile range for median values at 1-m depth intervals.

Page 5 of 14

Larval distributions of the Dungeness crab and red rock crab

(0.168 R 2 0.486). In S1, the majority of explained variation in


abundance of C. magister was accounted for by fraction [b] (R 2
0.427) for stages 1 2, and fractions [a] (R 2 0.153) and [c]
(R 2 0.179) for stages 3 4. With respect to C. productus, the majority of explained variation in abundance of stages 1 4 was allocated to fractions [a] (R 2 0.211) and [b] (R 2 0.181). In S2,
explained variation in abundance of C. magister was primarily
attributed to fraction [b] (R 2 0.470) for stages 1 2, and fractions
[a] (R 2 0.268) and [b] (R 2 0.256) for stages 3 4. With respect
to C. productus, explained variation in abundance of stages 1 2 and
3 4 was completely accounted for by fractions [a] (R 212 0.110;
R 234 0.127) and [b] (R 212 0.094; R 234 0.150). Fractions
[a], [b] and [c] were occasionally negative because they were
derived from adjusted R-squared values from the MRLs (Table 2).
Significant relationships between larval abundance and hydrographic data occurred in all cases except for C. productus in S2
(Table 2). Test statistics from each MLR are provided in the
Supplementary data. Results from hierarchical partitioning are

reported in Table 3. In S1, hydrographic variables with significant


independent effects were negatively related to the abundance of
C. magister (except Ts), and positively related to the abundance
of C. productus. In S2, relationships between the abundance of
C. magister and Sd and Os were consistent across stages; additional independent effects (Ss and Od) were significant for stages 12.
Independent effects of temperature were only significant in S1.
The abundance of stages 1 2 of C. magister was negatively related
to Td whereas the abundance of stages 3 4 of C. magister and 1 4
of C. productus were positively related to Ts. With respect to
salinity, Ss was negatively related to the abundance of stages 1 2
of C. magister in both surveys and positively related to the abundance of stages 1 4 of C. productus in S1. Independent effects of
Sd were only significant for C. magister in S2; Sd was positively
related to the abundance of stages 1 2 and 3 4. With respect to
oxygen, significant independent effects of Os were only detected
for C. magister; Os was negatively related to the abundance of
stages 1 2 in S2 and 3 4 in both surveys. In S1, Od was negatively

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 4. Horizontal distributions of salinity, temperature, and oxygen at 2 m depth from surveys 1 and 2.

Page 6 of 14

K. A. Sorochan and P. A. Quijon

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 5. Horizontal distributions of salinity, temperature and oxygen at 34 m depth from survey 1 and 48 m depth from survey 2 as well as
distributions of vertically integrated uorescence from 34 m to 2 m from survey 1 and 48 m to 2 m from survey 2.

related to the abundance of stages 1 2 and 3 4 for C. magister and


positively related to the abundance of stages 1 4 of C. productus,
whereas Od was positively related to the abundance of stages 1 2
of C. magister in S2. In all cases, independent effects of Fint were nonsignificant.

In S1, 40% of total larval abundance of each species was


accounted for by each of the second and third stages for C. magister
and first and second stages for C. productus (Figure 6). In S2, the
third stage of C. magister and second stage of C. productus accounted
for 60% of total larval abundance for each species. The CV values

Page 7 of 14

Larval distributions of the Dungeness crab and red rock crab

Table 1. Transformations of hydrographic variables used in each


analysis.
Variable
Ts,d
Ss
Sd
Os,d
Fint
Fint

Survey
S1, S2
S1,S2
S1, S2
S1, S2
S1
S2

Transformation
None
10x
None
None

X
Log(x)

Discussion
Relationships between the horizontal distribution of meroplankton
and hydrographic variables can be influenced by the distribution of
larval release locations, ocean circulation, and vertical swimming
behaviour (discussed by Isari et al., 2008; Ayata et al., 2011).
Variation partitioning between spatial variables (fraction [a]),
hydrographic variables (fraction [b]), and their intersection (fraction [c]), can be interpreted in several ways. For example, explained
variation allocated to fraction [a] or [b] may simply reflect the extent
of overlap between the distribution of larval release locations and
hydrographic variables. Alternatively, if larvae are transported passively from their release locations, larvae and hydrographic variables
can be correlated and appear randomly or non-randomly distributed in space depending on the heterogeneity of circulation at the
scale of the sampling resolution (Mackas et al., 1985). Overlap of
larvae and hydrographic variables that occurs in non-random or
random patterns could lead to variation being allocated to fractions
[a] or [c], respectively. Vertical migratory behaviour, which can
affect horizontal movements of decapod larvae (Queiroga and
Blanton, 2005), is probably capable of decoupling distributions of
larvae and hydrographic variables. This may result in non-random
patterns of larvae accounted for by fraction [a] or random patterns
of larvae contributing to unexplained variation (fraction [d]) (as
discussed in Ayata et al., 2011). The inability of spatial and hydrographic variables to account for more variation in larval abundance
was probably due to the lack of appropriate explanatory variables
and the simplicity of the surface trends used to simulate horizontal
patterns of larval abundance (Bocard et al., 1992).
Water properties probably did not directly contribute to variation
allocated to fraction [c] by affecting larval survival. Larvae likely
avoided entry into waters with salinities capable of causing physiological problems (,25 PSU; Reed, 1969), as swimming behaviour
of decapod zoeae has been demonstrated to be sensitive to salinity
gradients (Forward, 1989). The range of temperatures observed in
the present study (7.610.68C) has been demonstrated to be nonlethal to cancrid zoeae but could indirectly affect larval distributions
by influencing stage duration and mortality. However, these indirect
effects could not be evaluated because the thermal history of larvae
could not be inferred from the data collected.
The first and second larval stages of C. magister were least abundant in the northwest region and most abundant in the southern
region and along the eastern margin of the central strait. Fisheries
landings of C. magister have been most significant in the southern

Table 2. Adjusted R-squared values from multiple linear regression (MLR) relating larval abundance to spatial (Sp), environmental (Env),
and spatial and environmental (Sp + Env) explanatory matrices.
Partitioned variation
Species, stage
S1
C. magister, 1 2
C. magister, 3 4
C. productus, 1 4
S2
C. magister, 1 2
C. magister, 3 4
C. productus, 1 2
C. productus, 3 4

Sp
[a 1 b]

Env
[b 1 c]

Sp 1 Env
[a 1 b 1c]

[a]

[b]

0.395
0.240
0.392

0.471
0.265
0.274

0.440
0.418
0.486

20.031
0.153
0.211

0.427
0.087
0.181

0.044
0.179
0.093

0.560
0.582
0.514

0.652
0.524
0.204
0.278

0.505
0.265
0.058
0.099

0.687
0.533
0.168
0.227

0.181
0.268
0.110
0.127

0.470
0.256
0.094
0.150

0.035
0.009
20.036
20.051

0.312
0.467
0.832
0.773

[c]

[d ]

Bolded values indicate signicant relationships. The proportion of explained variation in larval abundance was divided into the following three fractions: [a],
spatial; [b], spatio-environmental intersection; [c], environmental; [d], unexplained.

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

were highest for the first larval stage of both species, except for C.
productus in S2 when CV values of the first and second stages were
low and of approximately the same magnitude (Figure 6).
In both surveys, the first and second larval stages of C. magister
were most abundant (10103 larvae 100 m23) along the central
SoGs eastern margin and least abundant (010 larvae 100 m23) in
the northwest region (Figure 7). Abundance of the third and fourth
stages of C. magister was highest (102 103 larvae 100 m23) at the
geographical centre of the strait in S1 and along the straits eastern
margin in S2 (Figure 7). Abundance of the first and second stages
of C. productus larvae was elevated (10102 larvae 100 m23) in the
straits northwest region in both surveys, but the abundance of
these stages was more widespread in S2 (Figure 8). In S2, abundance
of the third stage of C. productus was highest (10102 larvae 100 m23)
along the eastern margin of the strait (Figure 8).
Significant positive spatial dependence was not detected when
calculated from detrended larval abundance, but was detected in
the first distance class for Ts, Td, Ss and Sd in S1 and Sd and Fint in
S2 (Figure 9). In S1, significant negative spatial dependence occurred within the fifth distance class for Ss and the third distance
class for C. magister (stages 1 2 and 3 4) and Ts. In S2, significant
negative spatial dependence was detected in the fourth distance
class for C. productus (stages 1 2) and Fint (Figure 9).
Results from the ANOVA are summarized in Table 4. A significant interaction was detected between the factors species and
survey; C. magister was more abundant than C. productus in S1
but not in S2, and the abundance of C. productus was higher in S2
than in S1, but there was no difference in abundance between
surveys for C. magister (Figure 10). There was no effect of daylight
on larval abundance (Table 4). In addition, ANOSIMs indicated
no effect of daylight on larval stage composition for either species
in either survey (Table 5).

Page 8 of 14

K. A. Sorochan and P. A. Quijon

Table 3. Relative contributions of independent effects of explanatory variables from hierarchical partitioning analysis.
Variable
Ts
Td
Ss
Sd
Os
Od
Fint

S1, M1 2
9.7
25.2 (2)
31.7 (2)
7.4
3.8
14.5 (2)
7.5

S1, M3 4
20.0 (1)
11.0
6.8
10.7
24.6 (2)
20.1 (2)
6.7

S1, P1 4
19.5 (1)
24.6 (1)
28.1 (1)
5.9
1.9
16.6 (1)
2.6

S2, M1 2
5.9
4.7
22.6 (2)
19.6 (1)
23.8 (2)
20.5 (1)
2.8

S2, M3 4
7.8
4.2
13.4
23.6 (1)
34.7 (2)
12.7
3.6

Signicant independent effects are bolded and the sign of their correlation with larval abundance is indicated in parentheses. Column labels indicate survey
number, species, and stage from left to right. S, survey; M, C. magister; P, C. productus.

SoG (Jamieson and Phillips, 1993; Sulkin et al., 1996), and have contributed substantially to the overall value of the fishery (DFO,
2013a). It has been hypothesized that elevated temperatures
outside the thermal tolerance of juvenile instars (.188C) have prevented the establishment of stocks suitable for commercial fishing of
C. magister in the northwest region (Sulkin et al., 1996). Overlap
between the distribution of release locations and hydrographic variables was probably the primary reason why the majority of explained
variation in abundance of stages 12 of C. magister was allocated to
fraction [b]; however, passive larval transport may have also played
a role. The distribution of Ss in both surveys indicated that outflow
from the Fraser River was transported across the southern strait
and/or north along the eastern margin of the central strait. The
strong relationship between the abundance of stages 12 larvae and
Ss in both surveys suggests that near-surface residing larvae released
within the vicinity of the Fraser River were transported with this
low salinity signal. Although vertical distributions were not characterized in the present study, this transport scenario seems plausible given
that cancrid zoeae have been demonstrated to be largely confined to
the upper 30 m (Reilly, 1983; Shanks, 1986; Wing et al., 1998) and that
early stages of decapod larvae are typically located near the surface
(Queiroga and Blanton, 2005).

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 6. Relative stage-specic abundance and coefcient of variation


for C. magister (left panels) and C. productus (right panels) for surveys 1
(black) and 2 (grey). nc, not calculated.

Stages 3 4 larvae of C. magister were probably in the water


column for 40 50 days (Sulkin and McKeen, 1996); therefore,
their distribution was expected to be primarily a result of larval
transport rather than larval release. The relatively high amount of
explained variation in the the abundance of stages 3 4 accounted
for by fraction [a] in both surveys suggests that transport processes
were responsible for generating non-random horizontal patterns
decoupled from hydrographic variables. For example, the strong
relationship of stages 1 2 with Ss was no longer significant for
stages 3 4. In S1, fractions [a] and [c] were approximately the
same magnitude, indicating patterns of larvae and hydrographic
variables can be both correlated and complex. In S2, the cross-strait
abundance gradient of both stages 1 2 and 3 4 suggests that larvae
were retained along the central straits eastern margin.
Compared with C. magister, less information is available on
the distribution of adults of C. productus. Variation in abundance
of larvae of C. productus accounted for by fraction [b] in S1 was
high due to the spatial overlap between larvae, spatial variables,
and hydrographic variables. Larvae were most abundant in the
northwest region, an area where water properties are influenced
by input from the Johnstone Strait and largely unaffected by
freshwater input from the Fraser River (Thomson, 1981). Given
that C. productus was localized to the straits northwest region and
C. magister to the central and southern regions, it is not surprising
that larvae of C. productus and stages 1 2 of C. magister were both
significantly associated with Ss, Td and Od, but were correlated
with each variable in opposing directions. Abundance of the first
and second stages of C. productus was highest the northwest
region in both surveys indicating that important larval release
locations occur in this region for this species. In S2, the absence of
a significant relationship between larval abundance and hydrographic variables was caused by the widespread distribution of
larvae across different environments.
In S2, stages 1 2 of C. productus were characterized by low CV
values. The widespread distribution of these larvae suggests
release locations were less spatially restricted than C. magister.
With the exception of C. productus in S2, CV values decreased as
larvae of both species progressed from the first to the second
stage. A positive relationship between larval stage and dispersion
across sampling sites has been interpreted as dispersal from release
locations, whereas the opposite relationship has been attributed to
the formation of patches (Maynou et al., 2006 and references
therein). In S2, abundance of the third stage of C. productus developed a strong cross-strait gradient similar to that of C. magister
larvae, suggesting that dispersal processes had a similar effect on
the spatial pattern of larval abundance of both species over time.
In accordance with this observation, particle-tracking simulations
have demonstrated that subsurface particles accumulate on the

Larval distributions of the Dungeness crab and red rock crab

Page 9 of 14

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 7. Stage-specic distribution of C. magister from survey 1 (left panels) and survey 2 (right panels). Legend units: larvae 100 m23.

Page 10 of 14

K. A. Sorochan and P. A. Quijon

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 8. Stage-specic distribution of C. productus from survey 1 (left panels) and survey 2 (right panels). Legend units: larvae 100 m23.

Larval distributions of the Dungeness crab and red rock crab

Page 11 of 14

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 9. Morans I correlograms illustrating spatial dependence of C. magister, C. productus, and hydrographic variables from survey 1 (left panels)
and survey 2 (right panels). For C. magister and C. productus (in survey 2 only) black lines indicate Morans I from stages 1 2 combined whereas grey
lines indicate Morans I from stages 3 4 combined. For environmental variables black lines indicate Morans I from surface (2 m) and grey lines
indicate Morans I from the sampling depth (34 m in S1; 48 m in S2). Signicant values are indicated by open squares.

Page 12 of 14

K. A. Sorochan and P. A. Quijon

Table 4. Results from ANOVA testing for the effect of species, survey
and daylight on total larval abundance (all stages combined).
Source
(Intercept)
Species
Survey
Daylight
Species*Survey
Species*Daylight
Survey*Daylight
Species*Survey*Daylight
Residuals

Sum Sq
809.05
10.59
40.86
0.01
14.00
0.15
0.99
0.13
95.95

Df
1
1
1
1
1
1
1
1
228

F value
1 922.38
25.17
97.10
0.01
33.26
0.37
2.35
0.31

Pr (>F)
<0.001
<0.001
<0.001
0.911
<0.001
0.545
0.126
0.58

Signicant effects are indicated in bold.

SoGs eastern side (Snauffer, 2013). This phenomenon did not occur
in S1; however, distributions in both surveys cannot be directly compared due to differences in sampling methodology.
In both surveys, larval abundance of both species was anomalously high at only a few stations, resulting in strongly right-skewed
distributions. The combined effect of vertical swimming and convergent or divergent flows can generate spatial variation by concentrating larvae on small spatial scales (Shanks and Wright, 1987;
Franks, 1992). The presence of small-scale patchiness (meters to
kilometres) could be responsible for the observed high level of
spatial variability and strong abundance gradients between stations.
The lack of significant positive spatial dependence within the first
distance class (distances between stations 10.6 km) indicated
that a finer sampling resolution was required to properly resolve
larval patches. Global measures of spatial dependence have been
shown to decay at scales ranging from 4 16 km for fish eggs and
larvae (Maynou et al., 2006); and tens of metres to kilometres for individual patches of crab larvae (Natunewicz and Epifanio, 2001). At
these scales (metres to tens of kilometres), horizontal patterns of

Species, survey
C. magister, 1
C. magister, 2
C. productus, 1
C. productus, 2

R value
R 0.008
R 20.032
R 20.031
R 20.041

p
p 0.372
p 0.827
p 0.759
p 0.914

planktonic organisms are dependent on the nature of vertical and


horizontal transport processes (Mackas et al., 1985). In the central
SoG, low frequency currents have been estimated to have a spatial
extent of 10 km (Stacey et al., 1987). The presence and absence
of significant positive spatial dependence within the first distance
class for salinity (a passive tracer) and larval abundance,
respectively, further suggests the importance of vertical swimming
behaviour in shaping horizontal larval distributions.
The presence of significant spatial dependence of planktonic
larvae and hydrographic variables in the same distance class has
been used to suggest that physical forcing mechanisms influence
the spatial pattern of larvae and water properties at similar length
scales (Maynou et al., 2006). Results from the present study
suggest that this does not frequently occur for zoeae of C. magister
and C. productus, as overlap of significant (negative) spatial dependence of hydrographic variables and larval abundance only occurred
once for each species.
Sampling during day and night may have introduced unwanted
variability in abundance estimates caused by diel vertical migrations of larvae. The effect of vertical migrations on abundance
could not be properly tested because vertically stratified sampling
was not conducted, and samples were obtained only once from
each location. Nevertheless, the lack of differences in larval abundance and stage composition between samples collected during
daylight and darkness/twilight at least indicated that horizontal
patterns of larval abundance were not dominated by this source
of variability. Several studies have demonstrated that peak abundance of cancrid zoeae in the upper water column occurs during
the evening, early morning, or night (Reilly, 1983; Shanks, 1986;
Hobbs and Botsford, 1992; Park and Shirley, 2005), but cancrid
zoeae have also been demonstrated to be most abundant in the
upper 30 m in the day and night (Reilly, 1983; Wing et al., 1998).
The latter may explain the reduced larval abundance observed in S1
when larvae were sampled from an average depth of 34 m.
However, it should be noted that vertical distributions and migratory
patterns observed elsewhere may not apply to larvae sampled in the
present study. Given the uncertainty in the vertical distribution of
zoeae of C. magister and C. productus, it is impossible to predict the
accuracy of a given sampling methodology for abundance estimation.
Therefore, biases associated with different trawling methods must be
considered when comparing estimations of larval abundance and
spatial distributions between surveys, and may be responsible for
large differences in the abundance and distribution observed
between years. A detailed description of the vertical distribution of
zoeae is critical for interpretation of future work on larval crab distributions in this system.

Supplementary data
Supplementary data are available at ICES Journal of Marine Science
online.

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Figure 10. Box plots (median + interquartile range) of total larval


abundance (all stages combined for each species and survey. Whiskers
extend to 1.5 of interquartile range. S1, survey 1; S2, survey 2.

Table 5. Results from ANOSIM tests comparing stage resemblance


between darkness and daylight for C. magister and C. productus in
each survey.

Larval distributions of the Dungeness crab and red rock crab

Acknowledgements
We thank J. Dower and A. Metaxas for providing lab space, support,
and equipment. Comments from R. Allen, H. Hunt, A. Metaxas,
G. Pohle and K. Teather and two anonymous reviewers improved
earlier versions of the manuscript. We are also grateful for assistance
provided by I. Beverage, D. Grundle, L. Guan, E. Jenkins,
K. Meier, N. Philip, J. Rose, K. Suchy, K. Young and R. Vanderstichel.

Funding
Funding for this project was provided by the Canadian Healthy
Ocean Network (CHONe) and a Natural Sciences and
Engineering Research Council (NSERC) Discovery Grant.

References

Aegean Sea (eastern Mediterranean) during early summer. Estuarine,


Coastal and Shelf Science, 79: 607619.
Jamieson, G., and Phillips, A. 1993. Megalopal spatial distribution and
stock separation in Dungeness crab (Cancer magister). Canadian
Journal of Fisheries and Aquatic Sciences, 50: 416 429.
Johannessen, S. C., and McCarter, B. 2010. Ecosystem status and trends
report for the Strait of Georgia Ecozone. DFO Canadian Science
Advisory Secretariat Research Document 2010/010. vi +45 p.
Knudsen, J. W. 1964. Observations of the reproductive cycles and
ecology of the common Brachyura and crab-like Anomura of
Puget Sound, Washington. Pacific Science, 17: 3 33.
Kritzer, J., and Sale, P. 2004. Metapopulation ecology in the sea: from
Levins model to marine ecology and fisheries science. Fish and
Fisheries, 5: 131 140.
Legendre, P., and Fortin, M. 1989. Spatial pattern and ecological
analysis. Vegetatio, 80: 107 138.
Legendre, P., and Legendre, L. 2012. Numerical Ecology, 3rd edn, pp.
570 583. Elsevier, Amsterdam. 990 pp.
Lough, R. 1975. Dynamics of crab larvae (Anomura, Brachyura) off the
central Oregon coast, 1969 1971. PhD dissertation, Oregon State
University, Corvallis. 299 pp.
Mackas, D., Denman, K., and Abbott, M. 1985. Plankton patchiness:
biology in the physical vernacular. Bulletin of Marine Science, 37:
652 674.
Mac Nally, R. 2002. Multiple regression and inference in ecology and
conservation biology: further comments on identifying predictor
variables. Biodiversity & Conservation, 11: 1397 1401.
Masson, D., and Cummins, P. 2004. Observations and modeling of
seasonal variability in the Straits of Georgia and Juan de Fuca.
Journal of Marine Research, 62: 491 516.
Mayer, D. 1973. The ecology and thermal sensitivity of the Dungeness
crab, Cancer magister, and related species of its benthic community
on Similk Bay, Washington. PhD dissertation, University of
Washington, Seattle. 188 pp.
Maynou, F., Olivar, M., and Emelianov, M. 2006. Patchiness of eggs,
larvae and juveniles of European hake Merluccius merluccius from
the NW Mediterranean. Fisheries Oceanography, 15: 390 401.
Metaxas, A., and Saunders, M. 2009. Quantifying the bio- components in biophysical models of larval transport in marine benthic
invertebrates: advances and pitfalls. The Biological Bulletin, 216:
257 272.
Natunewicz, C., and Epifanio, C. 2001. Spatial and temporal scales of
patches of crab larvae in coastal waters. Marine Ecology Progress
Series, 212: 217 222.
Orensanz, J., and Galluchi, V. 1988. Comparative study of postlarval
life-history schedules in four sympatric species of Cancer
(Decapoda: Brachyura: Cancridae). Journal of Crustacean Biology,
8: 187 220.
Park, W., and Shirley, T. 2005. Diel vertical migration and seasonal
timing of the larvae of three sympatric cancrid crabs, Cancer spp.,
in southeastern Alaska. Estuaries, 28: 266 273.
Parks Canada. 2012. Feasibility study for the proposed southern Strait of
Georgia National Marine Conservation Area Reserve. http://www.
pc.gc.ca/eng/progs/amnc-nmca/dgs-ssg/index.aspx (last accessed
14 May 2014).
Peliz, A., Marchesiello, P., Dubert, J., Marta-Almeida, M., Roy, C., and
Queiroga, H. 2007. A study of crab larvae dispersal on the Western
Iberian Shelf: Physical processes. Journal of Marine Systems, 68:
215 236.
Poole, R. 1966. A description of laboratory-reared zoeae of Cancer
magister Dana, and megalopae taken under natural conditions
(Decapoda Brachyrua). Crustaceana, 11: 83 97.
Puls, A. 2001. An Identification Guide to the Larval Marine Invertebrates
of the Pacific Northwest, pp. 179250. Ed. by A. Shanks. Oregon State
University Press, Corvallis. 314 pp.

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Ally, J. 1975. A description of the laboratory-reared larvae of Cancer gracilis Dana, 1852 (Decapoda, Brachyura). Crustaceana, 28: 231 246.
Armstrong, D. A., Armstrong, J. L., and Dinnel, P. A. 1987. Ecology and
population dynamics of Dungeness crab, Cancer magister, in Ship
Harbour, Anacortes, Wa. Final Report to Leeward Development
Company and Washington State Department of Fisheries,
University of Washington College of Fisheries, Seattle Washington
(FRI-UW-8701).
Ayata, S., Stolba, R., Gomtet, T., and Thiebaut, E. 2011. Meroplankton
distribution and its relationship to coastal mesoscale hydrological
structure in the northern Bay of Biscay (NE Atlantic). Journal of
Plankton Research, 33: 1193 1211.
Bocard, D., Legendre, P., and Drapeau, P. 1992. Partialling out the spatial
component of ecological variation. Ecology, 73:1045 1055.
Burnham, K., and Anderson, D. 2002. Model Selection and Multimodel Inference: a Practical Information Theoretic Approach, 2nd
edn, pp. 322 326. Springer-Verlag, New York. 488 pp.
Chevan, A., and Sutherland, M. 1991. Hierarchical partitioning. The
American Statistician, 45: 90 96.
Christensen, R. 1992. Comment on Chevan and Sutherland. The
American Statistician, 46: 74.
Cowen, R., and Sponaugle, S. 2009. Larval dispersal and marine population connectivity. Annual Review of Marine Science, 1: 443 466.
DeBrosse, G., Sulkin, S., and Jamieson, G. 1990. Intraspecific morphological variability in megalopae of three sympatric species of the
genus Cancer (Brachyura: Cancridae). Journal of Crustacean
Biology, 10: 315 329.
DFO. 2013a. Economics of the fishery. In Pacific region integrated fisheries management plan, crab by trap, January 1, 2014 to December
31, 2014, pp. 12 17. http://www.dfo-mpo.gc.ca/Library/350564.
pdf (last accessed 21 April 2014).
DFO. 2013b. Rockfish Conservation Areas (RCAs) - Pacific. http://www.
pac.dfo-mpo.gc.ca/fm-gp/maps-cartes/rca-acs/index-eng.html (last
accessed 14 May 2014).
DFO. 2013c. A synthesis of the outcomes from the Strait of Georgia
ecosystem research initiative, and development of an ecosystem
approach to management. Canadian Science Advisory Secretariat
Science Advisory Report 2012/072.
Fisher, J. 2006. Seasonal timing and duration of brachyuran larvae in a
high-latitude fjord. Marine Ecology Progress Series, 323: 213 222.
Fortin, M., and Dale, M. 2005. Spatial Analysis: a Guide for Ecologists,
pp. 111 173. Cambridge University Press, New York. 265 pp.
Forward, R. 1989. Behavioural responses of crustacean larvae to rates of
salinity change. The Biological Bulletin, 176: 229 238.
Franks, P. 1992. Sink or swim: accumulation of biomass at fronts.
Marine Ecology Progress Series, 82: 1 12.
Hobbs, R., and Botsford, L. 1992. Diel vertical migration and timing of
metamorphosis of larvae of the Dungeness crab Cancer magister.
Marine Biology, 112: 417 428.
Isari, S., Fragopoulu, N., and Somarakis, S. 2008. Interranual variability
in horizontal patterns of larval fish assemblages in the northeastern

Page 13 of 14

Page 14 of 14

Sorochan, K. 2011. Spatial and temporal patterns of decapod larvae in


the Strait of Georgia, British Columbia. Masters thesis, University
of Prince Edward Island. 200 pp.
Stacey, M., Pond, S., Leblond, P., Freeland, H., and Farmer, D. 1987. An
analysis of low-frequency current fluctuations in the Strait of
Georgia, from June 1984 until January 1985. Journal of Physical
Oceanography, 17: 326 342.
Stevens, B., Armstrong, D., and Cusimano, R. 1982. Feeding habits of
the Dungeness crab Cancer magister as determined by the index of
relative importance. Marine Biology, 72: 135 145.
Sulkin, S., and McKeen, G. 1994. Influence of temperature on larval development of four co-occurring species of the brachyuran genus
Cancer. Marine Biology, 118: 593 600.
Sulkin, S., and McKeen, G. 1996. Larval development of the crab Cancer
magister in temperature regimes simulating outer-coast and inlandwater habitats. Marine Biology, 127: 235 240.
Sulkin, S., Mojica, E., and McKeen, G. 1996. Elevated summer temperature effects on megalopal and early juvenile development in the
Dungeness crab, Cancer magister. Canadian Journal of Fisheries
and Aquatic Sciences, 53: 2076 2079.
Thomson, R. 1981. Oceanography of the British Columbia Coast.
Canadian Special Publication of Fisheries and Aquatic Sciences,
56: 291 pp.
Trask, T. 1970. A description of laboratory-reared larvae of Cancer productus Randall (Decapoda, Brachyura) and a comparison to larvae of
Cancer magister Dana. Crustaceana, 18: 133 146.
Wing, S., Botsford, L., Ralston, S., and Largier, J. 1998. Meroplanktonic
distribution and circulation in a coastal retention zone of the northern California upwelling system. Limnology and Oceanography,
43: 1710 1721.
Yamada, S., and Boulding., E. 1996. The role of highly mobile crab
predators in the intertidal zonation of their gastropod prey.
Journal of Experimental Marine Biology and Ecology, 204:
59 83.

Handling editor: Claire Paris

All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.

Downloaded from http://icesjms.oxfordjournals.org/ at University of Prince Edward Island on June 7, 2014

Queiroga, H., and Blanton, J. 2005. Interactions between behaviour and


physical forcing in the control of horizontal transport of decapod
crustacean larvae. Advances in Marine Biology, 47: 107 214.
Rasmuson, L. 2012. The biology, ecology and fishery of the Dungeness
crab, Cancer magister. Advances in Marine Biology, 85: 95 148.
Reed, P. 1969. Culture methods and effects of temperature and salinity
on survival and growth of Dungeness crab (Cancer magister) larvae
in the laboratory. Journal of the Fisheries Research Board of
Canada, 26: 389 397.
Reilly, P. 1983. Dynamics of Dungeness crab, Cancer magister, larvae off
central and northern California. In Life History, Environment, and
Mericulture Studies of the Dungeness Crab, Cancer magister, with
Emphasis on the Central California Fishery Resource. Ed. by P. W.
Wild, and R. N. Tasto. Californian Department of Fish and Game
Fish Bulletin, 172: 57 84.
Rice, A., and Tsukimura, B. 2007. A key to the identification of brachyuran zoeae of the San Francisco Bay estuary. Journal of
Crustacean Biology, 27: 74 79.
Sale, P., and Kritzer, J. 2003. Determining the extent and spatial scale of
population connectivity: decapods and coral reef fishes compared.
Fisheries Research, 65: 153 172.
Shanks, A. 1986. Vertical migration and cross-shelf dispersal of larval
Cancer spp. and Randallia ornata (Crustacea: Brachyura) off the
coast of southern California. Marine Biology, 92: 189 199.
Shanks, A. 2009. Pelagic larval duration and dispersal distance revisted.
The Biological Bulletin, 216: 373 385.
Shanks, A., and Wright, W. 1987. Internal-wave-mediated transport
of cyprids, megalopae, and gammarids and correlated longshore
differences in the settling rate of intertidal barnacles. Journal of
Experimental Marine Biology and Ecology, 114: 1 13.
Shirley, S. M., Shirley, T. C., and Rice, S. D. 1987. Latitudinal variation in
the Dungeness crab, Cancer magister: zoeal morphology explained
by incubation temperature. Marine Biology, 95: 371 376.
Snauffer, E. 2013. Modeling herring and hake larval dispersal in the
Salish Sea. Masters thesis, University of British Columbia. 112 pp.

K. A. Sorochan and P. A. Quijon

You might also like