Full text of "Fishery bulletin"

See other formats

U.S. Department
of Commerce

Volume 108
Number 2
April 2010

U.S. Department
of Commerce

Gary Locke

Secretary of Commerce

National! Oceanic
and Atmospheric
Administration

Jane Lubchenco, Ph.D.

Administrator of NOAA

National Marine
Fisheries Service

James W. Balsiger, Ph.D.

Acting Assistant Administrator
for Fisheries

The Fishery Bulletin (ISSN 0090-0656)
is published quarterly by the Scientific
Publications Office, National Marine
Fisheries Service, NOAA, 7600 Sand
Point Way NE, BIN C15700, Seattle, WA
98115-0070. Periodicals postage is paid
at Seattle, WA. POSTMASTER: Send
address changes for subscriptions to Fish-
ery Bulletin, Superintendent of Docu-
ments, Attn.: Chief, Mail List Branch,
Mail Stop SSOM, Washington, DC 20402-
9373.

Although the contents of this publica-
tion have not been copyrighted and may
be reprinted entirely, reference to source
is appreciated.

The Secretary of Commerce has deter-
mined that the publication of this peri-
odical is necessary according to law for
the transaction of public business of this
Department. Use of funds for printing of
this periodical has been approved by the
Director of the Office of Management and
Budget.

For sale by the Superintendent of
Documents, U.S. Government Printing
Office, Washington, DC 20402. Subscrip-
tion price per year: $36.00 domestic and
$50.40 foreign. Cost per single issue:
$21.00 domestic and $29.40 foreign. See
back for order form.

Scientific Editor

Richard D. Brodeur, Ph.D.

Associate Editor
Julie Scheurer

National Marine Fisheries Service
Northwest Fisheries Science Center
2030 S. Marine Science Dr.
Newport, Oregon 97365-5296

Managing Editor

Sharyn Matriotti

National Marine Fisheries Service
Scientific Publications Office
7600 Sand Point Way NE
Seattle, Washington 98115-0070

Editorial Committee

John Carlson
Kevin Craig
Jeff Leis
Rich McBride
Rick Methot
Adam Moles
Frank Parrish
Dave Somerton
Ed Trippel
Mary Yoklavich

National Marine Fisheries Service, Panama City, Florida

Florida State University, Tallahassee, Florida

Australian Museum, Sydney, New South Wales, Australia

National Marine Fisheries Service, Woods Hole, Massachusetts

National Marine Fisheries Service, Seattle, Washington

National Marine Fisheries Service, Auke Bay, Alaska

National Marine Fisheries Service, Honolulu, Hawaii

National Marine Fisheries Service, Seattle, Washington

Department of Fisheries and Oceans, St. Andrews, New Brunswick, Canada

National Marine Fisheries Service, Santa Cruz, California

Fishery Bulletin web site: www.fisherybulletin.noaa.gov

The Fishery Bulletin carries original research reports and technical notes on investigations in
fishery science, engineering, and economics. It began as the Bulletin of the United States Fish
Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery
Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through
volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing
through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system
began in 1963 with volume 63 in which papers are bound together in a single issue of the
bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a
periodical, issued quarterly. In this form, it is available by subscription from the Superintendent
of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in
limited numbers to libraries, research institutions, State and Federal agencies, and in exchange
for other scientific publications.

U.S. Department
of Commerce

Seattle, Washington

Volume 108
Number 2
April 2010

Fishery

Bulletin

Contents

Articles

119-135 Tribuzio, Cindy A., Gordon H. Kruse, and Jeffrey T. Fujioka

Age and growth of spiny dogfish (Squalus acanthias)
in the Gulf of Alaska: analysis of alternative growth models

Companion articles

136-144 Rose Craig S., Carwyn F. Hammond, and John R. Gauvin

Effective herding of flatfish by cables with minimal seafloor contact

145-154 Ryer, Clifford H., Craig S. Rose, and Paul J. Iseri

Flatfish herding behavior in response to trawl sweeps:
a comparison of diel responses to conventional sweeps
and elevated sweeps

The National Marine Fisheries
Service (NMFS) does not approve,
recommend, or endorse any proprie-
tary product or proprietary material
mentioned in this publication. No
reference shall be made to NMFS,
or to this publication furnished by
NMFS, in any advertising or sales
promotion which would indicate or
imply that NMFS approves, recom-
mends, or endorses any proprietary
product or proprietary material
mentioned herein, or which has
as its purpose an intent to cause
directly or indirectly the advertised
product to be used or purchased
because of this NMFS publication.

The NMFS Scientific Publications
Office is not responsible for the con-
tents of the articles or for the stan-
dard of English used in them.

155-161 Mateo, Ivan, Edward G. Durbin, David A. Bengtson,

Richard Kingsley, Peter K. Swart, and Daisy Durant

Spatial and temporal variation in otolith chemistry for tautog
( Toutoga onitis) in Narragansett Bay and Rhode Island coastal ponds

162-173 Masuda, Reiji, Masami Shiba, Yoh Yamashita, Masahiro Ueno,
Yoshiaki Kai, Asami Nakanishi, Masaru Torikoshi,
and Masaru Tanaka

Fish assemblages associated with three types of artificial reefs: density
of assemblages and possible impacts on adiacent fish abundance

174-192 Lo, Nancy C. H., Beverly J. Macewicz, and David A. Griffiths

Biomass and reproduction of Pacific sardine (Sardinops sagax)
off the Pacific northwestern United States, 2003-2005

193-207 Hernandez Jr., Frank J., Sean P. Powers, and William M .Graham
Seasonal variability in ichthyoplankton abundance and assemblage
composition in the northern Gulf of Mexico off Alabama

Fishery Bulletin 108(2)

208-217

Stevenson, Duane E., and Kristy A. Lewis

Observer-reported skate bycatch in the commercial groundfish fisheries of Alaska

218-225

Fergusson, Emily A., Molly V. Sturdevant, and Joseph A. Orsi

Effects of starvation on energy density of juvenile chum salmon (Oncorhynchus keta)
captured in marine waters of Southeastern Alaska

226-232

Fruh, Erica L., Aimee Keller, Jessica Trantham, and Victor Simon
Accuracy of sex determination for northeastern Pacific Ocean thornyheads
( Sebastolobus altivelis and 5. alascanus)

233-247

Jacobson, Larry D., Kevin D. E. Stokesbury, Melissa A. Allard, Antonie Chute, Bradley P. Harris,
Deborah Hart, Tom Jaffarian, Michael C. Marino II, Jacob 1. Nogueira, and Paul Rago

Measurement errors in body size of sea scallops ( Placopecten magellanicus)
and their effect on stock assessment models

248

Guidelines for authors
Subscription form (inside back cover)

119

Age and growth of spiny dogfish
{Squalus accmthias ) in the GuSff of Alaska:
analysis of alternative growth models

Cindy A. Tribuzio (contact author)1
Gordon H. Kruse1
Jeffrey T. Fujioka2

Email address for contact author: cindy.tribuzio@noaa.gov

1 School of Fisheries and Ocean Sciences, Juneau Center
University of Alaska Fairbanks

17101 Pt. Lena Loop Road
Juneau, Alaska 99801

Present address for contact author: National Oceanic and Atmospheric Administration

National Marine Fisheries Service
Alaska Fisheries Science Center
Auke Bay Laboratories
17109 Pt. Lena Loop Road
Juneau, Alaska 99801

2 National Oceanic and Atmospheric Administration
National Marine Fisheries Service

Alaska Fisheries Science Center
Auke Bay Laboratories
17109 Pt. Lena Loop Road
Juneau, Alaska 99801

Abstract — Ten growth models were
fitted to age and growth data for spiny
dogfish ( Squalus acanthias ) in the
Gulf of Alaska. Previous studies of
spiny dogfish growth have all fitted
the t0 formulation of the von Berta-
lanffy model without examination
of alternative models. Among the
alternatives, we present a new two-
phase von Bertalanffy growth model
formulation with a logistically scaled
k parameter and which estimates L0.
A total of 1602 dogfish were aged
from opportunistic collections with
longline, rod and reel, set net, and
trawling gear in the eastern and cen-
tral Gulf of Alaska between 2004 and
2007. Ages were estimated from the
median band count of three indepen-
dent readings of the second dorsal
spine plus the estimated number of
worn bands for worn spines. Owing to
a lack of small dogfish in the samples,
lengths at age of small individuals
were back-calculated from a subsam-
ple of 153 dogfish with unworn spines.
The von Bertalanffy, two-parameter
von Bertalanffy, two-phase von Ber-
talanffy, Gompertz, two-parameter
Gompertz, and logistic models were
fitted to length-at-age data for each
sex separately, both with and without
back-calculated lengths at age. The
two-phase von Bertalanffy growth
model produced the statistically best
fit for both sexes of Gulf of Alaska
spiny dogfish, resulting in L00 = 87.2
and 102.5 cm and £ = 0.106 and 0.058
for males and females, respectively.

Manuscript submitted 17 February 2009.
Manuscript accepted 3 November 2009.
Fish. Bull. 108:119-135 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

The spiny dogfish (Squalus acanthias )
is a small long-lived shark common
among temperate coastal areas in
the Atlantic and Pacific oceans (Com-
pagno, 1984). This species has been
the target of commercial fisheries over
much of its range, in some cases for
over a century (Ketchen, 1986). In
some areas, severe declines in popu-
lation abundance and stock structure
have occurred (e.g., Rago et al., 1998).
Many elasmobranchs, including spiny
dogfish, are “equilibrium strategists”
that are highly susceptible to over-
fishing because of their slow growth
rates, low fecundity, and late matu-
ration (King and McFarlane, 2003),
all of which are directly related to
recruitment and parental stock sizes
(Holden, 1974; 1977). Off the west
coast of North America, spiny dogfish
were depleted by intense fisheries in
the 1940s, owing to the quantity and
quality of vitamin A in their livers
(Ketchen, 1986); the fishery demand
decreased by 1950 with the develop-
ment of synthetic vitamin A (Ketchen
et al., 1983). Since the 1970s, spiny
dogfish have continued to be targeted
by commercial fisheries in British

Columbia and the state of Washing-
ton for human consumption.

Although not targeted, spiny dog-
fish is a common bycatch species in
many fisheries in both state and fed-
eral waters off the coast of Alaska.
In the Gulf of Alaska (GOA) spiny
dogfish are taken in Pacific salmon
(Oncorhynchus spp.) gillnet fisher-
ies, sablefish ( Anoplopoma fimbria)
fisheries, Pacific halibut (Hippoglos-
sus stenolepis) longline fisheries, and
groundfish trawl fisheries (Boldt,
2003). Although an estimated aver-
age of 482.1 metric tons (t) of spiny
dogfish was taken annually from
1997 to 2007 in observed fisheries
(Tribuzio et al., 2008), the bycatch
in state waters is unknown and the
bycatch rates in federally managed
fisheries are likely underestimated
because of unobserved fisheries (e.g.,
the halibut individual fishing quota,
IFQ). Nearly all of this unintended
bycatch was and still is discarded at
sea. Even though estimated catch is
<1% of estimated spiny dogfish bio-
mass (Courtney et al., 2006), the po-
tential development of a commercial
fishery demands further investigation

120

Fishery Bulletin 108(2)

of the effect of total fishing mortality on biomass and
an investigation of spiny dogfish life history character-
istics in Alaska.

Biological reference points (e.g., BMS Y, F35%) are
benchmarks against which stock abundance or fishing
mortality rates can be compared to determine stock
status. Most commonly used reference points are func-
tions of stock productivity, such as growth, recruitment,
and natural mortality (Bonfil, 2005); thus accurate es-
timates of age and growth are important. For instance,
estimates of age and the growth coefficient (k) are criti-
cal for estimating natural mortality (M), where a lack
of data prevent direct estimation of M, abundance, and
appropriate harvest rates. In the GOA, biological refer-
ence points, such as those from age and growth models,
have yet to be determined for spiny dogfish.

Extension of life history parameters from other re-
gions to Alaska may be inappropriate because age and
growth characteristics of spiny dogfish vary widely over
its geographic range. For example, maximum age in the
northwest Atlantic Ocean is 35-40 years (Nammack
et al., 1985), but in the eastern North Pacific, spiny
dogfish have been aged to over 80 years (Saunders and
McFarlane, 1993). Growth characteristics also vary
widely throughout the North Pacific and North Atlantic
oceans (Ketchen, 1975; Nammack et al., 1985). Even
within the North Pacific basin, biological parameters,
such as k, can vary with latitude (Vega, 2006).

The selection of an appropriate growth model is im-
portant when estimating regionally specific parameters.
Elasmobranch age and growth studies have generally
focused on fitting length-at-age data to the von Berta-
lanffy (vB) growth equation, irrespective of goodness-
of-fit or alternative growth models (Carlson and Bare-
more, 2005). Despite its common use, the vB growth
equation may not be the best-fit growth model for all
elasmobranch species. For example, the logistic model
fitted best among four models tested for the spinner
shark ( Carcharhinus brevipinna, Carlson and Baremore,
2005), and a two-phase vB model fitted best among five
models for the piked spurdog ( Squalus megalops, Brac-
cini et al., 2007). A model that is not the best descriptor
of a species’ growth could have compounding effects on
demographic analyses, stock assessment, and fishery
management.

Typical growth models involve parameters of asymp-
totic length (L^), k, and t0 (Cailliet et ah, 2006). The t0
parameter is biologically difficult to interpret because
it is not measurable and testable in wild animals (Be-
verton and Holt, 1957). This parameter is the age at
which the animal is of zero length and is based on an
assumption of a fixed growth curve from fertilization
through life (Beverton and Holt, 1957). It is generally
interpreted to represent the period of gestation in tele-
ost fish species, but this assumption is violated for elas-
mobranchs (Driggers et al., 2004). For instance, when
considering males and females separately, models will
estimate different t0 values. If f0 is truly representative
of gestation time, then it leads to the incorrect infer-
ence that male and female pups have different gestation

periods. For these reasons, growth models that use size
at birth (L0) instead of t0 may be more appropriate for
elasmobranchs (Cailliet and Goldman, 2004).

The purpose of this study was to estimate best-fit
growth models for male and female spiny dogfish in
the GOA. Resultant growth equations provide critical
parameters for a better understanding of spiny dogfish
biology, estimation of biological reference points includ-
ing indirect estimates of M, improved stock assess-
ments, and development of sound fishery management
plans for this species in waters off Alaska.

Materials and methods
Sample collection

Spiny dogfish were collected by targeted sampling
cruises, state and federal assessment surveys, and oppor-
tunistic fishery bycatch samples between July 2004 and
April 2007 across the GOA (Fig. 1, Table 1 (delete bold
font after placing tables). All spiny dogfish were sexed
and length was measured to the nearest centimeter
(total length extended=T,Lej.,; total length natural^TL^;
precaudal length=PCL; and fork length=FL; Tribuzio et
al., 2009). Here, length measurements are reported as
total length extended (TLext). The posterior dorsal spine
was removed and stored frozen for laboratory analyses.
In the laboratory, spines were cleaned by thawing, by
boiling briefly, and the loose tissue was scraped free.
Spines were allowed to dry overnight and then stored in
individual paper envelopes for subsequent age reading.

Sampling bias was examined because we sampled
with multiple gear types in different locations. To test
for potential bias, a chi-squared (x2) test was conducted
to test for statistically significant (P<0.05) differences
in the mean length at age by sex for each gear (trawl,
setnet, longline, rod and reel) and region (Cook Inlet,
Prince William Sound, Yakutat Bay, and Gulf of Alas-
ka). Statistically significant differences among different
gears would provide evidence of sampling bias. However,
statistically significant differences among different geo-
graphic areas would provide equivocal evidence of bias
because the possibility of true underlying differences in
size distributions by area could not be dismissed.

Age determinations

The posterior dorsal spines were read in the laboratory
according to the methods of Ketchen (1975) and Beamish
and McFarlane (1985). Each band pair (hereafter termed
“band”), consisting of one dark and one light band, was
counted as one year or annulus (Cailliet et al., 2006).
Aging was conducted by two scientists at the Washington
Department of Fish and Wildlife’s age laboratory and by
the lead author at the University of Alaska Fairbankans.
Ease of age reading was categorized from 1 (easiest) to 3
(most difficult). Spines were photographed on a lxl mm
grid to standardize measurements. All measurements
were rounded to the nearest 0.01 mm by using Bersoft

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

121

Locations where spiny dogfish ( Squalus acanthias ) were sampled in the Gulf
of Alaska in 2004-07. The size of the circle is proportional to the number of
spiny dogfish sampled at each location.

Table 1

Locations, gear types, and sample sizes for male and female spiny dogfish (Squalus acanthias) collected during 2004-07. “Sport”
gear refers to hook-and-line fishing with rod and reel, “longline” refers to multiple hooks on a groundline, “trawl” denotes either
bottom or pelagic trawls, and “set net” refers to a stationary floating gill net, generally anchored at one end to the shore.

Year

Area

Gear

Males (n)

Females ( n )

2004

Yakutat Bay

Sport

2004

Gulf of A laska (GOA)

Longline

2005

Southeast Alaska (SEAK)

Longline

2005

Yakutat Bay

Longline

2005

Yakutat Bay

Sport

2005

Cook Inlet

Sport

2005

Yakutat Bay

Longline

2005

GOA

Longline

112

204

2005

Cook Inlet

Sport

2005

Yakutat Bay

Sport

2005

Prince William Sound

Longline

2005

GOA

Trawl

125

2006

Kamishak Bay

Trawl

2006

Cook Inlet

Set net

2006

Copper River

Set net

2006

Yakutat Bay

Set net

2006

Icy Point (SEAK)

Trawl

2006

Prince William Sound

Longline

2006

Cherikoff Island (SW GOA)

Trawl

2007

Cherikoff Island (SW GOA)

Trawl

122

Fishery Bulletin 108(2)

EBD
—

Figure 2

Measurements taken on spiny dogfish ( Squalus acanthias) spines.
Last readable point (LRP) is the point where the bands are no longer
visible on the leading edge of the spine (upper edge in this picture).
EBD = enamel base diameter, SBD = spine base diameter, BL = base
length, and TL = spine total length, which only applies to spines that
are unworn. All measurements were taken in millimeters.

Image Measurement vers 5.0 software
(Bersoft, Inc., http://bersoft.com). Mea-
surements included spine base diameter
(SBD), enamel base diameter (EBD), last
readable point (LRP, also called the no-
wear point); and, for nonworn spines, base
length (BL), and spine total length (TL,

Fig. 2) were also measured to the near-
est 0.01 mm. Nonworn spines were those
spines with a LRP<2Ab mm (McFarlane
and King, 2009), which is the EBD at
birth.

Aging bias and precision were evalu-
ated for all three readers. Pair-wise age-
bias plots were used to compare each
reader against the other two (Campana
et al., 1995) and a %2 test for symmetry
was used to test for statistically significant systematic
bias among the three readers (Hoenig et al., 1995).
Readers were considered to be in agreement when ages
were within 10% of each other rather than within some
fixed 1- or 2-year age interval. For instance, if reader
X counted 10 bands, then reader Y’s count would have
to have been between 9-11 bands to be in agreement,
but if reader X counted 40 bands, then reader Y’s count
would have to be between 36-44 to be in agreement.
We contend that the use of a percentage to define the
interval size is more appropriate for this long-lived spe-
cies. Finally, the coefficient of variation (CV) between
readers was calculated according to Campana’s methods
(2001).

Spiny dogfish ages are not always equal to the num-
ber of counted bands for two reasons: 1) bands are de-
posited during embryonic development, and 2) because
the external spines can become worn or can break off.
This problem was addressed by a correction method
for estimating the number of missing bands that was
based on a regression of band counts on the SBD of
unworn spines (Ketchen, 1975). This method was sub-
sequently re-examined and accepted as the best avail-
able method for the original samples plus additional
samples from the same geographic region (McFarlane
and King, 2009).

Various regression approaches were compared to de-
termine which method resulted in the best model for
estimating the number of worn bands in spiny dog-
fish collected from the GOA, including: nonlinear least
squares regression (NLS, Eq. 1), and ordinary least
squares (OLS, Eq. 2):

Band count = b0EBDt>1 (1)

In (Band count) = ln(60) + ln(EBD)61, (2)

where b0 and bx are estimated parameters (based on
Ketchen 1975, McFarlane and King 2009). Also, we
fitted parameters for Equations. 1 and 2 with weighted
nonlinear least squares (WNLS) and weighted ordi-
nary least squares (WOLS), where weights were applied
to the residuals as follows: spines in readability cat-

egory 1 were given a weight of 1, those in category 2
were weighted by 0.5, and those in category 3 by 0.3.
These values were chosen to discount the contribution
of individual length at-age data points to the estimation
process based on the degree of uncertainty in the age
estimates for difficult-to-read spines. As an alternative
to this weighting scheme, we explored the weighting
process by using the inverse of the variance in assigned
ages for each readability category. Ages of worn spines
were then estimated by equating the LRP to the EBD
in the best-fit model from Equations 1-4 and by adding
the resultant number of bands to the median band count
from the three readings and by subtracting two years
(for bands deposited during gestation) to obtain the final
estimated age of the animal (Ketchen, 1975). In the case
of nonworn spines, age was estimated by the median
band count minus two years. Data for males and females
were combined for these worn band models.

Fitting of growth models

A total of 10 growth model variations were fitted sepa-
rately to the length-at-age data for males and females
(Table 2). The growth models included 1) the vB growth
model for estimating /0; 2) the two-parameter vB with
fixed L0; 3) the two-phase vB with L0 (used in the present
study); 4) the Gompertz; 5) the two-parameter Gompertz;
and 6) the logistic. For comparison with previous studies
L0 is estimated for model 1 by setting /=0. An estimate
of L0 (i.e., the size at birth) for GOA spiny dogfish was
not available; therefore model 2 was run with L0 fixed
at 26.2 cm (size at birth for spiny dogfish from British
Columbia; Ketchen, 1972). Models 3 and 5 were run in
three different ways: 1) L0 was estimated by the model;

2) with L0 set at the value estimated from model 1; and

3) with L0 set at 26.2 cm. Model 3 is an adaptation of
the two-phase vB model (Soriano et al., 1992). Standard
fitting procedures with the two-phase model resulted in
the At parameter from Soriano et al. (1992) changing for
a brief time period and then returning to its original
value. To correct this we reformulated the At parameter
from Soriano et al. (1992); this treatment changes k,
depending on the age of the dogfish, so that At would

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

123

Table 2

Growth models fitted to spiny dogfish ( Squalus acanthias) length-at-age (Lt) data. Parameters are: asymptotic length (La), the
growth coefficient (k), length at birth (L0), age at size zero (f0), a phase change parameter (At) for the two-phase model, age at
transition (th), magnitude of the maximum difference between model 1 and the two phase model (h), time increment from previ-
ous t value (5), and the inflection point of the logistic curve (a).

Model number

Model name

Model equation

Reference

vB 1

= Z,oo(l-e"t('"'o))

von Bertalanffy (1938)

vB 2

Fabens (1965)

3a-3c

Two-phase vB with L0

= A-s+b.-Aj*( i-v4-'*""-'’).

This study

1 + /ope«h-<)

Gompertz

= Le 1 1

Ricker (1975)

5a-5c

Two-parameter Gompertz

= V ]>G =

In —
L

Mollet et al. (2002)

Logistic

, -k(t-a)

l + e [ 1

Ricker (1979)

follow a logistic pattern and remain in the second phase.
Another problem we encountered fitting the two-phase
model was that the typical differential form of the vB
equation can result in a decrease in length at the tran-
sition between phases. To prevent this unlikely result
the difference equation form of the vB equation (Gulland
1969) was used in this analysis.

Model parameters for equations describing the num-
ber of worn bands or growth were fitted by nonlinear
least-squares regression or ordinary least-squares re-
gression, and confidence intervals were estimated by
a bootstrap procedure with 5000 replicates by using R
statistical software (R, vers. 2.10.0, www.r-project.org).
Confidence intervals (95%) for parameter estimates
were based on the lower and upper 2.5th percentile of
the bootstrap replications. Parameters were considered
significantly different if the 95% confidence intervals
did not overlap. To evaluate best model fit for the male
and female datasets, Akaike information criteria (AIC)
and model summary statistics were calculated (Burn-
ham and Anderson, 2004).

Fraser-Lee back-calculation methods (Francis, 1990;
Campana, 1990; Goldman et al., 2006). The Fraser-Lee
method produced results that on an individual level
could be quite unreasonable (large negative ages), but
on average were more biologically reasonable than either
of the Dahl-Lea methods. Further, growth model results
with either of the Dahl-Lea methods were unreasonable
(Lx of >150 cm TLext), therefore, we used the Fraser-Lee
method for our data. Thus, the following equation was
used to estimate back-calculated length-at-age data:

TL, = TL, +

[EBD,-EBDc)(TLc-TLbh

birth ,

EBDc - EBDbirth

(3)

where TLi = the back calculated length;

TLC = the length at capture;

TL birth = the length at birth;

EBDi = the enamel base diameter at band /;

EBDc - the enamel base diameter at capture; and
EBDhirth = the enamel base diameter at birth.

Back-calculation methods

Owing to a paucity of specimens with EBD< 3.5 mm,
back-calculation methods were used to fill in the size
range missing from samples. The spine diameter at each
band along the spine (hereafter called “band diameters”)
was measured from a random subsample of 153 unworn
spines for use in the estimation of worn bands (Eqs. 1-4);
spiny dogfish with unworn spines tend to be smaller and
younger than those with worn spines. We examined the
Dahl-Lea, linear Dahl-Lea, and size at birth modified

Results

Sample collection

A total of 1713 spiny dogfish were sampled over the four
years of the study (585 males, 1128 females, Table 1) of
which 537 male and 1062 female spines were usable.
Lengths ranged from 56 to 99 cm TLext for males, and
56 to 123 cm TLext for females. The x2 test revealed no
significant differences between the mean length at age

124

Fishery Bulletin 108(2)

45
40 -
35 -
30 -
25
20 -
15 -
10 -
5
0

10 20 30

Band count reader 1

45
40
35
30 -
25 -
20
15
10 -
5 -
0

10 20 30

Band count reader 3

Figure 3

A comparison of age counts among readers. (A) Reader 2’s mean band counts (y-axis) in
relation to the band counts of reader 1; (B) Reader 3’s mean band counts in relation to
the band counts of reader 1; and (C) Reader 2’s mean band counts in relation to the band
counts of reader 3. Vertical lines are 95% confidence intervals and the diagonal line is
the 1:1 relationship line. (D) Percent agreement and coefficient of variation for reader 2
(Rd 2) compared to reader 1. The percent agreement (±10%) is represented by the solid
line and circles and the coefficient of variation (CV) by the dashed line and open circles.
(E) Percent agreement and coefficient of variation of reader 3 (Rd 3) compared to reader
1; and (F) Percent agreement and coefficient of variation of reader 2 (Rd 2) compared to
those of reader 3.

of any of the data groupings (P>0.99, 0.019<x2<4.525).
Thus, we failed to find evidence of sampling bias or
geographic differences in average size at age.

Age determinations

Sampled dogfish ranged in age from 8 to 50 years old.
The x2 test and the age-bias plots indicated no signifi-

cant systematic bias between the three readers (x2=241,
206, and 259 between readers 2 and 1, readers 2 and
3, and readers 3 and 1, respectively; all P>0.05; Fig.
3, A-C). The percent agreement between readers 2
and 1 (Fig. 3D) and readers 3 and 1 (Fig. 3E) was high
for band counts less than 30 but was more variable
or decreased for band counts greater than 30 (Fig. 3,
D-F). For readers 2 and 3, the percent agreement was

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

125

Table 3

Summary of the parameters used in the worn-band estimation models and model fits for spiny dogfish (Squalus acanthias). The
observed data are sample data, the band-diameter data were determined from a subsample of unworn spines where the diam-
eter of each band was measured to simulate bound count at spine size for younger animals that were not sampled in this study.
Regression models are ordinary least squares (OLS), weighted ordinary least squares (WOLS), nonlinear least squares (NLS)
and weighted nonlinear least squares (WNLS). Estimated model parameters (95% confidence intervals in parentheses) and
goodness-of-fit indicator AIC, the Akaike information criteria.

Model

Parameter

Observed sample data
ti = 685

Observed band-diameter data
71 = 3877

Estimate

AIC

Estimate

AIC

OLS

2.690(1.952-3.708)

6.205

0.211 (0.199-0.223)

3.738

1.135 (0.949-1.322)

2.867(2.825-2.910)

WOLS

2.471 (1.788-3.415)

6.219

0.212 (-0.201-0.224)

3.721

1.179(0.991-1.367)

2.856(2.814-2.898)

NLS

4.325 (3.400-5.444)

4.016

0.539(0.487-0.594)

3.781

0.955 (0.807-1.111)

2.241 (2.178-2.309)

WNLS

4.009 (3.106-5.231)

4.018

0.528 (0.475-0.586)

3.763

0.998(0.826-1.164)

2.247 (2.180-2.318)

more variable for band counts less than 20 (Fig. 3F).
The CV between all three readers was generally low
(<30%) for band counts less than 30, and there was
a notable increase in the variability and CV for band
counts greater than 30.

Spiny dogfish spines grow in a predictable pattern
with age (Fig. 4). The brownish-black banded, enameled
portion of the spine grows in length at a faster rate
than the white base portion.

Inclusion of the back-calculated band diameter
data dramatically changed the worn band estima-
tion models (Fig. 5), and therefore further worn band
estimations were made with both the observed and
back-calculated band diameter data. There were no
significant differences between the estimated worn-
band model parameters, but the WOLS model had
the lowest AIC value and therefore was chosen as the
best-fit model (Table 3). Alternative fits to the WOLS
and WLNS models, based on weightings by using the
inverse variance in assigned ages for each readability
category, yielded very similar parameter values and
nominally poorer fits indicated by slightly larger AIC
values (not shown). A high degree of natural varia-
tion resulted in wide 95% confidence intervals for all
parameters. Moreover, parameter confidence inter-
vals for the WOLS GOA model widely overlapped the
parameter confidence intervals for the Hecate Strait
and Strait of Georgia models (McFarlane and King,
2009). Although the parameters were not statistically
significantly different, the GOA, Hecate Strait, and
Strait of Georgia models appear to represent biologi-
cally meaningful differences in growth (Fig. 5). The
Hecate Strait and Strait of Georgia models tend to
overestimate the band count for larger spines and
underestimate for smaller spines of spiny dogfish col-
lected from the GOA.

0 10 20 30 40 50 60 70 80 90 100

Size class (cm)

Figure 4

Relationship between mean second dorsal spine
length and fish size determined from unworn spines
from spiny dogfish ( Squalus acanthias) collected
in the Gulf of Alaska. The top line is spine total
length ( TL ) and bottom line is base length ( BL ) in
millimeters. Numbers above upper line represent the
sample size for each 10-cm size class. Solid vertical
lines represent 95% confidence intervals. The dashed
vertical line represents the approximate size at birth
(Ketchen, 1972).

Fitting of growth models

The two-phase vB models fitted the observed data best
for males and females based on AIC values (Fig. 6, A
and D, Tables 4 and 5). For males, the two-phase model,
where L0 was used from model 1 (model 3b), was the
best fit and for females, it was the model where L0 was
estimated from model 1 (model 3b). Estimated (and 95%

126

Fishery Bulletin 108(2)

Enamel-base diameter (EBD) (mm)

Figure 5

Relationship of band count to enamel-base diameter for spiny
dogfish ( Squalus acanthias ) collected in the Gulf of Alaska (GOA)
between 2004 and 2007. The best-fit model (weighted ordinary
least squares [WOLS]) for (A) the observed data only and (B)
the observed data with the band-diameter data; both sections A
and B show the published best-fit relationships for spiny dogfish
collected from Hecate Strait and the Strait of Georgia, British
Columbia (McFarlane and King, 2009) for comparison.

confidence limits) asymptotic lengths (LJ were
87.2 cm (range 85.3-90.0 cm) and 102.5 cm
(range 99.9-106.3 cm) and growth coefficients
( k ) were 0.106 (range 0.097-0.117) and 0.058
(range 0.052-0.063) for males and females,
respectively. After including the back-calculated
data and the mean back-calculated data, the
two phase models were no longer the best fit
for males. The best-fit model with inclusion of
back-calculated data was model 2, and model
1 fitted best for the data including the mean
back-calculated data. Similarly, for females the
two-phase models were not the best-fit based
on AIC values after the inclusion of back-cal-
culated and mean back-calculated data: model
6 was the best fit with inclusion of back-calcu-
lated data, and model 5c (with L0 from model
1) was the best fit for the data including the
mean back-calculated data (Tables 4 and 5, Fig.

6, B, C, E, F).

Predicted length at age was similar for males
and females for the observed data, up to about
age 15, when a transition between growth
phases occurred (Fig. 6). After the transition,
females continued to grow at a faster rate and
to larger sizes than males (Fig. 6, A and D). At
the point of transition in the two-phase models
growth increased for about five years before
slowing, for both sexes.

Discussion

The model fits for all 10 examined growth
models were similar and had very small dif-
ferences in AIC, but the estimated parame-
ters differed substantially; for example, the
growth coefficient ( k ) was significantly different
between some models and thus could impact
estimates of natural mortality and subsequent
demographic analyses. The values of k tended
to fall into two groupings (in both data sets),
and those models that estimated the higher k
were also those that estimated lower estimates
for Lx. Interestingly, even with the significantly
different estimates of k, these estimates were still at the
lower range of reported growth rates for different types
of shark species (Cailliet and Goldman, 2004).

Cailliet et al. (2006) recommended considering more
than one form of evaluation of model performance and
considering biological interpretations along with statis-
tical fit when choosing the best model. Mean squared
error and the correlation coefficient (r2) were also cal-
culated for each model, but determinations of best fit
by the above criteria did not differ from those where
AIC was used and therefore are not reported. For the
observed data models 3a and 3b were the statistical
best fit for males and females, respectively. However,
the two-phase models tended to be unstable and would
converge at different localized minima, depending on

the starting value. A further consideration for the two-
phase models is that the growth curve indicates a pe-
riod of rapid growth immediately following the age at
transition.

The purpose of a two-phase model is to incorporate
changes in energy allocation as animals grow: imma-
ture fish use surplus energy for growth, whereas ma-
ture fish use surplus energy for reproduction (Soriano
et al., 1992). Thus, the rate of growth changes after
maturation. In our case, the transition between the two
growth phases occurred before the age at 50% maturity
for both males and females The early age at transition
and the period of rapid growth after transition indi-
cate that for female spiny dogfish there is a “growth
spurt” about 15 years before age at 50% maturity. For

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

127

Female

Age (years)

Figure 6

Model fits for male (A-C) and female (D-F) spiny dogfish ( Squalus acanthias) length-at-age data.
(A and D) Best-fit growth models based on the observed sample data; (B and E) best-fit growth
models based on the observed sample data and the back-calculated data; and (D and F) best-fit
growth models based on the observed sample data and the mean back-calculated data. nobs is the
number of samples, nback is the number of data points created through back calculation of the ages
from band-diameter data, and is the number of mean back-calculated data points.

’ mean 1

males, the pattern was similar, but occurred just before
age at 50% maturity. This finding does not follow the
theory behind the two-phase model and indicates that
a two-phase model may not be most appropriate in this
situation.

The two-phase vB model by Soriano et al. (1992) has
been examined with data sets from many species of
sharks to determine whether it is an adequate descrip-
tor of shark growth (Araya and Cubillos, 2006). Where-
as the two-phase model was better than the standard
vB model in 8 of 11 species for females and 7 of 11 for
males, the two-phase model did not perform better than
the vB (model 1 here) for spiny dogfish. Because Araya
and Cubillos (2006) included only one spiny dogfish
population (Black Sea), which appears to have different

age and growth characteristics from those in the GOA,
and only examined average length at age data (Avsar,
2001), we felt that it was worth while to investigate the
two-phase family of models in this study. Braccini et
al. (2007) found that the two-phase model was the best
statistical fit for the piked spurdog, which is a species
similar to spiny dogfish; however, the resultant mod-
els showed some of the same characteristic difficulties
that we encountered. Those results also indicated a de-
crease in length after transition (Fig. 7, Braccini et al.
2007) and that the At parameter appears to change only
briefly before returning to its original value. Braccini et
al. did not address these issues as we have attempted
here. A more comprehensive examination, which would
include multiple data sets from different regions for

128

Fishery Bulletin 108(2)

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

129

130

Fishery Bulletin 108(2)

0 10 20 30 40

Age (years)

Figure 7

Comparison of published spiny dogfish ( Squalus acanthias) female growth
models from sources listed in Table 4. (A) Growth models published for Pacific
Ocean spiny dogfish: “Alaska” includes the Gulf of Alaska (GOA) model from
this study and a Prince William Sound (PWS) model (Vega, 2006); “British
Columbia inshore” includes three models for dogfish sampled within the
Strait of Georgia and Hecate Strait (Ketchen, 1975; Saunders and McFar-
lane, 1993); “Puget Sound inshore” covers models based on samples collected
within the Puget Sound area south off British Columbia and east of the
Washington coast (Vega, 2006); “Pacific Coast South” includes four models
based on samples collected off Oregon and California (Vega, 2006); “Pacific
Coast North” includes models based on samples collected off of Washington
and the west coast of Vancouver Island (Ketchen, 1975; Jones and Geen,
1977; Vega, 2006); (B) The growth models from the Atlantic Ocean, North
Sea. and Black Sea (Holden and Meadows, 1962; Sosinski 1978; Nammack et
al., 1985; Fahy, 1989; Avsar, 2001; Henderson et al., 2002; Soldat [footnote 1
in Table 6]). Note the different x-axis scales.

each species, and a complete sample
of the size range may lead to a more
conclusive determination as to which
species exhibit two-phase growth.

Disregarding the two-phase mod-
els, the best-fit model was model 2
for males and model 5c for females.
In this situation, given the lack of
data and difficulties with the two-
phase models, it may be more ap-
propriate to select the best model
not based on the AIC criteria alone,
but to also consider the biological
soundness of the models. Model 2
(males) and model 5c (females) are
the statistical best fit of the more
biologically reasonable models. Both
of these best-fit models require L0 as
an input, not as an estimated pa-
rameter. The lack of data for spiny
dogfish <50 cm TLext likely causes
the models that estimate L0 to have
difficulty fitting the data and as a
result estimate L0 to be larger than
would be expected.

In the majority of published stud-
ies on spiny dogfish age and growth
the traditional von Bertalanffy
model is used. To facilitate a broad-
er comparison of our results with
growth parameter estimates for oth-
er regions of the geographic distri-
bution of spiny dogfish, we compared
parameters estimated from model 1
(Table 4) with growth curves fitted
by using the traditional vB formula-
tion, as reported in published stud-
ies (Table 5, Fig. 7). Clear differ-
ences in spiny dogfish growth exist
between the North Pacific and North
Atlantic oceans. For instance, we
found that male and female dogfish
reach larger asymptotic sizes (87.2
and 112.2 cm TLext, respectively) in
the GOA than off the northeastern
United States (82.5 and 100.5 cm
TLgxt, respectively; Nammack et al.,
1985). Indeed, virtually all stud-
ies have found large differences in
growth of spiny dogfish between
the North Pacific and North Atlan-
tic (Table 5, Fig. 7). Fish from the
North Atlantic tend to grow more
rapidly, achieve smaller asymptotic
sizes, and have shorter life spans
than those from the Pacific. Differ-
ences in growth also exist within
the Pacific (Table 5, Fig. 7). For ex-
ample, our GOA growth estimates
are similar to those for spiny dog-

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

131

Table 6

Summary of von Bertalanffy parameters (model 1) for growth models for female spiny dogfish ( Squalus acanthias ) from the
North Pacific and North Atlantic oceans and the North and Black seas. Parameters are asymptotic length (L„) size at birth
(L0), growth coefficient (k), and the theoretic age-at-size length zero (t0). Here, L0 was solved for from the published parameter
estimates for the purposes of comparison.

Location

Reference

Alaska, Gulf of Alaska

121.4

0.034

-12.1

40.9

This study

Alaska, Prince William Sound

110.4

0.038

-11.6

39.4

Vega (2006)

British Columbia, Hecate Strait

125.1

0.031

-10.6

35.0

Ketchen ( 1975)

British Columbia, Strait of Georgia

129.1

0.034

-7.3

28.4

Ketchen ( 1975)

British Columbia, Strait of Georgia

114.9

0.044

-3.6

16.8

Saunders and McFarlane (1993)

British Columbia, offshore

128.5

0.036

-6.9

28.3

Jones and Geen (1977)

U.S., inshore (WA north)

113.5

0.04

-5.2

21.3

Vega (2006)

U.S., inshore (WA south)

100.4

0.036

-8.4

26.2

Vega (2006)

U.S., offshore (WA)

123.6

0.027

-6.9

21.0

Vega (2006)

U.S., offshore (WA)

152.9

0.036

-6.7

32.8

Ketchen (1975)

U.S., offshore (OR)

101.9

0.027

-12.7

29.6

Vega (2006)

U.S., offshore (OR and CA combined)

90.9

0.031

-13.0

30.2

Vega (2006)

U.S., offshore (CA north)

158.9

0.009

-25.3

32.4

Vega (2006)

U.S., offshore (CA south)

123.6

0.027

-6.9

21.0

Vega (2006)

Northwest Atlantic (U.S.)

100.5

0.106

-2.9

26.6

Nammack et al. (1985)

Northeast Atlantic (Ireland)

98.8

0.090

-1.6

13.3

Fahy (1989)

Northeast Atlantic (Ireland)

112.0

0.150

-3.4

44.7

Henderson et al. (2002)

Northwest Atlantic

104.5

0.095

-3.7

31.0

Soldat2

North Sea

137.1

0.054

-4.7

30.7

Sosinski 1978 (as cited in Avsar, 2001)

North Sea

101.4

0.110

-3.6

33.2

Holden and Meadows (1962)

Black Sea

145.0

0.170

-0.7

16.3

Avsar (2001)

1 Soldat, V. T. 2002. Spiny dogfish ( Squalus acanthias L.) of the northwest Atlantic Ocean (NWA). NAFO Sci. Counc. Res Doc 02/84, 33 p.

fish from offshore Washington State waters (Fig. 7) but
greater than those caught in inshore Washington State
waters (Puget Sound) and British Columbia (Ketchen,
1975; Jones and Geen, 1977; Saunders and McFarlane,
1993; Vega, 2006). The age and growth studies from
British Columbia were conducted on spiny dogfish col-
lected in inshore waters (Strait of Georgia and Hecate
Strait); therefore the possibility cannot be ruled out
that spiny dogfish from the British Columbia offshore
region would have growth estimates similar to those
of Washington offshore and GOA spiny dogfish. The
vB growth model parameter estimates (Lx and k ) for
northern California spiny dogfish (defined as spiny dog-
fish between Point Conception to the Oregon border;
Vega, 2006) were radically different from our results
for the GOA, but the fits for California may have been
adversely affected by small sample size.

The wide variability in length-at-age contributes to
the lack of statistically significant differences among
growth models and worn-band estimation models. This
variability may be attributable to one or more of the
following factors: measurement error in either length
or age readings, sampling bias, true underlying vari-
ability in growth at age, and misidentification of worn
and unworn spines. We considered the potential role of
each of these factors.

Measurement error in the length measurements alone
is insufficient to explain the relatively large variabil-
ity in the size-at-age data. Aging errors may take two
forms: imprecision and bias. We found no bias among
the three readers tested, but imprecision of the band
counts among readers could contribute to variability
in the size-at-age data, especially for older ages. We
used the median band count (from the three readers)
to account for reduced precision because this measure
of central tendency is less sensitive to outliers than the
mean for small sample sizes (Dudewicz and Mishra,
1988). A more thorough analysis of the precision of age
estimates for spiny dogfish in the Pacific Ocean revealed
the overall coefficient of variation for aging estimates
among four laboratories to be 19% (Rice et al., 2009).
Systematic bias was found for two of the laboratories
(one biased high, the other biased low) in relation to
the other two, but relative bias did not always result
in statistically different parameters estimated from vB
growth curves (Rice et al., 2009).

Age validation is crucial for growth studies to assure
that physical structures used for aging are correctly in-
terpreted. For instance, a systematic aging error could
result if the periodicity of band formation is not an-
nual. Annual periodicity of band deposition on second
dorsal spines was validated for spiny dogfish in British

132

Fishery Bulletin 108(2)

Columbia (Beamish and McFarlane, 1985; McFarlane
and Beamish, 1987). Moreover, radioactive carbon iso-
topes absorbed into spiny dogfish spines provided age
estimates that agree with previous aging results for the
British Columbia spiny dogfish (Campana et ah, 2006)
and verified that periodicity is annual, even at old ages
(Campana, 2001). We assumed that this annual peri-
odicity of band formation in spiny dogfish, which was
confirmed for this species in British Columbia, also ap-
plies to fish from the GOA. Because spiny dogfish from
British Columbia have different age characteristics
(e.g., worn band curves, Fig. 5) from those of the GOA,
it is possible that the pattern of band deposition may
also differ.

Sampling bias was considered by examining potential
differences in average size at age among gear type and
location of capture. Because there were no significant
differences among the average size at age with the
different gear types used or the areas sampled, we do
not believe that sampling bias was a significant factor
affecting our results. However, the lack of significant
differences in our study should not be misconstrued to
rule out considerations of sampling bias in future spiny
dogfish studies, because this species may school by size
and sex (Nammack et al., 1985; Ketchen, 1986).

In the western North Atlantic Ocean commercial
fisheries target the largest and oldest age classes (Rago
et al., 1998). Thus, the size-frequency distributions
determined from commercial catches may not be repre-
sentative of the full size range of fish in the population.
Moreover, depletion of large old fish from the population
by heavy exploitation means that subsequent research
surveys may not catch a representative sample of the
full size and age ranges of the population. In the GOA,
spiny dogfish are taken as bycatch in multiple fisheries.
In some cases, dogfish bycatch is largely unaccounted
for, owing to the lack of observers on small (<60-ft) ves-
sels, such as those vessels with salmon gill nets, as well
as some longline vessels targeting halibut and sablefish,
resulting in an unknown level of total fishing mortal-
ity (Courtney et al., 2006). However, in the GOA, it is
unlikely that the fishing mortality has truncated the
size distribution of spiny dogfish because spiny dogfish
are not targeted and recent (2006) estimates of spiny
dogfish biomass are 80-100% of the estimated theoreti-
cal population carrying capacity (Rice, 2007). Therefore,
it is unlikely that the fishery has created size-selective
impacts that would lead to erroneous selection of the
two-phase models as the best-fit models (Braccini et
al., 2007).

One limitation of our size-frequency distributions
is the absence of spiny dogfish smaller than 50 cm
TLext. The lack of samples from smaller spiny dogfish is
likely due to fishery-dependent opportunistic sampling
which apparently occurs in areas devoid of juvenile
spiny dogfish. Examination of NMFS spring and fall
trawl surveys along the U.S. east coast revealed that
in spring most juveniles were caught in water between
50 and 150 m deep (range: 7-390 m) in offshore waters
from North Carolina to the eastern edge of Georges

Bank, whereas in fall most were caught between 25
and 75 m (range: 12-366 m) in various locations, such
as on Georges Bank, Nantucket Shoals, and throughout
the Gulf of Maine (McMillan and Morse, 1999). Spiny
dogfish smaller than 50cm TLext have been surveyed in
both Puget Sound, Washington (Tribuzio et al., 2009),
and in the northern Strait of Georgia (McFarlane et
al., 2006) by using bottom trawl gear. In this study,
we made numerous unsuccessful attempts to capture
juvenile dogfish smaller than 50 cm TLext in the GOA
using sport and longline gear in Yakutat Bay, long-
line gear with small (10/0 circle) hooks in Southeast
Alaska (K. Munk, personal commun.1), and commercial
bottom trawls off Kodiak Island (J. Gauvin, personal
commun.2).

A missing size group, such as small dogfish in our
case, may cause growth models to overestimate t0 or L0,
thus decreasing the k estimate. Further, this missing
size group may have caused the age of transition, th, in
the two-phase models to be underestimated. Also, the
lack of small animals may have limited our ability to
discriminate among competing growth models. We used
band-diameter data and back-calculated lengths derived
from unworn spines to attempt to address this data
gap. The inclusion of the band-diameter data greatly
improved the worn-band estimation models, but mini-
mally changed the growth models. Few of the estimated
growth model parameters based on the back-calculated
and mean back-calculated data were significantly dif-
ferent from those estimated from the observed data
alone.

Back-calculation methods are designed to be used
when sample sizes are small or if sampling has not oc-
curred each month (Cailliet and Goldman, 2004), but
in this case it was the entire smaller end of the size
range that was being estimated. With the modified
Fraser-Lee size-at-birth method, we had to assume that
average size at birth was known. We use 26.2 cm, which
is based on data collected from spiny dogfish inside the
Strait of Georgia, British Columbia (Ketchen 1972).
Sizes at birth are reportedly similar for the species
across the northern hemisphere, with ranges of 23-30
cm (Ketchen 1972, Tribuzio et al. 2009). We also as-
sumed that 2.45 mm was the spine diameter at birth,
based on studies of British Columbia spiny dogfish (Mc-
Farlane and King 2009). Because this is an average as
well, it is likely that some spines were classified as “un-
worn” when they were actually “worn.” Spines that are
classified as “unworn” can lead to underestimating the
age, and in the case of the back-calculation resulted in
instances where 20 cm or more of growth was predicted
in the first year. Back-calculations may not be appropri-
ate for this species when dorsal fin spines are used as
aging structures, but may work well if a structure such
as vertebrae are used.

1 Munk, Kristen. 2007. Alaska Department of Fish and
Game, Juneau, AK, 99801.

2 Gauvin, John. 2007. Gauvin and Associates, LLC. Burien,
WA 98166.

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

133

The relatively large variability in size at age of spiny
dogfish in the GOA could also reflect true underlying
variability in growth rates. Individuals experiencing
different thermal and feeding histories are expected to
have different growth characteristics. It is also conceiv-
able that our samples represent collections of dogfish
from multiple, mixed populations. For instance, 4 of
2940 recoveries (0.14%) of spiny dogfish tagged in Brit-
ish Columbia were recovered in Alaska (McFarlane and
King, 2003). Because the movements of spiny dogfish
from other areas to and from Alaska are unknown,
the degree of mixing is uncertain. However, there is
no evidence of genetic differentiation in the Northeast
Pacific based on analyses of eight microsatellite loci
from dogfish sampled from the Bering Sea, the Gulf of
Alaska, Strait of Georgia, Puget Sound, and the coasts
of Washington, Oregon, and California (Hauser, 2009).
Mixtures of spiny dogfish from other areas with growth
characteristics that are different from those of Alaska
resident dogfish could contribute to the variability in
size at age that we observed in the GOA. Nevertheless,
the existence of a statistically significant difference in
growth rates from different areas of the Northeast Pa-
cific (Vega, 2006; Table 4 this document) indicates that
mixing is incomplete.

Our findings have at least two important implications
for management of the species. First, for estimation of
stock productivity and biological reference points for
spiny dogfish in the GOA, it is important to use growth
curves that are fitted to size-at-age data from dogfish
captured in the GOA. Although alternative growth mod-
el parameters were not statistically significantly differ-
ent from one another in our study, the variation among
predicted length may be of biological significance. For
instance, the worn-band estimation curves for the GOA
and British Columbia resulted in very different esti-
mates of ages (Fig. 5); use of growth curves for British
Columbia would result in estimated numbers of worn
bands from dogfish spines in the GOA that would be
biased low. For example, for a spiny dogfish with a
1.8-mm EBD, the GOA model would estimate an age
of one year, whereas both of the British Columbia mod-
els would estimate an age of four years. A fish with a
6-mm EBD would be estimated to be 30 years old by the
GOA model and 46 and 37 years old by the two British
Columbia models. Such biases in growth estimates may
lead to biases in estimates of biological reference points
for fishery management.

Second, as in other portions of their range, the largest
spiny dogfish are the oldest females. Because commer-
cial fisheries for spiny dogfish select for the largest in-
dividuals, fishing mortality rates are disproportionately
higher for this reproductive segment of the population.
In the Northwest Atlantic Ocean, a sharp increase in
landings during 1987-1993 led to a fivefold increase in
fishing mortality rates on females from 0.016 to 0.26;
and fishing mortality rates exceeding 0.10 on large
(a80-cm) females resulted in negative pup replace-
ment, subsequently leading to stock decline (Rago et
al., 1998). Thus, to sustain spiny dogfish in the GOA,

fishery management plans should consider not only
slow growth rates, low fecundity, and late maturation
of this species (King and McFarlane, 2003), but also
the potentially disproportionate number of removals of
mature females from the stock by commercial fishing by
estimating size- and sex-specific fishing mortality rates
and biological reference points.

Future research should address the many uncer-
tainties remaining about spiny dogfish biology and
life history in Alaska. In particular, results from this
study indicate several areas of research needed to
improve our understanding of spiny dogfish age and
growth. First, although demonstrated for fish captured
off British Columbia (Beamish and McFarlane, 1985;
McFarlane and Beamish, 1987; Campana et al., 2006),
validation of annual band formation, as well as worn-
band properties, for spiny dogfish collected from the
GOA should be conducted to describe potential sources
of bias in the age estimates for spiny dogfish at this
northern portion of their range in the Pacific Ocean.
Second, the collection of juvenile dogfish (<50 cm) is
needed to provide more precise estimates of growth
over their full life history, as well as to help identify
statistically best-fit growth models. Third, tagging
studies, such as those conducted in British Columbia
(King and McFarlane, 2003), would help elucidate the
degree to which dogfish in Alaska represent mixed
stocks with different growth attributes; such tagging
results would help to delineate stock boundaries essen-
tial for fishery management. Fourth, controlled experi-
ments are necessary to fully examine the selectivity of
various fishing gears for spiny dogfish by size and sex.
This would be an important preliminary step toward
gear standardization, if long-term sampling programs
are envisioned for spiny dogfish. Finally, continued
sampling of spiny dogfish over small regional scales
is necessary to fully evaluate potential geographic
differences in growth and resultant parameters (i.e.,
natural mortality) within the GOA, as well as to more
broadly understand the life history of this species in
this portion of its range. Although our study would
not have been possible without the diversity of low-
cost sampling opportunities afforded to us, including
the valuable assistance of state and federal agencies
and sport and commercial fishermen, further progress
will be accelerated by a full-scale, directed field pro-
gram, which would be more successful at providing
an unbiased sample set of spiny dogfish in the waters
off Alaska, and which would aid in efforts to build a
more detailed stock assessment, and thus models of
population dynamics.

Acknowledgments

We are grateful for funding of this research by the
North Pacific Research Board (NPRB publication no.
227), the Rasmuson Fisheries Research Center, and the
Alaska Fisheries Science Center’s Population Dynamics
Fellowship through the Cooperative Institute for Arctic

134

Fishery Bulletin 108(2)

Research (CIFAR). We thank V. Gallucci, J. Rice, A.
Andrews, and W. Strasberger for field and laboratory
assistance, and G. Bargmann, S. Rosen, and J. Topping
at the Washington Department of Fish and Wildlife
for reading spines and training. We acknowledge the
National Marine Fisheries Service; Alaska Department
of Fish and Game; chartered vessels and crew of the
FVs Kingfisher , Winter King, and Sea View, commercial
fishermen in Yakutat, Cordova, and Kasilof; Gauvin
and Associates, LLC., and Alaska Pacific and Trident
Seafoods for kindly providing sampling opportunities.
Finally, we are grateful to T. Quinn II and K. Goldman
for considerable helpful analytical advice.

Literature cited

Araya, M., and L. A. Cubillos

2006. Evidence of two-phase growth in elasmobranchs. En-
viron. Biol. Fishes 77:293-300.

Avsar, D.

2001. Age, growth, reproduction and feeding of the
spurdog ( Squalus acanthias Linnaeus, 1758) in the
southeastern Black Sea. Estuar. Coast. Shelf Sci. 52:
269-278.

Beamish, R. J., and G. A. McFarlane.

1985. Annulus development on the second dorsal spine
of the spiny dogfish ( Squalus acanthias) and its valid-
ity for age determination. Can. J. Fish. Aquat. Sci.
42:1799-1805.

Beverton, R. J. H., and S. J. Holt.

1957. On the dynamics of exploited fish populations. U.K
Ministry of Agriculture and Fisheries, Fisheries Inves-
tigations 2, 533 p.

Boldt, J.

2003. Ecosystem considerations for 2004. In Stock
assessment and fishery evaluation report for the ground-
fish resources of the Bering Sea/Aleutian Islands and
Gulf of Alaska region, 335 p. [Available from North
Pacific Fishery Management Council, 605 W. 4th Ave.,
Suite 306, Anchorage, AK 99501.]

Bonfil, R.

2005. The purpose of stock assessment and the objectives
of fisheries management. In Management techniques
for elasmobranch fisheries ( J. A. Musick, and R. Bonfil,
eds.), p. 6-14. FAO Fisheries Tech. Paper 474.

Braccini, J. M., B. M. Gillanders, and T. I. Walker.

2007. Comparison of deterministic growth models fitted
to length-at-age of the piked spurdog ( Squalus mega-
lops) in south-eastern Australia. Mar. Freshw. Res.
58:24-33.

Burnham, K. P., and D. R. Anderson.

2004. Multimodel inference: understanding AIC and BIC
in model selection. Soc. Meth. Res. 33:261—304.

Cailliet, G. M., and K. J. Goldman.

2004. Age determination and validation in chondrich-
thyan fishes. In The biology of sharks and their rela-
tives (J. F. Carrier, J. A. Musick, and M. R. Heithaus,
eds.), p. 399-447. CRC Press, Boca Raton, FL.

Cailliet, G. M., W. D. Smith, H. F. Mollet, and K. J. Goldman.

2006. Age and growth studies of Chondrichthyan fishes:
the need for consistency in terminology, verification,
validation, and growth function fitting. Environ. Biol.
Fishes 77:211-228.

Campana, S. E.

1990. How reliable are growth back-calculations based
on otoliths? Can. J. Fish. Aquat. Sci. 47:2219-2227.

2001. Accuracy, precision and quality control in age deter-
mination, including a review of the use and abuse of age
validation methods. J. Fish Biol. 59:197-242.

Campana, S. E., C. Annand, and J. I. McMillan.

1995. Graphical and statistical methods for determining
the consistency of age determinations. Trans. Am.
Fish. Soc. 124:131-138.

Campana, S., C. Jones, G. A. McFarlane, and S. Myklevoll.

2006. Bomb dating and age validation using the spines
of spiny dogfish ( Squalus acanthias). Environ. Biol.
Fishes 77:327-336.

Carlson, J. K., and I. E. Baremore.

2005. Growth dynamics of the spinner shark (Carcha-
rhinus brevipinna) off the United States southeast and
Gulf of Mexico coasts: a comparison of methods. Fish.
Bull. 103:280_291.

Compagno, L. J. V.

1984. Sharks of the world: an annotated and illustrated
catalogue of shark species known to date, pt. 2: Car-
charhiniformes, 675 p. FAO Fish. Synop. 125 (vol.
4). FAO, Rome.

Courtney, D., C. A. Tribuzio, K. J. Goldman, and J. Rice.

2006. GOA sharks. In Stock assessment and fishery
evaluation report for the groundfish resources of the
Gulf of Alaska for 2007, Appendix E, p. 481-561. [Avail-
able from North Pacific Fishery Management Council,
605 W. 4th Ave., Suite 306, Anchorage, AK 99501.]

Driggers, W. B., J. K. Carlson, D. Oakley, G. Ulrich, B. Cullum,
and J. M. Dean.

2004. Age and growth of the blacknose shark, Carcharhi-
nus acronotus, in the western North Atlantic Ocean with
comments on regional variation in growth. Environ.
Biol. Fishes 71:171-178.

Dudewicz, E. J., and S. A. Mishra.

1988. Modern mathematical statistics, 864 p. John
Wiley & Sons, Inc., New York, NY.

Fabens, A. J.

1965. Properties and fitting of the von Bertalanffy growth
curve. Growth 29:265-289.

Fahy, E.

1989. The spurdog Squalus acanthias (L) fishery in
south west Ireland. Irish Fish. Invest. Ser. B (Mar.)
32, 22 p.

Francis, R. I. C. C.

1990. Back-calculation of fish length: a critical review.
J. Fish Biol. 36:883-902.

Goldman, K. J., S. Branstetter, and J. A. Musick.

2006. A re-examination of the age and growth of sand
tiger sharks, Carcharias Taurus, in the western North
Atlantic: the importance of ageing protocols and use of
multiple back-calculation techniques. Environ. Biol.
Fishes 77:241-252.

Gulland, J. A.

1969. Manual of methods for fish stock assessment: part
1, Fish population analysis. FAO manuals in fisheries
science, no. 4, 154 p. FAO Rome.

Hauser, L.

2009. The molecular ecology of dogfish sharks (Squalus
acanthias) . In Biology and management of dogfish
sharks (V. F. Gallucci, G. McFarlane, G. Bargmann,
eds.), p. 229-252. Am. Fish. Soc., Bethesda, MD.

Tribuzio et al.: Age and growth of Squalus acanthias in the Gulf of Alaska

135

Henderson, A. C., K. Flannery, and J. Dunne.

2002. Growth and reproduction in the spiny dogfish Squa-
lus acanthias L. (Elasmobranchii: Squalidae), from the
west coast of Ireland. Sarsia 87:350-361.

Hoenig, J. M., M. J. Morgan, and C. A. Brown.

1995. Analyzing differences between two age determi-
nation methods by tests of symmetry. Can. J. Fish.
Aquat. Sci. 52:364-368.

Holden, M. J.

1974. Problems in the rational exploitation of elasmo-
branch populations and some suggested solutions. In
Sea fisheries research (E. H. Jones, ed.), p. 187-215.
Logos, London.

1977. Elasmobranchs. In Fish population dynamics (J.
A. Gulland, ed.), p. 187-215. Wiley, London.

Holden, M. J., and P. S. Meadows.

1962. The structure of the spine of the spur dogfish ( Squa-
lus acanthias L.) and its use for age determination. J.
Mar. Biol. Assoc. U K. 42:179-197.

Jones, B. C., and G. H. Geen.

1977. Age and growth of spiny dogfish ( Squalus acanth-
ias) in the Strait of Georgia, British Columbia. Fish.
Mar. Serv. Res. Dev., Tech. Rep. 699, 16 p.

Ketchen, K. S.

1972. Size at maturity, fecundity, and embryonic growth of
the spiny dogfish (Squalus acanthias ) in British Colum-
bia waters. J. Fish. Res. Board Can. 29:1717-1723.

1975. Age and growth of dogfish Squalus acanthias in
British Columbia waters. J. Fish. Res. Board Can.
32:43-59.

1986. The spiny dogfish ( Squalus acanthias) in the north-
east Pacific and a history of its utilization. Can. Spec.
Publ. Fish. Aquat. Sci. 88, 78 p.

Ketchen, K.S., N. Bourne, and T. H. Butler.

1983. History and present status of fisheries for marine
fishes and invertebrates in the Strait of Georgia, British
Columbia. Can. J. Fish. Aquat. Sci. 48:1095-1119.

King, J. R., and G. A. McFarlane.

2003. Marine fish life history strategies: applications
to fisheries management. Fish. Manag. Ecol. 10:249-
264.

McFarlane, G. A., and R. J. Beamish.

1987. Validation of the dorsal spine method of age deter-
mination for spiny dogfish. In Age and growth in fish
(R. C. Summerfelt, G. E. Hall, eds.), p. 287-300. Iowa
State Univ., Ames, IA.

McFarlane, G. A., and J. R. King.

2006. Migration patterns of spiny dogfish ( Squalus
acanthias) in the North Pacific Ocean. Fish. Bull.
101:358-367.

2009. Re-evaluating the age determination of spiny dog-
fish (Squalus acanthias) using oxytetracycline and fish at
liberty up to twenty years. In Biology and management
of dogfish sharks (V. F. Gallucci, G. McFarlane, G. Barg-
mann, eds.), p. 77-88. Am. Fish. Soc., Bethesda, MD.

McFarlane, G. A., J. R. King, and V. R. Hodes.

2006. Biological results of the Strait of Georgia spiny
dogfish (Squalus acanthias) longline survey October
18-31, 2005. Can. Data Rep. Fish. Aquat. Sci. 1182,
24 p.

McMillan, D. G., and W. W. Morse.

1999. Essential fish habitat source document: Spiny
dogfish, Squalus acanthias, life history and habitat
characteristics. NOAA Tech. Memo. NMFS-NE-150, 19 p.

Mollet, H. F., J. M. Ezcurra, and J. B. O’Sullivan.

2002. Captive biology of the pelagic stingray, Dasyatis
violacea (Bonaparte, 1832). Mar. Freshw. Res. 53:531-
541.

Nammack, M. F., J. A. Musick, and J. A. Colvocoresses.

1985. Life history of spiny dogfish off the Northeastern
United States. Trans. Am. Fish. Soc. 114:367—376.

Rago, P. J., K. A. Sosebee, J. K. T. Brodziak, S. A. Murawski, and
E. D. Anderson.

1998. Implications of recent increases in catches on the
dynamics of Northwest Atlantic spiny dogfish ( Squalus
acanthias). Fish. Res. 39:165-181.

Rice, J.

2007. Population dynamics of spiny dogfish (Squalus
acanthias) in the Gulf of Alaska with an emphasis on
the analysis of bycatch data. M.S. thesis, 109 p. Univ.
Washington, Seattle.

Rice, J., V. F. Gallucci, and G. H. Kruse.

2009. Evaluation of the precision of age estimates for
spiny dogfish. In Biology and management of dogfish
sharks (V. F. Gallucci, G. McFarlane, G. Bargmann,
eds.), p. 161-168. Am. Fish. Soc,, Bethesda, MD.

Ricker, W. E.

1975. Computation and interpretation of biological sta-
tistics of fish populations. Bull. Fish. Res. Board Can.
191:1-382.

1979. Growth rates and models. In Fish physiology, vol.
IX (W. S. Hoar, D. J. Randall, and J. R. Brett, eds.), p.
677-743. Academic Press, New York.

Saunders, M. W., and G. A. McFarlane.

1993. Age and length at maturity of the female spiny
dogfish, Squalus acanthias, in the Strait of Georgia,
British Columbia, Canada. Environ. Biol. Fishes
38:49-57.

Soriano, M., J. Moreau, J. M. Hoenig, and D. Pauly.

1992. New functions for the analysis of two-phase growth
of juvenile and adult fishes, with application to Nile
perch. Trans. Am. Fish. Soc. 121:486-493.

Sosinski, J.

1978. Characteristics of the North Sea spurdog (Squalus
acanthias L.) stock. Fish. Biol. 8:9-22.

Tribuzio, C. A., V. F. Gallucci, G. Bargman.

2009. Reproductive biology and management implications
for spiny dogfish (Squalus acanthias) in Puget Sound,
WA. In Biology and management of dogfish sharks
(V. F. Gallucci, G. McFarlane, G. Bargmann, eds.), p.
181-194. Am. Fish. Soc., Bethesda, MD.

Tribuzio, C. A., C. Rodgveller, J. Heifetz, D. Courtney, and K. J.
Goldman.

2008. Assessment of the shark stocks in the Gulf of
Alaska. In Stock assessment and fishery evaluation
report for the groundfish resources of the Gulf of Alaska
for 2009, chapter 18, p. 557-612. [Available from North
Pacific Fishery Management Council, 605 W. 4th Ave.,
Suite 306, Anchorage, AK 99501.]

Vega, N. M.

2006. Biogeography of the spiny dogfish, Squalus
acanthias, over a latitudinal gradient in the North-
east Pacific. M.S. thesis, 117 p. Univ. Washington,
Seattle, WA.

von Bertalanffy, L.

1938. A quantitative theory of organic growth (inquiries
on growth laws II). Human Biol. 10:181-213.

136

Effective herding of flatfish by cables
with minimal seafloor contact

Carwyn F. Hammond1

Email address for contact author: craig.rose@noaa.gov

1 NOAA, National Marine Fisheries Service

Alaska Fisheries Science Center, Conservation Engineering Program
7600 Sand Point Way NE
Seattle, Washington 98115

2 Best Use Cooperative

4241 21st Avenue West, Suite 302
Seattle, Washington 98199

Abstract — Otter trawls are very
effective at capturing flatfish, but
they can affect the seafloor ecosys-
tems where they are used. Alaska
flatfish trawlers have very long
cables (called sweeps) between doors
and net to herd fish into the path
of the trawl. These sweeps, which
ride on and can disturb the seafloor,
account for most of the area affected
by these trawls and hence a large pro-
portion of the potential for damage
to seafloor organisms. We examined
modifications to otter trawls, such
that disk clusters were installed at
9-m intervals to raise trawl sweeps
small distances above the seafloor,
greatly reducing the area of direct
seafloor contact. A critical consider-
ation was whether flatfish would still
be herded effectively by these sweeps.
We compared conventional and modi-
fied sweeps using a twin trawl system
and analyzed the volume and com-
position of the resulting catches. We
tested sweeps raised 5, 7.5, and 10
cm and observed no significant losses
of flatfish catch until sweeps were
raised 10 cm, and those losses were
relatively small (5-10%). No size com-
position changes were detected in the
flatfish catches. Alaska pollock ( Ther -
agra chalcogramma ) were captured
at higher rates with two versions of
the modified sweeps. Sonar observa-
tions of the sweeps in operation and
the seafloor after passage confirmed
that the area of direct seafloor contact
was greatly reduced by the modified
sweeps.

Manuscript submitted 16 January 2009.
Manuscript accepted 13 November 2009.
Fish. Bull. 108:136-144 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

Craig S. Rose (contact author)1
John R. Gauvin2

Otter trawling is one of the most
effective methods for capturing com-
mercial quantities of flatfish and is
the principal method for flatfish har-
vest in Alaska waters. However, trawl
fisheries have received increasing
attention for their potential to affect
seafloor habitats. Changes to seafloor
ecosystems resulting from the pas-
sage of trawl gear have been have
described in a wide range of studies
(Barnes and Thomas, 2005; Lokke-
borg, 2005). These include changes
to infaunal (Tuck et al., 1998) and
epifaunal (Kaiser et al., 1998; Prena
et al., 1999; McConnaughey et al.,
2000) communities, as well as indi-
rect effects from changes to seafloor
structure and resuspension of sedi-
ments (Churchill, 1989). The most
common response to mitigate these
problems has been closures of sen-
sitive areas to trawling. When such
areas have rough, rocky substrates,
regulations requiring that trawl foot-
rope cross-sections be below a certain
size have been used to discourage
fishing in these areas; the smaller
footropes make nets more vulnerable
to damage (Hannah, 2003; Bellman
et al., 2005).

Alaskan commercial flatfish fisher-
ies, among the largest in the world,
are pursued almost exclusively with
demersal otter trawls. (The excep-
tion is the fishery for Pacific halibut
| Hippogossus stenolepis ], a large, pi-
scivorous species that is harvested
by longlines.) These otter trawls gen-

erally use very long cables, herein
called “sweeps,” that skim the sea-
floor ahead and to both sides of the
trawl net. In Alaska flatfish fisher-
ies, the fishermen have used progres-
sively longer sweeps to increase the
width of their gear and, hence, the
area from which flatfish are captured.
These sweeps now account for the
overwhelming majority of the seafloor
area swept by these trawlers to cap-
ture flatfish. Although these sweeps
greatly increase flatfish catches, they
also account for most of the negative
effects of trawling on the seafloor.

Although some reviews (Kaiser et
al., 2007) have recommended devel-
opment of modified fishing gear to
reduce the effects of trawling on sea-
floor communities, studies that test
such gear are just beginning to be
published. He (2007) reviewed such
efforts for all mobile fishing gears. A
substantial effort in Europe focused
on modifications for beam trawling
(van Marlen et al., 2005). Guyonnet
et al. (2008) described tests of modi-
fied gear that reduce the contact of
the cables between trawl doors and
nets with the seafloor. Although
their tests were accomplished with
different modifications to gear
(dangling chain sections attached
to neutrally bouyant rope) and in a
very different fishery, their concept
is very similar to the modifications
we tested.

Ryer (2008) has described flat-
fish behaviors that are important to

Rose et al. : Effective herding of flatfish by cables with minimal seafloor contact

137

their capture by trawls. Flatfish generally react to
approaching objects at much closer ranges (<1 m) than
do roundfish and remain very close to the seafloor
when avoiding such objects. To target these behaviors,
towing cables (angled toward the trawl net across the
sea floor) are used in both demersal seines and otter
trawls to herd flatfish into the path of a capture net.
Flatfish avoid the approaching cable by continuous or
burst-and-pause swimming, both of which move them
gradually into the path of the capture device. Conven-
tional sweep cables have equal diameters throughout,
and no structures to interrupt their contact with the
seafloor. Flere, we test whether effective herding re-
sponses could be stimulated if such cables were raised
a short distance above the seafloor.

Like most flatfish fisheries, those in Alaska operate
on seafloors consisting of unconsolidated mixtures of
sand and mud. The potential for reducing damage to
the physical and biological features of these habitats
by raising sweeps a short distance off the bottom is
dependent on the presence of low vertical relief or flex-
ible structures of the bottom relief. This modification
would likely not prevent damage to high relief and rigid
or fragile features more common on rockier substrates.
For the modifications tested here to be effective, their
effects on both catch rates and seafloor features need
to be examined.

To develop practical modifications for the trawl sys-
tems used in Alaska’s flatfish fisheries, we convened a
series of meetings with trawler captains and gear man-
ufacturers. For initial study, they recommended raising
the sweeps slightly above the seafloor, allowing small
and flexible animals and other habitat structure to pass
beneath. In the current study we examine the proposed
change, focusing on determining which adjustments

maintain catch rates and on using direct observations
to demonstrate reduced seafloor contact.

Methods

To test the effect of the modified sweeps on their ability
to herd flatfish, we used a twin trawl system (Fig. 1). A
twin trawl system tows two separate trawls, including
sweeps, simultaneously on parallel, adjacent tracks.
Close proximity and simultaneous operation assure that
both nets encounter very similar compositions of fish
species at similar abundances. Therefore differences in
catch are principally due to differences in the capture
effectiveness of the two trawls. The only difference
between the trawls in this experiment was the use of
the elevating disks on the sweeps of the trawls.

Twin trawl tests

Field experiments were conducted during September
2006 in the eastern Bering Sea onboard the FV Cape
Horn. The Cape Horn is a 47-m trawler processor,
active in the mixed groundfish fisheries of the Bering
Sea. This vessel was equipped for a twin trawling
system, with an extra winch and towing cable. The
sweeps and trawls were towed with conventional trawl
doors on each side and a weight (clump) in the middle.
Both doors and the clump were towed from three sepa-
rate cables that were adjusted so that both sides fished
evenly. Towing sites were selected to provide com-
mercial catch rates of a mixture of the four principal
flatfish species of the Bering Sea shelf: yellowfin sole
( Limanda aspera); northern rock sole (Lepidopsetta
polyxystra); flathead sole (Hippoglossoides elassodon );

138

Fishery Bulletin 108(2)

and arrowtooth flounder ( Atheresthes stomias ). Towing
continued through both day and night periods, reflect-
ing commercial practice. All of the tows were in areas
with bottom substrates composed of mixtures of sand
and mud (McConnaughey and Smith, 2000).

The trawls were identical two-seam nets with 200-
mm mesh in the forward portions and equipped with
130-mm codends. The mouth opening of each net was
much wider (25 m) than high (3 m). Similar nets in a
single trawl configuration are used to target flatfish on
the eastern Bering Sea shelf. The distances between
each of the doors and the central clump were monitored
for equality with acoustic measurement systems and
were each approximately 80 m. Both nets were equipped
with sensors that indicated the direction of water flow
in relation to the center of the headrope. The three
towing cables were adjusted to keep that flow perpen-
dicular to the headropes of both nets and to keep their
door-clump openings equal, assuring comparable fishing
characteristics for both fishing systems.

The sweeps were 180-m long and were composed of
5-cm (2-inch) diameter combination rope constructed of
steel cable covered with polyethylene fiber. This is the
most common sweep material currently used in U.S.
Bering Sea flatfish fisheries. Sweeps used on vessels to
target flatfish on the eastern Bering Sea shelf are 200
to 450 m long (C. Rose, unpubl. data). The sweeps of
the two adjacent trawls had to be about half as long as
those used with commercial single trawls, because the
entire twin trawl system is similar in width to a con-
ventional single trawl. The shorter sweep lengths were
necessary to assure that the angle of the test sweeps
to the direction of towing was similar to that common
in the fishery. In this field study, clusters of disks (disk
clusters) were attached over the experimental sweeps
at 9-m (30-ft) intervals (Fig. 2). The disks were either

15, 20, or 25 cm (6, 8, or 10 inch) in diameter attached
to 5-cm (2 inch) diameter sweeps, creating nominal
clearance between the cables and the seafloor of 5, 7.5,
and 10 cm (2, 3, and 4 inch), respectively. Nominal
clearances are those immediately adjacent to a disk
when the disk is resting on a hard surface. The press-
ing of disks into the seafloor and the sagging of sweeps
between elevating devices would affect actual clear-
ances. For stability, disk clusters were approximately
the same length as their diameter. These disk clusters
were fixed in position with a combination of clamps
and rope seizings, which were run through the sweep
cable to prevent the clusters from sliding along the
cable. Disk clusters were installed on the aft 90 m of
the modified sweeps. Halfway through each experiment,
the sweeps were switched between the two trawl nets,
but each trawl net remained in place.

Catches from each trawl were kept separate through-
out the sampling process. As catches entered the sam-
pling area, they passed across a motion-compensated
flow scale which provided a total catch weight. All indi-
viduals of four flatfish species (yellowfin sole, northern
rock sole, flathead sole, and arrowtooth flounder) and
two gadids (Pacific cod [Gadus macrocephalus ] and
Alaska pollock [ Theragra chalcogramma ]) were sorted
into separate holding bins. These are the principal
flatfish and gadid species harvested from the eastern
Bering Sea shelf. Fish from each bin were then run
across a second flow scale to measure the weight of each
of these species. During the sorting of catch from each
trawl, 50-150 fish of each species were sampled and
their fork lengths were measured to 1-cm intervals to
determine their size composition. These length samples
were periodically taken from the catch as it passed
through the sorting area to avoid bias in case fish size
varied between parts of each catch. Because of their
large size, limited holding space and handling
requirements precluded adequate length sampling
of Pacific cod.

Tows were planned to last 2 hours, unless catch
sensors indicated substantial catches (>8 met-
ric tons [t] per net) before that time. Actual tow
durations ranged from 33 to 150 minutes. We
eliminated hauls where debris (e.g., crab pots)
was large enough to clog the net, or where gear
components became entangled, because such con-
ditions could influence gear performance and the
size and composition of the resulting catch. Tow
locations were selected in order to encounter com-
mercial concentrations of the major flatfish species
of the eastern Bering Sea shelf. Environmental
parameters at the trawl, including depth, tempera-
ture and light level, were sampled throughout the
experiment with a Mk9 logger (Wildlife Comput-
ers, Redmond, WA) mounted at the center of the
trawl’s headrope.

We used a high-resolution, rapid-update sonar
(SoundMetrics DIDSON, Dual-frequency IDen-
tification SONar, Lake Forest Park, WA) to ob-
serve how the sweep modifications affected sea-

Schematic diagram of a cluster of disks (disk cluster) attached
to trawl sweeps to raise the sweeps above the seafloor to
test whether this gear modification reduces flatfish herding.
Rubber disks (A, 20 cm-diameter, and B, 15 cm-diameter)
were installed over the sweep cable, between clamps (D) that
fix their location on the cable. Steel washers (C) prevented
rubber disks from passing over clamps. Ropes seized over and
tucked through cable (E) blocked clamps from shifting.

Rose et al. : Effective herding of flatfish by cables with minimal seafloor contact

139

180°W 1 70°W 160°W

Figure 3

Fishing locations (•) in the eastern Bering Sea for the 2006 tests of
the effects of raised sweeps on flatfish herding. Regions shaded with
diagonal lines are areas of trawl closures around the Pribilof Islands
and in Bristol Bay. Contour lines indicate depths.

floor contact. This was mounted in
a protective sled, which was towed
both behind the sweeps, to show
interactions between the sweeps
and the seafloor, and, separate
from the trawl, across the track of
a previous haul, to show marks left
on the seafloor. These observations
were made only on sweeps with the
20-cm disks. The sled was also
towed across tracks from previous
trawl tows with conventional and
modified sweeps and was equipped
with a video camera for detailed
imagery.

To estimate the proportional
change in catch due to the sweep
modifications, the difference be-
tween the natural logarithms of
the catch weights from modified
and unmodified trawls (Log Dif)
was calculated for each species
from each twin-trawl haul:

Log Dif = In (Catch modlfied) -

In (Catch unmodified). (1)

This statistic, equivalent to the log-
arithm of the ratio between catches with modified and
unmodified nets, was appropriate because absolute catch
sizes were uncontrolled and varied widely. A statistic
based on subtracting the untransformed trawl catches,
like that for an ordinary paired t-test, would have varied
proportionally to absolute catch rates, whereas catch
ratios, as measured by Log Dif were independent of the
fish densities encountered during each tow. Averages
and confidence intervals of Lo gDif were computed for
each species and sweep modification. To report these
results as ratios, the averages and confidence intervals
were then back-transformed with the exponential func-
tion. Catch results were only used for species with more
than a minimal catch (>10 fish) in both nets. The null
hypothesis that the sweep modifications did not affect
catch was tested with a t-test of whether average Lo gDif
was different from 0, equivalent to a paired t-test for
differences between the log-transformed catches.

To test whether the sweep modifications affected the
size-selectivity for different fish species and to minimize
variability, we pooled fish into three size classes for
each species, except for arrowtooth flounder, where a
wide size range made four size classes more appropri-
ate. The size-class boundaries were set so that approxi-
mately one-third (one-fourth for arrowtooth flounder) of
the fish in the combined control catches were in each
category. To maintain consistency with the weight-
based analysis of overall catch, and because the Alaska
trawl fleet classifies fish sizes by weight, the boundaries
of the size classes were defined by individual weights
instead of lengths, and the catches of each size class
were computed as weights, instead of numbers. Length-

weight functions from the annual Bering Sea shelf trawl
survey (NMFS, unpubl. data1) were used to convert the
sampled lengths to their corresponding weights. The
catch of each size class was estimated by expanding the
proportion of that size class, by weight, from the sample
of catch for that species. As with the total catch data,
averages and confidence intervals were calculated. We
used analysis of variance to test for differences between
size classes for each combination of species and for each
sweep modification.

Results

From 6 to 23 September 2006, 61 successful twin trawl
hauls were conducted, including 19, 26, and 16 hauls
with experimental sweep clearances of 5, 7.5, and 10 cm,
respectively. Depths at these tow sites (Fig. 3) ranged
from 70 to 117 m, and bottom temperatures ranged from
2.5° to 5.5°C.

Sonar imagery during towing showed that unmodified
sweeps produced a continuous cloud of disturbed sedi-
ment due to contact with the seafloor. Variation in the
density of that cloud appeared to result from contact
with high and low spots on the seafloor, and rapid oscil-
lation of strong and weak cloud intensity appeared to
be due to vibration of the sweeps. In contrast, the sedi-
ment cloud from the modified sweep appeared only di-
rectly behind the disk cluster. The only clouds from the

1 NMFS, Alaska Fisheries Science Center, RACE Division,
7600 Sand Point Way NE, Seattle, WA

140

Fishery Bulletin 108(2)

Figure 4

(A) Sonar and ( B ) video imagery of the seafloor after passage of the raised otter trawl sweeps. Video picture was
taken as the seafloor sled passed over the location indicated on the sonar image. Otter trawl sweeps were raised
with widely spaced disk clusters, which caused the parallel tracks seen in the sonar image and the flattened
swath in the video image.

sweeps themselves were brief puffs after contact with
high spots on the seafloor. Areas covered by the modi-
fied sweeps showed marks from the disk clusters that

were approximately 10-cm wide separated by seafloor
indistinguishable from unaffected areas (Fig. 4A). This
disk cluster mark was approximately 5% of the 2-m
interval between marks. This spacing is
much shorter than the 9-m spacing on the
cable because sweeps are sharply angled
to their direction of movement (angle-of-
attack). Images of such tracks from the
video (Fig. 4B) showed a flattening of very
low-profile surface textures.

The use of 15-cm disks on the sweeps
did not cause significant differences in
catch rates (LogDif was not different
from 0) for any of the six species, and
only the pollock catch rate changed (12%
increase, P=0.007) with the 20-cm disks
(Fig. 5). Northern rock sole and flathead
sole catches both decreased significantly
(-11%, PcO.OOl, and -5%, P=0.02, respec-
tively) when the 25-cm disks were used,
whereas pollock catch increased again
(+12%, P=0.03). Decreases for the other
two flatfish were also observed — although
not statistically significant at the 0.05
level (P=0.08 for arrowtooth flounder and
P=0.07 for yellowfin sole). A consistent
decrease in the mean relative catch with
increasing disk size for all of the flatfish
species, although only significant for the
largest disks, indicates that smaller ef-
fects may have occurred for the smaller
disks that could not be statistically de-

1.30 -i

Yellowfin

sole

Pacific cod Alaska
pollock

Figure 5

Estimates of and 95% confidence intervals for the ratios of fish catches
with the modified trawl sweeps raised to three different heights off the
seafloor to fish catches with conventional sweeps for the four principal
flatfish species (yellowfin sole [Limanda aspera ]; northern rock sole
[Lepidopsetta polyxystra]\ flathead sole I Hippoglossoides elassodon J;
arrowtooth flounder [Atheresthes stomias ]); and two principal gadid
species: Pacific cod ( Gadus macrocephalus)\ and Alaska pollock ( Ther -
agra chalcogramma ) taken in Bering Sea trawl fisheries.

Rose et al.: Effective herding of flatfish by cables with minimal seafloor contact

141

Figure 6

Size compositions for flatfish and gadid species taken during tests of
whether raised trawl sweeps reduce herding of fish. Yellowfin sole (Limanda
aspera); northern rock sole ( Lepidopsetta polyxystra) ; flathead sole (Hippo-
glossoides elassodon)\ arrowtooth flounder ( Atheresthes stomias); Pacific cod
( Gadus macrocephalus ); and Alaska pollock ( Theragra chaleogramma).

tected in our experiment. Pacific cod
catches did not change significantly
with any of the modifications.

For evaluating the likelihood of
substantial losses of catch, the confi-
dence intervals provide more informa-
tion than the basic significance tests
alone. For example, the lower confi-
dence bounds for the effects of 20-cm
disks on flatfish catches leave only a
2.5% (1 of 40) probability that catch
losses would exceed 4-6%. Correspond-
ing “worst case” losses for the 15-cm
disks were even smaller. Similarly, al-
though none of the Pacific cod catch
results passed the threshold of a 95%
two-tailed probability of being differ-
ent from no change, all three of the
confidence intervals were almost en-
tirely above a value of 1. Therefore, a
trawler could implement one of these
modifications with little expectation of
catching fewer Pacific cod and with a
reasonable chance of slight increases
in Pacific cod catch.

The size composition of each spe-
cies from the unmodified nets (Fig. 6)
showed truncation at the lower end of
the size distribution, owing to use of
large mesh in the body of the net (20 cm, stretch mea-
sure), intermediates (14 cm) and codends (15 cm) that
release smaller fish. Although the proportions varied
somewhat between experiments, each study encoun-
tered a representative range of sizes available to the
commercial fishery.

ANOVA tests for differences in catch effects between
major size classes (thirds or quartiles of control size
frequencies) revealed no significant differences for any
of the flatfish species (Fig. 7). One significant difference
(P=0.04) was detected for pollock in sweeps with the
smallest disks (15 cm), attributable to a lower catch
rate of the smallest pollock. Confidence intervals were
included in Figure 7 to aid comparisons between size
groups within species and sweep modification classes.
Confidence intervals were wider for the largest and
smallest categories because few individuals from these
ranges were encountered in some tows, increasing vari-
ability, whereas all tows had substantial numbers of
fish in the central ranges.

Discussion

Flatfish can be effectively herded by trawl sweeps
and with greatly reduced seafloor contact. Signifi-
cant catch reductions, averaging 5% for flathead sole
and 11% for rock sole, were only detected when 25-cm
disks were installed that raised the sweeps 10 cm
above the substrate at the ends of each 9-m section. No
detectable catch reductions occurred during tests with

smaller clearances (5 and 7.5 cm). Confidence intervals
indicated only a 2.5% probability of catch reductions
greater that 5% with 7.5-cm clearances. Nor did sweeps
with such clearances appear to change size selectivity
significantly.

Flatfish exhibit predator avoidance behaviors that
allow them to be effectively herded by the sweeps. In
contrast to roundfish, flatfish cease movement when a
predator is detected and only flee upon very close ap-
proach (Ryer, 2008). Therefore, observed flatfish reac-
tions to trawl gear (Main and Sangster, 1981; Rose,
1996; Ryer and Barnett, 2006) mostly occur at horizon-
tal ranges of much less than 1 m. However, because con-
ventional fishing gear has either continuous or closely
spaced contact with the seafloor, there has been little or
no information to assess the role of gear contact or prox-
imity to the seafloor in either initiating or sustaining
the flight behaviors that result in herding. Given the
cryptic behaviors of flatfish, we could not assume that
stimuli several centimeters above the seafloor would be
as effective as those that would directly contact flatfish
on the seafloor. The current results demonstrate that
flatfish do respond with effective herding behaviors to
sweep cables displaced from the seafloor by 5 to 10 cm.
Even the largest of the flatfish encountered here would
not have contacted the raised sweeps if they remained
resting on the seafloor. At the highest clearance (10
cm), slightly reduced catches indicated that the flight
response began to break down and some of the flat-
fish were not herded as well as with the conventional
sweeps. Winger et al. (2004) found that flatfish size

142

Fishery Bulletin 108(2)

Yellowfin sole

Northern rock sole

Elevating disk diameter

Flathead sole

Elevating disk diameter

□ <1000 g
■ 1000-1500 g
£1 >1500 g

Elevating disk diameter Elevating disk diameter

Figure 7

Estimates of and 95% confidence intervals for ratios of fish catches during tests with modified trawl sweeps raised to
three different heights off of the seafloor to catches with conventional sweeps for broad size classes of four principal flat-
fish species and a principal gadid species taken in Bering Sea trawl fisheries: yellowfin sole ( Liman da aspera); northern
rock sole (Lepidopsetta polyxystra)\ flathead sole (Hippoglossoides elassodon)\ arrowtooth flounder ( Atheresthes stomias );
and Alaska pollock (Tlieragra chcilcogramma) .

affected behavioral responses to approaching sweeps,
including tailbeat frequency and swimming endurance.
Although any of these behaviors could affect herding-
related capture rates, the current study did not indicate
behavioral differences between size classes in response
to the elevated sweeps.

We followed commercial practices in the gear type
used, weight-based catch metrics, towing durations,
catch handling, and round-the-clock operations. This

procedure was undertaken to increase the relevance of
our results to those with the greatest stake in deciding
on the use of these modifications: the fishermen and
fishing companies. Fishermen actively participated in
designing the gear modifications and in conducting the
research.

To examine consequences of using modified sweeps
in the fishery and to improve precision, all tows were
analyzed together, including day and night tows, even

Rose et al.: Effective herding of flatfish by cables with minimal seafloor contact

143

though light levels affect the herding process (Ryer and
Barnett, 2006). The effects of light on flatfish herding
are analyzed and reported in a separate paper (Ryer
et al., 2010).

Although not the focus of this study, an unexpected
result was the increase in pollock catches that occurred
with two of the sweep modifications. Pollock herd differ-
ently from flatfish, reacting to stimuli at much greater
distances (Rose, 1996). The forward sections of the
most modern pollock trawls have “meshes” that are
more than 25-m long. Although large groups of pol-
lock could easily swim through such meshes, they still
avoid the netting and are eventually herded into parts
of the net that physically restrain them. These nets
would not work if pollock herded only at short ranges.
Separation of the sweeps from the seafloor, or the disk
clusters themselves, could have increased visibility of
the sweeps, which may have enhanced pollock herding.
Both factors would be reduced at the smallest disks,
where herding improvement was not detected.

Sonar observations of the elevated sweeps showed
that their interaction with the seafloor was radically
changed. The continuous sediment clouds produced
along the entire length of the unmodified sweeps were,
for the modified sweeps, reduced to isolated clouds be-
hind each disk, indicating substantial reductions in the
area of direct contact. Therefore, any effects based on
direct contact, as well as resuspension of sediments,
should have been greatly reduced. The sonar images
of the seafloor after passage of the sweep showed that
the contact area of the disks was approximately 5% of
the total swept area. Seafloor texture between the disk
tracks was indistinguishable from unaffected areas,
but areas covered by conventional sweeps showed slight
smoothing. The seafloor directly contacted by the disks
was uniformly smoothed. Although the texture change
due to conventional sweeps appeared slight, the resus-
pension observed during fishing indicated some distur-
bance of the bottom and we believe that the substantial
reduction of contact due to using the disks more than
compensates for any increased disturbance to the small
area directly under the disks.

In another recent study (Guyonnet et al., 2008), the
concept of slightly raising trawl sweeps, therein called
“legs,” was also applied to reduce their impact on the
seafloor. Instead of disk clusters, Guyonnet et al. used
neutrally buoyant sweep material that was weighted
only by dangling chains attached every 50 cm. They
also found no significant effects on catch composition
or size selectivity for target animals. They found that
damage to benthic animals was reduced with the al-
ternative gear.

Our results alone, although promising, do not address
the full potential of sweep modifications to reduce the
effects on the seafloor of trawling for Bering Sea flat-
fish. Although creating several centimeters of separation
between the sweeps and the seafloor greatly reduces the
potential for damage to infauna and small epifauna,
it does not prevent contact with seafloor features and
animals larger than that spacing. The vulnerability of

ecosystem features to trawling operations is a function
of the amount of damage caused by each trawl exposure
(e.g., the proportion of a particular species in the path
of a trawl that dies due to trawl contact) and the fre-
quency and coverage of the trawling effort. An analysis
of such factors for the Bering Sea shelf highlighted
structure-forming animals as the seafloor feature most
vulnerable to trawling.2 The structure-forming animals
of the eastern Bering Sea shelf are generally small and
flexible; therefore it is quite conceivable that creating
a space below the sweeps could also reduce damage to
these animals. That potential is being examined by
the authors in a subsequent study that will focus on
how these sweep modifications change damage rates to
structure-forming animals of the Bering Sea shelf.

Successful gear modifications for reducing trawling
effects on seafloor habitats would add a habitat pro-
tection option in addition to area closures and gear
switching. Closures of areas to trawling can move fish-
ing effort from productive grounds, and therefore can
increase the total effort required or concentrate fishing
and its effects in the remaining fishing grounds (Fu-
jioka, 2006). The list of alternative gear for harvesting
these flatfish is quite limited and none are without
some negative effects on habitat. With beam trawling,
herding sweeps are not used to concentrate fish into the
path of the capture device. Therefore, the entire area
from which fish are collected is swept with the capture
net itself. Studies to reduce the effects of beam trawls
on habitat have focused on other stimuli to move fish
from the seafloor into the net (van Marlen et al., 2005).
The capture process for demersal seines is similar in
many ways to that of Alaska otter trawls with long
sweeps — weighted cables are pulled across the seafloor
to herd fish into the path of a capture net. Demersal en-
tangling nets depend on natural movements of the fish
to bring them to the gear, and therefore they are effec-
tive only during periods when fish are actively moving.
They are still unlikely to produce catch rates similar
to those produced with trawls unless vast fleets of nets
are deployed. Such extensive net deployments would
exacerbate the most notable problem with demersal
entangling nets — ghost fishing of derelict and lost gear.
Finally, although longline fishing is the foundation for
one of the most successful commercial flatfish fisheries
(Pacific halibut), most flatfish species are not of the size
and do not have a predatory diet that make longlines
particularly effective.

Implementing the trawl gear modifications described
here would require some adaptations in equipment and
handling methods for fishermen. The volume of the
elevating devices would require additional space on
deployment reels or net drums, thus requiring either
that sweep lengths be shortened to fit onto the reels
or larger reels be installed on vessels. The disks would

2 Final environmental impact statement for essential fish
habitat identification and conservation in Alaska. April
2005 [online], http://www.fakr.noaa.gov/habitat/seis/efheis.
htm.

144

Fishery Bulletin 108(2)

also complicate deployment and retrieval because they
do not wrap as evenly onto reels as unmodified sweeps.
Potential advantages with the use of disks would in-
clude longer usability of sweeps and reduced drag (im-
proved fuel efficiency), both due to reduced contact of
the sweeps with the seafloor. An important factor in
identifying these implementation and operational issues
early, as well as in the development of potential solu-
tions, has been the direct participation of the fishing in-
dustry in this research and our ability to conduct these
tests under conditions identical to most of the important
operational aspects of the commercial fishery.

Acknowledgments

The authors thank K. Hjelm, captain of the FV Cape
Horn , and his crew for their tireless work and creativ-
ity in helping with this research. We also appreciate
the support of numerous Bering Sea trawl captains and
companies in discussing, motivating, and moving our
study forward. Scott McEntire developed the systems
necessary for collection of video and sonar data, a criti-
cal contribution to this project. We are very grateful to
our sampling crew: D. Benjamin, N. Roberson, E. Acuna,
and H. Kenney, and those from a pilot study, J. Olsen,
J. Hagga, and C. Shavey. Our thanks are also extended
to the many reviewers both anonymous and within the
Alaska Fisheries Science Center, whose thoughtful com-
ments and suggestions greatly improved this article.

Literature cited

Barnes, P. W., and J. P. Thomas (eds.).

2005. Benthic habitats and the effects of fishing. Am.
Fish. Soc. Symp. 4, 890 p.

Bellman, M. A., S. A. Heppell, and C. Goldfinger.

2005. Evaluation of a US west coast groundfish habi-
tat conservation regulation via analysis of spatial and
temporal patterns of trawl fishing effort. Can. J. Fish.
Aquat. Sci. 62:2886-2900.

Churchill, J. H.

1989. The effect of commercial trawling on sediment
resuspension and transport over the Middle Atlantic
Bight continental shelf. Cont. Shelf Res. 9:841—865.

Fujioka, J. T.

2006. A model for evaluating fishing impacts on habitat
and comparing fishing closure strategies. Can. J. Fish.
Aquat. Sci. 63:2330-2342.

Guyonnet, B., J. Grail, and B. Vincent.

2008. Modified otter trawl legs to reduce damage and
mortality of benthic organisms in North East Atlantic
fisheries (Bay of Biscay). J. Mar. Syst. 72:2-16.

Hannah, R. W.

2003. Spatial changes in trawl fishing effort in response
to footrope diameter restrictions in the U.S. west coast
bottom trawl fishery. N. Am. J. Fish. Manag. 23:693-
702.

He, P.

2007. Technical measures to reduce seabed impact of

mobile fishing gears. In Bycatch reduction in the world’s
fisheries (S. J. Kennelly ed.), p. 141-179. Springer,
New York.

Kaiser, M. J., D. B. Edwards, P. J. Armstrong, K. Radford,
N. E. L. Lough, R. P. Flatt, and H. D. Jones.

1998. Changes in megafaunal benthic communities in
different habitats after trawling disturbance. ICES J.
Mar. Sci. 55:353-361.

Kaiser, M. J., N. Graham, C. S. Rose, and P. H. Weibe.

2007. Ecosystem-sensitive approaches to fishing: reconcil-
ing fisheries with conservation through improvements in
fishing technology. ICES J. Mar. Sci. 64:1610-1611.

Lokkeborg, S.

2005. Impacts of trawling and scallop dredging on ben-
thic habitats and communities. FAO Fish. Tech. Paper
472, 58 p. FAO, Rome.

Main, J., and G. I. Sangster.

1981. A study of the fish capture process in a bottom
trawl by direct observations from a towed underwater
vehicle. Scott. Fish. Res. Rep. 23, 24 p. Marine
Scotland, Aberdeen.

McConnaughey, R. A., C.B. Dew, and K. Meir.

2000. An examination of chronic trawling effects on soft
bottom benthos in the eastern Bering Sea. ICES J. of
Mar. Sci. 57:1377-1388.

McConnaughey, R. A., and K.R. Smith.

2000. Associations between flatfish abundance and sur-
ficial sediments in the eastern Bering Sea. Can. J.
Fish. Aquat. Sci. 57:2410-2419.

Prena, J., P. Schwinghamer, T. W. Rowell, D. C. Gordon, K. D.
Gilkinson, W. P. Vass, and D. L. McKeown.

1999. Experimental otter trawling on a sandy bottom
ecosystem of the Grand Banks of Newfoundland: analysis
of trawl bycatch and effects on epifauna. Mar. Ecol.
Progr. Ser. 181:107-124.

Rose, C. S.

1996. Behavior of North Pacific groundfish encounter-
ing trawls: applications to reduce bycatch. In Solv-
ing bycatch: considerations for today and tomorrow, p.
235-241. Alaska Sea Grant College Program Report
96-03, Univ. Alaska, Fairbanks.

Ryer, C. H.

2008. A review of flatfish behavior relative to trawls.
Fish. Res. 90:138-146

Ryer, C. H., and L. A. K. Barnett.

2006. Influence of illumination and temperature upon
flatfish reactivity and herding behavior: Potential
implications for trawl capture efficiency. Fish. Res.
81:242-250.

Ryer, C. H., C. S. Rose, and P. S. Iseri.

2010. Flatfish herding behavior: diel patterns of trawl
sweep efficiency as inferred from field and laboratory
manipulations. Fish. Bull. 108:145-154.

Tuck, I. D., S. J. Hall, M. R. Robertson, E. Armstrong, and D. J.
Basford.

1998. Effects of physical trawling disturbance in a previ-
ously unfished sheltered Scottish sea loch. Mar. Ecol.
Progr. Ser. 162:227-242.

van Marlen, B, M. J. N. Bergman, S. Groenwold, and M. Fonds.

2005. New approaches to the reduction of non-target
mortality in beam trawling. Fish. Res. 72:333-345.

Winger, P. D., S. J. Walsh, P. He, and J. A. Brown.

2004. Simulating trawl herding in flatfish: the role of fish
length in behavior and swimming characteristics. ICES
J. Mar. Sci. 61:1179-1185.

145

Flatfish herding behavior in response

to trawl sweeps: a comparison of diet responses

to conventional sweeps and elevated sweeps

Email address for contact author: cliff.ryer@noaa.gov

1 Fisheries Behavioral Ecology Program

Resource Assessment and Conservation Engineering Division

Alaska Fisheries Science Center, NOAA Fisheries

Hatfield Marine Science Center

2030 Marine Science Drive

Newport, Oregon 97365

2 Resource Assessment and Conservation Engineering Division
Alaska Fisheries Science Center, NOAA Fisheries

7600 Sand Point Way
Seattle, Washington 98115

Abstract — Commercial bottom trawls
often have sweeps to herd fish into
the net. Elevation of the sweeps off
the seafloor may reduce seafloor
disturbance, but also reduce herd-
ing effectiveness. In both field and
laboratory experiments, we examined
the behavior of flatfish in response
to sweeps. We tested the hypotheses
that 1) sweeps are more effective at
herding flatfish during the day than
at night, when fish are unable to see
approaching gear, and that 2) eleva-
tion of sweeps off the seafloor reduces
herding during the day, but not at
night. In sea trials, day catches were
greater than night catches for four
out of six flatfish species examined.
The elevation of sweeps 10 cm sig-
nificantly decreased catches during
the day, but not at night. Laboratory
experiments revealed northern rock
sole ( Lepidopsetta polyxystra ) and
Pacific halibut ( Hippoglossus stenol-
epis ) were more likely to be herded
by the sweep in the light, whereas in
the dark they tended to pass under or
over the sweep. In the light, elevation
of the sweep reduced herding, and
more fish passed under the sweep. In
contrast, in the dark, sweep elevation
had little effect upon the number of
fish that exhibited herding behavior.
The results of both field and labo-
ratory experiments were consistent
with the premise that vision is the
principle sensory input that controls
fish behavior and orientation to trawl
gear, and gear performance will differ
between conditions where flatfish can
see, in contrast to where they cannot
see, the approaching gear.

Manuscript submitted 8 June 2009.
Manuscript accepted 15 December 2009.
Fish. Bull. 108:145-154 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

Clifford H. Ryer1 (contact author)
Craig S. Rose2
Paul J. Iseri1

Trawl catches for many fish species
exhibit diel patterns (Casey and Myers,
1998). This is generally viewed as a
product of two independent factors:
availability and catchability of the
fish species. Many gadids exhibit diel
changes in availability associated with
vertical migration (Beamish, 1965;
Casey and Myers, 1998; Schabets-
berger et al., 2000). Gadids aggregate
close to the bottom during the day and
are highly available to bottom trawls.
At night, dispersal into the overly-
ing water renders them less available.
Interestingly, for many flatfish species
the opposite pattern, higher catches
at night, has been observed (Walsh,
1991; Walsh and Hickey, 1993; Casey
and Myers, 1998). Seasonal migra-
tions will occasionally take flatfish
into the water column (Metcalfe et al.,
1990; Nichol and Sommerton, 2009),
as will the occasional exploitation of
pelagic prey. However, under normal
circumstances many flatfish species
appear to remain on the bottom and
are consistently available to trawl
gear, day or night. As a consequence,
greater flatfish catches at night are
thought to be associated with higher
catchability, that is, with a decreased
ability to evade capture (Ryer, 2008).

Video cameras mounted on trawls,
as well as supplemental bag nets be-
hind the main net, have documented

extensive flatfish escapement beneath
the footrope during the day (Main
and Sangster, 1981; Walsh, 1988). For
obvious technical reasons, behavior
in front of the footrope, or sweeps, at
night has not been observed in field
studies, except with flash photogra-
phy (Walsh and Hickey, 1993). How-
ever, laboratory experiments indicate
that northern rock sole ( Lepidopsetta
polyxystra), Pacific halibut ( Hippo-
glossus stenolepis), and English sole
( Parophrys vetulus) are more likely
to rise or hop into the water column
during darkness, than to herd (Ryer
and Barnett, 2006). By moving off the
bottom, these fish remove themselves
from the “zone of influence” of the
ground gear, and as they cease swim-
ming they pass over the footrope and
into the net. This behavior potentially
explains why flatfish are captured in
greater numbers at night.

This paradigm, i.e., higher flatfish
catches at night, stems largely from a
series of published studies (Main and
Sangster, 1981; Walsh, 1988, 1991;
Walsh and Hickey, 1993; Casey and
Myers, 1998; and references therein),
based on survey trawls. On survey
trawls, the combined length of bridles
and sweeps is typically minimized. In
contrast, on commercial flatfish trawls
lengthy sweeps are used to herd fish
inward toward the net (Winger et

146

Fishery Bulletin 108(2)

al., 1999, 2004). On some modern flatfish trawls these
sweeps may be up to 400 m in length, and as much
as 90% of the seafloor is subject to the action of gear
which is designed to affect capture by manipulating
flatfish swimming behavior. But for the very reason that
footropes are more efficient in the dark, sweeps may
be less efficient. If flatfish, unable to see the approach-
ing sweep, rise or hop into the water column, rather
than herding as happens during the day, they will pass
over the sweep and be lost to the catch. This situation
raises the possibility that flatfish trawls that rely upon
sweep herding may capture more flatfish during the day
than during the night — a pattern not seen with survey
trawls, which have minimal sweeps.

In this study we investigated the performance of
trawls equipped with sweeps under day and night con-
ditions, using a combination of manipulative at-sea and
laboratory procedures. For our at-sea experiment, we
used a data set acquired during a series of cruises in
the eastern Bering Sea, the goal of which was to evalu-
ate sweeps designed to reduce damage to benthic habi-
tat (Rose et al., 2010). In brief, trawling was conducted
with sweeps that were elevated, to various degrees, off
the seafloor to evaluate the trade-off between reductions
in habitat disturbance and decreased flatfish herding
efficiency. Here we test hypotheses related to our prin-
ciple premise: flatfish behavior initiated by ground-gear
is principally controlled by ambient light levels. More
specifically, first we test the hypothesis that trawls
configured with control (commercial type) sweeps in
contact with the bottom, will catch more flatfish dur-
ing the day than during the night. Following from this,
we test a second related hypothesis: the elevation of
sweeps off the bottom will have differential effects,
day as opposed to night. During the day, elevation will
reduce sweep efficiency, resulting in lower flatfish catch.
During the night, because sweeps are already rela-
tively ineffective, elevation of the sweeps will have no
influence upon their efficiency, as reflected by flatfish
catch. Lastly, we conducted comparable experiments
under both light and dark conditions, using simulated
ground-gear in the laboratory where behavior could be
quantified, to ascertain whether the proposed effects
of elevated sweeps on catch are directly attributable to
ambient-light-mediated differences in flatfish behavior
in relation to ground gear.

Methods

At-sea experiments

Tows of paired trawls (control and elevated sweeps)
were conducted during September 2007 in the eastern
Bering Sea onboard the FV Cape Horn. Details of gear
and onboard procedures can be found in Rose et al.
(2010). Briefly, the Cape Horn is a 47-m trawler proces-
sor, configured so as to allow twin trawling, i.e., fishing
with two identical nets side-by-side. Each net had a
set of independent 180-m sweeps, being spread by one

otter board on each side of the vessel, and connected
in the middle by a towed weight (clump). The sweeps
were composed of 5-cm diameter combination rope, con-
structed of steel cable and covered by polyethylene fiber.
Modifying the sweeps on one net, while keeping all other
trawl characteristics consistent, allowed the difference
between the two catches to reflect the effect of the modi-
fication. In this field study, disk clusters were attached
to the experimental sweeps at 9-m intervals. The disks
were either 15, 20, or 25 cm in diameter. This created
a nominal spacing between the sweeps and the seafloor
of 5, 7.5, and 10 cm, respectively. Test tows were made
with modified sweeps on one net and unmodified sweeps
on the other. Halfway through each experiment, the
modified sweeps and unmodified sweeps were switched
(left to right, right to left).

Catches from each trawl were kept separate until the
entire catch had been sampled. As catches entered the
sampling area, they were passed across a motion-com-
pensated flow scale to determine total catch weight. The
five or six most abundant species were then completely
sorted into holding bins. Fish from each bin were then
run across a second flow scale to measure the weight of
each of those species. To estimate the weight of other
species, samples of the unsorted catch were taken at
intervals, sorted, and weighed by species. The com-
position of these samples was then expanded to the
weight of the entire catch by calculating the fraction
of the sample weight to the total catch weight. For the
species cited in this paper, Pacific halibut and Alaska
plaice catches were estimated from the samples and
all other species were fully weighed on the second flow
scale. During the sorting phase, samples of 50-150 fish
of each species were drawn and measured to determine
their length composition. Length samples were taken
from throughout the catch as it passed through the
sorting area and the length of each individual in the
sample was measured

Sixty-one paired hauls were made over depths rang-
ing from 70 to 117 m. Ambient light on the bottom is
greatly influenced by water depth. To minimize poten-
tial depth effects upon ambient light, we limited our
analysis to hauls where depth was between 79 and
94 m: a 15-m range. In addition, we eliminated hauls
where large debris (crab pots, etc.) were encountered,
or where gear components became entangled, assuming
that such conditions would influence gear performance
and catch. After examining in situ light measurements
(Wildlife Computers, MK9 light meter, Redmond, WA)
we further eliminated daytime hauls where light levels
fell below l.OxlO-4 pmol photons/m2/s, and nighttime
hauls exceeding 1.0xl0~5 pmol photons/m2/s. This step
eliminated hauls made around dusk or dawn and set
a clear differentiation between daytime and nighttime
light. In the resulting data set (36 hauls), mean tow
depth did not differ between nighttime and daytime
tows (day: n= 7, mean [x] = 82 m, standard error [SE] = 1;
night: n =19, x=84 m, SE = 1; t(34]=1.54, P=0.133). Tow
durations ranged from 33 to 150 min, being somewhat
longer at night (x=115.8, SE = 5.9) than during the day

Ryer et al.: Flatfish herding behavior in response to trawl sweeps

147

(x- 87.5, SE = 6.3, £[34] = 3.28, P = 0.003). During long
tows, accumulating catch can distort meshes and back
up into the intermediate portion of the net, altering
gear selectivity (Herrmann, 2005). However, catches in
this study were small compared to net capacity, never
filling the codend. Hence, we assume that differences
in duration between day and night did not influence
net performance or fish catchability in a manner that
would bias our results. Similarly, during long tows
proportionately more fish will tire and fall back into
the net, particularly so for many roundfish species,
which can swim for prolonged periods in front of the
net (Main and Sangster, 1981). However, flatfish typi-
cally swim for less than 1 minute in front of nets ( Ryer,
2008), and thus this source of bias was also unlikely
in our study.

For our first analysis, we compared daytime and
nighttime catches from the control nets only; where
sweeps were in contact with the bottom along their
entire length. Catch per unit of effort (CPUE: kg/min)
was calculated for total catch (all species) as well as for
six flatfish species: yellowfish sole ( Limanda aspera );
flathead sole ( Hippoglossoides elassodon); arrowtooth
flounder ( Atheresthes stomias)\ rock sole ( Lepidopsetta
spp. ); Alaska plaice ( Pleuronectes quadrituberculatus);
and Pacific halibut. CPUE values were natural log (In)
transformed and tested for day and night differences
with t-tests (Sokal and Rohlf, 1969). Where variances
were heteroscedastic, Satterthwaite’s adjusted degrees
of freedom were used (Snedecor and Cochran, 1980). Be-
cause CPUE was based upon weight, we also compared
mean total length between daytime and nighttime hauls
for each flatfish species.

For our second analysis, we used the subset of samples
from trawls where 25.4-cm disks were attached to el-
evate sweeps of the experimental net to an approximate
height of 10 cm (the distance between sediment surface
and bottom of the sweep material). For this analysis,
catch of the experimental net was compared to that of
the paired control net (with bottom contact sweeps) by
using a paired /-test (Sokal and Rohlf, 1969). Separate
analyses were conducted for daytime (rc = 10 pairs) and
nighttime (n = 5 pairs) hauls. Similar analysis was con-
ducted for flatfish lengths.

Laboratory experiments

Northern rock sole and Pacific halibut were collected
as age-0 juveniles by using a 2-m plumb-staff beam
trawl from Chiniak Bay, Kodiak, Alaska. Fish were
transported to the Hatfield Marine Science Center in
Oregon and reared in 2. 2-m (diameter) circular tanks
with flow-through seawater (28-35%o, 9°C [± 1°]) on a
diet of krill and gelatinized food. After reaching age 1,
fish were transferred to 3-m diameter pools for contin-
ued growth.

Simulated sweep exposure took place in an elongated
tank (10.7x1.5x1.2 m) filled to a depth of 0.9 m. This
tank was provided with flow-through seawater (28-35%e)
and located in a light-proof room, allowing for control of

illumination by an overhead bank of fluorescent lamps.
The tank bottom was covered to a depth of 4 cm with
sand, allowing flatfish to completely bury themselves.
Details of this apparatus are presented elsewhere (Ryer
and Barnett, 2006) and will only be described briefly
here. By means of a moveable carriage a simulated
sweep was propelled down the length of the tank. This
sweep consisted of a piece of 5-cm diameter PVC pipe,
painted green to resemble the actual sweep used in the
field study. It could be positioned so that it ran down
the tank in contact with the bottom, or elevated so that
it was approximately 10 cm off the bottom.

Fish were maintained on a 12/12 h photo period
during all experiments, with lights turned on at 0700
and off at 1900. At 1600 on the day before the trials,
the length of the tank was subdivided into three equal
3-m sections, by means of four removable partitions,
of which two of these partitions prevented fish from
moving to the extreme ends of the tank. Next, fish
were introduced to each of the three main sections of
the tank. This sectioning assured that fish would not
aggregate in a single area of the tank. At 0800 on the
day of trials, the footrope carriage was lowered into
the tank, behind one of the end partitions and secured
to its tracks. Then the lighting was either turned off
(dark trials) or kept on (light trials), and fish were
allowed 2 h acclimation before a trial. Illumination at
the sand surface was measured once at the beginning
of the study. For light trials, illumination was approxi-
mately 1.5 pmol photons/m2/s (-125 lux), whereas, for
dark trials illumination was <lxl0-8 pmol photons/m2/s
(~10-6 lux). Both species used in this study have the
same light thresholds ( 10 5 pmol photons/m2/s) for vi-
sual discrimination of small motile prey (Hurst et al.,
2007), and we assumed they would see approaching
footrope in the light trials, but not in the dark trials.
Illumination was measured with a research radiometer
(International Light Inc., Model IL1700, Peabody, MA)
equipped with a 2ji PAR (photosynthetically active ra-
diation) sensor. Water supply to the tank was filtered
through sand, making it unlikely that water clarity,
and hence light levels, changed appreciably from day
to day. At 1000 h, immediately before a trial, the parti-
tions were removed; for dark trials red flashlights were
used during this process, and care was taken to avoid
shining the lights directly into the tank. Five minutes
later the footrope carriage was pulled from one end of
the tank to the other at a speed of 1.0 m/s (± 0.1 m/s), a
speed roughly equal 3.6 km/h or 2 knots; flatfish trawls
are commonly towed at 2-5 knots. Afterwards, the
lights in the room, if turned off, were turned back on
and rakes were used to herd fish back into each of the
three main sections of the tank, after which the parti-
tions were put back in place and the footrope carriage
was removed from the tank. This entire process was
repeated in the afternoon, using the opposite lighting
from that of the morning: at 1200 h, a footrope carriage
was lowered into the tank and lighting was adjusted;
at 1400 h, partitions were removed and the footrope
carriage was pulled. We assume that this alternation

148

Fishery Bulletin 108(2)

in treatment order precluded any bias attributable to
flatfish habituation or learning.

Positioned behind and above the footrope were three
(50W) infrared LED (light emitting diode) lamps, aimed
forward and down, so that they illuminated the footrope
and tank bottom immediately in front of the footrope.
The wavelength of light emitted by these lamps peaked
at 880 nm, and emissions dropped to 0 below 760 nm.
Most fish are insensitive to light at those wavelengths
(Douglas and Hawryshyn, 1990) and results from light-
threshold feeding studies for all three flatfish species
used in this study are consistent with this generaliza-
tion (Hurst et al., 2007). Two underwater video cameras
(Aqua-Vu, model ZT-120, Crosslake, MN ) were mounted
alongside the lamps, also directed at the area in front
of the footrope. This arrangement allowed for visual
monitoring out to 1.1 m in advance of the footrope. The
video footage was captured from a remote location by
digital mini-DV recorders.

Trials were conducted with three age classes of Pa-
cific halibut: age-1, age-2, and age-3, as well as age-
2 northern rock sole. For age-3 Pacific halibut, three
groups of five fish each were examined. Trials took
place over two consecutive days. On the first day sweep
height was randomly set to either the “in contact” or
“elevated” position. On the second day the alternative
position was used. During each day, fish were exposed
to the simulated sweep approach twice; once in the light
and once in the dark. The order of application of light
vs. dark trials was also randomly determined. After the
second day fish were then removed from the tank, their
total length was measured, and they were replaced by
a new group. Age-3 Pacific halibut ranged from 37-52
cm in total length.

For age-2 Pacific halibut, age-1 Pacific halibut, and
age-2 rock sole, groups consisting of 10 fish each were
trialed differently. Each group was trialed for only a
single day, at one sweep height. For age-2 Pacific hali-
but, six groups were trialed at each sweep height. For
age-1 Pacific halibut and age-2 northern rock sole, five
groups were trialed at each sweep height. As before,
the order of light and dark trials was randomized. Age-
2 Pacific halibut ranged from 19-31 cm, age-1 hali-
but from 8-14 cm, and age-2 northern rock sole from
9-17 cm.

Fish behavior was quantified by using the slow-mo-
tion playback of digital video. First, the number of
fish encountered, i.e., observed, as the sweep made
its transit from one end of the tank to the other, was
recorded from each trial. Then the initial behavioral
response of each observed fish was assigned to one
of four categories: 1) pass under, 2) hop, 3) rise, and
4) herd. Fish characterized by “under” either did not
react at all to the approaching sweep, or reacted when
contacted by the sweep, but passed under the sweep
as it progressed down the tank. “Hop” characterized
fish that reacted to the sweep with one or two sinu-
soidal body undulations, typically after being struck
by the sweep, which resulted in the fish “hopping” off
the substrate. However, this initial startle reaction

10 ’

•

Night

10 2
10 3

O O
■ Qr-O-

Day

10 4

o c
o

10 s

•

10 6

• • X

•

• •

• %

•

• —

Depth

Figure 1

In situ natural log-transformed light data for trawl tows
conducted during day and night, plotted by mean depth
over the course of each tow. Regression analysis indi-
cated no effect of depth upon ambient light over this
relatively narrow range of depths and hence, regressions
are plotted as zero-slope lines.

was not followed by any further swimming, such that
the fish tended to hang stationary in the water, and
passed over the sweep as it progressed down the tank.
“Rise” characterized the motion of fish that departed
the bottom with sustained swimming in an upward
direction, such that the distance between fish and bot-
tom continuously increased as the fish swam. This was
in contrast to fish characterized by “herd” where fish
maintained a distance of less than one body length be-
tween themselves and the bottom as they swam along
in front of the sweep, i.e. herding behavior. Ryer and
Barnett (2006) investigated whether initial orienta-
tion, i.e., the direction fish were facing, influenced
behavioral response. No relationship was observed, and
consequently, no data on fish orientation were recorded
in this study. Categorical data on behavioral response
were pooled across replicate groups and analyzed by
contingency table analysis by using log-linear models
(Fienberg, 1980).

Results

At-sea experiment

Mean ambient light on the seafloor (Fig. 1) was greater
during daytime tows (2.0xl0~3 pmol photons/m2/s)
than during nighttime tows (8.4xl0~7 pmol photons/
m2/s, F(1 33]=352.76, P<0.001). However, over the rela-
tively narrow range of tow depths used in this analy-
sis, depth had no influence upon bottom ambient light
level (Fu 33]=0.27, P=0.607). Mean total catch (CPUE)
in terms of weight (kg/min) was greater during the
day than at night (Table 1, day: x=100.6 kg, SE = 9.61;
night: x=53.07 kg, SE = 6.14). This pattern of diurnally

Ryer et al.: Flatfish herding behavior in response to trawl sweeps

149

C Arrowtooth flounder

q_ E Pacific halibut

Figure 2

Mean catch per unit of effort (CPUE) ±1 standard error (SE) for daytime and nighttime
hauls for each of six flatfish species from control nets where the sweep was in contact with
the seafloor: (A) yellowfin sole (Limanda aspera); (B) flathead sole ( Hippoglossoides elas-
sodon)\ (C) arrowtooth flounder (Atheresthes stomias ); (D) rock sole (Lepidopsetta spp.);
(E) Alaska plaice ( Pleuronectes quadrituberculatus); and (F) Pacific halibut ( Hippoglossus
stenolepis ).

Table 1

Statistics for a comparison of day and night trawl catches, by total catch, and catch of six individual species of flatfish. For both
day and night tows, trawl nets were equipped with control sweeps (that had contact with the bottom). Where needed, Satterth-
waite’s adjusted degrees of freedom were used to mitigate for nonhomogeneity of variance.

Species

t-test statistic

P value

Total catch

4.85

31.3

<0.001

Yellowfin sole (Limanda aspera )

1.71

30.6

0.097

Flathead sole (Hippoglossoides elassodon )

-7.44

<0.001

Arrowtooth flounder (Atheresthes stomias)

-3.26

0.003

Rock sole (Lepidopsetta spp.)

-2.38

29.3

0.024

Alaska plaice (Pleuronectes quadrituberculatus)

-3.74

26.4

0.001

Pacific halibut (Hippoglossus stenolepis)

1.58

0.126

larger catches was also exhibited by four out of six
flatfish species examined (Table 1, Fig. 2). Flathead
sole, arrowtooth flounder, rock sole, and Alaska plaice
were all characterized by higher CPUE during the day.
Yellowfin sole and Pacific halibut exhibited no signifi-
cant differences in catch between day and night. Of the
four species for which fish total length was measured
in catch subsamples (i.e., yellowfin sole, flathead sole,
arrowtooth flounder, and rock sole), fish tended to be
slightly larger at night. This was only statistically sig-
nificant for yellowfin sole (f[24 4]=3.93, P=0.001), where
fish averaged 1 cm longer during the night (x=32.8 cm,
SE = 0.2) than during the day (x=31.8 cm, SE = 0.1), and
rock sole (f[32]=2.91, P=0.006), where fish averaged 0.9

cm longer during the night (*=33.3 cm, SE = 0.2) than
during the day (5;=32.4 cm, SE = 0.2).

The effect of elevating sweeps 10 cm off the bottom
differed, depending upon whether tows were made dur-
ing the day or night (Fig. 3). During the day, total catch
tended to decrease when sweeps were elevated (Table
2, elevated: 5c=93.4, SE = 8.7; control: 3c=100.6, SE = 9.6).
However, during the night, elevation of sweeps had little
influence upon catch (elevated: x=55.1, SE = 6.8; control:
x=53.1, SE = 6.1). This same pattern was evident for four
out of six flatfish species examined. Species for which
daytime elevation of sweeps decreased catch included
flathead sole, arrowtooth flounder, rock sole, and Alaska
plaice. Sweep configuration had no significant effect on

150

Fishery Bulletin 108(2)

A Yellowfin sole B Flathead sole C Arrowtooth flounder

Figure 3

Mean catch per unit of effort (CPUE) ±1 standard error (SE) for daytime and nighttime
catches of each of six flatfish species from both control nets, where the sweep was in contact
with the seafloor, as well as experimental nets where the sweep was elevated 10.2 cm off
the seafloor: (A) yellowfin sole ( Limanda aspera); (B) flathead sole ( Hippoglossoides elas-
sodon)\ (C) arrowtooth flounder ( Atlieresthes stomias); (D) rock sole ( Lepidopsetta spp.);
(E) Alaska plaice ( Pleuronectes quadrituberculatus)\ and (F) Pacific halibut (Hippoglossus
stenolepis).

Table 2

Statistics for comparison of total catch and catch of six individual species of flatfish between trawl nets equipped with control
(bottom contact) and those equipped with elevated (10 cm off bottom) sweeps, from both day and night tows.

Species

Paired t-test statistic

P value

Total catch

Day

2.11

0.064

Night

-0.22

0.834

Yellowfin sole ( Limanda aspera )

Day

1.84

0.099

Night

0.09

0.935

Flathead sole (Hippoglossoides elassodon)

Day

2.33

0.045

Night

-0.78

0.481

Arrowtooth flounder (Atheresthes stomias )

Day

4.35

0.002

Night

-0.71

0.519

Rock sole (Lepidopsetta spp.)

Day

5.42

<0.001

Night

0.23

0.830

Alaska plaice (Pleuronectes quadrituberculatus)

Day

2.39

0.041

Night

-0.67

0.539

Pacific halibut (Hippoglossus stenolepis)

Day

-0.59

0.753

Night

0.29

0.785

daytime catches of yellowfin sole or Pacific halibut. In
contrast to daytime results, elevated sweeps had no ef-
fect upon nighttime catches for any species. Of the four
species that were measured, fish lengths did not differ
between tows with elevated sweeps and control tows,
regardless of time of day (P>0.05 for each species, day
and night).

Laboratory experiment

Overall, 28% of fish initiated herding behavior in
response to simulated sweep disturbance. Herding was
most prevalent in the light, and tended to be replaced
by fish passing under the sweep, as well as hopping or
rising off the bottom in the dark (Fig. 4). There was also

Ryer et al.: Flatfish herding behavior in response to trawl sweeps

151

A Age-3 Pacific halibut — light

Age-3 Pacific halibut — dark

B Age-2 Pacific halibut — light

Age-2 Pacific halibut — dark

C Age-1 Pacific halibut — light

Age-1 Pacific halibut — dark

D Age-2 Northrock sole — light Age-2 Northrock sole — dark

Figure 4

Behavioral response of flatfish, under light and dark conditions, with the simulated sweep both
in contact (control) and elevated 10 cm off the bottom: (A) age-3 Pacific halibut ( Hippoglossus
stenolepis)\ (B) age-2 Pacific halibut; (C) age-1 Pacific halibut; and (D) age-2 northern rock sole
( Lepidopsetta polyxystra). “Pass under” represents fish that either did not react to the sweep,
or reacted late, such that they passed under the sweep as it progressed down the tank. “Hop”
characterized fish that reacted to the sweep with one or two body undulations, but almost
immediately pass over the sweep. “Rise” characterized fish in which the initial jump off the
bottom was followed by sustained swimming in an upward direction, such that the distance
between fish and bottom continuously increased as the fish swam. “Herd” characterized fish
which, after reacting to the gear, swam along in front of the sweep, close to the bottom, typically
maintaining a distance of less than one body length between themselves and the bottom.

a tendency for herding in the light to decrease when
the sweep was elevated. These observations are sup-
ported by results of log-linear model analysis, in which
ambient light (light, dark) mediated the influence of

sweep height upon behavioral response (G[3| = 9.96,
P=0.019). All three age classes of Pacific halibut, and
northern rock sole, behaved comparably; there were
no significant effects of species or age on the type of

152

Fishery Bulletin 108(2)

response displayed, or interactions with light level or
sweep height (P>0.05 for all). Examination of Figure 4
could lead one to conclude that age-3 halibut behaved
somewhat differently than the other species and age
groups. However, the number of age-3 halibut tested
(;z=15) was small compared to each of the other species
and age groups (n >50 for each), and as a consequence,
had little influence upon our statistical model. We pooled
data across species and collapsed response categories
down to those fish that herded in contrast to those that
did not (pass under, hop, and rise combined), so as to
render the data into a form most similar to our at-sea
trawl-catch experiments. Again, ambient light (light
or dark) mediated the influence of sweep height upon
behavioral response (Gtlj = 5.75, P=0.017). In Figure 5
we have simplified this relationship by graphing the
percentage of fish herding under the two light and sweep
height treatments. In addition to a conspicuous decrease
in herding in the dark, elevation of the sweep decreased
herding in the light but had little influence in the dark-
ness— results consistent with those observed in the at-
sea experiment.

Discussion

Ambient illumination controls many aspects of fish
behavior, from feeding and habitat use (Janssen, 1978;
Helfman and Schultz, 1984; Ryer and Olla, 1999; De
Robertis et ah, 2003; Petrie and Ryer, 2006) to social and
antipredator behavior (Shaw, 1961; Ryer and Olla, 1998).
Similarly, light has a pervasive influence upon interac-
tions between fish and trawls. In this study, field data
were largely consistent with our principal hypothesis;
that trawls configured with sweeps that are in contact
with the seafloor would catch more flatfish during the

60 r

50 -

E 30 '

9 20 -

10 -

0 L-1

Light Dark

Figure 5

Percentage of fish that herded in response to simulated
trawl sweep disturbance under both light and dark con-
ditions, with the sweep both in contact (control) and
elevated 10 cm off the bottom. Data were pooled across
species and age classes.

day than during the night. This pattern was observed
for four out of six flatfish species examined: flathead sole,
arrowtooth flounder, rock sole, and Alaska plaice. Herd-
ing, as seen in both roundfish and flatfish, is an ordered
behavioral response in which fish move away from an
approaching threat, i.e., the doors, sweeps, bridles, and
wings of the net. Through either continuous swimming,
or sudden swimming bursts, interspersed with rests on
the bottom (Winger et al., 1999, 2004), fish then funnel
to the center of the gear, where they concentrate before
tiring and falling back into net. Several studies have
demonstrated that both roundfish (Olla et al., 2000; Ryer
and Olla, 2000) and flatfish (Ryer and Barnett, 2006)
lose the ability to orient themselves in relation to gear
and initiate herding when ambient light falls below the
threshold for visual perception of the gear (Kim and
Wardle, 1998a, 1998b).

Given the brief evolutionary time during which fish
have interacted with towed fishing gear, approximate-
ly 100 years, it is unlikely that specific gear avoid-
ance behavior has evolved. Rather, we consider it most
parsimonious to assume gear avoidance is rooted in
antipredator behavior. Although flatfish may initially
erupt from the seafloor upon being disturbed by trawl
ground-gear, as when attacked by a predator, subse-
quent herding behavior is consistent with “distance
keeping” behavior, during which the fish attempts to
maintain a safe distance between itself and a slowly
pursuing predator. Scuba and skin divers who have at-
tempted to follow fish along the seafloor are certainly
familiar with this behavior. For flatfish, movement in
the vertical dimension also plays a critical role during
herding. It has been observed that flatfish remain close
to the bottom during herding, usually less than half a
body length (Ryer, 2008). Staying close to the bottom
reduces drag, lessening thrust requirements to achieve
a given speed — the ground effect (Videler, 1993; Gib-
son, 2005). Rising off the bottom makes flatfish more
conspicuous, and due to the location of a flatfish’s eyes,
also interferes with visual tracking of a pursuing preda-
tor, in this case, the trawl ground-gear. Although they
herd close to the bottom in the light, Pacific halibut
and northern rock sole respond differently to ground-
gear in the darkness, as demonstrated by laboratory
experiments (Ryer and Barnett, 2006). Unable to see,
the fish respond to contact with the ground-gear ini-
tially by hopping or swimming upward and away from
the bottom. Similarly, in this study the percentage of
fish moving off the bottom increased from 4% in the
light to 21% in darkness, for all species and bar heights
combined. Moving off the bottom in darkness probably
functions as an antipredator tactic, making the flatfish
more difficult to follow and may simply be the flatfish
version of the Mauthner-cell triggered (lateral line)
startle response (Eaton and Hackett, 1984).

Our second hypothesis, that elevation of sweeps off
the bottom, 10 cm in this case, would decrease catch
during daylight, but not at night, was also partial-
ly supported by our field experiment. Again, four of
six flatfish species examined displayed the predicted

Control

Elevated

Ryer et al.: Flatfish herding behavior in response to trawl sweeps

153

catch pattern. Arguably, our analysis is based upon
a small set of paired tows, particularly at night (n = 5
pairs). Taken alone, these at-sea trials might not be
convincing. However, these results were mirrored by
our laboratory experiments, where the elevation of
sweeps decreased herding to a greater extent in the
light, compared to darkness. The elevation of sweeps
had several consequences, all of which were likely to
have influenced flatfish behavior. First, because most
flatfish react to ground gear at a very short distance,
often only after being struck, the likelihood that fish
would simply not react and be passed over by sweeps
was probably increased by sweep elevation. Further,
part of the visual stimulus to herd that is associated
with ground gear is the sand and mud cloud that is
kicked up by the gear. This visual stimulus would be
absent or greatly diminished by sweep elevation, fur-
ther decreasing the likelihood of flatfish response. Our
laboratory experiments with rock sole exhibited a pat-
tern of response nearly identical to that seen in the
field and indicated that passage under or over the gear
was probably responsible for the decline in herding as-
sociated with sweep elevation during the day; in the
light, fish passing beneath the sweep increased by 24%
when the sweep was elevated. Lastly, even when herd-
ing is initiated, it must be maintained. Flatfish will
sometimes dive under ground gear when they perceive
a gap between the gear and the bottom — a trait that
has been used to reduce flatfish bycatch (DeAlteris et
al., 1997). Sweep elevation probably facilitated such
escape. Unfortunately, our laboratory data were of little
aid in evaluating this possibility. Because of the physi-
cal limitations of our apparatus, we characterized only
the initial behavioral response of fish — not prolonged
behavioral sequences that would characterize such
deliberate escape tactics.

Our field data indicate that Pacific halibut could have
a different pattern of availability or catchability, com-
pared to that of the other flatfish species we examined.
By virtue of size, Pacific halibut stand apart from most
other flatfish. Beyond three or four years of age, their
size likely renders them immune to most predators. This
may make them more likely to venture from the bottom,
as may their piscivorous diet. Consequently, they may
be more likely than other species to rise off the bottom
and swim back over sweeps. If so, it follows that most
of the fish captured are those directly in the path of
the net, excluding the area swept by the sweeps. Our
trawling operations tended to produce larger, albeit not
significant, Pacific halibut catches at night — a trend re-
ported by commercial fishermen as well. It may be that
with their greater speed and endurance, many halibut
escape trawls during the day, but at night cannot see
the gear to coordinate their escape. In contrast to the
halibut results, the nonsignificant differences for yel-
lowfin sole were similar in direction and magnitude to
the significant differences detected for the other small
flatfishes. This finding opens the possibility that these
flatfishes had similar reactions, but our experiment just
did not have the statistical power to detect them.

Diel patterns of catch in trawl fisheries and surveys
reflect not only patterns in fish availability, but gear-
specific behavioral influences upon catchability that
are directly controlled by ambient illumination. Results
of our laboratory experiments, along with earlier ex-
periments (Ryer and Barnett, 2006), indicate that trawl
footropes are likely to be more efficient at displacing
flatfish from the bottom and rapidly transitioning them
to the net under conditions of darkness (Ryer, 2008). In
contrast, sweeps are probably more effective at herding
flatfish inwards to the path of the net under daylight
conditions. This disparity is probably responsible for
the observed pattern of higher flatfish catches at night
with survey nets, where bridles and sweeps are kept to
minimal length, as compared to higher daytime catch-
es with commercial flatfish nets and lengthy sweeps .
These differences, as explained by the results of this
work, highlight the importance of fish behavior for fish
capture technology.

Acknowledgments

We wish to thank C. Hammond and J. Gauvin for assis-
tance with the at-sea portion of this project, as well as
the captain and crew of the FV Cape Horn. M. Ottmar,
and S. Haines assisted with animal husbandry and labo-
ratory experiments. A. Stoner, M. Davis, B. Laurel, and
T. Hurst provided helpful comments and discussion of
ideas explored in this research, and R. Hannah and W.
Wakefield provided helpful critiques of an early draft of
this manuscript. C. Sweitzer assisted with manuscript
preparation.

Literature cited

Beamish, F. W. H.

1965. Vertical migration by demersal fish in the North-
west Atlantic. J. Fish. Res. Board Can. 23:109-139.

Casey, J. M., and R. A. Myers.

1998. Diel variation in trawl catchability: is it as clear
as day and night? Can. J. Fish. Aquat. Sci. 55:2329-
2340.

DeAlteris, J., H. Milliken, and D. Morse.

1997. Bycatch reduction in the Northwest Atlantic
small-mesh bottom-trawl fishery for silver hake ( Merluc -
cius bilinearis). In Developing and sustaining world fish-
eries: the state of science and management. 2nd World
Fisheries Congress (D. A. Hancock, D. C. Smith, A.
Grant, and J. P. Beumer, eds.), p. 568-573. CSIRO,
Collingwood, Australia.

De Robertis, A., C. H. Ryer, A. Veloza, and R. D. Brodeur.

2003. Differential effects of turbidity on prey consump-
tion of piscivorous and planktivorous fish. Can. J. Fish.
Aquat. Sci. 60:1517-1526.

Douglas, R. H., and C. W. Hawryshyn.

1990. Behavioural studies of fish vision. In The visual
system of fish (Douglas, R. H., and M. B. A. Djamgoz,
eds.), p. 373-418. Chapman and Hall, London.

Eaton, R. C., and J. T. Hackett.

1984. The role of the Mauthner cell in fast-starts involv-

154

Fishery Bulletin 108(2)

ing escapes in teleost fishes. In Neural mechanisms of
startle behavior (Eaton, R.C., ed.), p. 213-266. Plenum
Publ. Corp., New York.

Fienberg, S. E.

1980. The analysis of cross classified categorical
data. MIT Press, Cambridge, MA.

Gibson, R. N.

2005. The behaviour of flatfishes. In Flatfishes: biology
and exploitation (Gibson, R. N., ed.), p. 213-239. Black-
well Science, Oxford.

Helfman, G. S., and E. T. Schultz.

1984. Social transmission of behavioural traditions in
a coral reef fish. Anim. Behav. 2:379-384.

Herrmann, B.

2005. Effect of catch size and shape on the selectivity of
diamond mesh cod-ends I. Model development. Fish.
Res. 71:1-13.

Hurst, T. P., C. H. Ryer, J. M. Ramsey, and S. A. Haines.

2007. Divergent foraging strategies of three co-occurring
north Pacific flatfishes. Mar. Biol. 151:1087-1098.

Janssen, J.

1978. Will alewives ( Alosa pseudoharengus ) feed in the
dark? Environ. Biol. Fish. 3:239-240.

Kim, Y. -H„ and C. S. Wardle.

1998a. Modelling the visual stimulus of towed fishing
gear. Fish. Res. 34:165-177.

1998b. Measuring the brightness contrast of fishing
gear, the visual stimulus for fish capture. Fish. Res.
34:151-164.

Main, J., and G. I. Sangster.

1981. A study of the fish capture process in a bottom
trawl by direct observations from a towed underwater
vehicle. Scott. Fish. Res. Rep. 23.

Metcalfe, J. E., G. P. Arnold, and P. E. Webb.

1990. Energetics of migration by selective tidal stream
transport: an analysis for plaice tracked in the southern
North Sea. J. Mar. Biol. Assoc., U.K. 70:149-162.

Nichol, D., and D. A. Somerton.

2009. Evidence of the selection of tidal streams by
northern rock sole (Lepidosetta polyxystra) for trans-
port in the eastern Bering Sea. Fish. Bull. 107:221-
234.

Olla, B. L. M. W. Davis, and C. Rose.

2000. Differences in orientation and swimming of
walleye pollock Theragra chalcogramma in a trawl net
during light and dark conditions: concordance between
field and laboratory observations. Fish. Res. 44:261-
266.

Petrie, M. E., and C. H. Ryer.

2006. Hunger, light level and body size affect refuge
use by post-settlement lingcod Ophiodon elongatus. J.
Fish. Biol. 69:957-969.

Rose, C. S., J. R. Gauvin, and C. F. Hammond.

2010. Effective herding of flatfish by cables with minimal
seafloor contact. Fish. Bull. 108:136-144

Ryer, C. H.

2008. A review of flatfish behavior relative to trawls.
Fish. Res. 90:138-246.

Ryer, C. H., and L. A. K. Barnett.

2006. Influence of illumination and temperature upon
flatfish reactivity and herding behavior: Potential
implications for trawl capture efficiency. Fish. Res.
81:242-250.

Ryer, C. H., and B. L. Olla.

1998. Effect of light on juvenile walleye pollock shoaling
and their interaction with predators. Mar. Ecol. Prog.
Ser. 167:215-226.

1999. Light-induced changes in the prey consumption
and behavior of two juvenile planktivorous fish. Mar.
Ecol. Prog. Ser. 181:412-51.

2000. Avoidance of an approaching net by juvenile walleye
pollock Theragra chalcogramma in the laboratory: the
influence of light intensity. Fish. Res. 45:195-199.

Schabetsberger, R., R. D. Brodeur, L. Ciannelli, J. M. Napp, and
G. L. Swartzman.

2000. Diel vertical migration and interaction of zoo-
plankton and juvenile walleye pollock (Theragra chal-
cogramma) at a frontal region near the Pribilof Islands,
Bering Sea. ICES J. Mar. Sci. 57:1283-1295.

Shaw, E.

1961. Minimal light intensity and the dispersal of school-
ing fish. Bull. Inst. Oceanogr. 1213:1-8.

Snedecor, G. W., and W. G. Cochran.

1980. Statistical methods, 593 p. Iowa State Univ.
Press, Ames, IA.

Sokal, R. R., and F. J. Rohlf.

1969. Biometry, 776 p. W.H. Freeman and Co., San
Francisco.

Videler, J. J.

1993. Fish swimming, 260 p. Chapman & Hall,
London.

Walsh, S. J.

1988. Diel variability in trawl catches of juvenile and
adult yellowtail flounder on the Grand Banks and the
effect on resource assessment. N. Am. J. Fish. Manag.
8:373-381.

1991. Diel variation in availability and vulnerability of
fish to a survey trawl. J. Appl. Ichthyol. 7:147-159.

Walsh, S. J., and W. M. Hickey.

1993. Behavioural reactions of demersal fish to bottom
trawls at various light conditions. ICES J. Mar. Sci.
Symp. 196:68-76.

Winger, P. D., P. He, and S. J. Walsh.

1999. Swimming endurance of American plaice ( Hippoglos -
soides platessoides ) and its role in fish capture. ICES
J. Mar. Sci. 56:252-265.

Winger, P. D., S. J. Walsh, P. He, and J. A. Brown.

2004. Simulating trawl herding in flatfish: the role of fish
length in behaviour and swimming characteristics. ICES
J. Mar. Sci. 61:1179-1185.

155

Spatial and temporal variation in otolith chemistry
for tautog ( Tautoga onitis ) in Narragansett Bay
and Rhode Island coastal ponds

Ivan Mateo (contact author)1
Edward G. Durbin2
David A. Bengtson1
Richard Kingsley2
Peter K. Swart3
Daisy Durant4

Email address for contact author: imateo32@hotmail.com

1 University of Rhode Island

Department of Fisheries, Animal and Veterinary Sciences
Kingston, Rhode Island 02881

2 University of Rhode Island
Graduate School of Oceanography
Narragansett, Rhode Island 02882

3 Division of Marine Geology and Physics

Rosenstiel School of Marine and Atmospheric Sciences
University of Miami
Miami, Florida 33149

4 Narragansett Bay National Estuarine Research Reserve
P.O. Box 151

Prudence Island, Rhode Island 02872

Abstract — The elemental composi-
tion of otoliths may provide valuable
information for establishing connec-
tivity between fish nursery grounds
and adult fish populations. Concen-
trations of Rb, Mg, Ca, Mn, Sr, Na,
K, Sr, Pb, and Ba were determined
by using solution-based inductively
coupled plasma mass spectrometry in
otoliths of young-of-the year tautog
(Tautoga onitis ) captured in nursery
areas along the Rhode Island coast
during two consecutive years. Stable
oxygen (6180) and carbon (<513C) iso-
topic ratios in young-of-the year oto-
liths were also analyzed with isotope
ratio mass spectrometry. Chemical
signatures differed significantly
among the distinct nurseries within
Narragansett Bay and the coastal
ponds across years. Significant dif-
ferences were also observed within
nurseries from year to year. Classi-
fication accuracy to each of the five
tautog nursery areas ranged from 85%
to 92% across years. Because accu-
rate classification of juvenile tautog
nursery sites was achieved, otolith
chemistry can potentially be used as
a natural habitat tag.

Manuscript submitted 16 June 2009.
Manuscript accepted 14 December 2009.
Fish. Bull. 108:155-161 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

The dependence of fish production
and population dynamics on disper-
sal and migration among multiple
habitats, referred to as “connectiv-
ity,” is a critical property of marine
populations. Connectivity rates deter-
mine colonization patterns for new
habitats, the resiliency of popula-
tions to harvest, and can be used in
the design of marine protected areas
(MPAs). Quantifying connectivity
rates in marine organisms is, how-
ever, extremely difficult because the
natal and nursery origins of adults are
almost unknown. Recently, tagging
techniques with natural isotopic and
elemental markers have been devel-
oped for species that were not able to
be tagged or recaptured by conven-
tional approaches. Chemical natural
habitat tags in the otoliths of juvenile
fish have been used to differentiate
individuals from different estuarine
and riverine systems (Thorrold et al.,
1998a; Thorrold et al., 1998b; Gilland-
ers and Kingsford, 2000; Gillanders,
2002b) and other types of nearshore
habitats, such as estuary as opposed
to rocky reef (Gillanders and Kings-

ford, 1996) and estuary as opposed to
exposed coastal habitats (Yamashita
et al., 2000; Forrester and Swearer,
2002). In addition, through chemical
analysis of the juvenile core region of
adult otoliths, natural habitat tags
have been used to determine the pro-
portion of the adult population that
resided in different juvenile habitats
(Yamashita et al., 2000; Thorrold et
al., 2001; Gillanders, 2002a).

The tautog (Tautoga onitis ) is an
economically and ecologically impor-
tant species found in the waters of
eastern North America from the Gulf
of Maine to North Carolina. Juvenile
tautog are known to depend on shal-
low water habitats where they are
safe from high levels of predation
and can find necessary food resourc-
es (Dorf and Powell, 1997; Arendt,
1999). However, the relative impor-
tance of open coastline and enclosed
bays and lagoons as nursery habitat
for tautog is still poorly understood
(Sogard et al., 1992). In light of the
fact that the northeastern coast of
the United States has experienced a
major loss of its estuarine habitats

156

Fishery Bulletin 108(2)

because of human alteration of the coastal zone (Brom-
berg and Bertness, 2005), data are needed to quantify
the importance of specific coastal habitat types in sus-
taining tautog populations.

Our long-term goal is to investigate the utility of
naturally occurring habitat tags to determine habitat
linkages in Narragansett Bay and other nearby es-
tuarine systems by juvenile tautog. This is an initial
crucial step to quantify the relative contribution of
estuarine habitats for the population connectivity of
adult tautog.

Materials and methods

Sampling of juvenile fish

In Rhode Island, young-of-the-year (YOY) tautog of 45-
64 mm fork length (FL) were sampled from three sites
in Narragansett Bay: Mt. Hope Bay (MH), Gaspee Point
(GP), and Rose Island (RS); and from two sites from the
coastal ponds along the Rhode Island southern shore:
Point Judith, lower pond (PJ), and Charlestown Pond
(CP) (Fig. 1). The samples were obtained in coopera-
tion with Rhode Island Department of Environmental
Management, Division of Marine Fisheries (RIDEM),
during August and September of 2005 and 2006. The
sampling stations were selected to include different
nursery areas and possibly different chemical back-
grounds and according to information on juvenile tautog
abundance from RIDEM. Average monthly surface tem-
peratures and salinities at Gaspee Point for 2005 were
22°C and 24.9%e, and for 2006 were 20.6°C and 22.5%e.
For Mount Hope Bay, average surface temperatures
and salinities were 21.7°C and 27.0 %c, and for 2006
were 20.5°C and 24.9%c. Data from the closest point
to Rose Island showed average surface temperatures
and salinities for 2006 were 17.4°C and 30.8 %c. Twenty
juveniles per site per year were captured for analysis.
Sampled fish were kept frozen until dissection for the
removal of otoliths.

Laboratory processing of samples

Before dissection, each fish was weighed (wet weight
to the nearest 0.1 g) and measured (FL and standard
length [SL] to the nearest 0.1 mm). Both sagittal oto-
liths were removed from each fish, cleaned of adhering
tissue, rinsed 3x with Milli-Q-filtered (Millipore Corp.,
Billerica, MA) water, and allowed to dry in a class-100
laminar-flow hood. The left sagittal otolith was used
for trace metal analysis and the right otolith was used
for stable isotope analysis. A total of 164 otoliths were
prepared for trace metal analysis. Each otolith was
weighed on a Thermo Cahn microbalance (± 0.01 mg)
(Thermo Fisher Scientific, Waltham, MA). Samples were
then placed in acid-washed 2.5-mL snap-cap polypropyl-
ene containers. The otolith weights ranged from 0.08
to 0.34 mg and averaged 0.18 mg. Otoliths for trace
metal analysis were transferred to 5-mL clean polypro-

pylene tubes and 0.2 mL of triple-distilled 17% HN03
was added to insure complete dissolution (in about 30
seconds). An internal thulium single-element standard
spike was added (to correct for variable matrix effects
during the inductively coupled plasma mass spectrom-
etry analyses) and then the solution was diluted to 1.8
mL with triple-distilled water. This dilution resulted
in a Ca concentration of approximately 40 ppm in the
analyzed otolith solution.

Otolith chemistry

Elemental concentrations of YOY otoliths were deter-
mined through solution-based ICPMS at the University
of Rhode Island Graduate School of Oceanography. All
measurements were carried out on a Finnigan ele-
ment high-resolution inductively coupled plasma mass
spectrometer (HR-ICPMS) (Thermo Fisher Scientific,
Waltham, MA). A procedural blank was prepared in the
same manner as had been used for the other samples,
but with no otolith present. The procedural blank was
compared to the system blank to determine if contami-
nation occurred during processing. System blanks were
made from the same acid used for sample dissolution
and were run every four samples. A drift-correction
standard was prepared by gravimetrically spiking a
CaC03 standard solution with the appropriate concen-
trations of Na, K, Rb, Mg, Ca, Mn, Ni, Cu, Zn, Sr, Ba,
Co, and Pb to match the typical elemental composition
of the otoliths. This drift-correction standard was ana-
lyzed every four samples to track and correct for varia-
tions in instrument sensitivity during each analytical
time period. The choice of these thirteen elements for
our study was based on previous studies of elemental
composition of juvenile fish otoliths. Analytical results
were expressed as absolute concentrations of elemental
molar ratios with respect to calcium: Element:Ca ratios,
expressed as units of mmol/mol or pmol/mol.

The elements that were always above detection lim-
its (Rb, Mg, Ca, Sr, and Ba) were used for subsequent
analysis. The average relative standard deviations
were as follows: Rb (3%), Mg (10%), Ca (1%), Sr (1%),
and Ba (5%). The limits of detection were as follows
(values in ppm): Rb (0.007), Mg (0.02), Sr (0.077), and
Ba (0.014). The detection limits for the whole otolith
dissolution-solution-based method were calculated as
three times the standard deviation of the counts per
second (cps) of the isotope of interest in acid blanks
divided by the sensitivity in cps/ppm of the CRM22
carbonate standard. For every isotope, these were in
the sub-ppm range — a result that compares with the
3 to 2000 ppm range of the elements of interest in the
sample otoliths.

Stable carbon and oxygen isotopes of these otolith
samples were determined at Rosenstiel School of Marine
and Atmospheric Sciences, University of Miami, by us-
ing an automated carbonate device (Kiel III) attached
to a thermo Finnigan delta-plus stable isotope mass
spectrometer (Thermo Fisher Scientific, Waltham, MA).
Data were expressed by using conventional d notation

Mateo et al.: Otolith chemistry for Tautoga onitis in Narragansett Bay and Rhode Island coastal ponds

157

71°25'W 71°15'W

71 °40'W 71 °30W

(A) Map of tautog (Tautoga onitis) sampling stations in Narragansett Bay for 2005
and 2006 that were surveyed for juvenile otolith element concentrations and isotopic
signatures. Sampling stations are shown in arrows. (B) Map of the south coast of Rhode
Island showing two coastal ponds (Charlestown, Point Judith) that were surveyed for
tautog for years 2005 and 2006 to determine otolith element concentrations and isotopic
signatures. Sampling stations are shown in arrows.

in relation to V-PDB (Vienna Peedee Belemnite). Data
were corrected for the usual isobaric interferences. The
external precision (calculated from replicate analyses of
an internal laboratory calcite standard) was 0.04% for
513C and 0.08% for d180.

Statistical analysis

Two-way analysis of variance (ANOVA) was used to test
for differences in fish body length among stations and
years. We also examined relationships between otolith

158

Fishery Bulletin 108(2)

Table 1

Average size distribution of tautog ( Tautoga onitis) collected in Rhode Island for analysis of otolith elemental concentrations and
stable isotopic signatures. The numbers of fish measured at each station ( n ) to obtain average fork lengths (FL in mm) in each
year are shown. Numbers in parentheses are standard errors.

Station

2005

2006

Gaspee Point (GP)

59.6(1.2)

Mount Hope Bay (MH)

59.1 (2.1)

63.0(2.9)

Rose Island (RS)

52.1 (1.2)

45.3(3.5)

Point Judith, lower pond (PJ)

49.4 (2.1)

57.3 (1.5)

Charlestown Pond (CP)

50.2(3.0)

54.9(2.7)

weight and otolith elemental composition and isotopic
signatures with analysis of covariance (ANCOVA). If
there was a significant relationship, we removed the
effect of size (otolith weight used as a proxy for fish size)
to ensure that differences in fish size among samples did
not confound any site-specific differences in otolith chem-
istry. Concentrations of elements were weight-detrended
by subtraction of the product of the common within-
group linear slope multiplied by the otolith weight from
the observed concentration (Campana et al., 2000).

To detect differences in the concentrations of par-
ticular elements and multi-element fingerprints among
stations and between years, we performed ANOVA and
multivariate analyses of variance (M ANOVA). Pillai’s
trace statistic was chosen as the multivariate test sta-
tistic because it is more robust than other multivari-
ate statistics (Wilkes’s lamda, Hotelling’s T-test) to
small sample sizes, unequal cell sizes, and situations in
which covariances are not homogeneous. Tukey’s HSD
test was used to detect a posteriori differences among
means (a=0.05). Before statistical testing, residuals
were examined for normality and homogeneity among
stations. To meet model assumptions, all analyses were
performed on natural log-transformed data. We also
used linear discriminant function analyses (DFAs) on
tautog juvenile data to visualize spatial differences in
juvenile otolith chemistry data within sites and to ex-
amine classification success for juveniles from different
sites or regions. Classification success is the percentage
of fish that are correctly assigned to their actual region
given the information on location where the fish was
collected and the chemical signature of each fish. Cross
validations were performed by using jackknife (“leave
one out”) procedures in SYSTAT (vers. 11, Systat Soft-
ware, Inc., Chicago, IL).

Results

Size distribution

Mean (FL) of juvenile tautog at stations in Rhode Island
ranged from 45 to 63 mm (Table 1). There were sig-

nificant differences in mean length among stations
(ANOVA, PcO.OOl) and between years (ANOVA, P<0.05)
within Rhode Island stations. There were no significant
differences in mean FL among stations within Narra-
gansett Bay. However, in 2005, mean FL from all sta-
tions within Narragansett Bay were significantly longer
than that for individuals caught in the coastal ponds
(Point Judith, lower pond, Charlestown Pond) (Tukey
test, P<0.05). In 2006; only Mount Hope Bay had fish
significantly longer than those from Rose Island (Tukey
test, P<0.05).

Otolith chemistry

Results of MANOVA showed that the chemical signa-
tures of trace metals and stable isotopes combined in
tautog otoliths differed significantly among stations
(MANOVA, F18 384 = 20.72, PcO.OOl) and years (MANOVA,
P6 126 = 9.05, P<0.001) within Rhode Island. Signifi-
cant interaction between station and year (MANOVA,
F18 3g4=5.18, P<0.001) implied that chemical signatures
differed between years depending on the station studied.
Classification success for tautog by using both trace
metals and stable isotopes for stations within Rhode
Island for each of the two years ranged from 85% to
92% (Table 2).

Individual elemental concentrations

In Rhode Island, one trace element (Rb) and one stable
isotope (d13C) showed significant relationships with the
covariable otolith weight in the ANCOVA (P<0.001)
and therefore required the effect of otolith weight be
removed for subsequent ANOVA analysis. The ele-
mental concentrations and isotope signatures varied
significantly among stations (ANOVA, P<0.001), and
between years (ANOVA, P<0.001) (Fig. 2). Significant
interaction between station and year (ANOVA, PcO.OOl)
indicated that concentration of individual elements dif-
fered between years depending on the station studied.
In Rhode Island, elemental concentrations of Sr, Ba,
Mg, Rb, and the stable isotopes d13C and <5180 varied
significantly among stations in 2005, whereas only Ba

Mateo et al.: Otolith chemistry for Tautoga onitis in Narragansett Bay and Rhode Island coastal ponds

159

Table 2

Classification success (as a percentage) results determined by jack-knife cross validation procedure for linear discriminant func-
tion analysis of chemical concentrations in tautog (Tautoga onitis) otoliths collected at Rhode Island stations in 2005 and 2006,
with the use of solution-based inductively coupled plasma mass spectrometry for the combined trace metals (Sr, Ba, Mg, Rb):
[Sr/Ca], [Ba/Ca], [Rb/Ca], [Mg/Ca]) and for d13C and stable isotopes. Names of the stations are Gaspee Point (GP), Mount
Hope Bay (MH), Rose Island (RS), Point Judith, lower pond (PJ), Charlestown Pond (CP).

Classification success {%)

2005

Total

2006

Total

and 6180 varied significantly among stations in 2006
(ANOVA, P<0.001) (Fig. 2). For example, 6180 was high-
est at Rose Island at the mouth of Narragansett Bay,
whereas <513C magnitudes were similar across years for
all Narragansett Bay stations. Sr concentrations within
Narragansett Bay and the coastal ponds also remained
similar in magnitude throughout the years of study.

Discussion

The elemental composition of juvenile tautog otoliths
varied considerably within and among estuaries and
between years. We found very strong differences in the
concentrations of Mg, Sr, Ba, and Rb, as well as in the
stable isotopic signatures of d 13C and 6 180, among sta-
tions within RI. High classification success rates (gener-
ally >85%) of the discriminant functions derived from
trace element and stable isotope signatures together
confirmed their use as an effective natural tag of the
estuarine nursery area of juvenile tautog. Although
most of the variance in trace element signatures was
concentrated among estuaries, we also found signifi-
cant differences in elemental fingerprints and stable
isotopes in tautog otoliths among sites about 10 to 25
km2 apart within Narragansett Bay resulting in 100%
classification success within that water body. These
data indicate that the physicochemical characteristics
of specific sections of the estuaries may vary enough to
generate the differences in otolith chemistry that we
observed within each estuary.

Elemental fingerprints, however, should not be regard-
ed as permanent markers of actual estuarine habitat or
environment (Forrester and Swearer, 2002; Swearer et

ah, 2003). Estuarine habitats are very dynamic; seawa-
ter properties and composition at a particular location
can vary over tidal to annual time scales (Peters, 1999).
As a result, it may be expected that the magnitude of
variations in elemental fingerprints in otoliths among
estuaries will not remain constant over time. The sig-
nificant interannual differences we report among year
classes in age-0 tautog otolith elemental signatures is
similar to interannual differences in otolith chemis-
try reported for other marine fishes (Gillanders and
Kingsford, 2000; Gillanders, 2005). Thus, interannual
differences indicate that age-0 tautog elemental signa-
tures must be analyzed on a year-class-specific basis
because there were stations where concentrations were
not consistent between years.

It is not surprising to see such clear differences
in otolith chemical signatures among the stations
sampled in Narragansett Bay. Data from RIDEM show
that there were also significant differences in salinity
regimes in these regions during the late spring and
summer of 2005 and 2006 (H. Stoffel, and J. McNa-
mee, unpubl. data1). The proximity of Rose Island
station to the mouth of Narragansett Bay meant that
high salinities (up to 30 %c) would be observed. On
the other hand, the lower-salinity stations within the
upper region of Narragansett Bay are located much
closer to the industrial area and watershed and there-
fore potentially more prone to terrestrial influences
from freshwater runoff resulting in reduced salinities
(20-25%*).

1 Stoffel, H., and J. McNamee. 2008. Rhode Island Dept.
Environmental Management (RIDEM), Jamestown, RI
02879.

160

Fishery Bulletin 108(2)

Sampling year

Figure 2

Variation in trace elements and stable isotopes concentrations measured in otoliths of young-of-
the-year tautog ( Tautoga onitis ) collected in Rhode Island in 2005 and 2006. All trace element
data (element/CaxlO6) are ln(x+l) transformed. Rhode Island station codes are GP=Gaspee Point,
MH = Mount Hope Bay, RS = Rose Island, PJ = Point Judith, lower pond, and CP= Charlestown
Pond.

Successful discrimination between estuarine nurs-
eries in the present study was accomplished through
otolith elemental fingerprints, fulfilling one of the re-
quirements for their possible use as natural tags (Cam-
pana et al., 2000). The estuarine nursery origin of
juvenile tautog was accurately identified based on oto-
lith elemental fingerprints and stable isotopes. Several
methods based on laser ablation (Thorrold et al., 2001;
Gillanders, 2002a) or micromilling techniques (Gil-

landers and Kingsford, 1996; Gillanders, 2005; Brown,
2006) could be used to determine elemental fingerprints
found in the otolith cores of adult tautog for comparison
with the juvenile estuarine fingerprints that we have
established. We believe solution-based techniques are
more suitable than microprobe techniques for analysis
of tautog otolith elemental concentrations because 1)
solution-based techniques tend to have higher sensitiv-
ity, accuracy, and precision compared to microprobe

Mateo et al.: Otolith chemistry for Tautoga onitis in Narragansett Bay and Rhode Island coastal ponds

161

techniques (Campana, 1999; Campana et al., 2000); and
2) solution-based techniques can provide more precise
natural tags on fish with limited movement within habi-
tats during their first year of life. For example, tautog
have a short larval period of 15 to 20 days (Sogard et
al., 1992; Dorf and Powell 1997) and once larvae have
settled, they have small home range of approximately
20 meters (Able et al., 2005) during their first year of
life. Thus, juvenile cores samples from age classes rep-
resenting fish born in 2005 and 2006 could be extracted
by micromilling procedures and their chemical elements
can be analyzed by solution ICPMS. Present results are
a step towards establishing juvenile movement to adult
habitats, which must be examined in nursery studies
(Beck et al., 2001). Identifying links between juvenile
and adult habitats, and understanding connectivity
between estuarine nurseries and adult populations,
has the potential to aid fishery managers and aid in
the management and conservation of estuarine fish
nursery habitats.

Acknowledgments

We would like to thank C. Powell, M. Burnett, and B.
Murphy from RIDEM; as well as P. Stout from Camp
Fuller, and R. Dickau from Pond Shore Association for
helping to collect fish. Special thanks go to B. Taplin, R.
Pruell and the late L. Meng from U.S. Environmental
Protection Agency, and to K. Castro from University of
Rhode Island Sea Grant Fisheries Extension for support
and inspiration for this project. This study was funded
through University of Rhode Island Sea Grant Program
and the Nature Conservancy Global Marine Initiative.

Literature cited

Able, K. W„ L. S. Hales, and S. M. Hagan

2005. Movement and growth of juvenile (age 0 and 1+)
tautog ( Tautoga onitis [ L . ] ) and cunner ( Tautogola -
brus adspersus [Walbaum]) in a southern New Jersey
estuary. J. Exp. Mar. Biol. Ecol. 327:22-35.

Arendt, M. D.

1999. Seasonal residence, movement, and activity of adult
tautog ( Tautoga onitis) in lower Chesapeake Bay. M.
S. thesis, 104 p. School of Marine Science, College of
William and Mary, Gloucester Point, VA.

Beck, M. W„ K. L. Heck, Jr., K. W. Able, D. L. Childers, D. B. Egg-
leston, B. M. Gillanders, B. Halpern, C. G. Hays, K. Hoshino,
T. J. Minello, R. J. Orth, P. F. Sheridan, and M. P. Weinstein.

2001. The identification, conservation, and manage-
ment of estuarine and marine nurseries for fish and
invertebrates. BioScience 51:633-641.

Bromberg, K. D, and M. D. Bertness.

2005. Reconstructing New England salt marsh losses
using historical maps. Estuaries 28:823-832.

Brown, J. A.

2006. Using the chemical composition of otoliths to
evaluate the nursery role of estuaries for English sole
Pleuronectes vetulus populations. Mar. Ecol. Prog. Ser.
306:269-281.

Campana, S. E.

1999. Chemistry and composition of fish otoliths: path-
ways, mechanisms and applications. Mar. Ecol. Prog.
Ser. 188:263-297.

Campana, S. E., G. A. Chouinard, J. M. Hanson, A. Frechet, and
J. Brattey.

2000. Otolith elemental fingerprints as biological tracers
of fish stocks. Fish. Res. 46:343-357.

Dorf, B. A., and J. C. Powell.

1997. Distribution, abundance, and habitat characteris-
tics of juvenile tautog (Tautoga onitis , Family Labridae)
in Narragansett Bay, Rhode Island, 1988-1992. Estu-
aries 20:589-600.

Forrester, G. E., and S. E. Swearer.

2002. Trace elements in otoliths indicate the use of open-
coast versus bay nursery habitats by juvenile California
halibut. Mar. Ecol. Prog. Ser. 241:201-213.

Gillanders, B. M.

2002a. Connectivity between juvenile and adult fish
populations: do adults remain near their recruitment
estuaries? Mar. Ecol. Prog. Ser. 240:215-223.

2002b. Temporal and spatial variability in elemental
composition of otoliths: Implications for determining
stock identity and connectivity of populations. Can.
J. Fish. Aquat. Sci. 59:669-679.

2005. Using elemental chemistry of fish otoliths to
determine connectivity between estuarine and coastal
habitats. Estuar. Coast. Shelf Sci. 64:47-57.

Gillanders, B. M., and M. J. Kingsford.

1996. Elements in otoliths may elucidate the contribu-
tion of estuarine recruitment to sustaining coastal reef
populations of a temperate reef fish. Mar. Ecol. Prog.
Ser. 141:13-20.

2000. Elemental fingerprints of otoliths of fish may
distinguish estuarine ‘nursery’ habitats. Mar. Ecol.
Prog. Ser. 201:273-286.

Peters, H.

1999. Spatial and temporal variability of turbulent
mixing in an estuary. J. Mar. Res. 57:805-845.

Sogard, S. M., K. W. Able, and M. P. Fahay.

1992. Early life history of the tautog, Tautoga onitis, in
the Mid-Atlantic Bight. Fish. Bull. 90:529-539.

Swearer, S. E., G. E. Forrester, M. A. Steele, A. J. Brooks, and
D. W. Lea.

2003. Spatio-temporal and interspecific variation in oto-
lith trace-elemental fingerprints in a temperate estua-
rine fish assemblage. Estuar. Coast. Shelf. Sci. 56:
1111-1123.

Thorrold, S. R., C. M. Jones, S. E. Campana, J. W. McLaren, and
J. W. H. Lam.

1998a. Trace element signatures in otoliths record natal
river of juvenile American shad ( Alosa sapidissima).
Limnol. Oceanogr. 43:1826-1835.

Thorrold, S. R., C. M. Jones, P. K. Swart, and T. E. Targett.

1998b. Accurate classification of juvenile weakfish
Cynoscion regalis to estuarine nursery areas based on
chemical signatures in otoliths. Mar. Ecol. Prog. Ser.
173:253-265.

Thorrold, S. R., C. Latkoczy, P. K. Swart, and C. M. Jones.

2001. Natal homing in a marine fish metapopulation.
Science 291:297-299.

Yamashita, Y., T. Otake, and H. Yamada.

2000. Relative contributions from exposed inshore and
estuarine nursery grounds to the recruitment of stone
flounder, Platichthys bicoloratus, estimated using otolith
Sr:Ca ratios. Fish. Oceanogr. 9:316-327.

162

Fish assemblages associated
with three types of artificial reefs:
density of assemblages
and possible impacts
on adjacent fish abundance

Reiji Masuda (contact author)1 Yoshiaki Kai1

Masami Shiba2 Asami Nakanishi1

Yoh Yamashita1 Masaru Torikoshi1

Masahiro Ueno1 Masaru Tanaka3

Email address for contact author: reiji@kais. kyoto-u.ac.jp

1 Maizuru Fisheries Research Station
Kyoto University

Nagahama, Maizuru
Kyoto 625-0086, Japan

2 Ashiu Forest Research Station
Kyoto University

Miyama, Nantan
Kyoto 601-0703, Japan

3 University of Malaysia Sabah
Locked Bag No. 2073
88999, Kota Kinabalu
Sabah, Malaysia

Abstract— We evaluated the effective-
ness of wooden artificial reefs (ARs)
as fish habitat. Three types of ARs,
made of cedar logs, broadleaf tree
logs, and PVC pipes, respectively,
were deployed in triplicate at 8-m
depth off Maizuru, Kyoto Prefec-
ture, Sea of Japan, in May 2004. Fish
assemblages associated with each of
the nine ARs were observed by using
SCUBA twice a month for four years.
Fish assemblages in the adjacent
habitat were also monitored for two
years before and four years after reef
deployment. In the surveyed areas
(ca. 10 m2) associated with each of the
cedar, broadleaf, and PVC ARs, the
average number of fish species was
4.14, 3.49, and 3.00, and the average
number of individuals was 40.7, 27.9,
and 20.3, respectively. The estimated
biomass was also more greater when
associated with the cedar ARs than
with other ARs. Visual censuses of the
habitat adjacent to the ARs revealed
that the number of fish species and
the density of individuals were not
affected by the deployment of the ARs.
Our results support the superiority
of cedar as an AR material and indi-
cate that deployment of wooden ARs
causes no reduction of fish abundance
in adjacent natural reefs.

Manuscript submitted 29 January 2009.
Manuscript accepted 14 December 2009.
Fish. Bull. 108:162-173 (2019).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

Habitat complexity plays a major role
in the survival of young demersal
fishes by providing a refuge from pre-
dation (Ferreira et al., 2001; Scharf
et al., 2006; Hamilton and Konar,
2007). Fish species richness is highly
dependent on the rugosity and vari-
ety of growth forms in the habitat,
whereas the height of vertical struc-
tures is an important predictor of
total fish abundance (Gratwicke and
Speight, 2005). In this respect, arti-
ficial reefs (ARs) are often deployed
to improve the quality of habitat
(Gorham and Alevizon, 1989). In
addition to their role as refuges, ARs
host encrusting invertebrates that
can be consumed as prey by fishes
(Seaman and Jensen, 2000). Fish are
often more abundant at ARs than at
natural reefs, probably because the
vertical structures potentially allow
more varied refuges for fish settle-
ment and recruitment than the usual
more moderately sloped bottoms of
natural reefs (Rilov and Benayahu,
2000; Reed et al., 2006).

Although the deployment of struc-
tures functioning as ARs may well
have started long ago by fishermen
in various localities around the globe,
research on this subject is relatively
recent (Seaman and Sprague, 1991).
Two countries, United States and Ja-
pan, have relatively long histories of
nationwide projects on ARs. In the
case of the United States, the main
goal of deploying ARs has been to
improve catch for recreational fish-
ermen. Common materials used for
these ARs have been waste products,
such as automobiles, tires, and oil
and gas platforms. The use of such
products has caused environmental
concerns, resulting in a shift toward
the construction of ARs with concrete
(Collins et al., 2002). In contrast, the
purpose of Japanese deployments of
ARs have primarily been to improve
commercial fishery production, and
governmental agencies have invested
heavily in the construction of large
ARs made of concrete and steel to be
deployed in coastal areas.

Masuda et al.: Fish assemblages associated with three types of artificial reefs

163

The recent trend for ARs in Japan has shifted
from concrete to wooden construction. This has
been partly due to funding shortages, but also
because fishermen have found that wooden ARs
attract fish more rapidly than those made of con-
crete or steel. Indeed, most coastal prefectures
in Japan deploy wooden ARs with or without
governmental subsidies under the supervision
of local fishermen’s cooperatives. The materials
and shape of wooden ARs differ depending on
each fishery cooperative. As much as 70% of the
land area in Japan is forested, half of which is
plantation forests of conifers, such as Japanese
cedar (Cryptomeria japonica) and hinoki cypress
( Chamaecyparis obtusa). Although these forests
require occasional thinning, many of them lack
such maintenance because of the decline in the
market price of timber. Therefore, the construc-
tion of wooden ARs also has the socioeconomic
potential to stimulate the demand for forestry
materials.

The primary goal of the present study was to
confirm the efficacy of wooden ARs, especially
those made of cedar tree logs as fish habitat. For
this purpose, fish assemblages associated with
ARs made from cedar trees were compared to
those made from broadleaf trees and those made
with polyvinyl chloride (PVC) pipes. There is a
debate whether ARs merely attract fishes from adjacent
areas or whether they do improve fishery productivity
(Grossman et al., 1997; Pickering and Whitmarsh,
1997). We therefore tested the possibility that ARs at-
tract fishes from adjacent areas and thus concentrate
fish abundance at the ARs, rather than fish abundance
is spread over the fishing ground as a whole. A visual
census had been conducted twice a month for more
than two years before the deployment of these ARs in
adjacent areas; hence the fish fauna was compared in
the area before and after the deployment of ARs.

Materials and methods

Deployment and visual census of artificial reefs

Three types of ARs were prepared. The design of the ARs
was modified from that designed by the Atake Forestry
Association, Yamaguchi, Japan (http://www.geocities.jp/
abu_kikori/katsudou/gyosyou/gyosyou2.html, accessed
on December 2003; also see Fig. 1). The first type of AR
(cedar AR) was constructed of 16 log sections (1.5 m long,
6.9-18.4 cm diameter) of Japanese cedar ( Cryptomeria
japonica) arranged in a parallel cross formation. Each
corner was tied with rope and fixed with a stainless
steel rod. Diagonal wires helped maintain the rectan-
gular shape. The second type of AR (broadleaf AR) was
constructed from six species of broadleaf trees harvested
from the Ashiu Forest Research Station, Kyoto Univer-
sity, and assembled with the same dimensions as those
used for the cedar AR. The broadleaf tree species used

were Japanese cherry birch ( Betula grossa), hornbeam
( Carpinus laxiflora), Japanese beech ( Fagus crenata),
Chinese chestnut (Castanea crenata), redvein maple
( Acer rufinerve), and macropoda holly ( Ilex macropoda).
The diameter of broadleaf and cedar logs ranged from
7.5 to 19.2 cm. The third type of AR (PVC AR) was made
of hollow PVC pipes (11.8 cm diameter, 3 mm thickness)
and was assembled in the same manner as that used for
the other two types of ARs.

These three types of ARs were constructed in trip-
licate and deployed at a depth of 8 m off the Maizuru
Fisheries Research Station (MFRS), Nagahama, Maiz-
uru, Kyoto (35°29'N lat. and 135°22'E long.) on 21 May
2004 (Fig. 2). The shore in this area is a concrete bank
and its subtidal zone consists of natural rocks, concrete
blocks, both partly covered by live oyster ( Crassostrea
gigas) and their dead shells, and sandy silt with some
macroalgal vegetation. The substrate in the research
area consisted of muddy silt with no macroalgae veg-
etation. Each AR was sunk with 240 kg of sand bags
(60 kg attached to each corner of the AR). ARs were
set 15 m apart.

Twice monthly visual censuses of fish assemblages as-
sociated with each AR were conducted for four consecu-
tive years after AR deployment. All census observations
were made by the first author with SCUBA equipment.
The area in and around each AR was observed for about
three minutes and the species, size, and number of
fish were recorded. A census commenced from one of
the lateral sides of an AR and extended out to about
1 m from each side. The observer then swam around
and above the AR, and the fish inside the AR were

164

Fishery Bulletin 108(2)

Figure 2

Map of study area for artificial reef deployment off Maizuru, Kyoto, in 2004. Upper-left
map shows location of Wakasa Bay (in box) along the Sea of Japan. The arrow in the
upper-right map represents the location of the research area in Maizuru Bay. Lower
map shows the research area off the Maizuru Fisheries Research Station (MFRS),
Kyoto University, with the nine artificial reefs (three typesxthree replicates) deployed
in a line. Observations were conducted after the visual census of the adjacent habitat
(transects 1-3). Census lines are expressed by thick dotted lines, and -2 m, -5 m, and
-10 m isobaths are expressed by thin dotted lines.

recorded. Fish were considered as associating with an
AR if they were swimming or dwelling within 1 m of
the AR (Sherman et ah, 2002), and thus fish in an
area of about 10 m2 were counted for each AR. Fish
standard length (SL) was estimated with the help of a
scale marked on a clipboard and was recorded. Length

estimates were occasionally calibrated by capturing
and measuring fish. These calibrations revealed that
visual SL estimates were within 10% error of the actual
measured SL. Water temperature and visibility during
observations ranged from 10.1° to 28.8°C and from 1
to 5 m, respectively. Biomass calculation for each AR

Masuda et al.: Fish assemblages associated with three types of artificial reefs

165

was conducted according to the method of San-
tos et al. (2005) and Friedlander et al. (2007).

The estimated average length of each species for
each sample was converted to mass by using the
length-mass relationship

M=aSL6,

where a and b - constants for allometric growth;

SL = standard length; and
M = mass.

Length-mass parameters were obtained from Fish-
Base (www.fishbase.org, accessed on July 2008)
and calibration was based on our own samples.

The number of fish species (species richness),
total number of fish individuals (abundance), to-
tal fish biomass, and number of individuals of
each fish species associated with each type of AR
were compared among the three types of ARs by
repeated measures ANOVA followed by Tukey’s
HSD test. Data for the number of fish individuals
and their biomass were log (x+l) transformed to
obtain homoscedasticity.

Estimation of the impact of AR deployment
on fish abundance in the adjacent area

Fish assemblages in the area surrounding the ARs
were compared before and after AR deployment.

Data from the twice monthly visual censuses in
each area were used for this purpose (Masuda,

2008; Fig. 2). The number and size of fish of each spe-
cies found along three 400-m2 belt transects have been
recorded twice a month since 1 January 2002. One
transect was close to the location of the ARs that we
deployed in the present study (transect 1), and the other
two were relatively distant (transects 2 and 3). There-
fore, species richness and fish abundance in transect 1
would decline after AR deployment if fish were simply
attracted from the adjacent natural reef to these ARs.
Each of the three transects included areas of rocky reef,
live oysters and their dead shells, a sandy or muddy
silt bottom, and an artificial vertical structure made
of concrete blocks that had been deployed more than
20 years earlier. The size (length x width x height) of
the concrete structures along transects 1, 2, and 3 were
0. 5x3x2. 4 m, 1. 8x3x1 m, and 2. 5x2. 5x2 m, respectively.
Data from 23 May 2002 to 15 May 2004, and those from
29 May 2004 to 8 May 2008 were used to compare the
fish assemblages before and after deployment of the
ARs. Analyses of covariance (ANCOVA) was used to
compare species richness and fish abundance in each
transect before and after deploying the wooden or PVC
ARs, and bottom water temperature was used as a
covariant because fish species richness and abundance
increase almost linearly with the increase of bottom
water temperature in this habitat (Masuda, 2008). The
number of individuals of each species was also compared
by ANCOVA before and after deployment of the ARs. All

Table 1

The mean (± standard error) number of species, individuals, and
estimated biomass of fish attracted to the cedar, broadleaf, and
PVC artificial reefs over the entire observation period (2004-08)
and for each of the four years (n = 3 ARs per type). Different let-
ters represent significant differences among AR types (P<0.01,
Tukey’s HSD test).

Cedar ARs

Broadleaf ARs

PVC ARs

No. of species

Whole period

4.14 ±0.138°

3.49 ±0.1076

3.00

±0.113°

1st year

5.14 ±0.332°

3.44 ±0.2456

2.51

±0.201°

2nd year

4.10±0.289°

3.49 ±0.244h

2.83

±0.232°

3rd year

3.93 ±0.226°

3.63±0.225°6

3.28

±0.2176

4th year

3.38 ±0.196

3.40 ±0.193

3.38

±0.195

No. of individuals

Whole period

40.7 ±4.43°

27.9 ±2.88fe

20.3

±2.18°

1st year

84.5 ±12.9°

36.8 ±5.86h

29.6

±6.23°

2nd year

24.1 ±5.00°

28.0 ±5.88“

10.9

±2.25h

3rd year

32.1 ±8.31°

24.7 ±6.80ft

19.0

±4.00b

4th year

22.0 ±4.59

22.1 ±4.15

21.9

±3.82

Fish biomass (grams)

Whole period 284 ±34.7°

143 ±19.1*

157

±40. 76

1st year

498 ±89.8°

113 ±24.4ft

243

±1576

2nd year

222 ±51.6

134 ±38.7

89.1

±19.3

3rd year

310 ±82.0

179 ±44.8

141

±28.4

4th year

108 ±29.9fc

148 ±41.8ft

155

±28.2°

statistical analyses were conducted with the software
JMP (vers. 5.0. 1J, SAS Institute, Inc., Cary, NC) with
an alpha level of 0.01.

Results

Fish assemblages associated with the ARs

Both species richness and fish abundance were high-
est associated with the cedar ARs, intermediate with
the broadleaf ARs, and lowest with the PVC ARs when
compared over the entire sampling period (Table 1).
These differences were significant among the three AR
types in both of these measurements (repeated mea-
sures ANOVA followed by Tukey’s HSD test: PcO.Ol).
The greater effectiveness of the cedar ARs was promi-
nent in the first year after deployment but decreased
with time and became nonsignificant in the fourth
year (Table 1; Fig. 3). Fish biomass was greatest in
the cedar and PVC ARs in the first and fourth year,
respectively, but did not differ significantly in the
second and third years.

A total of 62 fish species were observed in 96 dives
on these nine ARs, among which six species were
found most frequently in the cedar ARs, two in the
broadleaf ARs, and two in the PVC ARs (Table 2).
Five most commonly observed fish species in the ARs

166

Fishery Bulletin 108(2)

were black rockfish ( Sebastes inermis), jack mackerel
( Trachurus japonicus), bambooleaf wrasse (Pseudola-
brus sieboldi), chameleon goby (Tridentiger trigono-
cephalus) and whitespotted pigmy filefish ( Rudarius
ercodes ) (Fig. 4); the former three species are tar-
geted in commercial fisheries, whereas the latter two
are prey species of other commercial species. Jack
mackerel is pelagic and migratory, and the other four
species are demersal and relatively sedentary. The
typical fishes showing high preference for the cedar
ARs were black rockfish, sunrise sculpin (Pseudo-
blennius cottoides), black sea bream (Acanthopagrus
schlegelii), whitespotted pigmy filefish, thread-sail
filefish ( Stephanolepis cirrhifer), and finepatterned
puffer ( Takifugu poecilonotus). Two species of goby (7s-
tigobius hoshinonis and T. trigonocephalus) were most
abundant in the broadleaf ARs (Fig. 4). Redspotted
grouper ( Epinephelus akaara) and barface cardinalfish
( Apogon semilineatus) were most abundant in the PVC

ARs. Jack mackerel and bambooleaf wrasse were the
most abundant species during the entire census period
(Table 2), but they did not show any clear preference
for a particular type of AR.

Maximum, minimum, and average body length in
two highly abundant and commercially important spe-
cies, black rockfish and jack mackerel, are plotted for
each type of artificial reef in Figure 5. Black rockfish
generally had a wide range (1.5-16 cm) of body length,
whereas jack mackerel had a smaller body size range
(4-12 cm). This was prominent in cedar ARs, especially
shortly after the deployment of the AR (Fig. 5A).

A bryozoan community was established within two
to three months of deploying the cedar ARs. Other en-
crusting epibenthic assemblages, such as Porifera, Cni-
daria, Mollusca, and Annelida, gradually formed on the
broadleaf and PVC ARs after one year. The upper sec-
tions of the ARs attracted these encrusting organisms
more rapidly than the lower sections. In the fourth year,
some of the upper sections of the cedar and
broadleaf ARs began to decay because of foul-
ing by encrusting organisms, particularly wood
boring piddock (Martesia striata). Crabs ( Cha -
rybdis japonica) and sea cucumbers ( Stichopus
japonicus) were common in all types of ARs.
At least four fish species, black sea bream,
Temminck’s surfperch (Ditrema temmincki),
whitespotted pigmy filefish, and thread-sail
filefish, were observed feeding on the encrust-
ing organisms on and around the cedar ARs.
Conger eel ( Conger myriaster), two species of
groupers, and large individuals of bambooleaf
wrasse resided inside the PVC pipes. Some fish,
such as thread-sail filefish and redfin velvetfish
( Paracentropogon rubripinnis), overwintered,
showing minimal movement in the cedar ARs
through the winter.

Fish assemblages in the adjacent habitat

Visual censuses of the areas adjacent to the
ARs revealed that both fish species richness
and abundance showed clear seasonal changes
corresponding to variations in sea bottom water
temperature (Fig. 6). A total of 73,922 fish indi-
viduals from 90 species were recorded from
23 May 2002 to 8 May 2008 in transects 1-3.
There was no significant change in fish species
richness or abundance along any of the three
transects after the deployment of ARs (P>0.5,
ANCOVA; Table 3). Species-to-species analy-
sis revealed that although there were several
cases of increases or decreases in abundance
after deployment, there was no evidence of a
systematic decrease in species richness along
transect 1, in which one species decreased and
four species increased after the deployment (see
far-right column in Table 2). The average (±SE)
number of individuals in the entire census area
of the adjacent habitat was 171 ±12.6 per 400 m2.

_CO

§ 100
"D
>

d 10
z

1000
3

« 100
C/5
03

1 10

Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr
2004 2004 2004 2004

Figure 3

Species richness, fish abundance, and fish biomass associated
with each type of artificial reef on each observation day between
May 2004 and April 2008. Plotted data are averages of the two
monthly observations carried out at each triplicate artificial
reef. Note log scale for individuals and biomass plots.

Masuda et al.: Fish assemblages associated with three types of artificial reefs

167

168

Fishery Bulletin 108(2)

-H>

1— 1

i—l

-O

-ch

-o

1—1

i—i

PVC

i— H

1—1

i—l

±SE

C+-.

»o

-o

T— 1

i— 1

T— 1

'-a

i—i

i— 1

1— 1

1—1 1

i—l

i—i

-o

i—i

i— i

i—i

T— 1

i — i

+ 1

i— 1

1—1

i-H

i—i

T— 1

1— 1

r— i

Cti

< 1 )

!— 1

i— 1

i — 1

i— i

i—i

r— 1

»o

i—i

1—1

r— I

rH 1

-H>

£h

bJD

i-H

•K.

<33

.co

»o

<33

ty)

UjO

~£

-a

<33

^<3>

<33

.g,

<33

•2

tuo

-o

f a

<33

• 2?

'•?

<33

►5

_£

E-h

4— <

4— >

r£

<33

Discussion

The greater effectiveness of cedar ARs

We found that ARs made from logs of cedar trees had a
higher fish species richness and abundance than those
made of broadleaf trees or PVC pipes. The greater effec-
tiveness of the cedar ARs can be attributed to the direct
or indirect effects of cedar wood as an AR material.
Qualitative observations support the latter because we
observed some fish feeding on encrusting organisms on
the cedar ARs. Cedar emits volatile compounds that
repel terrestrial invertebrates to protect the living tree
(Morisawa et al., 2002), but such chemicals might not be
effective as repellants in seawater, making it a suitable
habitat for fouling marine organisms. The rapid growth
of cedar trees results in relatively soft tissues that can
further make the wood a suitable substrate for fouling
organisms. A comparison of the abundance of epibenthic
assemblages between cedar and broadleaf logs will be
required to confirm this hypothesis.

Redspotted grouper was significantly more abundant
in PVC ARs than in the other two types of ARs. The
body length of this species was an average of 14 cm
and ranged from 10 to 19 cm (Table 2), and the inner
diameter of the PVC pipes was 11 cm. ARs with holes
are expected to host more fish (Kellison and Sedberry,
1998), especially large predators (Hixon and Beets,
1989). Indeed, PVC pipes, because of their size, pro-
vided a suitable shelter for redspotted groupers. Yel-
lowspotted grouper ( E . awoara), conger eel, and some
large individuals of bambooleaf wrasse also used the
cavities of the PVC pipes.

Two species of goby were more abundant in the broad-
leaf ARs than in the other two ARs. Most of these go-
bies ranged from 1 to 5 cm. Predation pressure by the
abundant sunrise sculpin and black rockfish in the
cedar ARs, and groupers in the PVC ARs, may have
reduced the survival of gobies in these two types of
ARs, resulting in the relatively higher abundance of
gobies in the broadleaf ARs.

Black rockfish associated with cedar ARs ranged from
1.5 to 16 cm SL. Black rockfish is a viviparous fish
and matures at 12 cm BL in 1-2 years after birth, and
1.5 cm and 16 cm SL individuals represent 1.5-month
and 4-5 year-old individuals, respectively (Hisada et
al., 2000). Whitespotted pigmy filefish associated with
cedar ARs ranged from 1 to 5 cm SL. Whitespotted
pigmy filefish mature at 3 cm SL (Ishida and Tanaka,
1983). Therefore these species use ARs as settlement
sites, nurseries, and adult habitats. Jack mackerel as-
sociated with ARs ranged from 4 to 12 cm SL. Jack
mackerel mature at 14 cm SL (Ochiai et al., 1983) and
attain 4 cm in 2 months (Xie et al., 2005). Therefore
they use ARs mainly as nursery habitat and are loosely
associated with ARs. This finding is in agreement with
that of Rooker et al. (1997) who reported that some mid-
water pelagic fishes, such as carangids and scombrids,
were transient members of the AR fish assemblages.
Considering that there are both pelagic predators, such

Masuda et al.: Fish assemblages associated with three types of artificial reefs

169

2004 2005 2006 2007 2008

Figure 4

The monthly average of individuals of black rockfish iSebastes inermis), jack mackerel ( Trachurus japonicus ), bambooleaf
wrasse ( Pseudolabrus sieboldi), chameleon goby ( Tridentiger trigonocephalus), and whitespotted pigmy filefish (Rudarius
ercodes) associated with each type of artificial reef installed off Maizuru, Kyoto, in 2004.

Table 3

The number of species and number of individuals of fish recorded during
observations along transects 1, 2, and 3 before and after the deployment of
the artificial reefs, expressed as the mean ±standard error (n = 48 and 96
observations for before and after deployment, respectively).

Transect 1

Transect 2

Transect 3

No. of species

Before deployment

9.69 ±0.61

9.67 ±0.60

8.88 ±0.62

After deployment

9.40 ±0.43

9.44 ±0.47

8.40 ±0.42

No. of individuals

Before deployment

116.9 ±21.5

237.6 ±44.6

178.0 ±37.4

After deployment

165.7 ±22.7

225.9 ±34.8

171.1 ±28.9

as Japanese seabass (Lateolabrax ja-
ponicus), and benthic predators, such
as Japanese flounder ( Paralichthys oli-
vaceus ), in this area (Masuda, 2008),
these ARs may well be used as refuges
from predators.

Because the size of ARs was 1.5x1. 5
m and fish were counted within a dis-
tance of 1 m, the survey area repre-
sented about 10 m2 for each AR. The
density of fish associated with the AR
was estimated as 4.07, 2.79, and 2.03
fish per m2 in and around the cedar,
broadleaf, and PVC ARs, respectively
(Table 1). Santos et al. (2005) stud-
ied fish assemblages associated with
ARs made of concrete blocks located
at a similar latitude but deeper depth
(17-22 m) in south Portugal (37°00'N lat., 7°45' and
8°00'E long.), and estimated the mean fish density as
2.01 ±0.74 fish per m2 and fish biomass as 123.6 ±77.4 g
per m2. Fish density on our cedar ARs was about twice

as much but the biomass was much less than the value
reported by Santos et al. This finding was probably the
result of the cedar ARs hosting more recruited juveniles
than adults.

170

Fishery Bulletin 108(2)

Dq Cedar

Trachurus japonicus

: Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr

1 2005 2006 2007 2008

2004 2005

onr\7 200ft

Figure 5

Monthly maximum (rectangles), minimum (triangles), and average (circles) body length of black
rockfish ( Sebastes inermis) found associated with (A) cedar, (B) broadleaf, and (C) PVC artificial
reefs, and those of jack mackerel (Trachurus japonicus) found associated with (D) cedar, (E)
broadleaf, and (F) PVC artificial reefs.

The cedar ARs hosted fish assemblages within the
first two to three months of deployment. These recruits
may have come from the adjacent coastal habitat or
from offshore. Rapid colonization of ARs was also re-
ported by Bohnsack et al. (1994) who observed that fish
species, number of individuals, and biomass reached
peak levels within two months of deploying concrete
ARs in Florida.

There was only one species, Acentrogobius pflaumii,
that decreased in abundance in transect 1 after the de-
ployment of ARs. This goby is the fifth most frequently
observed fish in the adjacent natural reef (Masuda,
2008), but relatively few were associated with ARs.
Therefore it is unlikely that the attraction to ARs in-
duced the decline in the population along transect 1.
The relative stability of fish species and abundance
observed among the three transects supports the con-
cept of an inshore migration and is in agreement with
data of Connell (1997) who found that the number of

recruits did not differ between ARs located close to and
far from a natural reef. Sanchez-Jerez and Ramos-Espla
(2000) also confirmed that antitrawling reefs deployed
in a seagrass habitat had little effect on seagrass fish
assemblages in the surrounding area. We therefore
conclude that the three types of ARs deployed in this
study provided additional habitat for young fish without
any significant depletion of numbers in the existing fish
community.

The average number of fish in the adjacent habitat
was 171 individuals per 400 m2, or 0.43 individuals per
m2. Fish density on the cedar reef was thus 10 times
larger than that of the adjacent area. Bohnsack et al.
(1991) reviewed experimental studies, where fish densi-
ties at natural reefs were compared with those at arti-
ficial reefs, and found that in some cases the latter can
host densities of more than 10 times that of the former.
Therefore, our results of fish density on cedar ARs are
within the range of previously reported ARs.

Masuda et al.: Fish assemblages associated with three types of artificial reefs

171

Deployment of wooden ARs as a tool
for ecosystem-based fishery management

The major anthropogenic impacts on coastal ecosys-
tems include overfishing, loss of physical complexity
induced by construction or trawling, and eutrophica-
tion induced by water discharge. ARs made of cedar
and other materials have the potential to attenuate at
least some of these problems. ARs are useful in that
they preclude trawling, protect juveniles in nursery
grounds, and provide fishing sites for artisanal fisher-
men (Polovina, 1991). Our study site had also been a
trawl fishing ground for bivalves and sea cucumbers,
but fishermen could not trawl at our ARs. The preven-
tion of trawling resulted in the accumulation of rela-
tively large individuals of sea cucumber in our ARs
(R. Masuda, unpubl. data). Habitat complexity, such
as vertical relief and holes, can be a positive factor
for the survival of juvenile fish. For instance, Gorham
and Alevizon (1989) showed that the attachment of
polypropylene rope to ARs significantly increases the
abundance of juvenile fish. Wooden ARs not only pro-
vide vertical relief but also provide a porous substrate
for boring and attachment by encrusting organisms,
such as boring sponges, oysters, and wood boring pid-
dock. Some demersal fishes, such as black rockfish,
wrasses, and gobies might well use these encrusting
organisms for both refuge and as prey.

Most of the encrusting organisms on ARs are
plankton feeders that can use a wide size range of
phytoplankton and zooplankton. For example, a sin-
gle oyster filters several liters of sea water per day
and produces pseudofeces that contain about half
of the organic content of that trapped on the gills
(Deslous-Paoli et al., 1992). Most juvenile and young
demersal fish feed on benthic organisms in addition
to relatively large zooplankton. Therefore, encrusting
organisms on ARs can transform phytoplankton and
microzooplankton to a usable energy source for fish-
es. Fabi et al. (2006) demonstrated that ARs provide
the main food source (e.g., encrusted organisms and
crustaceans) for the three major fish species ( Sciaena
umbra, Diplodus annularis, and Lithognathus mor-
rnyrus) they studied. Furthermore, improved water
clarity due to the filtering function of the encrusting
organisms is likely to result in the better growth of
primary producers, such as macroalgae. The use of fish
reefs as biofilters for nutrient removal has also been
proposed by Seaman and Jensen (2000).

The efficacy of wooden ARs is of a short duration (up
to 3-5 years) compared to those made of concrete, which
can last decades (Yabe, 1995). However, fishermen have
observed that wooden ARs attract fish sooner than other
types of AR. Although wooden ARs biodegrade sooner
than concrete ARs, from an ecological point of view of
providing immediate refuge, habitat, and a source of
food, they have long-term effects on the marine environ-
ment. Simple wooden ARs that combine logs and con-
crete blocks sink easily in a muddy substrate, and their
life as an effective AR can be as short as one year (R.

Figure 6

Seasonal changes in the mean (±standard error) number
of fish species and individuals per transect in the area
immediately adjacent to the artificial reefs, and the surface
and bottom water temperatures measured in those areas
during the surveys from January 2002 to June 2008, at
Nagahama, Maizuru, Japan. Vertical arrows represent the
date (21 May 2004) of artificial reef deployment.

Masuda, personal observ. ). The shape of wooden ARs
presented in this article, with a double-cross formation
(Fig. 1), provides an open and stable vertical relief that
can attract more fish recruits. This formation can also
act as a stable substrate for encrusting organisms that
can function as powerful biofilters, and has a longer
durability than other wooden constructs.

The recruitment of reef fishes is often limited by
the availability of suitable nearshore nursery habitats,
which tend to be vulnerable to anthropogenic impacts.
The decrease of reef fish populations is therefore partly
attributable to the loss of nursery habitats, such as
natural rocky reefs and seagrass beds. The deployment
of wooden ARs may provide an opportunity to mitigate
this trend of decline in nursery quality and because
they are highly biodegradable, the risks of unexpect-

172

Fishery Bulletin 108(2)

ed negative impacts on the environment are minimal.
Stock enhancement, defined as the release of cultured
juveniles into wild populations to augment harvest, has
been used as a strategy to reconstruct depleted fisheries
resources (Bell et al., 2008). We suggest that the release
of reef-associating fish juveniles, such as black rockfish,
combined with the deployment of wooden ARs would
be an efficient approach for the recovery of depleted
coastal fisheries.

A major problem of deploying ARs is that they at-
tract fishermen as well as fishes. There is always the
possibility that fishermen will catch more fish than the
increase of production because fish attracted to ARs
are generally more easily exploitable than those spread
over natural reefs (Powers et al., 2003). Indeed, we of-
ten observed local anglers fishing at our experimental
reefs. Therefore, a management strategy is critically
important in controlling the harvesting pressure at AR
sites (Pickering and Whitmarsh, 1997). As our long-
term goal is to improve the productivity of local inshore
fishing grounds, we would suggest that part of the ar-
eas to be enhanced should have ARs distributed within
them and be managed as marine protected areas.

Acknowledgments

We are grateful to H. Fujii and other technical staff
in the Ashiu Forest Research Station for providing the
materials and construction for the wooden ARs, and
I. Shiga, K. Sato, and graduate students at Maizuru
Fisheries Research Station (MFRS) for help in deploy-
ing the ARs. D. Robert of MFRS, W. Seaman of Univer-
sity of Florida, and three anonymous reviewers kindly
provided constructive and insightful comments on the
manuscript. This research was partly supported by a
Grant-in-Aid for Scientific Research from the Japan
Society for the Promotion of Science.

Literature cited

Bell, J. D., K. M. Leber, H. L. Blankenship, N. R. Loneragan,
and R. Masuda.

2008. A new era for restocking, stock enhancement and
sea ranching of coastal fisheries resources. Rev. Fish.
Sci. 16:1-9.

Bohnsack, J. A., D. L. Johnson, and R. F. Ambrose.

1991. Ecology of artificial reef habitats and fishes. In
Artificial habitats for marine and freshwater fisher-
ies (W. Seaman Jr., and L. M. Sprague, eds.), p. 61-
107. Academic Press, San Diego, CA.

Bohnsack, J. A., D. E. Harper, D. B. McClellan, and M. Hulsbeck.
1994. Effects of reef size on colonization and assemblage
structure of fishes at artificial reefs off southeastern
Florida, U.S.A. Bull. Mar. Sci. 55:796-823.

Collins, K. J., A. C. Jensen, J. J. Mallinson, V. Roenelle, and
I. P. Smith.

2002. Environmental impact assessment of a scrap tyre
artificial reef. ICES J. Mar. Sci. 59:S243-S249.
Connell, S. D.

1997. The relationship between large predatory fish and

recruitment and mortality of juvenile coral reef-fish
on artificial reefs. J. Exp. Mar. Biol. Ecol. 209:261-
278.

Deslous-Paoli, J. — M., A. — M. Lannou, P. Geairon, S. Bougrier,
O. Raillard, and M. Heral.

1992. Effects of the feeding behaviour of Crassostrea
gigas (bivalve Molluscs) on biosedimentation of natural
particulate matter. Hydrobiologia 231:85-91.

Fabi, G., S. Manoukian, and A. Spagnolo.

2006. Feeding behavior of three common fishes at an
artificial reef in the north Adriatic Sea. Bull. Mar.
Sci. 78:39-56.

Ferreira, C. E. L., J. E. A. Confalves, and R., Coutinho.

2001. Community structure of fishes and habitat com-
plexity on a tropical rocky shore. Environ. Biol. Fish.
61:353-369.

Friedlander, A. M., E. K. Brown, and M. E. Monaco.

2007. Coupling ecology and GIS to evaluate efficacy
of marine protected areas in Hawaii. Ecol. Appl.
17:715-730.

Gorham, J. C., and W. S. Alevizon.

1989. Habitat complexity and the abundance of juvenile
fishes residing on small scale artificial reefs. Bull.
Mar. Sci. 44:662-665.

Gratwicke, B., and M. R. Speight.

2005. The relationship between fish species richness,
abundance and habitat complexity in a range of shallow
tropical marine habitats. J. Fish Biol. 66:650-667.

Grossman, G. D., G. P. Jones, and W. J. Seaman Jr.

1997. Do artificial reefs increase regional fish production?
A review of existing data. Fisheries 22(4): 17 — 2 3 .

Hamilton, J., and B. Konar.

2007. Implications of substrate complexity and kelp
variability for south-central Alaskan nearshore fish
communities. Fish. Bull. 105:189-196.

Hisada, T., T. Inoue, and Y. Hamanaka.

2000. Age, growth and maturity of a black rockfish in
the western Wakasa Bay. Bull. Kyoto Inst. Ocean.
Fish. Sci. 22:44-49.

Hixon, M. A., and J. P. Beets.

1989. Shelter characteristics and Caribbean fish assem-
blages: experiments with artificial reefs. Bull. Mar.
Sci. 44:666-680.

Ishida, Y., and S. Tanaka.

1983. Growth and maturation of the small filefish Rudar-
ius ercodes in Odawa Bay. Bull. Japan. Soc. Sci. Fish.
49:547-553.

Kellison, G. T., and G. R. Sedberry.

1998. The effects of artificial reef vertical profile and
hole diameter on fishes off South Carolina. Bull. Mar.
Sci. 62:763-780.

Masuda, R.

2008. Seasonal and interannual variation of subtidal
fish assemblages in Wakasa Bay with reference to the
warming trend in the Sea of Japan. Environ. Biol.
Fish. 82:387-399.

Morisawa, J., C.-S. Kim, T. Kashiwagi, S. Tebayashi, and
M. Horiike.

2002. Repellents in the Japanese cedar, Cryptome-
ria japonica, against the pill-bug, Armadillidium
vulgare. Biosci. Biotechnol. Biochem. 66:2424-2428.

Ochiai, A., K. Mutsutani, and S. Umeda.

1983. On the first year’s growth, maturity and artificial
spawning of cultured jack mackerel. Bull. Japan. Soc.
Sci. Fish. 49:541—545.

Masuda et al.: Fish assemblages associated with three types of artificial reefs

173

Pickering, H., and D. Whitmarsh.

1997. Artificial reefs and fisheries exploitation: a review
of the ‘attraction versus production’ debate, the influ-
ence of design and its significance for policy. Fish.
Res. 31:39-59.

Polovina, J. J.

1991. Fisheries applications and biological impacts of
artificial reefs. In Artificial habitats for marine and
freshwater fisheries ( W. Seaman Jr., and L. M. Sprague,
eds.), p. 153-176. Academic Press, San Diego, CA.

Powers, S. P., J. H. Grabowski, C. H. Peterson, and W. J.

Lindberg.

2003. Estimating enhancement of fish production by off-
shore artificial reefs: uncertainty exhibited by divergent
scenarios. Mar. Ecol. Prog. Ser. 264:265-277.

Reed, D. C., S. C. Schroeter, D. Huang, T. W. Anderson, and R.

F. Ambrose.

2006. Quantitative assessment of different artificial reef
designs in mitigating losses to kelp forest fishes. Bull.
Mar. Sci. 78:133-150.

Rilov, G., and Y. Benayahu.

2000. Fish assemblage on natural versus artificial reefs:
the rehabilitation perspective. Mar. Biol. 136:931-942.

Rooker, J. R., Q. R. Dokken, C. V. Pattengill, and G. J. Holt.

1997. Fish assemblages on artificial and natural reefs in
the Flower Garden Banks National Marine Sanctuary,
USA. Coral Reefs 16:83-92.

Sanchez-Jerez, P., and A. Ramos-Espla.

2000. Changes in fish assemblages associated with
the deployment of an antitrawling reef in seagrass
meadows. Trans. Am. Fish. Soc. 129:1150-1159.

Santos, M., N., C. C. Monteiro, and G. Lasserre.

2005. Observations and trends on the intra-annual varia-
tion of the fish assemblages on two artificial reefs in
Algarve coastal waters (southern Portugal). Sci. Mar.
69:415-426.

Scharf, F. S., J. P. Manderson, and M. C. Fabrizio.

2006. The effects of seafloor habitat complexity on
survival of juvenile fishes: Species-specific interac-
tions with structural refuge. J. Exp. Mar. Biol. Ecol.
335:167-176.

Seaman, W., Jr., and A. C. Jensen.

2000. Purposes and practices of artificial reef
evaluation. In Artificial reef evaluation (W. Seaman
Jr., ed), p. 1-19. CRC Press, Boca Raton, FL.

Seaman, W., Jr., and L. M. Sprague.

1991. Artificial habitat practices in aquatic systems. In
Artificial habitats for marine and freshwater fisheries
( W. Seaman Jr., and L. M. Sprague, eds.), p. 1-29. Aca-
demic Press, San Diego, CA.

Sherman, R. L., D. S. Gilliam, and R. E. Spieler.

2002. Artificial reef design: void space, complexity, and
attractants. ICES J. Mar. Sci. 59:S196-S200.

Xie, S., Y. Watanabe, T. Saruwatari, R. Masuda, Y. Yamashita,
C. Sassa, and Y. Konishi.

2005. Growth and morphological development of sag-
ittal otoliths of larval and early juvenile Trachurus
japonicus. J. Fish Biol. 66:1704-1719.

Yabe, K.

1995. The note of evaluation of artificial fish reefs on
the sand beaches at Haboro, Hokkaido. Bull. Hokkaido
Tokai Univ. 8:101-108.

Abstract — The Pacific sardine ( Sar -
dinops sagax) is distributed along the
west coast of North America from Baja
California to British Columbia. This
article presents estimates of biomass,
spawning biomass, and related biolog-
ical parameters based on four trawl-
ichthyoplankton surveys conducted
during July 2003-March 2005 off
Oregon and Washington. The trawl-
based biomass estimates, serving as
relative abundance, were 198,600 t
(coefficient of variation [CV] = 0.51)
in July 2003, 20,100 t (0.8) in March
2004, 77,900 t (0.34) in July 2004,
and 30,100 t (0.72) in March 2005
over an area close to 200,000 km2.
The biomass estimates, high in July
and low in March, are a strong indi-
cation of migration in and out of this
area. Sardine spawn in July off the
Pacific Northwest (PNW) coast and
none of the sampled fish had spawned
in March. The estimated spawn-
ing biomass for July 2003 and July
2004 was 39,184 t (0.57) and 84,120 t
(0.93), respectively. The average active
female sardine in the PNW spawned
every 20-40 days compared to every
6-8 days off California. The spawning
habitat was located in the southeast-
ern area off the PNW coast, a shift
from the northwest area off the PNW
coast in the 1990s. Egg production in
off the PNW for 2003-04 was lower
than that off California and that in
the 1990s. Because the biomass of
Pacific sardine off the PNW appears
to be supported heavily by migratory
fish from California, the sustainabil-
ity of the local PNW population relies
on the stability of the population off
California, and on local oceanographic
conditions for local residence.

Manuscript submitted 22 October 2008.
Manuscript accepted 29 December 2009.
Fish. Bull. 108:174-192 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

Biomass and reproduction of

Pacific sardine ( Sardinops sagax )

off the Pacific northwestern United States,

2003-2005

Nancy C. H. Lo (contact author)

Beverly J. Macewicz
David A. Griffith

Email address for contact author: Nancy.Lo@noaa.gov

Southwest Fisheries Science Center
8604 La Jolla Shores Dr.

La Jolla, California 92037

Pacific sardine ( Sardinops sagax ; here-
after “sardine”) are distributed widely
off the west coast of North America
from Baja California, Mexico, to Brit-
ish Columbia, Canada; the majority of
the population is located off California
(Felin, 1954; Murphy, 1966; Emmett
et ah, 2005; McFarlane et ah, 2005;
Smith, 2005). Tagging studies have
shown that sardine migrate along the
west coast (Janssen, 1938; Clark and
Janssen, 1945). The sardine popu-
lation reached a peak in the early
1930s at 3.5 million metric tons (t)
and declined rapidly in the mid-1950s
(Marr, 1950). The sardine fishery off
California and British Columbia dates
from 1916 (Fig. 1). Pacific sardine
was one of the economically impor-
tant species off California and British
Columbia in the 1930s when a fishery
began off Oregon and Washington (the
Pacific Northwest: PNW). The PNW
catch peaked at nearly 50,000 t in
1938 (Marr, 1950; Mosher and Eckles,
1954; Murphy, 1966). In the 1960s,
however, a moratorium on sardine
fishing was established in U.S. waters
because of low catches (Murphy, 1966;
MacCall, 1976). In the mid 1980s, sar-
dine became common as bycatch in
fisheries off Baja California and Cali-
fornia state (Wolf, 1992; Deriso et al.,
1996) and reappeared from Oregon
to British Columbia in 1992 (Emmett
et al., 2005; McFarlane et al., 2005),
apparently in response to the 1992-93
El Nino event. The sardine population
now supports a relatively large fish-
ery with annual catches over 50,000
t in recent years (Fig. 1). Sardine

also serve as important food for tuna,
salmon, marlin, mackerel, sharks, and
some groundfish species, as well as
many seabirds, seals, sea lions, dol-
phins, and whales (Snodgrass and
Lowry, personal commun.1) (Preti et
al., 2001, 2004; Emmett et al., 2005).
The reappearance of sardine popula-
tions in the north California Current
ecosystem adds another forage base for
predators and an emerging resource of
consumer interest to the ecosystem.

Pacific sardine off the PNW are
considered to be a part of the north-
ern subpopulation, the majority of
which is distributed off the western
United States and Canada (Smith,
2005), as determined from historical
tagging studies (Clark and Janssen,
1945), size at age, and other biological
characters. Historical tagging studies
indicated that some large sardine mi-
grate from California to the PNW in
late spring and early summer to feed,
and that the majority of the large
sardine off the PNW move south to
California in the winter to spawn in
the spring (Clark and Janssen, 1945).
The major spawning area of this
northern subpopulation was believed
to be located off southern California
before the 1960s (Ahlstrom, 1948;
Marr, 1960; Smith, 2005). Spawning
also may have occurred in the PNW
because young fish were caught by
commercial boats in Canadian wa-
ters in 1940 (Hart, 1943). However,

1 Snodgrass, Owyn. 2009. Southwest
Fisheries Science Center, La Jolla, CA.
Lowry, Mark. 2009. Southwest Fisher-

ies Science Center, La Jolla, CA.

Lo et al. Biomass and reproductive status of Sardmops sagax off the Pacific coast.

175

the importance of the PNW as a spawning area
has not been studied. After the resurgence of
Pacific sardine off California, ichthyoplankton
and fishery-independent trawl surveys have been
conducted off California to assess the biological
characteristics of the sardine population since the
mid 1980s, when the estimated sardine biomass
approached 20,000 t (Wolf, 1992; Lo et al., 2005).
Beginning in the mid 1990s, sardine abundance,
distribution, and ecological relationships off the
PNW and Canada were analyzed with data from
salmon surface-rope trawl surveys off the PNW
and trawl surveys off Vancouver Island, Canada
(Bentley et al., 1996; Emmett et al., 2005; Mc-
Farlane et. al., 2005); however, very few of those
surveys were designed specifically to assess the
biological characteristics of Pacific sardine.

Four trawl surveys off the PNW were conducted
in July 2003, March and July 2004, and March
2005 to provide fishery-independent measures of
biological characteristics of sardine in this area,
and to answer the following questions: 1) Do sar-
dine migrate between the PNW and California?

2) To what extent does Pacific sardine spawning
in the PNW depend on the sardine population off
California? and 3) How much does the Pacific sar-
dine egg production in the PNW contribute to that of the
whole population? To answer these questions, we esti-
mated spring and summer biomasses with length distri-
butions to serve as signals of migration; the location and
spatial extent of spawning habitat to examine the fol-
lowing: the effect of the reduction of the spawning area
in the PNW to the local population; daily egg production
and its contribution to the total egg production; adult
reproductive parameters to estimate rates of spawning,
fecundity and maturity; and spawning biomass. These
measurements were compared with available PNW mea-
surements from the mid-1990s and those off California
in the same time period, to facilitate our understanding
of the population dynamics of the Pacific sardine off
the northern west coast of the North American con-
tinent, and to better manage the entire population.

Materials and methods

Survey

In order to obtain unbiased estimates of the biologi-
cal characteristics of Pacific sardine off the PNW, the
Fisheries Resources Division of the Southwest Fisheries
Science Center, conducted four surveys in July 2003,
March and July 2004, and March 2005 aboard the FV
Frosti. Multiple gear types were used: a surface trawl to
collect adult samples, the CalVET plankton net (Califor-
nia Cooperative Oceanic Fisheries Investigation verti-
cal-egg-tow net; Smith et al., 1985), and the continuous
underway fish egg sampler (CUFES; Checkley et al.,
1997) to collect ichthyoplankton samples and record
hydroacoustics. The survey region encompassed the area

of the northeast Pacific Ocean from 42° to 48°N latitude
and from inshore out to 128°W longitude.

The basic survey pattern comprised seven transect
lines oriented on the parallels at a spacing of 60 nauti-
cal miles (111 km). Stations were spaced at 30 nautical
miles (55.5 km) along each transect measured from
the offshore station. Forty-two predetermined stations
were sampled by trawl and CalVET tow during each
survey. For the July 2003 survey, the primary goal was
to estimate the spawning biomass of Pacific sardine. In
the offshore area, few trawls were undertaken because
both acoustics and CUFES samples showed little sign
of sardine schools and eggs. The inshore sampling was
discontinued close to the 100-m isobath during July
2003 to avoid net damage in shallow water. All fish-
ing was conducted at night, when Pacific sardine are
distributed in the upper 50 m of the water column and
oceanographic conditions at depths greater than 50 m
would have little influence on the spatial and vertical
distributions of sardine schools. Moreover, within 60 km
from the shore, the densities of fish were not related to
the distance from shore (Emmett et al., 2005). Therefore
we expected little bias introduced from sampling along
the 100-m isobath. With more experience, we found that
we could tow the net at a shallower depth than initially
expected, and during subsequent surveys we occasion-
ally fished inshore at shallower depths (see below, Figs.
2-5). For the remaining three surveys, most trawls
were evenly distributed along the transect line and be-
tween transect lines in the inshore area. Occasionally,
trawls were made during transit between transect lines.

Both trawl and CalVET samples were collected dur-
ing all four surveys and CUFES samples were collected
during July surveys only (Figs. 2-5). Trawl-related

176

Fishery Bulletin 108(2)

48°N

46°N

44°N

42°N

40°N

48°N

46°N

44°N

42°N

40°N

• Positive CalVET
o Negative CalVET
▲ 0<egg density from CUFES<0.5 eggs/min!
J i Egg density from CUFES>0 5 eggs/min
|~j High density spawning region j
Temperature contours are °C

California

100

200

128°W

126°W

124°W

122°W

128°W

126°W

124°W

122“W

Figure 2

(A) Locations of trawls (stars) used for the estimation of biomass of Pacific sardine (Sardinops sagax), excluding added
trawls in the inshore area (Table 1), and (B) locations of California vertical egg tows (CalVET: circles) and where con-
tinuous underway egg samples were taken (measured in eggs per minute (CUFES: triangles). Contours are sea surface
temperatures (°C). The dashed vertical line is the 125°W longitude divider of the two sampling strata. The offshore
shaded area in (B) is the major spawning habitat. Positive tows were those tows during which sardine were caught.
Negative tows were tows when sardine were not caught.

station activities were performed between twilight and
dawn, whereas CalVET and CUFES samples were col-
lected throughout all 24 hours. At each station, a Cal-
VET sample was collected and sea surface temperature
(SST) was recorded, whereas between stations, CUFES
samples and water temperature were taken at a fixed
3-m depth (Figs. 2 and 4). The CUFES data were used
primarily to map the spawning area based on the den-
sity of sardine eggs.

A Nordic 264 trawl (NET Systems, Bainbridge Is-
land, WA), with a vertical opening of 20 m, a mouth
area of approximately 360 m2, and a 7-mm codend
mesh (Emmett et ah, 2005), was towed to sample the

upper 18-20 m of the water column. The distance trav-
eled by each trawl was recorded and was later con-
verted to the volume sampled. The swept area (m2)
is the volume (m3) divided by 20 m. During the July
2003 survey, few trawls were taken in the offshore
area. Additional trawls were taken inshore to collect
extra samples to determine reproductive parameters
in areas of sardine spawning activity identified by
sardine egg densities in CUFES samples or the pres-
ence of schools as indicated by acoustic signals (Fig. 2).
Similarly, during July 2004, trawls were taken in the
southern spawning area off Port Orford, OR (Fig. 4),
in addition to the prepositioned and between-transect

Lo et al. Biomass and reproductive status of Sardinops sagax off the Pacific coast.

177

trawls. Data from the added trawls
were excluded in estimating the total
biomass to avoid bias. For the two
March surveys, all locations (fixed
stations and between-transects) were
trawled regardless of spawning or
acoustic signals (Figs. 3 and 5). The
total number of trawls for each sur-
vey was close to 50 (Table 1).

For each trawl, the total weight
(kg) of the Pacific sardine catch was
recorded and up to 50 Pacific sardine
were randomly sampled from each
trawl where sardine were caught
(hereafter referred to as a “positive
trawl”). Sex was determined for each
fish, and standard length (SL) and
weight were measured. For the female
fish, the ovaries were first examined
for torpedo shape and or development
of visible oocytes (yolking or hydrat-
ing). When oocytes were not visible
and the ovary was small, clear, and
torpedo shaped, the ovary was re-
corded as code 1 (clearly immature).
Otherwise, the additional ovarian
codes 2 (intermediate), 3 (active), or
4 (hydrated) (Table 2) were used to
identify potentially mature females —
because only histological analysis can
verify sardine maturity with certain-
ty (Macewicz et al., 1996). All ovaries
were removed and preserved in 10%
neutral buffered formalin. If a 50-
fish subsample did not have 25 po-
tentially mature females (ovary codes
2-4), more females were sampled to
attain 25 per trawl for estimation
of reproductive parameters used for
computing spawning biomass. Addi-
tional females were also processed to
estimate batch fecundity, but were
not included in the original random
subsample for length distributions.
We also obtained length distribu-
tions based on data from commercial
purse seine catches off the PNW in
the summer seasons and from a test
purse seine set in March 2005.

48°N

46°N

44°N

42°N

29 February-19 March 2004

hf.

*&:: ■ : ■

Astoria . WASHINGTON
OREGON

/Coos Bay

OREGON

CALIFORNIA

40>^
38^4?

36°N

34°N

32°N

30°N

# Positive CalVET
Negative CalVET

a 0<egg density from CUFES<1 eggs/min
& Egg density from CUFES>1 eggs/min

■ j High density spawning regi

★ Positive sardine trawls
☆ Negative sardine trawls

Temperature contours are °C
I 1

128°W 126°W 124°W 122°W 120°W 118°W 116°W

Seasonal biomass

A swept-area method was used to
estimate the total biomass of Pacific
sardine in summer and spring based
on July and March trawl data, respec-
tively. Because the efficiency of the
trawl catch has not been evaluated,
the biomass estimates must be con-
sidered as relative and minimum

Figure 3

Locations of trawl (stars) and California vertical egg tows (CalVET: circles),
for 2004 March ichthyoplankton-trawl survey off the Pacific Northwest
(top map), and locations of trawls, CalVET tows (circles), and continuous
underway egg sampling (CUFES: triangles) for the March-April 2004
California Cooperative Oceanic Fisheries Investigations (CalCOFI) daily
egg production survey (bottom map). Solid symbols indicate that Pacific
sardine ( Sardinops sagax) were captured in the sample at that site. Con-
tours are sea surface temperatures (°C). The dashed vertical line at 125°W
longitude (seen in top map) is the divider of the two sampling strata. The
shaded area on the bottom is the identified spawning habitat.

178

Fishery Bulletin 108(2)

Figure 4

(A) Locations of trawls (stars) used for biomass estimation of Pacific sardine ( Sardinops sagax), excluding added trawls
in the inshore area (Table 1), and (B) California vertical egg tows (CalVET: circles) and continuous underway egg
sampling in eggs/minute (CUFES: triangles) for 2004 July trawl-ichthyoplankton survey off the Pacific Northwest.
Contours are sea surface temperatures (°C). The dashed vertical line is the 125°W longitude divider of the two sam-
pling strata. The shaded area is the major spawning habitat. Positive tows were those tows during which sardine were
caught. Negative tows were tows when sardine were not caught.

abundances. A stratified sampling design was used
to estimate biomass and spawning biomass, because
more stations were assigned close to the shore than
offshore. Otherwise, estimates would be biased toward
the inshore area (Holt and Smith, 1979). The survey
area was divided into an inshore area (stratum 1) and
an offshore area (stratum 2) with 125°W longitude as
the dividing line. For the July 2003 survey, we excluded
the nonpredetermined trawls (i.e., those trawls locations
of which were not determined before the survey) taken
in the vicinity of positive trawls to prevent an overesti-
mate of the total biomass. The catch for each tow was
expressed as kg/m2 ( = catch [kg]/swept areaT m2]=catch
[kg]/volume of water [m3]/depth 20 m), where the volume

of water filtered was computed as the distance covered
by each tow multiplied by the area of the vertical trawl
mouth opening of approximately 360 m2 (with 20 m as
diameter). We estimated relative total biomass (B) and
its standard error (SE) for each survey as follows:

S = ^ ( A, 106 ) / 1000, (1)

SE(B) = ( ^ ( var( Xl ) (A, 106 )2 )1/2/ 1000 ( 2 )

where B = the estimate of the total biomass (t);

Xi = the mean catch (kg/m2); and

Lo et al. Biomass and reproductive status of Sardmops sagax off the Pacific coast.

179

A( = the area (km2) in stratum
i, i=l (inshore) and 2 (off-
shore).

Note: the coefficient of variance (CV)
of the estimate is CV(B) - SE (B)/B.
Bootstrap simulation was used to esti-
mate the bias of the estimate (Eq. 1),
and the bias-corrected estimate (Bc)
as Bc=B-(Bb-B), where B is computed
from Equation 1, Bh is the estimate
from the bootstrap simulation, and the
mean square error (MSE -variance +
bias2) of the biomass estimates (Eq. 2).

We also computed a crude esti-
mate of the recruit biomass (age-
zero year or incoming year class) as
ancillary information for compara-
tive purposes for spring in 2004 and
2005, based on the biomass of fish
<120 mm SL because 120 mm was
the break point for the length-fre-
quency distribution in March surveys
from this study (Fig. 6) and it was
reported that age-0 sardine in the
PNW were <110 mm (measured by
fork length) (Emmett et al., 2005).
Recruit biomass (BR) was estimated
by using Equations 1 and 2, where
Xt= the mean catch (kg/m2) of fish
<120 mm SL in the tth stratum ( XR ■ ).
The catch of recruits for each trawl
would be obtained as

XR,ij=Xi*Uij, length <120mm’ where Xij =

the total catch from the yth trawl,
and U lengths 120 mm =the Weight of fish

<120 mm SL divided by the total fish
weight based on our random samples
with a maximum of 50 fish from each
tow.

Spawning habitat

The spawning habitat was defined as
the area of relatively high egg densi-
ties during early summer, because
June-July was the peak spawning
time for Pacific sardine off the PNW
as determined from egg and larval
data collected in the mid-1990s (Bent-
ley et al., 1996). Because the number
of positive CalVET tows was low (four
of 54 tows during July 2003 and 3 of
48 tows during July 2004), we chose to
use data from CUFES sampling. The
spawning habitat area was defined
as the area where the majority of egg
densities exceeded a threshold of 0.5
eggs/min because the egg densities
were generally low. Off California,

48°N -

46°N

44°N -

42°N

36°N

34°N -

32°N

128°W

126°W

124°W

122°W

120°W

118°W

116°W

40'X,

38°^

30°N

Temperature contours are °C

• Positive CalVET
Negative CalVET

* 0<egg density from CUFES<1 eggs/min
± Egg density from CUFES>1 eggs/min

fH High density spawning region

★ Positive sardine trawls
☆ Negative sardine trawls

)

1 5 April—

1 May 2005

Long Beach

2-21 March 2005

Figure 5

Locations of trawl (stars) and California vertical egg tows (CalVET:
circles) for the 2005 March ichthyoplankton-trawl survey off the Pacific
Northwest (top) and trawls, CalVET tows (circles), and continuous
underway egg samples (CUFES: triangles) for the April-May 2005
California Cooperative Oceanic Fisheries Investigations (CalCOFI)
daily egg production survey (bottom). Solid symbols indicate that Pacific
sardine (Sardinops sagax) were captured in the sample at that site.
Contours are sea surface temperatures (°C). The dashed vertical line
is the 125°W longitude divider of the two sampling strata. The shaded
area is the spawning habitat.

180

Fishery Bulletin 108(2)

Table 1

Estimates of biomass of Pacific sardine ( Sardinops sagax), biomass-related parameters for each survey (stratum 1, stratum 2
[with latitude 125°W as the dividing line between them], and the entire survey area), and confidence intervals for biomass and
recruits: incoming year class (fish <120 mm standard length) in tons (t). Either the coefficient of variation (CV) or number of
positive (pos) trawls is in parentheses.

Stratum 1

Stratum 2

Entire survey area

July 2003

Mean density (kg/m3)

2.41e-°04

2.2e-°06

4.85e-°05

Biomass (t) (CV)

192,801(0.57)

7207(0.85)

200,008(0.56)

No of trawls1 (pos)

38(34)

10(2)

48(36)

No of trawls used for biomass (positive)

17(14)

5(2)

22(16)

Survey area (km2) (% of entire survey area)

40,043(19)

166,333(81)

206,377(100)

Bootstrap results

Biomass (t) (CV)

193,946(0.53)

7406(0.82)

201,360(0.51)

Mean square error (MSE)1/2

102,174

6094

102,378

Bias-corrected
Confidence interval (t)

191,656

7009

198,656

44,286-421,321

March 2004

Mean density(kg/m3)

2.6e-°05

4 6e-°08

5.2e-°06

Biomass (t) (CV)

21,243(0.83)

155(0.7)

21,398(0.82)

Recruits (t)( CV )

21,030(0.83)

69(1.0)

21,099(0.83)

No of trawls (pos)

25(7)

34(2)

59(9)

Survey area (km2) (%)

40,043(19)

166,334(81)

206,377(100)

Bootstrap results

Biomass (t) (CV)

22,494(0.81)

155(0.69)

22,650(0.8)

MSE1/2

18,260

106

18260

Bias-corrected
Confidence interval (t)

19,992

155

20,147

0-63,017

Recruits (t)(CV)

21,629(0.81)

69(1.02)

21,698(0.81)

MSE1/2

17,576

17,578

Bias-corrected
Confidence interval (t)

20,432

20,501

4749-45,808

July 2004

Mean density (kg/m3)

9.0e-005

2.3e-°06

2 ie-°°5

Biomass (t) (CV)

72,206(0.405)

6989(0.992)

79,194(0.379)

No of trawls1 (pos)

20(16)

38(11)

58(27)

No of trawls used for biomass (pos)

17(15)

30(14)

47(19)

Survey area (km2) (%)

40,043(21)

150,932(79)

190,975(100)

Bootstrap results

Biomasst t) (CV)

73,186(0.41)

7299(0.95)

80,485(0.38)

MSE1/2

29,723

6928

30,605

Bias-corrected
Confidence interval (t)

71,226

6678

77,903

30,474-146,176

March 2005

Mean density (kg/m3)

3.7e-°05

2.3e-007

7.9e~006

Biomass (t)(CV)

29,488(0.69)

700(0.57)

30,188(0.68)

Recruits (t) (CV)

55(1.0)

0(0)

54.80(1.0)

No of trawls (pos)

15(11)

34(9)

49(20)

Survey area (km2) (%)

40,043(21)

150,932(79)

190,976(100)

Bootstrap results

Biomass (t)(CV)

29,573(0.73)

705(0.57)

30,278(0.72)

MSE1/2

21,713

402

21,714

Bias-corrected
Confidence interval (t)

29,403

695

30,098

1800-86,035

Recruits (t) (CV)

56.6(0.98)

0(0)

57(0.98)

MSE1/2

Bias-corrected
Confidence interval (t)

70-1640

1 During the July 2003 cruise, data from 22 out of 48 trawls were used for biomass computation. The total 48 trawls included 38 (34 with sardine)
in stratum 1 and 10 (2 with sardine) in stratum 2. During the July 2004 cruise, only data from the first 47 trawls out of 58 trawls were used. The
total 58 trawls included 20 trawls (16 with sardine) in stratum 1 and 38 trawls (11 with sardine) in stratum 2.

Lo et al. Biomass and reproductive status of Sardmops sagax off the Pacific coast.

181

Table 2

Gross anatomical classification of female and male Pacific sardine ( Sardinops sagax) gonads.

Gonad code

Female: Ovary description

Clearly immature : Oocytes are not visible. Ovary is very small, translucent or clear, and thin, but with
rounded edges (torpedo shaped).

Intermediate'. Individual oocytes are not visible to the unaided eye (no visible yolk or hydrate oocytes in
the ovaries), but ovary is not clearly immature. Includes possible maturing and regressed ovaries.

Active : Yolked oocytes in ovaries visible to the unaided eye in any size or amount, including the smaller
opaque oocytes (around 0.4-0. 5 mm) to the large yellowish oocytes (about 0.6-0. 8mm).

Hydrated : Hydrated oocytes are present, yolked oocytes may also be seen. Hydrated oocytes (large and
transparent), from few to many, or even if loose or “oozing” or “running” from ovary, qualify for this
class

Male:

Testis description

Clearly immature'. Testis is very small, knife shaped, translucent or clear, and thin with a flat ventral
edge.

Intermediate'. No milt is evident and testis is not clearly immature (includes maturing or regressed
testes).

Active: Milt is present either oozing from the gonopore, in the duct, or in the testis (observed when the
testis was cut).

the threshold was one egg/min. We obtained the SST
for CUFES samples with >0.5 eggs/min as a proxy for
the oceanographic conditions. No biological variables
such as zooplankton volume (Lynn, 2003) were collected
during these surveys.

Daily egg production

The daily egg production (P0) is defined as the newly
spawned eggs produced per 0.05 m2 per day, where 0.05
m2 was the surface area covered by the CalVET net tow
The daily rate of egg production and the daily specific
fecundity rate from adult parameters (Lasker, 1985)
are needed to compute spawning biomass. In California
waters, sardine egg data from CalVET tows and yolksac
larval data from both CalVET tows and bongo nets, and
sardine ages were used to model the embryonic mortal-
ity curve, a negative exponential curve (Lo et al., 1996,
2005):

Pl = P0e^zt\ (3)

where Pt = the daily production rate at age t (days);

2 = the daily instantaneous embryonic mortality
rate; and

P 0 = the intercept, is the daily egg production at
age zero.

Because few eggs were caught during CalVET net tows
in July surveys and no eggs were caught in March
surveys (Fig. 2-5, Table 3), no attempt was made to
estimate egg production for the March surveys. For July
surveys, it was impossible to model the egg mortality
curve because the mortality curve requires sufficient

data on egg abundance for each egg stage and age.
Instead, we used an alternative algorithm to estimate
P0, an integral method (P0 1) based on the standing stock
of eggs from CalVET tows.

The estimate of P0 (P0 7) was based on the relation-
ship between the mean catch of eggs from CalVET
tows (Y) and egg production (P0) through the inte-
gral of Pt over the period from spawning to hatching
(th). The mean catch of eggs from CalVET tows was
a weighted average with the area in each stratum as
weight. This method requires prior knowledge of the
egg mortality rate and the temperature-dependent
hatching time:

*h th

Y=^Ptdt = ^ P0e~ztdt. (4)

o o

Integrating the above equation yields the estimate of
P0 as a function of the mean egg density, Y, incubation
time, th , and the daily instantaneous mortality rate, 2:

with variance calculated by using the delta method:

<3Pa J 9 d Pv 1 9

var( P0 7 ) = ( J r var( 2) + ( - )" var( Y),

az oY

Y[1 - exp(-2^ )(1 + zth )] 2 , ,

= ( - 0 ) var(2)

[1-exp {-zth)Y

+( )2 var(Y).

1 - exp( -zth )

182

Fishery Bulletin 108(2)

The 2 value was the estimate from the daily egg pro-
duction method (DEPM) surveys off California in 2003
(0.48 [CV=0.08]) and 2004 (0.25 [CV=0.04]) (Lo et ah,
2005) because of the lack of sufficient data to estimate
2 off the PNW. Age at hatching (in days) was 2.5 days
computed from the temperature-dependent sardine
egg development model for stage XII given in Lo et al.
(1996): ^ = 30.65* exp(-0.145*temp-0.037*12)*121 41/24,
where temp is the average temperature from positive
CUFES collections during the July surveys, and equals

16.4°C and 16.3°C for 2003 and 2004, respectively. This
integral estimate is biased upward on the basis of a com-
parison of P0 j and the P0 from the nonlinear regression
from four California daily egg production surveys and
a simple theoretical population. Both cases indicated
that the relative bias (rb = ( P0I — P0 )/ P0I) was close
to 20% of P0 j Thus the bias-corrected egg production
(P o,f) would be P0c=P0I (l-rb)=_P0 7(0.8).

The mean density of eggs (Y-) (eggs/0.05 m2) was
estimated for each of two strata (i=l,2), with 125°W
latitude as the dividing line.
The overall mean density (Y)
for the whole survey area was a
weighted average with the area
in each stratum as the weight
and was used to estimate the
daily egg production. No esti-
mate of egg production for each
stratum was obtained because
of the small sample sizes.

To understand the relative
contribution of egg production
from the PNW area, we com-
puted a ratio of the total egg
production in the PNW to the
total egg production in the
whole area (PNW and Califor-
nia) as P0 j Aj/^P0 j Aj , where
P0 ■ is the daily egg production
during the peak spawning time
in the survey area Ay, j= 1 re-
fers to the PNW area in July
and j- 2 refers to California in
April.

Adult reproductive state
and parameters

For all four surveys, we used
histological analysis of all ovar-
ian tissues, along with trawl and
female data, to provide accurate
assessment of adult parameters
and reproductive state such as
maturity, spawning period,
recent spawning activity, post-
spawning condition, or iden-
tification of advanced oocyte
development for a selection of
females for batch fecundity esti-
mation. In the laboratory, each
preserved ovary was blotted and
weighed to the nearest mg. A
piece of each ovary was removed,
a histological slide was pre-
pared, and the tissue sections
were stained with hematoxylin
and eosin. We analyzed oocyte
development, atresia, and post-
ovulatory follicle age to assign

Washington-Oregon

California

-p

0.35
0.30
0 25
0.20
0 15
0.10
0.05
0.00

March 2004

-M-

0.35
0 30
0.25
0.20
0 15
0.10
0.05
0.00

March-April 2004

80 100 120 140 160 180 200 220 240 260 280

Standard length class (mm)

Figure 6

Length frequency distribution of Pacific sardine ( Sardinops sagax) off Washington
and Oregon, and California during 2003, 2004, and 2005 from fishery independent
trawl surveys (black bars) and port sampling of commercial purse seine catches
(gray bars). Catch data were provided by California Dept, of Fish and Game, Oregon
Dept, of Fish and Wildlife, and Washington Dept, of Fish and Wildlife.

Lo et al. Biomass and reproductive status of Sardinops sagax off the Pacific coast.

183

Table 3

Estimated Pacific sardine (Sardinops sagax ) egg densities, egg production ( P0 c, Eq. 5) with coefficient of variation (CV) in paren-
theses, and number of collections with positive collections in parentheses from California Cooperative Oceanic Fisheries Inves-
tigation vertical egg tow net (CalVET) and continuous underway fish egg sampler (CUFES) samples in two strata with dividing
latitude of 125°W and the entire survey area for the July 2003 and July 2004 surveys. Dashes indicate where statistics were not
computed because of small or zero catches.

Stratum 1

Stratum 2

Entire survey area

July 2003
CalVET

Egg density (eggs/0.05 m2)(CV)

0.388(0.51)

0.073(0.51)

P0 c (Egg production /0.05 m2/day)(CV)

—

0.04(0.51)

No. of CalVET tows (positive)

18(4)

36(0)

54(4)

CUFES

Eggs/min (CV)

0.148(0.61)

0.05(0.74)

0.069(0.49)

No of CUFES samples (positive)

316(102)

166(15)

482(117)

Survey area (km2)(%)

40,043(19)

166,334(81)

206,377(100)

July 2004
CalVET

Egg density (eggs/0.05 m2)(CV)

0.0(— )

0.088(0.56)

0.070(0.56)

P0 c (Egg production /0.05 m2/day)(CV)

—

0.037(0.58)

No. of CalVET tows (positive)

14(0)

34(3)

48(3)

CUFES

Eggs/min (CV)

0.11(0.42)

0.097(0.67)

0.1(0.53)

No of CUFES samples (positive)

197(65)

450(64)

647(129)

Survey area (km2)(%)

40,043(21)

150,932(79)

190,975(100)

Table 4

Percentage and average size in each maturity class of Pacific sardine ( Sardinops sagax ) females in the random samples from
trawls conducted during four research surveys in 2003-05 off Oregon and Washington. Maturity was based on histological
analysis of ovaries.

Survey dates
(n females)

Maturity class

Percentage
of females

Mean standard
length (mm)

Mean whole
body weight (g)

6-25 July 2003

Immature

0.7

204

124

(690)

Mature

98.3

238

194

29 Feburary-19 March 2004

Immature

97.2

108

(108)

Mature

2.8

207

105

6-25 July 2004

Immature

62.2

147

(410)

Mature

37.8

240

200

2-21March 2005

Immature

89.2

161

(241)

Mature

10.8

195

female maturity and reproductive state (Macewicz et
al., 1996; Lo et al., 2005).

Sufficient numbers of immature and mature females
in the random 50-fish subsample of a positive trawl
for estimation of the length at which 50% were ma-
ture were collected during July 2004 and March 2005
(Table=4). Females were grouped into 10-mm length
classes and the length at which 50% were mature was
estimated by logistic regression: y=l/(l + e~<a+bL)), where
y = the proportion of mature female sardine and L = the

standard length in mm. The length-specific maturation
relationships were compared to those off California in
April 1994, 2004, and 2005 (Macewicz et al., 1996; Lo
et al., 2005).

Because the spawning season occurs in early summer,
we used the two sets of July survey data to estimate
the following adult reproductive parameters, which were
used in the spawning biomass computation based on
the daily egg production method (Lasker, 1985; Parker,
1985; Lo et al., 2005): the daily spawning fraction (S)

184

Fishery Bulletin 108(2)

or the fraction of mature females spawning per day; the
average batch fecundity (number of eggs per spawning
per mature female: F ); the fraction of mature fish that
were female by weight (sex ratio: R ); and the average
weight of mature females (g) (WJ. The reproductive
parameters were estimated from the data on the first
25 mature females per trawl or all mature females if
there were <25 by following the methods in Macewicz
et al. (1996). Females with ovaries histologically iden-
tified as containing hydrated oocytes (hydrated ovary)
have temporarily inflated ovary weights. For each July
survey, the relation between wet weight (y) and ovary-
free wet weight (x) from mature females lacking hy-
drated oocytes was determined as y=-9.0998 + 1.0758x
in 2003 and y=-6.316 + 1.05608x in 2004. Thus, the
observed female weight was adjusted downward for fe-
males with hydrated ovaries when calculating average
mature female weight (Wf) for each collection by year.
During March of 2004 and 2005, adjustments were not
necessary and fecundity was not estimated from mature
females caught because none of them had ovaries with
oocytes in the migratory-nucleus or hydrated stages.
Mean batch fecundity was estimated by the gravimet-
ric method for 54 females from 21 trawls from the July
surveys. The relationship of batch fecundity to female
weight (without ovary) was then determined.

Reproductive adult parameters were summarized for
each trawl. Population values were estimated by meth-
ods in Picquelle and Stauffer (1985), where estimation
of each adult parameter (S, F, W, R) was based on a
ratio estimator (Picquelle and Stauffer, 1985; Lo et ah,
1996) and used to calculate spawning biomass and its
covariance for the July 2003 and July 2004 surveys.

Spawning biomass

The denominator (. RSF/W* ) is referred to as the daily
specific fecundity (number of eggs/population weight
[g]/day).

The variance of the spawning biomass estimate (Bs)
was computed from the Taylor expansion in terms of
the coefficient of variation (CV) for each parameter
estimate and covariance for adult parameter estimates
(Parker, 1985; Picquelle and Stauffer, 1985; Lo et al.,
1996; 2005):

var(bs)=bs

cv(p0f +cv(w/.)2 +cv(s)2 +
cv (r)2 + CV (f)2 + 2 covs

(8)

The last term, involving the covariance term, on the
right-hand side is

COVS =

i i<j

COV ( Xj , Xj j
xtx-

(9)

where x;=the z'th adult parameter estimate, e.g., xt=F
and Xj=Wf. The sign of any two terms is positive if they
are both in the numerator of Bs or denominator of Bs
(Eq. 7); otherwise, the sign is negative. The covariance
term is

COV( Xj Xj )

[n I (n-l)\^mk{xlk- x^g^Xj k- Xj)
* , (10)

( \

x**

V k ,

\ k J

The DEPM is a well-accepted method used for estimating
spawning biomass for fish with indeterminate fecundity,
i.e. multiple spawners (Hunter and Lo, 1993; Stratouda-
kis et al., 2006) and was used to estimate the spawning
biomass of Pacific sardine in this area in 1994 (Bentley
et al., 1996). The spawning biomass was computed with
the following equation:

where k = kth tow, and k=l, ... , n\
mk and gk = sample sizes; and
x£ k and Xj k = sample means from the kth tow for x; and
Xj respectively.

Results

, PqAC
>s RSF/Wf

(7)

where P0 =

A =
C =

R =

S =

F =

the daily egg production/0.05 m2 at hatch-
ing;

the survey area in units of 0.05 m2;

the conversion factor from grams (g) to

metric tons (t);

the fraction of mature fish that is female, by
weight (sex ratio);

the daily spawning fraction: fraction of

mature females spawning per day;

the average batch fecundity (number of eggs

per spawn per mature female); and

the average weight of mature females (g).

Seasonal biomass

The relative abundance of Pacific sardine was higher
in summer than in the following spring off the PNW.
The bias-corrected seasonal biomass estimates were
198,600 t (CV=0.51) for July 2003, 20,100 t (CV=0.80)
for March 2004, 77,900 t (CV=0.38) for July 2004, and
30,100 t (CV=0.72) for March 2005 over an area close
to 200,000 km2 (Table 1). The inshore stratum 1 made
up 20% of the survey area. Yet, for all years stratum 1
had over 80% of the biomass. The recruit biomasses (fish
<120 mm SL) in spring of 2004 and 2005 were quite dif-
ferent: 20,500 t (CV=0.81) for the 2003 year class and
53 t (CV=0.72) for the 2004 year class, respectively. The
2004 point estimate of the recruit biomass, 20,500 t,
was greater than that of the total biomass of 20,100 t

Lo et al Biomass and reproductive status of Sardinops sagax off the Pacific coast.

185

but this was primarily due to the bias correction based
on the bootstrap simulation and the difference was not
statistically significant.

The relatively large 2003 year class constituted a
major proportion of the total biomass in March 2004,
whereas the 2004 year class constituted a very small
proportion of the fish in 2005 (Fig. 6). Therefore, the
relative abundance of Pacific sardine in the spring of
2004 and 2005 was primarily supported by the strong
year class of 2003.

Spawning habitat

The spawning habitat was located east of 125. 5°W longi-
tude in July 2003 and 2004 (Figs. 2 and 4), and between
43° and 44.5°N latitude in July 2003, and between 42°
and 44.5°N latitude in July 2004. The location of the
spawning center, computed as the weighted latitude and
longitude with the eggs/min (>0.5) as the weight, was
124. 7°W and 43.7°N in 2003 and 125.13°W and 42.9°N
for 2004. Therefore, the spawning habitat shifted south-
westward from 2003 to 2004. Because the eggs from the
CUFES samples were distributed more to the west, the
size of the spawning habitat was 10,716 km2 for 2003 and
14,260 km2 for 2004. The spawning habitat, determined
from CUFES data, crossed the dividing line of 125°W
between two strata based on trawl allocation. For both
July cruises, the range of SST in the spawning habitat
was 13.4-18.5°C with a mean close to 16°C (15.7°C and
16.0°C for 2003 and 2004). Note that the overall mean
SST for July 2003 was 16.2°C (range 9.4-25.3°C) and
the mean temperature was 16.8°C (range 9.7-19.9°C)
in July 2004. The number of positive CUFES samples
was 117 out of 482 in July 2003 and 129 out of 647 in
July 2004. Therefore, the proportion of positive samples
(24% in 2003, 19% in 2004) was similar during these
two years.

Daily egg production

The mean density of eggs was 0.388 eggs/0.05 m2
(CV=0.51) in stratum 1 and no eggs were caught by
CalVET net tows in stratum 2 during the July 2003
survey. The opposite was true for the July 2004 survey:
no eggs were caught in stratum 1 and the mean density
in stratum 2 was 0.088 eggs/0.05 m2 (CV=0.56) (Table
3). The overall mean densities were 0.073 eggs/0.05 m2
(0.51) and 0.07eggs/0.05 m2 (0.49) for 2003 and 2004,
respectively. The bias-corrected estimates of the daily
egg production from the integral method (P0c) (Eqs.
5 and 6) in July were 0.04 eggs produced/0.05 m2/
day(CV=0.51) for 2003 and 0.037 eggs produced/0.05
m2 /day (CV=0.58) for 2004. The mean egg capture
rates from CUFES samples for 2003 and 2004 were
0.069 eggs/min (CV=0.49) and 0.1 eggs/min (CV=0.53)
(Table 3).

The ratio of the total egg production in the PNW to
the total egg production off the U.S. west coast (PNW
and California) was 1.46% and 2.2% for 2003 and 2004
and therefore Pacific sardine off the PNW contrib-

uted approximately to 1.8% of the total egg production
(Table 5).

Adult sardine reproductive parameters
and spawning biomass

During the four surveys, 92 of the 214 trawls (Figs. 2-5,
Table 1) captured adults or subadults. In the random
subsamples from these trawls, 2862 sardine were mea-
sured (Fig. 6); standard length ranged from 99-289 mm
for females, 106-281 mm for males, and 75-146 mm for
individuals of indeterminate sex (where it was difficult
to accurately determine sex without microscopic exami-
nation). Nearly all females were mature in July 2003 and
nearly all were immature in March 2004 (Table 4). Using
logistic regression we computed the standard length at
which 50% were mature as 195.1 mm and 199.8 mm for
July 2004 and March 2005, respectively (Fig. 7).

Mean batch fecundity was estimated for 35 females
caught in July 2003 and 19 from July 2004 (Fig. 8).
Analysis of covariance showed no differences in the
relationship between female weight (without ovary, W0f)
and batch fecundity (Fb) among years (P=0.531). Com-
bining the data from July 2003 and 2004, we found
that the relationship between female weight and batch
fecundity, as determined by simple linear regression,
was Fft = -16755 + 372.1Wo/ with the r2 = 0.47. Because
the intercept did not differ from zero (P=0.165), we
chose the regression without the intercept, which yield-
ed the relationship Fb = 295.83Wo ^ , where W ’<■ ranged
from 111-322 g (Fig. 8). The latter equation was used
to calculate batch fecundity for each mature Pacific
sardine female in the July trawl samples.

The population sex ratio ( R ) for mature fish was
0.534 female (CV=0.04) in July 2003 and 0.568 female
(CV=0.05) in July 2004 (Table 5). The 657 mature fe-
male Pacific sardine analyzed from July 2003 and 196
from July 2004 were considered a random sample of
the population in the area trawled. Population-level
estimates of the other adult reproductive parameters
were as follows: average batch fecundity {F)= 55,986
eggs/spawning event (CV=0.04) in July 2003 and
55,883 eggs/spawning (CV=0.06) in July 2004; daily
spawning fraction (S) = 0.027 (CV=0.31) in 2003 and
S = 0.010 (CV=0.74) in 2004; and mean mature female
fish weight (W^= 194.36 g (CV=0.02) in 2003, and 193.16
g (CV=0.03) in 2004 (Table 5). The daily specific fecun-
dity was calculated as 4.21 and 1.68 eggs/gm/day in
2003 and 2004, respectively (Table 5). The proportion
of active females spawning was 0.05 and 0.025 for July
2003 and 2004, respectively, which meant that the av-
erage female was spawning roughly once every 20 to
40 days. None of the three mature females caught in
March 2004 or the 37 mature females caught in March
2005 had histological evidence of imminent or recent
spawning (hydrating oocytes or postovulatory follicles),
and thus S = 0; hence, spawning biomass was not esti-
mated for either March (Table 5).

The estimated spawning biomass based on biased
corrected egg production from the integral method (P0 c)

186

Fishery Bulletin 108(2)

and the adult reproductive parameters for July 2003
and July 2004 (Eq. 7, Table 5) was 39,184 t (CV=0.57)
and 84,120 t (CV=0.93), respectively, for an area close
to 200,000 km2 from 42°N to 48°N off Oregon and
Washington.

Discussion

Dynamics of biomass

Off the PNW, the seasonal relative abundances of Pacific
sardine based on the swept area method are nonsta-
tionary (i.e, not static): high in summer and low in
spring. Fish residing in the PNW in spring are those

over-wintering, and in the summer the majority of fish
>190 mm SL are likely those migrating from Califor-
nia. The spatial distribution of the Pacific sardine was
similar between summer and spring: high in the inshore
area and low in the offshore area, except during March
2005 when small numbers of sardine were caught in
the northern offshore area (Fig. 5). This distribution is
quite different from that off California where the spatial
distribution varied among years (Lo et al., 2005). The
PNW biomass estimates, high in July and low in March,
together with the differential length distributions are
consistent with the conceptual migration schedule of
Pacific sardine (a migration route that appears to be
similar to that of Pacific hake, Merluccius productus),
namely of movement to the PNW from California before

Table 5

Trawl information, estimated female adult parameters, egg production, and spawning biomass (estimated by the daily egg pro-
duction method (DEPM)) for Pacific sardine ( Sardinops sagax) from July and March surveys conducted from 2003 through 2005
off Washington and Oregon (Pacific Northwest) and from April surveys conducted from 2003 through 2005 off California and
in 1994 off California and Mexico. Either the coefficient of variation (CV) or number of positive trawls is in parentheses. na=not
available.

Pacific Northwest

California

2003

July

2004

March

2004

July

2005

March

1994

April

2003

April

2004

April

2005

April

No. trawls (positive)

48(36)

59(9)

58(27)

49(20)

79(43)

25(17)

19(14)

Ave. surface temperature (°C)
at sardine locations

15.4

10.4

15.6

10.4

14.36

13.59

14.18

Fraction of females by weight

0.534

0.568

0.538

0.618

0.469

Ave. mature female weight
(g) with ovary
(g) without ovary
Average batch fecundity0

Wof

194.36

189.25

55,986

105

102.7

193.16

188.90

55,883

102.5

100.2

82.53

79.33

24,283

166.99

156.29

55,711

65.34

63.11

17,662

Relative batch fecundity (oocytes/g)

288

289

294

334

270

No. mature females analyzed

657

196

583

290

175

No. active mature females

374

327

290

148

Fraction of mature females6
spawning per day (CV)

0.027

(0.31)

0.010

(0.74)

0.074

(0.23)

0.131

(0.17)

0.124

(0.31)

Fraction of active females0

0.050

0.025

0.131

0.155

spawning per day
Daily specific fecundity

RSF

4.21

1.68

11.7

27.04

15.67

Egg production/0.05 m2/day
(CV) (Eq. 5)

0.04d

(0.51)

0.037d

(0.58)

0.193

(0.21)

1.520

(0.18)

0.960

(0.24)

1.916

(0.42)

Survey area (km2)

206,037

190,975

380,175

365,906

320,620

253,620

Spawning biomass (t) (CV)

39,184

(0.57)

84,120

(0.93)

127,102

(0.32)

485,121

(0.36)

281,639

(0.30)

621,657

0.54

Eggs/min from CUFES sample (CV)

0.069

(0.49)

0.1

(0.53)

1.57

(0.27)

0.78

(0.11)

0.62

(0.15)

a Mature females: 1994 estimate was calculated with ^=-10858 + 439.53 W0/- (Macewicz et al., 1996), in 2004 with Fb= 356.46Wo/:(Lo et al., 2005),
in 2005 with F^-6085 + 376.28 W,^, and for Pacific Northwest in 2003 and 2004 with Fh= 295.83 Wa ^
b Mature females included females that were active and those that were postbreeding (incapable of further spawning during the season).
c Active mature females were capable of spawning and had oocytes with yolk or postovulatory follicles less than 60 hours old.
d Calculated by the integral method and corrected for bias (P0 c).

Lo et al. Biomass and reproductive status of Sardinops sagax off the Pacific coast.

187

summer to feed, and a return to the south before
spring to spawn (Clark and Janssen, 1945; Dorn,

1995; Emmett et al., 2005; Smith, 2005).

The U.S. stock biomass of age 1+ Pacific sardine
increased from 1981 to a peak of one million tons
in 2000 and, according to the stock assessment,
began to decline in 2003 (Hill et al., 2007). The
high biomass off the PNW in 2003 was most likely
due to the accumulation of migrant survivors from
1999 through 2002, when the stock assessment
reported that biomasses were high. The PNW sar-
dine biomass, estimated from surface rope-trawl
surveys for salmon off the Columbia River, has
been decreasing since 2003 (R. Emmett, personal
commun.2). This decrease is likely due to 1) the
decline of migratory fish as a result of the de-
creasing biomass since 2003 off California, 2) a
decline in successful spawning off the PNW, or 3)
the continued sardine movement northward into
Canadian waters, or a combination of the three
events.

The July 2003 survey indicated that the major-
ity of fish were large (>190 mm SL), whereas the
July 2004 survey showed the opposite because
most of the small fish were from the strong 2003
year class. The presence of large sardine off Or-
egon in July 2003 and California in March-April
2004 is consistent with the concept of the migra-
tion of large fish from the PNW to California
before spawning. However, the large sardine off
Oregon in July 2004 did not show up off either
California or the PNW during March-April 2005
(Fig. 6). This finding may have been due to a
lower total biomass and a smaller proportion of
large fish off the PNW in July 2004 (Table 1, Fig.

6), or because during the 2005 California survey,
few trawls were taken north of 34°N latitude
where most migrants had resided according to
the 2004 DEPM survey off California, or it could
have been due to a combination of both factors
(Fig. 5).

Although the summer PNW biomass estimates
were different between years, the spring biomass
estimates were stable. March surveys clearly revealed
the relative magnitude of the migratory and the local
PNW stocks during the survey years. The change in
biomass off the PNW among years can be due to mul-
tiple reasons: a change in the biomass of the resident
PNW fish, or a change in the biomass off California, or
a change in the migration pattern due to food availabil-
ity and oceanographic conditions, or both (MacFarlane
et al., 2005). To better understand the dynamics of the
Pacific sardine off the west coast of North America,
spring and summer synoptic surveys from Baja Cali-
fornia, Mexico, to British Columbia, Canada, and from
tagging studies are necessary.

2 Emmett, Robert. 2009. Northwest Fisheries Science Center,
Newport, OR.

Figure 7

Fraction of Pacific sardine ( Sardinops sagax) females that
were sexually mature (y) as a function of standard length (L)
fitted to logistic curves for Oregon and Washington in July
2004 and March 2005, and for California in April of 1994,
2004, and 2005. Symbols represent the actual fraction mature
within 10-mm length classes.

Spawning habitat and daily egg production

The spawning habitats off the PNW in the summer of
2003 and 2004 were similar in size between 42-44. 5°N
and east of 125. 4°W. The spawning area occupied 5-7%
of the survey area, much smaller than that off California
(20-25% of the survey area in 2003-04). The spawning
habitat in the mid-2000s (2003 through 2005) seemed to
contract southward and shoreward compared to the mid-
1990s (1994 through 1998) when it extended to 46°N and
close to 126°W (Emmett et al., 2005). The temperature
range in the offshore spawning habitat in the 1990s
(14-16°C) was similar to that in the 2003-04 inshore
area (13— 18°C); therefore, the change of oceanographic
conditions may have caused the apparent contraction of
spawning habitat between the mid-1990s and mid-2000s
off the PNW. Because no adult samples were taken in
the mid-1990s, we were unable to compare the adult

188

Fishery Bulletin 108(2)

120000-

+ July 2003 +

O July 2004

Number of oocytes

"vl

i i 1 1 i i

+0 o

. 0+ «+ +
+ +0

++ ++

20000 -

I 1 1 1 1 1 1 1 1 1 1 1

100 200 300

Body weight (gm without ovary)

Figure 8

Batch fecundity ( Fb ), the estimated number of hydrated or migra-
tory-nucleus oocytes, of Pacific sardine ( Sardinops sagax) as a
function of ovary-free fish weight (WJ. Pacific sardine females were
collected from trawl samples off Oregon and Washington during July
2003 and 2004. Equation for regression line is F6 = 295.83W0^.

spawning characteristics during these two periods. The
spawning habitats of sardine off the PNW in 2003 and
2004 were similar, whereas the spawning habitats off
California were quite different: concentrated off central
California in 2004 and distributed through the whole
survey area in 2005. Note, no eggs were caught during
the July California Cooperative Oceanic Fisheries Inves-
tigations (CalCOFI) surveys off California in either
CalVET or bongo net tows.

The daily egg production off the PNW was low in
both 2003 and 2004 (0.04 and 0.037eggs produced/0.05
m2/day, respectively), lower than that in 1994 (0.50
eggs produced/0.05 m2/day) (Bentley et al., 1996), and
lower than those off California (1.52 and 0.96 eggs
produced /0.05 m2/day) in 2003, 2004, and other years.
This low egg production in the PNW contributed only
1.8% of the total U.S. west coast egg production in
2003-04. The low PNW egg production estimates could
be the result of the July surveys occurring after the
spawning peak, possibly in June, as the SST was high
(Emmett et al., 2005), or the result of the egg mortal-
ity of Pacific sardine off the PNW being different from
that off California, or both. Future ichthyoplankton
surveys with large sample sizes are needed to obtain
direct estimates of the daily egg production and egg
mortality off the PNW.

The egg production estimates from July 2003 and
2004 were very similar even though the relative
abundances were quite different. With similar egg
production in two years, one might expect that the
biomass of recruits would be similar. However, the
2003 year class was much stronger than that of 2004.
This difference would be most likely due to the more

favorable environmental conditions in 2003 than in
2004.

One interesting question to ask is what effect a
reduction of the spawning habitat or egg production
would have on the PNW Pacific sardine population. The
sustainability of the Pacific sardine population off the
PNW depends greatly on the Pacific sardine population
off California, oceanographic conditions, and food avail-
ability (MacFarlane et al., 2005) because most of the
spawners (>190 mm SL) off the PNW in the summer
are migrants from California. As long as the Pacific
sardine population off California is large enough to al-
low adequate migration to the PNW in the summer to
spawn, the population off the PNW will be sustained.
Of course, if environmental conditions are unfavorable,
the proportion of spawners may be reduced, affecting
both the recruits to the local population and the size
of the population in the following spring. If the popu-
lation off California decreases to the level of collapse,
the population off the PNW may have been diminished
well before the collapse off California. This status of the
PNW population was evident from the history of land-
ings in the last Pacific sardine collapse (Fig. 1). Dur-
ing the waning years of sardine population, the PNW
commercial landings ended in 1949, 16 years before the
California catch ended in 1965. The sardine population
began recovering in the late 1970s-early 1980s off Cali-
fornia. Incidental landings off California began in 1981,
11 years before an incidental catch of sardine off the
PNW in 1992 due to the favorable El Nino conditions,
and 17 years before directed landings in 1998.

Application of proper management strategies to pre-
serve the population off California, and thus the mi-

Lo et al. Biomass and reproductive status of Sardmops sagax off the Pacific coast.

189

grants, is essential because most of the migrants are
mature fish and the leaders of migration imprints: older
fish lead younger fish to migrate. This recent entrain-
ment hypothesis (Petitgas et al., 2006) is a step for-
ward from the theory that fish population life cycles are
controlled only by physical conditions (Sinclair, 1988).
The entrainment hypothesis implies that the older fish
are essential to ensure the sustainability of popula-
tion and fisheries of Pacific sardine off the PNW and
thus along the entire west coast of the North American
continent.

Adult reproductive parameters and spawning biomass

Pacific sardine spawn off the PNW, contrary to beliefs
in the 1930s and 1940s that they spawn only off Cali-
fornia. Although sardine eggs, larvae, and adults have
been caught in surveys off the PNW since 1994 (Bentley
et al., 1996; Emmett et al., 2005), only with the four
surveys during 2003-05 were the reproductive param-
eters for female Pacific sardine off the PNW examined
in detail.

The spawning season of Pacific sardine off the PNW
apparently occurs primarily in the early summer, al-
though a few fish possibly spawn in spring. If July
is the spawning peak off the PNW, then spawning is
less intense than during the peak off California in
April. The daily spawning fraction of mature females
(S=0.027 and 0.01) was much lower than that off Cali-
fornia (0.07-0.17). Previous work has indicated that ac-
tive mature females of Sardinops spp. worldwide spawn
once every eight days (Macewicz et al., 1996). Recent
results off California are similar (once every 6-8 days),
where as active mature Pacific sardine females off the
PNW spawned much less frequently (only once every
20-40 days). In addition, females in July produced
about 288 eggs per gram of female weight (relative
batch fecundity) off the PNW — few eggs than similar
females off California that spawned 334 eggs per gram
of female weight in April 2004 (Table 5). According to
the April 2004 DEPM sardine survey off California,
the large mature females, in particular those >200 mm
SL, were spawning very vigorously (S= 0.131) and these
migratory females may not have recovered sufficient-
ly to spawn at higher rates off the PNW during July
2004, a phenomenon similar to that which occurred
with Pacific sardine off Chile, which were less active
during a second annual spawning period (Tascheri and
Claramunt, 1996). The presence of a high percentage
of inactive mature females off the PNW in July (43%
in 2003 and 59% in 2004) indicates two other possible
explanations for the low level of spawning: July is not
the peak spawning time for sardine off the PNW be-
cause they may be similar to northern anchovy where
ovaries with high levels of atresia (indicating cessa-
tion of reproductive activity) are common at the end
of the spawning season (Hunter and Macewicz, 1985);
or, Pacific sardine in the PNW may behave like chub
mackerel (Scomber japonicus) whose individuals spawn
only for a short period and inactive mature females are

common throughout the spawning season (Dickerson et
al., 1992). If so, it may be necessary in future surveys
to analyze reproductive samples collected over a longer
time to better define the peak spawning period, and to
determine whether the peak spawning fraction is simi-
lar to the rate off California (about 0.13 spawning per
day) or whether it remains low (<0.03).

Few mature Pacific sardine females were caught off
the PNW during March and it seems that they may
have followed warmer water south. The majority of the
40 mature females were inactive (postbreeding or rest-
ing) and none had spawned. It was surprising that we
caught 12 females of 202-260 mm SL that were active
(their ovaries contained some oocytes with yolk) and
were potentially capable of spawning in the near future
(3-30 days). We examined the locations where females
were caught and their associated water temperatures.
The average SST of trawls during March was 10.4°C.
During March 2004, the three mature females (one
active) were caught farthest south (42°N) in 11.1°C
water. One inactive mature female was caught near
Astoria, OR, in 11.1°C water during March 2005, and
the other 36 (11 active) mature females were caught
inshore, south of 44.5°N in 11.5°C (10.7-12.5°C) wa-
ter. Immature female Pacific sardine were generally
found north of 44.5°N in cooler water; on average 10.2°C
(9.6-10.7°C) in March 2004 and 10.3°C (9.0-12. 5°C) in
March 2005. Thus, in the winter, the older fish were
able to move south following the warmer water, while
the younger fish, due to a lack of stored energy for long
distance swimming, remained in the cold water. Over-
wintering immature females seem to tolerate water as
cold as 9.0°C. The PNW generally has warmer coastal
SSTs in the winter (from downwelling) than in summer.
However temperatures in the estuaries can be very cold
and die offs of age-0 sardine in the Columbia River and
other estuaries have been observed during the winter
(E. Dorval, personal commun.3).

Female Pacific sardine in the PNW mature at lengths
greater than those off California. Fifty percent of the
females caught off the PNW matured at around 195
mm and > 90% off California were mature at the size
of the smallest mature PNW female (182 mm). A major-
ity of sardine > 200 mm off the PNW migrate during
fall-winter (Clark and Janssen, 1945; Nottestad et al.,
1999). During the April 2004 DEPM survey, Pacific
sardine were collected off central California between
34.8°N and 37.3°N and a majority were the large, mi-
gratory size (those >200 mm), whereas in 2005, the
majority of positive adult samples were collected in the
inshore area of Southern California between 32°N and
36°N and most sardines were <200 mm. The length of
females at 50% maturity off the PNW was similar to
the length estimate (193 mm) in April 2004 off Cali-
fornia which indicated that the large Pacific sardines
off central California likely were winter migratory fish.
This conclusion is consistent with the historical tagging

3 Dorval, Emanis. 2008. Librairie La Lumiere, Rue Baussan,
# 34, Turgeau, Port-au-Prince, Haiti, W.I.

190

Fishery Bulletin 108(2)

results, which indicated that the majority of the tags
released off the PNW were recovered off central Cali-
fornia (Clark and Janssen, 1945).

The point estimates of spawning biomass of Pacific
sardine off the PNW differed, but were not statistically
different because of a large coefficient of variation:
39,184 t and 84,120 t for July 2003 and 2004, respec-
tively. They were close to 50,000 t in 1994 (Bentley et
al., 1996). Theoretically, the spawning biomass should
constitute a good proportion of the total biomass, which
was not so for July 2003. This could be due to an under-
estimate of P0, to an overestimate of the spawning frac-
tion, or both. The overestimate of the spawning fraction
could be due to the movement of the postspawners out of
the spawning area. A DEPM study is needed to evaluate
such effects and model the effects of fish movement on
estimates of spawning rate. The effect of the timing of
the survey in relation to spawning and movement cycles
needs to be studied with new data and modeling.

The difference between the spawning biomass es-
timates in 2003 and 2004 was primarily due to the
difference in the estimated spawning fractions (0.027
in contrast to 0.01), because the estimates of daily egg
production (P0) were similar. The large coefficients of
variation of spawning biomass estimates were mainly a
result of the uncertainty in estimates of P0 and the dai-
ly spawning fraction (S) in July 2004. For low values of
P0 and S, the number of samples has to be substantially
increased to obtain a more precise estimate (Picquelle
and Stauffer, 1985). Estimated spawning biomass for off
the PNW in July was much smaller than estimates for
off California during April in recent years. The smaller
fish length at 50% maturity off California means that
the more numerous smaller resident Pacific sardine are
able to participate in local spawning at the same time
as the larger migratory sardine.

Future work

The Pacific sardine spawning habitat and season in the
PNW are loosely defined in this study and the magnitude
and scope of the coastal migration are not fully explored.
To better characterize these, we need to conduct syn-
optic trawl-ichthyoplankon-acoustic surveys from Baja
California, Mexico, to British Columbia, Canada, during
spring and early summer at three to five year intervals.
To better characterize the spawning habitats in this
area, we need to obtain physical and biological oceano-
graphic data (Lynn, 2003; Emmett et al., 2005; Reiss et
al., 2008) and demographic data of Pacific sardine over
a broader geographic range because the Pacific sardine
is a migratory species.

For trawl swept-area-based biomass estimates, the
efficiency of the trawl needs to be calibrated. Biomass
estimates from acoustic surveys would be another fish-
ery-independent source of relative abundance. Because
the coefficients of variation of all estimates are large,
the number of trawls needs to be increased or other
statistical estimation procedures should be explored, or

both, to improve the precision of estimates. To obtain
a representative length distribution of the population,
fishery-independent surveys covering the whole west
coast area are essential, and length data from com-
mercial vessels should be used with caution for both the
PNW and California. For spawning biomass, we need to
understand the maturation schedules of females and the
spawning season off Oregon and Washington. Numerous
plankton net tows are needed to obtain direct estimates
of the daily egg production and egg mortality rates in
early summer. Currently, only the spawning biomass
of Pacific sardine off California is estimated from the
annual April DEPM survey. Because mature females
were caught during two March surveys off the PNW,
efforts should be made to obtain trawl data off the
PNW in April. Data for mature females collected off the
PNW could then be combined with the April data set
off California to estimate reproductive parameters and
the spawning biomass of Pacific sardine off the whole
west coast of the United States. To better understand
the relationship between the sardine populations off
California and the PNW, we need to examine migration
characteristics (i.e, migration range, pattern and sched-
ule) and the effect of fishing pressure on the migratory
fish because most of these fish are mature and leaders
of migration imprints. We need a long time series of
abundance for all regions together, along with ocean-
ographic and biological data, to enhance our under-
standing of the dynamics of the entire Pacific sardine
population to provide information for the development
of future strategies to sustain the population.

Acknowledgments

We thank two anonymous reviewers for their construc-
tive comments. We thank the captain and crew members
of the FV Frosti and the support for the charter provided
by NMFS Cooperative Research Program. These surveys
would not have been possible without the cooperation
of the Northwest Fisheries Science Center, NOAA, the
Washington Department of Fish and Wildlife (WDFW),
Oregon Department of Fish and Wildlife (ODFW),
and the Pacific Fishery Management Council (PFMC).
We thank all those who participated in the surveys:

D. Waldeck (PFMC), Todd Miller (Oregon State Univer-
sity), J. McCrae (ODFW), A. Thurman (WDFW), and E.
Acuna and N. Bowlin of Southwest Fisheries Science
Center. We thank J. Hunter, W. Watson, S. Picquelle, E.
Dorval, K. Hill, A. Takasuka, S. McClatchie, A. MacCall,

E. Weber, and R. Emmett for reviewing the manuscript
and R. Sanford for organizing the manuscript.

Literature cited

Ahlstrom, E. H.

1948. A record of pilchard eggs and larvae collected
during surveys made in 1939 to 1941. U.S. Fish Wildl.
Serv., Spec. Sci. Rep. no. 54, 76 p.

Lo et al. Biomass and reproductive status of Sardmops sagax off the Pacific coast.

191

Bentley, J. R., R. L. Emmett, N. C. H. Lo, and H. G. Moser.

1996. Egg production of the Pacific sardine ( Sardinops
sagax ) off Oregon in 1994. Calif. Coop. Oceanic Fish.
Invest. Rep. 37:193—200.

Checkley, D. M., Jr., P. B. Ortner, L. R. Settle, and S. R. Cummings.

1997. A continuous, underway fish egg sampler. Fish.
Oceanogr. 6(21:58-73.

Clark, F. N., and J. F. Janssen Jr.

1945. Movements and abundance of the sardine as mea-
sured by tag returns. Fish Bull. Calif. Dep. Fish Game
61:7-42.

Deriso, R. B., J. T. Barnes, L. D. Jacobson, and P. R. Arenas.
1996. Catch-at-age analysis for Pacific sardine ( Sardinops
sagax), 1983-1995. Calif. Coop. Oceanic Fish. Invest.
Rep. 37:175-187.

Dickerson, T. L., B. J. Macewicz, and J. R. Hunter.

1992. Spawning frequency and batch fecundity of chub
mackerel, Scomber japonicus, during 1985. Calif. Coop.
Oceanic Fish. Invest. Rep. 33:130-140.

Dorn, M. W.

1995. The effects of age composition and oceanographic
conditions on the annual migration of Pacific whiting,
Merluccius produetus. Calif. Coop. Oceanic Fish. Invest.
Rep. 36:97-105.

Emmett, R. L., R. D. Brodeur, T. W. Miller, S. S. Pool, P. J.
Bentley, G. K. Krutzikowsky, and J. McCrae.

2005. Pacific sardine ( Sardinops sagax) abundance, dis-
tribution, and ecological relationships in the Pacific
northwest. Calif. Coop. Oceanic Fish. Invest. Rep.
46:122-143.

Felin, F. E.

1954. Population heterogenity in Pacific pilchard. Fish.
Bull. 54:201-225.

Hart, J. L.

1943. The pilchard, Sardinops caerulea (Girard) on
Canadian fishing grounds with special reference to
an unusual abundance of young fish. Trans. R. Soc.
Can., ser. 3, 37(51:55-73.

Hill, K. T., E. Dorval, N. C. H. Lo, B. J. Macewicz, C. Show, and
R. Felix-Uraga.

2007. Assessment of Pacific sardine resource in 2007
for U.S. management in 2008. NOAA Tech. Memo.
NMFS-SWFSC-413, 157 p.

Holt, D., and T.M. F. Smith.

1979. Post stratification. J. R. Stat. Soc. A. 142:33-
46.

Hunter, J. R., and N. C. H. Lo.

1993. Icthyoplankton methods for estimating fish bio-
mass introduction and terminology. Bull. Mar. Sci.
53(21:723-727.

Hunter, J. R., and B. J. Macewicz.

1985. Rates of atresia in the ovary of captive and wild
northern anchovy, Engraulis mordax. Fish. Bull.
83:115-136.

Janssen, J. F., Jr.

1938. Second report of sardine tagging in California. Fish
Bull. Calif. Dep. Fish Game 24:376-389.

Lasker, R., ed.

1985. An egg production method for estimating spawn-
ing biomass of pelagic fish: application to the northern
anchovy, Engraulis mordax. NOAA Tech. Rep. NMFS
36, 99 p.

Lo, N. C. H., Y. A. Green Ruiz, M. J. Cervantes, H. G. Moser, and
R. J. Lynn.

1996. Egg production and spawning biomass of Pacific
sardine ( Sardinops sagax), in 1994, determined by the

daily egg production method. Calif. Coop. Oceanic Fish.
Invest. Rep. 37:160-174.

Lo, N. C. H., B. J. Macewicz, and D. A. Griffith.

2005. Spawning biomass of Pacific sardine ( Sardinops
sagax), from 1994-2004 off California. Calif. Coop.
Oceanic Fish. Invest. Rep. 46:93-112.

Lynn, R. J.

2003. Variability in the spawning habitat of Pacific
sardine (Sardinops sagax) off southern and central
California. Fish. Oceanogr. 12(31:1-13.

MacCall, A. D.

1976. Density dependence of catchability coefficient in
the California Pacific sardine, Sardinops sagax caerulea,
purse seine fishery. Calif. Coop. Oceanic Fish. Invest.
Rep. 18:136-148.

Macewicz, B. J., J. J. Castro-Gonzalez, C. E. Cotero Altamirano,
and J. R. Hunter.

1996. Adult reproductive parameters of Pacific sardine
(Sardinops sagax) during 1994. Calif. Coop. Oceanic
Fish. Invest. Rep. 37:140-151.

Marr, J. C.

1950. Apparent abundance of the pilchard ( Sardinops
caerulea) off Oregon and Washington, 1935-43, as mea-
sured by the catch per boat. Fish. Bull. 51:385-394.

1960. The causes of major variations in the catch of the
Pacific sardine, Sardinops caerulea (Girard). In Pro-
ceedings of the world scientific meeting on the biology
of sardine and related species, vol. 3 (Rosa, H., and G.
I. Murphy, eds.), p. 667-791. FAO, Rome.

McFarlane, G. A., J. Schweigert, L. MacDougall, and C. Hrabok.

2005. Distribution and biology of Pacific sardine
(Sardinops sagax 1 off British Columbia, Canada. Calif.
Coop. Oceanic Fish. Invest. Rep. 46:144-160.

Mosher, K. H. and H. H. Eckles.

1954. Age determination of Pacific sardine from
otoliths. U.S. Dept. Int., Fish Wild. Serv. Res. Rep.
37, 40 p.

Murphy, G. L.

1966. Population biology of the Pacific sardine (Sardinops
caerulea). Proc. Calif. Acad. Sci. series 4, 34(11:1-84.

Nottestad, L., J. Giske, J. C. Hoist, and G. Huse.

1999. A length-based hypothesis for feeding migrations
in pelagic fish. Can. J. Fish. Aquat. Sci. 56 (suppl.
11:26-34.

Parker, K.

1985. Biomass model for egg production method. In An
egg production method for estimating spawning biomass
of pelagic fish: application to the northern anchovy,
Engraulis mordax (R. Lasker, ed.), p. 5-6. NOAA Tech.
Rep. NMFS 36.

Petitgas, P., D. Reid , B. Planque, E. Nogueira, B. O’Hea, and
U. Cotano.

2006. The entrainment hypothesis: an explanation for
the persistence and innovation in spawning migrations
and life cycle spatial patterns. ICESCM:2006/B:07, 9
p. [Available at http://www.ices.dk/products/cmdoc-
sindex.asp]

Picquelle, S., and G. Stauffer.

1985. Parameter estimation for an egg production method
of anchovy biomass assessment. In An egg production
method for estimating spawning biomass of pelagic fish:
application to the northern anchovy, Engraulis mordax
(R. Lasker, ed.1, p. 7-16. NOAA Tech. Rep. NMFS 36.

Preti, A., S. E. Smith, and D. A. Ramon.

2001. Feeding habitats of the common thresher shark
( Alopias vulpinus) sampled from the California-based

192

Fishery Bulletin 108(2)

drift gill net fishery, 1998-1999. Calif. Coop. Oceanic
Fish. Invest. Rep. 42:145-152.

2004. Diet differences in the thresher shark (Alopias
vulpinus) during transition from a warm-water regime
to a cool-water regime off California-Oregon, 1998-
2000. Calif. Coop. Oceanic Fish. Invest. Rep. 45:118-125.

Reiss, C. S., D. M. Checkley Jr., and S. J. Bograd.

2008. Remotely sensed spawning habitat of Pacific sar-
dine ( Sardinops sagax) and Northern anchovy ( Engru -
aulis mordax) within the California Current. Fish.
Oceanogr. 17:126-136

Sinclair, M.

1988. Marine populations: an essay on population regu-
lation and speciation, 252 p. Univ. Washington Press,
Seattle, WA.

Smith, P. E.

2005. A history of proposals for subpopulation struc-
ture in Pacific sardine ( Sardinops sagax ) population
off western North America. Calif. Coop. Oceanic Fish.
Invest. Rep. 46:75-82.

Smith, P. E., W. Flerx, and R. P. Hewitt.

1985. The CalCOFI vertical egg tow (CalVET) net. In An
egg production method for estimating spawning biomass
of pelagic fish: application to the northern anchovy,
Engraulis mordax, (R. Lasker, ed.), p. 27-32. NOAA
Tech. Rep. NMFS 36.

Stratoudakis, Y., M. Bernal, K. Ganias, and A. Uriarte.

2006. The daily egg production method: recent advances,
current applications and future challenges. Fish Fish.
7:35-57.

Tascheri, R. and G. Claramunt.

1996. Aproximacion a los cambios intra-annuales en el
contenido de energia del ovario de sardina ( Sardinops
sagax Jenyns, 1842) en el norte de Chile. Invest. Mar.
24:51-66. [In Spanish.]

Wolf, P.

1992. Recovery of the Pacific sardine and the Califor-
nia sardine fishery. Calif. Coop. Oceanic Fish. Invest.
Rep., vol. 33:76-86.

193

Seasonal variability

in ichthyoplankton abundance

and assemblage composition

in the northern Gulf of Mexico off Alabama

Email address for contact author: fhernandez@disl.org

1 Dauphin Island Sea Laboratory
101 Bienville Boulevard
Dauphin Island, Alabama 36528

2 Department of Marine Sciences
University of South Alabama

307 University Boulevard, LSCB Rm 25
Mobile, Alabama 36688

Abstract — Multiyear ichthyoplankton
surveys used to monitor larval fish
seasonality, abundance, and assem-
blage structure can provide early indi-
cators of regional ecosystem changes.
Numerous ichthyoplankton surveys
have been conducted in the north-
ern Gulf of Mexico, but few have had
high levels of temporal resolution and
sample replication. In this study, ich-
thyoplankton samples were collected
monthly (October 2004-October 2006)
at a single station off the coast of
Alabama as part of a long-term bio-
logical survey. Four seasonal periods
were identified from observed and
historic water temperatures, includ-
ing a relatively long ( June-October)
“summer” period (water tempera-
ture >26°C). Fish egg abundance,
total larval abundance, and larval
taxonomic diversity were significantly
related to water temperature (but not
salinity), with peaks in the spring,
spring-summer, and summer periods,
respectively. Larvae collected during
the survey represented 58 different
families, of which engraulids, sciae-
nids, carangids, and clupeids were
the most prominent. The most abun-
dant taxa collected were unidenti-
fied engraulids (50%), sand seatrout
(Cynoscion arenarius, 7.5%), Atlantic
bumper ( Chloroscombrus chrysurus,
5.4%), Atlantic croaker ( Micropogo -
nias undulatus, 4.4%), Gulf menha-
den ( Brevoortia patronus, 3.8%), and
unidentified gobiids (3.6%). Larval
concentrations for dominant taxa were
highly variable between years, but
the timing of seasonal occurrence for
these taxa was relatively consistent.
Documented increases in sea surface
temperature on the Alabama shelf
may have various implications for
larval fish dynamics, as indicated by
the presence of tropical larval forms
(e.g., fistularids, labrids, scarids, and
acanthurids) in our ichthyoplankton
collections and in recent juvenile sur-
veys of Alabama and northern Gulf
of Mexico seagrass habitats.

Manuscript submitted 8 October 2009.
Manuscript accepted 8 January 2010.
Fish. Bull. 108:193-207 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National
Marine Fisheries Service, NOAA.

Frank J. Hernandez Jr. (contact author)’
Sean P. Powers' 2
William M. Graham1

Ichthyoplankton surveys provide fish-
eries-independent information that is
inherently “ecosystem-based”; entire
larval fish assemblages are collected
(i.e., early stages of both exploited and
unexploited finfish species) along with
zooplankton predators and prey, and
often with a suite of environmental
observations (e.g., salinity, tempera-
ture). At the ecosystem level, infor-
mation on larval assemblages can be
used to detect changes in marine fish
community composition and abun-
dances over time (Sherman et ah,
1984). Previous studies have indicated
that larval assemblages are the result
of convergent spawning strategies by
multiple species taking advantage of
favorable environmental conditions for
larval fish survival (Doyle et ah, 1993;
Sherman et al., 1984). The composi-
tion of larval fish assemblages varies
spatially and temporally because of
the behaviors of the larvae (Gray and
Miskiewicz, 2000; Hare and Govoni,
2005) and the spawning adults (Sher-
man et al., 1984; Hernandez-Miranda
et al., 2003), as well as oceanographic
transport and mixing processes (Auth,
2008; Muhling et al., 2008). Variabil-
ity in any of these factors, therefore,
may result in a different structure
of larval fish assemblages. Because
larval fish survival is closely tied with
primary and secondary productivity

in coastal oceans, changes in larval
fish assemblage structure (over larger
time scales) can be an early indica-
tor of climate-related environmental
shifts (Auth, 2008; Brodeur et al.,
2008).

Despite the importance of the re-
gion to fisheries, seasonal variabil-
ity in larval fish assemblages in the
northern Gulf of Mexico has been
examined in relatively few studies.
Much of the previous ichthyoplankton
research has focused on estuarine as-
semblages (Raynie and Shaw, 1994;
Tolan et al., 1997) or on relatively
short-term interactions between as-
semblages and specific oceanograph-
ic features, such as the Mississippi
River plume (Sogard et al., 1987; Go-
voni et al., 1989) or the Loop Current
(Richards et al., 1993). Other studies
have used ichthyoplankton survey
data from the National Marine Fish-
eries Service’s (NMFS’s) gulf-wide
Southeast Monitoring and Assess-
ment Program (SEAMAP), but these
studies are typically focused on a sin-
gle species (Scott et al., 1993; Lycz-
kowski-Shultz and Ingram, 2003; Ly-
czkowski-Shultz and Hanisko, 2007).
Ditty et al. (1988) summarized the
available ichthyoplankton literature
at the time to provide information on
larval fish seasonality for the entire
northern Gulf of Mexico, and more

194

Fishery Bulletin 108(2)

Figure I

Location of the sampling station used during the October 2004-October 2006 ichthyo-
plankton monitoring survey (star symbol) and the NOAA National Data Buoy Center
oceanographic data buoy (NDBC 42007) used to determine the 10-year (1993-2003)
mean monthly water temperature estimates for the region (diamond symbol).

recently, Lyczkowski-Shultz et al.1 reported on larval
fish seasonality and distribution for the northeastern
Gulf of Mexico.

Although these latter studies provided information
on multiple species, no analyses of larval fish assem-
blages and environmental variability were presented.
Here we report on the seasonality and concentrations
of larval fishes in relation to water temperature based
on data collected during an intensive two year (October
2004-October 2006) ichthyoplankton survey conducted
off the coast of Alabama. The objectives of this study
were 1) to examine the seasonal variability in ichthyo-
plankton diversity and taxon-specific abundances off
the coast of Alabama; and 2) to examine variability
in the relationship between larval fish assemblages
and seasonal changes in water temperature. These
objectives would contribute to our overall goal of un-
derstanding the oceanographic factors that maintain
larval fish assemblages.

1 Lyczkowski-Shultz, J., D. S. Hanisko, K. J. Sulak, and G.
D. Dennis III. 2004. Characterization of ichthyoplankton

within the U.S. Geological Survey’s northeastern Gulf of
Mexico study area — based on analysis of Southeast Area
Monitoring and Assessment Program (SEAMAP) sampling
surveys, 1982-1999, 136 p. NEGOM Ichthyoplankton Synop-
sis Final Report, U.S. Dep. Interior, U.S. Geological Survey,
USGS SIR-2004-5059.

Materials and methods

Data collection

The sampling station was located on the inner continen-
tal shelf of the northern Gulf of Mexico, approximately
18 km south of Dauphin Island, Alabama, at a water
depth of approximately 20 m (Fig. 1). Ichthyoplank-
ton sampling was conducted during monthly day-time
surveys (n=26) and quarterly diel surveys (n- 8) from
October 2004 to October 2006 (Table 1). All samples
were collected with a Bedford Institute of Oceanography
Net Environmental Sampling System (BIONESS) (Open
Seas Instrumentation, Inc., Musquodoboit Harbour,
Nova Scotia, Canada), with a 0.25-m2 mouth opening
fitted with seven (during quarterly surveys) or eight
(during monthly surveys) plankton nets. During monthly
surveys, six depth-discrete samples (18-15 m, 15-12
m, 12-9 m, 9-6 m, 6-3 m, and 3-1 m) and one oblique
sample (18-1 m) were collected during eight replicate
tows at the study site with 202-pm mesh nets. An addi-
tional oblique sample was collected during each tow with
a 333-pm mesh net for a nominal total of 64 samples per
monthly cruise. All eight replicate tows were collected
during daylight hours, generally during a single day.
During the quarterly surveys, a set of six depth-discrete
samples (same depth bins as monthly survey) and one

Hernandez et al.: Variability in ichthyoplankton abundance and composition in the northern Gulf of Mexico

195

Table 1

Station data for ichthyoplankton samples collected during a larval fish monitoring survey at a site located approximately 18 km
south of Dauphin Island, Alabama (October 2004-October 2006). Seasonal classification is based on historic (10-year average)
and observed monthly mean temperatures for the region (see Fig. 2).

Year

Cruise date

Survey type

Seasonal classification

Number of samples

2004

22 Oct

monthly

Summer

2004

16-17 Nov

diel

Fall

2004

29 Nov

monthly

Fall

2004

08 Dec

monthly

Fall

2005

06-07 Jan

monthly

Winter

2005

18-21 Jan

diel

Winter

2005

16 Feb

monthly

Winter

2005

29 Mar

monthly

Spring

2005

05 Apr

monthly

Spring

2005

19 Apr

monthly

Spring

2005

09-13 May

diel

Spring

2005

17 May

monthly

Spring

2005

09 Jun

monthly

Summer

2005

13 Jul

monthly

Summer

2005

09 Aug

monthly

Summer

2005

14 Sep

monthly

Summer

2005

27-29 Sep

diel

Summer

2005

11 Oct

monthly

Summer

2005

09 Nov

monthly

Fall

2005

29 Nov-02 Dec

diel

Winter

2005

16 Dec

monthly

Winter

2006

12 Jan

monthly

Winter

2006

07-10 Feb

diel

Winter

2006

17 Feb

monthly

Winter

2006

16 Mar

monthly

Spring

2006

12-13 Apr

monthly

Spring

2006

01-04 May

diel

Spring

2006

17 May

monthly

Spring

2006

15 Jun

monthly

Summer

2006

05 Jul

monthly

Summer

2006

10 Aug

monthly

Summer

2006

08 Sep

monthly

Summer

2006

19-22 Sep

diel

Summer

2006

12 Oct

monthly

Summer

oblique sample were collected with 202-pm mesh nets
at dawn, noon, dusk, and midnight (local time) over
the course of three diel periods for a nominal total of
84 samples per quarterly cruise. Contents of nets were
rinsed with seawater, sieved, and preserved in 4% forma-
lin for 48 hours before being transferred to 70% ethanol.
A conductivity-temperature-depth probe (CTD) (SBE19,
Sea-Bird Electronics, Inc., Bellevue, WA) was integrated
into the BIONESS system and provided temperature,
salinity, and depth profiles for each plankton tow. A flow-
meter (General Oceanics, Miami, FL) mounted within
the BIONESS frame estimated the volume of water
filtered for each sample. Filtered volume estimates for
each sample were compared with measurements from a
second, externally mounted flowmeter to estimate filtra-

tion efficiency. In all, 1634 ichthyoplankton samples were
processed and used in subsequent analyses. Although all
fish larvae were collected from a single station, Alabama
has a relatively short coastline (<85 km), thus the larval
fishes collected likely represent the ichthyofauna of the
entire Alabama inner shelf region.

Preparation of environmental data

CTD data were processed using the manufacturer’s
software (SEASOFT, Seabird Electronics, Inc., Bellevue,
WA) and averaged into 0.5-m bins. Seasonal patterns
in water temperature were examined using depth-inte-
grated monthly mean temperatures recorded during each
sampling month. For historic comparisons, the 10-year

196

Fishery Bulletin 108(2)

average for water temperature was calculated for each
month with data from a coastal observing buoy (NOAA
National Data Buoy Center Station 42007) located
approximately 54 km west of our sampling station at a
water depth of approximately 15 m (Fig. 1). Although the
temperature values from the buoy were measured near
the surface (0.6-m depth), these observations serve as
good indicators of seasonal shifts in water-column ther-
mal structure, as indicated by our own CTD comparisons
of sea surface temperature and depth-integrated tem-
perature (correlation coefficient, r2=0.98; slope, m=0.90;
P<0.0001). Together, these data were used to define
ecologically relevant “seasons” (rather than calendar
date) for multivariate analyses.

Preparation of ichthyoplankton data

Ichthyoplankton samples were sorted and larval fish
were identified to the lowest possible taxonomic level
at the Plankton Sorting and Identification Center (Szc-
zecin, Poland) and at the Dauphin Island Sea Laboratory
(Dauphin Island, Alabama). Many larval fishes were not
identified to the species level, owing to the relatively
small sizes of larvae collected in the 202-pm mesh nets
and the overall diversity of larval forms present in the
western central Atlantic region, which includes the
Gulf of Mexico (Marancik et ah, 2005). Most identifica-
tions were at the family level (52%), followed by species
(22%), order (14%), and genus (7%) level identifications.
Five percent of the larvae collected were damaged or
unidentified.

Unidentified clupeiforms (engraulids and clupeids)
were excluded from further analyses because their ex-
treme concentrations and taxonomic ambiguity can
often mask abundance and assemblage trends (Tolan et
al., 1997; Hernandez et al., 2003). Order-level taxa and
unidentified larvae were removed from consideration for
similar reasons. Further taxonomic analyses, therefore,
were limited to taxa that represented at least 1% of the
total catch during any individual sampling event, where
the proportion of the total catch for each taxonomic
group was determined after removing unidentified lar-
vae, order-level larvae, and all unidentified clupeiforms.
Following Marancik et al. (2005), we further modified
the data sets to exclude genus-level groupings in in-
stances where many congeners could potentially mask
any seasonal trends. The following genus-level group-
ings were retained because each represented relatively
few congeners with likely similar early life histories in
the northern Gulf of Mexico: Auxis spp. (A. rochei and
A. thazard), Centropristis spp. (C. philadelphica, C.
ocyurus, and C. striata), Diplectrum spp. ( D . bivattatum
and D. formosum), Microdesmus spp. (M. lanceolatus
and M. longipinnis), and Paralichthys spp. (P. albigutta,
P. lethostigma, and P. squamilentus) . Similarly, all fam-
ily-level groups were removed except Gerreidae (most
likely Eucinostomus gula or E. argentus) and Labridae
(most likely Xyrichtys novacula ). In all, 30 taxa were
considered for analyses (Table 2). Because the objective
of this study was to examine the seasonal variability of

larval fish occurrence and relative larval fish concentra-
tions and not size-selectivity or vertical distribution,
our analyses included ichthyoplankton data collected
from all surveys (monthly and quarterly diel), mesh
sizes (202 pm and 333 pm), and depth bins. Depth
stratification and gear selectivity will be addressed in
separate analyses in forthcoming publications.

Analyses

All fish egg and larval fish abundances were standard-
ized by the volume filtered to determine concentration
estimates (no./m3). Taxonomic diversity was calculated
for each sample by taking the exponential of Shannon
entropy, exp (H), following the method of Jost (2006).
Monthly mean observations of total fish eggs, total fish
larvae, and taxonomic diversity were compared to mean
temperature and salinity data by using least squares
regressions. Two approaches were used to examine
larval fish seasonality. First, monthly mean concentra-
tions (no./lOO m3) were calculated for the dominant
taxa to examine monthly trends in abundance. Second,
observed and historic water temperature observations
were used to define distinct seasons for the sampling
region. Seasonality in fish egg concentrations, total
larval fish concentrations, and taxonomic diversity was
examined (after log+1 transformation) by using one-way
ANOVAs with season as a factor and Tukey’s honesty
significant difference (HSD) tests. Lastly, larval con-
centrations for dominant taxa were square-root trans-
formed and analyzed by using Bray Curtis similarity
and cluster analysis with the PRIMER statistical pack-
age (PRIMER, vers. 6, Plymouth Marine Laboratory,
Plymouth, U.K.).

Results

Mean monthly water temperature varied seasonally
over the two year period, with a low of 16.5°C (January
2005) and a high of 30.2°C (August 2006) (Fig. 2). The
general pattern of our monthly temperature observations
was similar (±2°C) to that of recent historical values
(Fig. 3). Notable deviations were relatively cooler tem-
perature observations in May during our study (mean
differences of 3.2°C and 2.4°C during 2005 and 2006,
respectively) and warmer temperatures in October (mean
differences of 2.6°C and 3.0°C during 2005 and 2006,
respectively) and December (mean difference of 3.0°C
in 2004). Even with these disparities, both data sets
were in agreement to define seasonal breaks in water
temperature. (Fig. 3). Sampling periods with mean water
temperature values <18°C were classified as winter, and
those with mean water temperatures above 26°C were
classified as summer. The transitional periods of spring
and fall had mean water temperatures between 18°C
and 26°C. In general, the observed seasonal pattern
comprised three-month winter (December-February)
and spring (March-May) seasons, a relatively long five-
month summer period ( July-October), and a relatively

Hernandez et at: Variability in ichthyoplankton abundance and composition in the northern Gulf of Mexico

197

Table 2

Seasonal (X) and peak (*) occurrence of the dominant larval fish taxa collected in plankton samples (n=1634) off the coast of
Alabama from October 2004 to October 2006. Seasonal classification is based on historic (10-year average) and observed monthly
mean temperatures for the region, (see Fig. 2).

Family

Taxon

Winter

Season

Spring Summer

Fall

Elopidae

El ops saurus

Ophichthidae

Myrophis punctatus

Clupeidae

Brevoortia patronus

Etrumeus teres

Harengula jaguana

Opisthonema oglinum

Serranidae

Centropristis spp.

Diplectrum spp.

Serraniculus pumilio

Carangidae

Chloroscombrus chrysurus

Decapterus punctatus

Lutjanidae

Lutjanus campechanus

Gerreidae

Unidentified

Sciaenidae

Cynoscion arenarius

Cynoscion nothus

Larimus fasciatus

Leiostomus xanthurus

Micropogonias undulatus

Sciaenops ocellatus

Labridae

Unidentified

Microdesmidae

Microdesmus spp.

Scombridae

Auxis spp.

Euthynnus alletteratus

Scomberomorus maculatus

Stromateidae

Peprilus alepidotus

Peprilus burti

Paralichthyidae

Citharichthys spilopterus

Etropus crossotus

Paralichthys spp.

Syacium papillosum

short one-month fall period (November). In one instance,
the interannual variability in water temperature at
our sampling site allowed for the same month to be
designated as a different season during different years
(December was classified as “fall” in 2004 and “winter”
in 2005) (Table 1).

No seasonal pattern in salinity was observed at the
sampling station (Fig. 3). Salinity observations were
generally lower and more variable during the first year
of the study, with values fluctuating between 30.4 and
34.6 between October 2004 and September 2005. Sa-
linity was generally higher and less variable between
October 2005 and October 2006, with values ranging
between 33.0 and 34.8.

A total of 504,478 fish eggs and 311,970 fish larvae
were collected over the course of the survey. Total fish
egg concentrations during the survey ranged from 0.16
to 48.3 eggs/m3 (Fig. 3). Egg concentrations were sig-

nificantly higher in the spring than in other seasons
(F=271.3, P<0.0001, spring>summer>fall>winter). Total
fish larvae concentrations ranged from 0.15 to 35.0 lar-
vae/m3 (Fig. 3). Larval concentrations were significantly
higher during summer and spring seasons (F=206.1,
P<0.0001, spring=summer>fall>winter). The diversity
of ichthyoplankton assemblages, exp (H), ranged from
1.32 to 9.48 and was also highest during the summer
seasons (F=299.3, P<0.0001, summer>spring>fall>w
inter) (Fig. 3). Species diversity was significantly re-
lated to temperature as determined by a least squares
regression (F=34.7, P<0.001, r2 = 0.60). Although also
significantly correlated, the relationships between tem-
perature and fish egg concentrations (F=4.4, P<0.05,
r2=0.16) and total larval concentrations (F=6.9, P<0.05,
r2= 0.23) were not as strong. No significant relationships
were observed between salinity and fish eggs (F=0.22,
P=0.64, r2=0.01), total fish larvae (F<0.01, P=0.94,

198

Fishery Bulletin 108(2)

r2<0.01), and taxonomic diversity (,F=0.16, P= 0.69,
r2 = 0.01).

Excluding order-level larvae and unidentified larvae,
unidentified engraulids dominated our collections and
represented approximately 50% of the total (overall)
catch (Table 3). Engraulid larvae were present year-
round and likely comprised several commonly occurring
species in the region, including Anchoa hepsetus, A. na-
suta, A. mitchilli, and Engraulis eurystole. No attempt
was made to examine these fishes beyond the family
level because many were relatively small (<10 mm) and
damaged, and engraulid identifications are problem-
atic in our region (Farooqi et al., 2006a). Other taxa
that represented over 1% of the overall catch included
Cynoscioti arenarius (7.5%), Chloroscombrus chrysurus
(5.4%), Micropogonias undulatus (4.4%), Brevoortia pa-
tronus (3.8%), unidentified Gobiidae (3.6%), unidentified
Sciaenidae (2.8%), unidentified Ophidiidae (2.5%), Sym-
phurus spp. (2.1%), Menticirrhus spp. (1.2%), unidenti-
fied Clupeidae (1.2%), Syacium spp. (1.2%), and Etropus
crossotus (1.0%).

Larval fish specimens collected during the survey
represented 58 different families. Larvae belonging to
22 of these families could not be identified beyond the
family level, usually because published descriptions of

representative species in our region are either lacking
or are insufficient to discern between different species
within the family (e.g., Gerreidae, Sparidae, Haemu-
lidae, Echeneidae, Labridae, Scorpaenidae). Several
families were well represented with numerous species
or genera, including Ophichthidae (11 identified spe-
cies), Sciaenidae (9 species), Carangidae (7 species),
Myctophidae (6 genera), Paralichthyidae (5 genera), and
Clupeidae (5 species). Overall, the dominant families
collected during our survey (e.g., Engraulidae, Sciaeni-
dae, Carangidae, and Clupeidae) are the same as those
from previous surveys in the general vicinity (Table
3). In general, the taxonomic richness observed in our
survey falls between that found in surveys of shorter
duration and in limited spatial-scale surveys (e.g., Wil-
liams,1983; Rakocinski et al., 1996) and from SEAMAP
surveys that encompass a larger area and longer (20
years) time scales (ENTRIX, 2006).

Seasonal patterns were observed for most of the domi-
nant taxa collected (Fig. 4). Lutjanus campechanus and
Chloroscombrus chrysurus were collected only during the
summer periods ( June-October). Similarly, Sciaenops
ocellatus larvae were collected only during late summer
(September-October). In contrast, Citharichthys spilop-
terus was collected in almost every sampling event, in-
dicating year-round spawning or extended
pelagic larval durations. Although sev-
eral species had winter peaks, none were
present exclusively during winter months.
Brevoortia patronus and Paralichthys spp.,
for example, peaked in concentration dur-
ing November-December, but were also
collected in fall-spring. Similar patterns
were observed for Elops saurus and Micro-
pogonias undulatus (late summer-winter)
and Peprilus burti and Leiostomus xan-
thurus (late summer-spring). Etrumeus
teres differed in that larvae were collected
during winter-spring periods. Most of the
dominant taxa, however, were collected
primarily during the late spring-late
summer months (May-October), such as
Myrophis punctatus, Harengula jaguana,
Opisthonema oglinum, Centropristis spp.,
Diplectrum spp., Serraniculus pumilio, De-
capterus punctatus , Auxis spp., Euthynnus
alletteratus, Scomberomorus maculatus,
Peprilus alepidotus, Syacium spp., ger-
reids, and microdesmids. The remaining
taxa ( Cynoscion arenarius, C. nothus,
Larimus fasciatus, labrids, and Etropus
crossotus) were collected during the same
period, but inclusive of the early spring
months (March-April).

Larval concentrations among the domi-
nant taxa varied widely throughout the
survey period (Fig. 4). Several taxa were
present in low numbers throughout the
survey. For example, mean densities of E.
saurus, O. oglinum, Diplectrum spp., S.

-*-1993-2003 Mean

♦ 2004
■ 2005

• 2006

JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Month

Figure 2

Mean monthly temperature observations (depth-integrated) at the
ichthyoplankton sampling station and the 10-year average temperature
(1993-2003). Sampling station means are derived from temperature
profile observations recorded by the Bedford Institute of Oceanography
Net Environmental Sampling System (BIONESS). The 10-year mean
was determined from near-surface (0.6 m depth) temperature observa-
tions (T ) recorded by an oceanographic buoy located approximately
54 km west of the sampling station. The plotted depth-integrated
temperature estimates (T-) were calculated through the relation ship
T. = 0.90*77 + 2.37.

Hernandez et al.: Variability in ichthyoplankton abundance and composition in the northern Gulf of Mexico

199

-sir

ONDJ FMAMJ JASONDJ FMAMJ JASO

Total fish eggs j

r£

r3rj

-i . .1 .

r“n

i — i

ONDJ FMAMJ J ASONDJ FMAMJ J ASO
40 r Total fish larvae

30 -

20 -

c 10 -
a)
a

iQ.

.rinn

nti

jffl

Q. 5

ONDJ FMAMJ JASONDJ FMAMJ JASO
10 f Larval fish assemblage

rj|j

nfln

r3ri

nnnn

ONDJ FMAMJ J ASONDJ FMAMJ J ASO
Month (2004-2006)

Figure 3

Mean temperature and salinity, fish egg and larval fish concentrations,
and diversity indices for larval fish assemblages for October 2004-October
2006. Temperature and salinity are depth-integrated mean values for each
month. Egg and larval fish concentrations are standardized by volume of
water filtered (error bars denote ±1 standard error). Calculation of diversity
follows Jost (2006) and depicts the exponential function of Shannon entropy,
H (error bars denote ±1 standard error).

pumilio, L. campechanus, Gerreidae,

S. ocellatus, Labridae, Auxis spp., E.
alletteratus, P. burti, C. spilopter-
us, Paralichthys spp., and Syacium
spp. did not exceed 10 larvae/100
m3 during any sampling event.

Other taxa were characterized by
relatively high concentrations, either
during a single sampling event (e.g.,

E. teres, C. chrysurus, C. arenarius,

L. xanthurus, Microdesmus spp., S.
maculatus, P. alepidotus ) or dur-
ing a single year (e.g., H. jaguana).

The remaining taxa ( M . punctatus,

B. patronus, Centropristis spp., D.
punctatus, C. nothus, L. fasciatus,

M. undulatus, E. crossotus) were
present during multiple years in
relatively similar concentrations.

Results from the cluster analysis
were largely in agreement with the
observed seasonal patterns previ-
ously defined by water temperature
(Fig. 5). Taxonomic assemblages
from fall and winter periods were
clustered separately from spring
and summer periods. All summer
months ( June-October) were clus-
tered together with the exception
of August 2005 and October 2004.

Larval collections in August 2005
were characterized by atypically
high concentrations of a few spe-
cies, most notably C. chrysurus and

C. arenarius, which were present
in concentrations exceeding >500
larvae/100 m3 (Fig. 4), resulting
in relatively low species diversity
(Fig. 2) for the summer period. The
October 2004 sampling event was
included in the summer period, al-
though the mean temperature was
marginally below the 26°C criterion
for the summer period (Fig. 3) and
indicative of a seasonal transitional
period. Similarly, the assemblages
from the May sampling events were
relatively distinct from the ear-
lier spring period sampling events
(March and April).

Discussion

Although numerous ichthyoplankton surveys have
been conducted in the northern Gulf of Mexico, most
have been conducted off the coasts of Texas, Louisi-
ana, and Florida (Ditty et al., 1988), and few have been
conducted with a high level of temporal resolution and
sample replication. The Alabama shelf region, although

relatively small, is unique in that it is bounded by two
major topographic features (Mississippi River Delta to
the west and DeSoto Canyon to the east) that poten-
tially inhibit alongshore transport of larvae (Johnson
et al., 2009). In addition, the Alabama continental shelf
receives freshwater outflow from the Mobile River system,
which drains the fourth largest watershed in the United
States and has the sixth largest freshwater discharge
on the North American continent (Park et al., 2007).
As a result, the inner shelf environment off Alabama
is a highly productive region that supports valuable

200

Fishery Bulletin 108(2)

Table 3

Summary (90% cumulative percentage and abundance ranking) of the dominant family groups collected during the 2004-2006
ichthyoplankton survey in the northern Gulf of Mexico off the coast of Alabama and from other ichthyoplankton surveys in the
general vicinity.

Family

This study

ENTRIX (2006b

Rakocinski et al. (1996)2

Williams (1983)3

% (Rank)

Engraulidae

50.5 (1)

32.3 (1)

82.0(1)

69.3(1)

Sciaenidae

15.9(2)

10.2 (3)

4.0(3)

14.0(2)

Carangidae

5.4(3)

2.7 (8)

5.0(2)

2.8(4)

Clupeidae

5.0(4)

15.5 (2)

4.3 (3)

Paralichthyidae

3.9(5)

8.5(4)

Gobiidae

3.6(6)

4.1 (6)

Ophidiidae

2.5(7)

3.6(7)

Cynoglossidae

2.1 (8)

5.6(5)

Synodontidae

0.9(9)

1.9(9)

Triglidae

0.8 (10)

0.8(13)

Serranidae

1.9(10)

Bregmacerotidae

1.6(11)

Labridae

1.0(12)

Callionymidae

0.7(14)

Stromateidae

0.4(15)

Scombridae

0.3 (16)

Lutjanidae

0.2(17)

Congridae

0.2(18)

Ophichthidae

0.2 (19)

Tetraodontidae

0.2(20)

Cumulative %

90.6

91.9

91.0

90.4

1 Samples (oblique) were collected as part of the SEAMAP ichthyoplankton survey (Rester el al., 2000) during the months of June-November from
1982 to 2002 by using a 61-cm bongo net fitted with 333-pm mesh. Sample stations were limited to the Mississippi and Alabama inner-shelf
region.

2 Samples (upper and lower water column) were collected monthly from November 1979 to October 1980 in Mississippi Sound by using a 1-m
diameter opening-closing conical-ring plankton net with 335-pm mesh.

3 Samples (surface and demersal) were collected monthly from March 1979 to February 1980 in lower Mobile Bay by using a 1x0. 5-m rectangular
opening plankton net with 505-pm mesh.

fisheries resources (Shipp, 1992). The establishment of
our survey is the first to specifically target larval fish
assemblages in Alabama shelf waters and is the only
survey from the northern Gulf of Mexico to combine
frequent sampling effort (monthly) with high temporal
replication (64+ samples/month) for a relatively long
duration (25 months). Few ichthyoplankton surveys have
been conducted near our sampling location, including a
one-year survey of lower Mobile Bay (Williams, 1983), a
one year survey of Mississippi Sound (Rakocinski et al.,
1996), and a summary of SEAMAP ichthyoplankton data
collected on the Mississippi and Alabama shelf during
1982-2002 (ENTRIX, 2006). The fisheries-independent
data collected during our survey, therefore, provide a
baseline for future comparisons with respect to vari-
ability in local oceanographic and climatic features (e.g.,
warming water temperatures), water and land usage
(e.g., Mobile Bay nutrient loading and water outflow),
and habitat modifications (e.g., artificial reef programs).

A comparison of results among multiple ichthyoplank-
ton surveys is complicated because the motives for sur-
veys often differ, resulting in survey-specific protocols
and sampling biases. For example, the summary of
larval fish seasonality reported by Ditty et al. (1988) for
the northern Gulf of Mexico included over 30 separate
surveys covering a wide range of spatial extent (Gulf-
wide to individual bays and passes), sampling depths
(neuston to 200 m), mesh sizes (0.086-1.05 mm), gear
types (eight different samplers), sampling frequency
(biweekly to quarterly), and survey duration (weeks
to years). In addition, the taxonomic level to which
ichthyoplankton are identified and at which they are
reported varies with larval fish size, condition after
capture, and availability of adequate descriptions. Our
decision to use a 202-pm mesh size (as opposed to more
standard sizes, e.g., >333 pm) is the factor that most
likely biases our survey results when compared with
previous studies. The effect of mesh size on the reten-

Hernandez et al.: Variability in ichthyoplankton abundance and composition in the northern Gulf of Mexico

201

0.5

S. pumilio

t I

4—H

ONDJ FMAMJ J ASONDJ FMAMJ J ASO

600

500

400

300

200

100

C. chrysurus

-t- i. .i i. . | -t-nfal..

_ cfaA i 1 ■ _l_

ONDJ FMAMJ JASONDJ FMAMJ JASO

Month 2004-2006

Figure 4

Mean larval concentrations (no./lOO m3) of dominant taxa for each month during the ichthyoplank-
ton survey (October 2004-October 2006). Error bars denote ±1 standard error. Figure panels are
presented in taxonomic order, as listed in Table 2.

tion of larvae has been documented in numerous stud-
ies, with the general conclusion that larger mesh sizes
may efficiently collect the late larval stages but under-
estimate the smaller size classes because of extrusion
(Houde and Lovdal, 1984; Leslie and Timmins, 1989).
Conversely, smaller mesh nets may collect smaller size

classes of larvae, but are prone to clogging, thus reduc-
ing their effectiveness in sampling ichthyoplankton,
particularly late-stage fish larvae (Smith et al., 1968;
Tranter and Smith, 1968). In our study the smaller
mesh size enabled us to achieve better estimates of
fish egg and preflexion larval fish concentrations, which

202

Fishery Bulletin 108(2)

Month 2004-2006

Figure 4 (continued)

are indicative of nearby adult spawning activity. The
tradeoff, however, was that many of the larvae were
too small to identify to the genus or species level. As
a result, most fish larvae collected in this survey were
identified to the order and family level only (14% and
52%, respectively).

Fifty-eight different families of fishes were collected
in our ichthyoplankton collections, the adult forms of

which represent diverse zoogeographic regions and
ecological niches. As expected, larvae of nearshore
and inner shelf species were the most dominant, such
as coastal pelagic (e.g., engraulids, carangids, clupe-
ids, stromateids, gerreids) and coastal demersal (e.g.,
sciaenids, paralichthyids, gobiids, cynoglossids, syn-
odontids) species. Unidentified engraulids were the
most abundant larval fish group in our survey (ap-

Hernandez et al.: Variability in ichthyoplankton abundance and composition in the northern Gulf of Mexico

203

Microdesmus
spp.

0 1 I I I I I "H-H

1111

.7(7.. I

1 — 1 — 1 — 1 — 1 — i I 1 r

ON n .1 F M AM .1 .1 A

Month 2004-2006

Figure 4 (continued)

proximately 50%) and in the aforementioned regional
surveys (Table 3). Engraulid larvae appear to be more
abundant in protected coastal waters, as indicated by
their higher dominance in the surveys of Mobile Bay
(82%) and Mississippi Sound (69%), both of which are
shallow estuarine regions. On the basis of identification
of larger specimens, most of the engraulids collected
in Mobile Bay and Mississippi Sound were Anchoa

mitchilli and A. hepsetus (Williams, 1983; Rakocinski
et al., 1996), whereas our collections contained these
species as well as the coastal species A. nasuta and
Engrciulis eurystole. The inner shelf taxa Brevoortia
patronus, Cynoscion arenarius, Micropogonias undula-
tus, Chloroscombrus chrysurus, and unidentified gobies
were among the most dominant ichthyoplankton in all
surveys, including ours. As adults, these fishes are ex-

204

Fishery Bulletin 108(2)

Similarity

Figure 5

Dendrogram depicting relationships (based on Bray Curtis similarities) of
the dominant taxonomic assemblages between months. Larval concentrations
for dominant taxa were square-root transformed before analyses.

tremely abundant in estuarine and inner shelf waters
and serve important ecological roles as forage fishes
(e.g., B. patronus, C. chrysu/'us) and as predators link-
ing primary consumers to higher trophic levels (e.g., M.
undulatus, C. ai'enarius ) (Naughton and Saloman, 1981;
Overstreet and Heard, 1982; Sheridan et al., 1984;
Franks et al., 2003). The larvae of these relatively few
taxa often comprise the majority of ichthyoplankton in
surveys throughout the northern Gulf of Mexico (Ditty,
1986; Cowan and Shaw, 1988; Tolan et al., 1997).

Flatfish larvae (e.g., paralichthyids and cynoglossids)
represented another dominant coastal group. Cynoglos-
sids (Sytnphurus spp.) were common year-round in our
study, which indicates that our collections contained
multiple species. These fishes are commonly reported in
ichthyoplankton surveys throughout the Gulf of Mexico,
but identification of larvae (and adults) is problematic
owing to high species richness and overlapping mer-
istics (Farooqi et al., 2006b). Similarly, Citharichthys
spp. were abundant year-round, as were C. spilopterus.
Again, identification down to species is problematic
because five species (C. arctifrons, C. cornotus, C. gym-
norhinus, C. macrops, and C. spilopterus) are found in
the study region (Lyczkowski-Shultz and Bond, 2006).
Although efforts were made to identify larvae conser-
vatively, some of our C. spilopterus may have included
congeners. This issue of questionable identification ap-
pears less likely for the Etropus species complex, which
was also abundant, primarily E. crossotus and E. mi-
crostomus.

Equally notable in our survey was the absence (or
rarity) of larvae from taxa that are common in our
sampling region as adults. For example, serranine
(seabasses) serranid larvae were collected, but epi-

nepheline (grouper) larvae were not. Similarly absent
(or rare) were larvae from other recreational and com-
mercially important species such Coryphaena hippurus
(Coryphaenidae), Rachycentron canadum (Rachycentri-
dae), Balistes eapriscus ( Balistidae), Lobotes surina-
mensis (Lobotidae), Chaetodipterus faber (Ephippidae),
and Mugil cephalus (Mugilidae), all of which spawn in
coastal or offshore waters of Alabama. The fact that
we did not collect some of these taxa is not surprising
(e.g., B. eapriscus, M. cephalus) because they are more
commonly collected in the neuston (which we did not
sample). The absence of grouper larvae is perplex-
ing, even though the rarity of epinepheline larvae has
been documented in the northern Gulf of Mexico. For
example, only 37 grouper larvae were collected in gulf-
wide SEAMAP ichthyoplankton surveys between 1982
and 1999 (>7000 samples) (Lyczkowski-Shultz et al.1).
Most of the grouper larvae were collected at offshore
SEAMAP sampling stations, which indicates that their
occurrence in nearshore environments may be rare.
It is possible that the limited spatial extent of our
survey (i.e., a single station) may have influenced our
estimates of larval fish concentrations and variability,
because coastal marine processes that influence larval
fish dynamics are often site-specific (e.g., local wind
regimes, tidal flows, river discharge), but the overall
seasonal supply of larvae available at our sampling
station is likely representative of the ichthyofauna from
a larger northcentral Gulf of Mexico region between
the 87°W and 89°W longitude (Boschung, 1992).

The main objective of this study was to describe
taxon-specific seasonality for larval fishes collected in
the survey region. For several reasons, we limited our
seasonal analyses to water temperature, as opposed to

Hernandez et al.: Variability in ichthyoplankton abundance and composition in the northern Gulf of Mexico

205

a suite of environmental parameters. First, tempera-
ture has long been proposed as an important factor in
the initiation of spawning for marine fishes (Orton,
1920), and numerous field and laboratory (primarily
aquaculture-related) studies have provided support for
temperature as a primary influence (Arnold et al., 2002;
Sheaves, 2006). Second, water temperature varies pre-
dictably at seasonal scales (e.g., months), as opposed
to other factors that vary at shorter time scales. Our
salinity data (Fig. 3), for example, showed no seasonal
trends and were not correlated with egg or larval fish
concentrations. The monthly mean salinity values cal-
culated during each cruise likely reflect short-term
variability related to tidal flow, riverine outflow, local
wind conditions, and related factors that affect salinity
at our sampling station. In addition, salinity, although
an important factor for many estuarine-spawning spe-
cies, is generally considered less important than tem-
perature to the timing of marine fish spawning (Bye,
1984; Sheaves, 2006).

Defining seasonality in terms of water temperature
also provides a framework for monitoring fisheries dy-
namics with respect to anticipated rises in sea tem-
perature due to global climate change. Our monthly
observed depth-integrated temperatures were relatively
consistent with those for the previous ten-year aver-
age for the region, although winter (December- Janu-
ary) and late summer (August-October) values were
generally higher (Fig. 2). Fodrie et al. (2009) noted a
significant increase in sea surface temperature near
the mouth of Mobile Bay over a 20-year period (1987-
2007). The authors also noted a concurrent increase in
the number and occurrence of juvenile subtropical and
tropical fishes collected in seagrass meadows along the
northern Gulf of Mexico. For example, in 2006-2007
surveys, juveniles of tropical species such as Chaetodon
ocellatus (Chaetodontidae), Fistularia tabaccu'ia (Fistu-
laridae), Ocyurus chrysurus (Lutjanidae), Thalassoma
bifasciatum (Labridae), Sparisoma viride (Scaridae),
and unidentified acanthurids were collected in coastal
habitats where they were not collected during previous
surveys (1971-79) (Livingston, 1985). Notably, in our
ichthyoplankton survey larvae from all of these fami-
lies, except Chaetodontidae, were collected but regret-
tably, comparable ichthyoplankton data from the 1970s
were not available and our identifications were made
only to the family level.

Conclusions

Increases in regional water temperatures may have sig-
nificant impacts on the reproductive success of marine
fishes and the subsequent survival of early life stages,
including early gonad maturation and spawning in
adults, altered larval transport pathways, extended
pelagic larval durations, changes in larval assemblage
structure, and mismatched timing of larval fish occur-
rence with food resources and physiological optima,
among other effects (Sheaves, 2006; O’Conner et al.,

2007; Genner et al., 2009). Establishment of long-term
baseline surveys provides a means of monitoring larval
fish assemblages and the factors that influence larval
fish dynamics in order to provide early indicators of
ecosystem changes due to environmental perturbations.
The ichthyoplankton survey efforts described here for
the October 2004-October 2005 period have since con-
tinued and expanded to include near monthly (depth-
discrete) ichthyoplankton sampling at five stations along
a cross-shelf transect from inside Mobile Bay extending
offshore to a station approximately 54 km south of Dau-
phin Island. The expanded survey program (Fisheries
Oceanography of Coastal Alabama, or FOCAL) will allow
us to estimate and monitor the variability in ichthyo-
plankton seasonality, abundance, assemblage structure,
and vertical distribution over multiple temporal and
spatial scales.

Acknowledgments

We would like to thank the technicians and graduate
students that participated on our research cruises:
A. Beck, S. Bosarge, L. Chiaverano, T. Clardy, D. del
Valle, N. Geraldi, J. Goff, E. Goldman, J. E. Herrmann,
J. M. Herrmann, J. Higgins, L. Kramer, B. Lacour, C.
Martin, M. Miller, S. Muffelman, C. Newton, C. Pabody,
D. Ploetz, C. Schobernd, Z. Schobernd, R. Shiplett, and
D. Vivian. We especially thank the captains and crew of
the RV Verrill and RV E.O. Wilson (R. Collier, T. Guoba,
C. Lollar, and R. Wilson) and the Dauphin Island Sea
Laboratory technical support team (M. Dardeau, A.
Gunter, and K. Weiss). We also thank M. Konieczna
and the scientific staff at the Plankton Sorting and
Identification Center in Szczecin, Poland, for larval
fish identifications. K. Park provided assistance with
the CTD data. H. Fletcher and L. Hu provided data-
base management support. S. Bosarge produced the
station map (Fig. 1). Valuable comments and guidance
throughout the course of our survey were provided by J.
Lyczkowski-Shultz (NOAA/NMFS/SEFSC, Pascagoula
Laboratory, MS) and S. Heath (Alabama Department of
Conservation and Natural Resources, Marine Resources
Division, Dauphin Island, AL). We also thank R. Bro-
deur and three anonymous reviewers for comments
on a previous version of this manuscript. A portion of
these data were collected as part of contract # 2004-
GPS-MSA-NC-0085 from ConocoPhillips Corporation,
Houston, TX.

Literature cited

Arnold, C. R., J. B. Kaiser, and G. J. Holt.

2002. Spawning of cobia Rachycentron canadum in
captivity. J. World Aquacult. Soc. 33:205-208.

Auth, T.D.

2008. Distribution and community structure of ichthyo-
plankton from the northern and central California Cur-
rent in May 2004-06. Fish. Oceanogr. 17:316-331.

206

Fishery Bulletin 108(2)

Boschung, H. T.

1992. Introduction: geographical area of coverage. In
Catalog of freshwater and marine fishes of Alabama
(H. T. Boschung, ed.), p. 3-5. Alabama Mus. Nat. Hist.
14, Univ. Alabama, Tuscaloosa, AL.

Brodeur, R. D., W. T. Peterson, T. D. Auth, H. L. Soulen, M. M.
Parnel, and A. A. Emerson.

2008. Abundance and diversity of coastal fish larvae
as indicators of recent changes in ocean and climate
conditions in the Oregon upwelling zone. Mar. Ecol.
Prog. Ser. 366:187-202.

Bye, V. J.

1984. The role of environmental factors in the timing
of reproductive cycles. In Fish reproduction: strate-
gies and tactics (G. W. Potts and R.J . Wooton, eds.),
p. 132-148. Academic Press, London.

Cowan, J. H., and R. F. Shaw.

1988. The distribution, abundance, and transport of larval
sciaenids collected during winter and early spring from
the continental shelf waters off west Louisiana. Fish.
Bull. 86:129-142.

Ditty, J. G.

1986. Ichthyoplankton in neritic waters of the northern
Gulf of Mexico off Louisiana: Composition, relative abun-
dance, and seasonality. Fish. Bull. 84:935- 946.

Ditty, J. G., G. G. Zieske, and R. F. Shaw.

1988. Seasonality and depth distribution of larval fishes
in the northern Gulf of Mexico above latitude 26°00'N.
Fish. Bull. 86:811-823.

Doyle, M. J., W. W. Morse, and A. W. Kendall Jr.

1993. A comparison of larval fish assemblages in the
temperate zone of the northeast Pacific and northwest
Atlantic oceans. Bull. Mar. Sci. 53:588-644.

ENTRIX.

2006. Final environmental impact statement for
the Compass Port LLC Deepwater Port License
Application. United States Coast Guard, Docket number
USCG-2004-17659.

Farooqi, T. W., J. G. Ditty, and R. F. Shaw.

2006a. Engraulidae: anchovies. In early stages of Atlan-
tic fishes: an identification guide for the western Central
North Atlantic (W. J. Richards, ed.), p. 101-127. Taylor
and Francis Group, Boca Raton, FL.

Farooqi, T. W., R. F. Shaw, J. G. Ditty, and J. Lyczkowski-
Shultz.

2006b. Cynoglossidae: tonguefishes. In Early stages of
Atlantic fishes: an identification guide for the western
Central North Atlantic (W. J. Richards, ed.), p. 2367-
2379. Taylor and Francis Group, Boca Raton, FL.

Fodrie, F. J., K. L. Heck, S. P. Powers, W. M. Graham, and K. L.
Robinson.

2009. Climate-related, decadal-scale assemblage changes
of seagrass-associated fishes in the northern Gulf of
Mexico. Glob. Change Biol. 16:48-59.

Franks, J. S., K. E. VanderKooy, and N. M. Garber.

2003. Diet of tripletail, Lobotes surinamensis, from Mis-
sissippi coastal waters. Gulf Caribb. Res. 15:27-32.

Genner, M. J., N. C. Halliday, S. D. Simpson, A. J. Southward,
S. J. Hawkins, and D. W. Sims.

2009. Temperature-driven phenological changes within
a marine larval fish assemblage. J. Plankton Res.
doi:10.1093/plankt/fbp082.

Govoni J. J., D. E. Hoss, and D. R. Colby.

1989. The spatial distribution of larval fishes about the
Mississippi River plume. Limnol. Oceanogr. 34:178-
187.

Gray, C. A., and A. G. Miskiewicz.

2000. Larval fish assemblages in south-east Australian
coastal waters: seasonal and spatial structure. Estuar.
Coast. Shelf Sci. 50:549-570.

Hare, J. A., and J. J. Govoni.

2005. Comparison of average larval fish vertical distri-
butions among species exhibiting different transport
pathways on the southeast United States continental
shelf. Fish. Bull. 103:728-736.

Hernandez, F. J., Jr., R. F. Shaw, J. S. Cope, J. G. Ditty,
T. W. Farooqi, and M. C. Benfield.

2003. The across-shelf larval, postlarval and juvenile
fish community associated with offshore oil and gas
platforms west of the Mississippi River Delta. In Fish-
eries, reefs and offshore development (D. Stanley, and
A. Scarborough-Bull, eds.), p. 39-72. Am. Fish. Soc.
Symp. 36, Bethesda, MD.

Hernandez-Miranda, E., A. T. Palma, and F. P. Ojeda.

2003. Larval fish assemblages in nearshore coastal waters
off central Chile: temporal and spatial patterns. Estuar.
Coast. Shelf Sci. 56:1075-1092.

Houde, E. D., and J. A. Lovdal.

1984. Seasonality of occurrence, foods and food prefer-
ences of ichthyoplankton in Biscayne Bay, Florida. Es-
tuar. Coast. Shelf Sci. 18:403-419.

Johnson, D. R., H. M. Perry, J. Lyczkowski-Shultz, and
D. Hanisko.

2009. Red snapper larval transport in the northern Gulf
of Mexico. Trans. Am. Fish. Soc. 138:458-470.

Jost, L.

2006. Entropy and diversity. Oikos 113:363-375.

Leslie, J. K., and C. A. Timmins.

1989. Double nets for mesh aperture selection and sam-
pling in ichthyoplankton studies. Fish. Res. 7:225-232.

Livingston, R. J.

1985. Organization of fishes in coastal seagrass system:
the response to stress. In Fish community ecology in
estuaries and coastal lagoons: towards an ecosystem inte-
gration (A. Yanez-Arancibia, ed.), p. 367-382. UNAM
Press, Mexico City, Mexico.

Lyczkowski-Shultz, J., and P. J. Bond.

2006. Paralichthyidae: sand flounders. In Early stages
of Atlantic fishes: an identification guide for the western
Central North Atlantic (W.J. Richards, ed.), p. 2291-
2325. Taylor and Francis Group, Boca Raton, FL.

Lyczkowski-Shultz, J., and D. S. Hanisko.

2007. A time series of observations on red snapper larvae
from SEAMAP surveys, 1982- 2003: seasonal occur-
rence, distribution, abundance and size. Am. Fish.
Soc. Symp. 60:3—23.

Lyczkowski-Shultz, J., and G. W. Ingram Jr.

2003. Preliminary guide to the identification of the
early life stages of balistid fishes of the western Central
North Atlantic. NOAA Tech. Memo. NMFS-SEFC-
493, 13 p.

Marancik, K. E., L. M. Clough, and J. A. Hare.

2005. Cross-shelf and seasonal variation in larval fish
assemblages on the southeast United States continen-
tal shelf off the coast of Georgia. Fish. Bull. 103:
108-129.

Muhling, B. A., L. E. Beckley, J. A. Koslow, and A. F. Pearce.

2008. Larval fish assemblages and water mass struc-
ture off the oligotrophic south- western Australian
coast. Fish. Oceanogr. 17:16-31.

Naughton, S. P, and C. H. Saloman.

1981. Stomach contents of juveniles of king mackerel

Hernandez et al.: Variability in ichthyoplankton abundance and composition in the northern Gulf of Mexico

207

(Scomberomorus cavalla) and Spanish mackerel ( S .
maculatus). Northeast Gulf Sci. 5:71-74.

O’Connor, M. I., J. F. Bruno, S. D. Gaines, B. S. Halpern,
S. E. Lester, B. P. Kinlan and J. M. Weiss.

2007. Temperature control of larval dispersal and
the implications for marine ecology, evolution, and
conservation. Proc. Natl. Acad. Sci. USA 104:1266—
1271.

Orton, J. H.

1920. Sea-temperature, breeding and distribution in
marine animals. J. Mar. Biol. Assoc. U.K. 12:339-
366.

Overstreet, R. M., and R. W. Heard.

1982. Food contents of six commercial fishes from Mis-
sissippi Sound. Gulf Res. Rep. 7:137-149.

Park, K., Kim, C.-K., and W. W. Schroeder.

2007. Temporal variability in summertime bottom hypoxia
in shallow areas of Mobile Bay, Alabama. Estuaries
Coasts 30:54-65.

Rakocinski, C. F., J. Lyczkowski- Shultz, and S. L. Richardson.

1996. Ichthyoplankton assemblage structure in Missis-
sippi Sound as revealed by canonical correspondence
analysis. Estuar. Coast. Shelf Sci. 43:237-257.

Raynie, R. C., and R. F. Shaw.

1994. Ichthyoplankton abundance along a recruitment
corridor from offshore spawning to estuarine nursery
ground. Estuar. Coast. Shelf Sci. 39:421-450.

Rester, J. K., D. Hanisko, N. Sanders, Jr., and B. Pellegrin.

2000. SEAMAP environmental and biological atlas of
the Gulf of Mexico, 1998. Gulf States Marine Fisher-
ies Commission, Ocean Springs, MS.

Richards, W. J., M. F. McGowan, T. Leming, J. T. Lamkin, and
S. Kelley.

1993. Larval fish assemblages at the loop current bound-
ary in the Gulf of Mexico. Bull. Mar. Sci. 53:475-
537.

Scott, G. P, S. C. Turner, C. B. Grimes, W. J. Richards, and
E. B. Brothers.

1993. Indices of larval bluefin tuna, Tliunnus thynnus,
abundance in the Gulf of Mexico: modeling variability
in growth, mortality, and gear selectivity. Bull. Mar.
Sci. 53:912-929.

Sheaves, M.

2006. Is the timing of spawning in sparid fishes a
response to sea temperature regimes? Coral Reefs
25:655-669.

Sheridan, P. F., D. L. Trimm, and B. M. Baker.

1984. Reproduction and food habits of seven species of
northern Gulf of Mexico fishes. Contrib. Mar. Sci.
27:175-204.

Sherman, K. W., W. Smith, W. Morse, M. Berman, J. Green, and
L. Ejsymont.

1984. Spawning strategies of fishes in relation to cir-
culation, phytoplankton production, and pulses in zoo-
plankton off the northeastern United States. Mar.
Ecol. Prog. Ser. 18:1-19.

Shipp, R. L.

1992. Introduction: biogeography of Alabama’s marine
fishes. In Catalog of freshwater and marine fishes of
Alabama (H. T. Boschung, ed.), p. 7-9. Alabama Mus.
Nat. Hist. 14, Univ. Alabama, Tuscaloosa, AL.

Smith, P. E., R. C. Counts, and R. I. Clutter.

1968. Changes in filtering efficiency of plankton nets
due to clogging under tow. J. Cons. Int. Explor. Mer
32:232-248.

Sogard, S. M., D. E. Hoss, and J. J. Govoni.

1987. Density and depth distribution of larval gulf men-
haden, Brevoortia patronus, Atlantic croaker, Micropogo-
nias undulatus , and spot, Leiostomus xanthurus, in the
northern Gulf of Mexico. Fish. Bull. 85:601-609.

Tolan, J. M., S. A. Holt, and C. P. Onuf.

1997. Distribution and community structure of ichthyo-
plankton in Laguna Madre Seagrass meadows: potential
impact of seagrass species change. Estuaries Coasts
20:450-464.

Tranter, D. J. and P. E. Smith.

1968. Filtration performance. In Reviews on zooplank-
ton sampling methods, part I (D. J. Tranter, ed.), p.
27-56. Monogr. Oceanogr. Methodol. 2, UNESCO
Press, Paris.

Williams, L. W.

1983. Larval fish assemblages of lower Mobile Bay.
M.S. thesis, 55 p. Dep. Biology, Univ. South Alabama,
Mobile, AL.

208

Observer-reported skate bycatch

in the commercial groundfish fisheries of Alaska

Kristy A. Lewis

Email address for contact author: duane.stevenson@noaa.gov

NMFS, Alaska Fisheries Science Center

Fisheries Monitoring and Analysis Division

7600 Sand Point Way NE

Seattle, Washington 98115

Abstract — We analyzed skate catch
data collected by observers in the
North Pacific Groundfish Observer
Program (NPGOP) from 1998 through
2008 to document recent changes in
the identification of skates by observ-
ers and to examine the species com-
position of observed skate catch in
Alaska’s groundfish fisheries as well
as recent trends in skate retention
by commercial fishermen. Histori-
cally, almost all skate bycatch has
been reported by NPGOP observers as
“skate unidentified.” However, since
2004 observers have been trained to
identify skates to the genus and spe-
cies level. In 2008 over 95% of all
skates were identified at least to the
genus level, and over 50% were iden-
tified to species. The most common
species of skates identified by observ-
ers in groundfish fisheries are Bathy-
raja parmifera (Alaska skate), Raja
binoculata (big skate), and Bathyraja
aleutica (Aleutian skate). Species com-
position of reported skate catch gen-
erally reflects recent survey-derived
biomass estimates, with B. parmifera
dominating the catches in the Bering
Sea and, to a lesser extent, in the
Aleutian Islands region, and species
of the genus Raja dominating catches
in the Gulf of Alaska. A relatively
high percentage of the skate catch
on longline vessels is still reported
at the family or genus level because
of difficulties in the identification
of skates not brought onboard the
vessel. For the larger skate species,
the proportion retained for processing
has increased in recent years as the
market price for skate product has
increased. Although observed skate
catch does not give a complete account
of skate bycatch in the fisheries of the
region, observer data provide criti-
cal information for the appropriate
management of skate populations in
Alaska.

Manuscript submitted 28 September 2009.
Manuscript accepted 19 January 2010.
Fish. Bull. 108:208-217 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

Duane E. Stevenson (contact author)

Skates are large, long lived fishes with
relatively slow growth rates and low
reproductive potential (Ebert, 2005;
Ebert et al., 2008). These aspects of
their life history, combined with their
relatively low mobility and benthic
habitat, make skates particularly
vulnerable to fishing pressure and
slow their recovery from population
declines; yet few countries have man-
agement plans for skates or other
chondrichthyan species (Stevens et ah,
2000). In cases where skates are tar-
geted by fisheries, population declines
can be rapid (Agnew et ah, 2000).
Moreover, because discard mortality
can be high (Stobutzki et al., 2002;
Laptikhovsky, 2004), skate popula-
tions may be dramatically affected
by fishing activity even if they are
not targeted directly (Brander, 1981;
Casey and Myers, 1998; Dulvy et ah,
2000; Stevens et al., 2000). In addition
to population declines, fishing pres-
sure may lead to significant shifts in
community structure because declines
in some species of a skate assemblage
may be masked by increases in other,
more resilient species (Agnew et al.,
2000; Dulvy et al., 2000; Stevens et
al., 2000). Therefore, effective man-
agement of skate populations requires
species-specific data on abundance
trends and exploitation rates.

Skates are regularly caught in
nearly all of the commercial ground-
fish fisheries currently prosecuted in
Alaska waters, including fisheries
targeting Pacific cod ( Gadus mac-
rocephalus), walleye pollock ( Ther -
agra chalcogramma), yellowfin sole
(Limanda aspera), and other species
(Ormseth et al., 2009). In addition to
their ubiquitous presence as bycatch

species, skates have been targeted in
Alaska waters on a short-term region-
al basis. An unregulated fishery tar-
geting Raja binoculata (big skate), R.
rhina (longnose skate), and assorted
species of Bathyraja (including Alaska
skates) developed in the central Gulf
of Alaska (GOA) in February 2003.
Shifting economic conditions and fish-
ing seasons soon made other target
species more valuable, but this short-
lived fishery revealed that skates can
quickly become an attractive alterna-
tive target when other fisheries are
closed (Matta, 2006). More recently,
the Alaska Department of Fish and
Game (ADF&G) approved a pilot fish-
ery for big and longnose skates in
the state-managed waters of Prince
William Sound (ADF&G Emergency
Order #2-G-E-04-09) in 2009. Else-
where in Alaska skates are still man-
aged as part of a large nontarget spe-
cies complex, although beginning in
2011 skates in the Bering Sea and
Aleutian Islands will be managed as
a separate unit.

Recent advances in the taxonomy
of the skates of the North Pacific and
Bering Sea (Ishiyama and Ishihara,
1977; Ishihara and Ishiyama, 1985,
1986; Stevenson et al., 2004, 2007,
2008) have facilitated increasingly
detailed identification of skates by
observers in the commercial fisher-
ies of Alaska. The resulting wealth
of detailed catch data now permits
an examination of skate bycatch on
a level that was not previously pos-
sible. The objectives of this study are
1) to document recent changes for the
identification of skates in the NPGOP,
and 2) to provide an overview of po-
tential management concerns by ex-

Stevenson and Lewis: Skate bycatch in the commercial groundfish fisheries of Alaska

209

amining the species composition of observed skate catch
(OSC) in Alaska’s groundfish fisheries and recent trends
in skate retention by commercial fishermen.

Materials and methods

All data used for this study were extracted from the
North Pacific Groundfish Observer Program (NPGOP)
database maintained by the Fisheries Monitoring and
Analysis (FMA) Division of the National Marine Fisher-
ies Service’s (NMFS) Alaska Fisheries Science Center.
This database houses all biological data collected by
groundfish observers onboard commercial fishing vessels
operating in the waters of Alaska’s federal Exclusive Eco-
nomic Zone (EEZ). For an overview of the database, see
the FMA Division website (National Marine Fisheries
Service, http://www.afsc.noaa.gov/FMA/fma_database.
htm, accessed November 2009).

Federal law requires observers to be present at all
times on commercial fishing vessels of 125 ft (38.1 m)
or more in length overall (LOA) operating in the fed-
eral EEZ. For vessels from 60 to 124 ft (18.3 to 37.8
m) LOA, observer coverage is required for only 30% of
fishing days and for vessels less than 60 ft (18.3 m)
LOA no observer coverage is required. The catch data
used for this study were taken from trawl hauls and
longline sets during which an observer was present and
was sampling, so that the catch statistics presented
here do not represent the total catch of the fisheries in
this region, nor do they represent biomass estimates.
For some commercial fisheries in the area, pot gear is
used, but observers rarely encounter skates in these
fisheries, and therefore such data are not included in
this study.

The process used by observers to determine the spe-
cies composition and catch weights of sampled hauls
depends on gear type. Observers on trawlers may de-
termine the species composition of a haul by identifying
and weighing the entire catch, which is usually not
possible, or by choosing a random sample (generally 300
kg or more) of the catch and identifying and weighing
all taxa within the sample. The proportion by weight
of each taxon in the sample is then extrapolated to the
total catch weight, which may be determined by a num-
ber of methods, including flow scale readings, codend
measurements, or bin volume estimates. On longline
vessels, observers randomly select a “tally period” as
the gear is being retrieved. During this tally period,
the observer identifies and counts specimens, including
specimens that drop off the line or are intentionally dis-
carded. A subset of the specimens identified during the
tally period (15 or more per species, when possible) is
retained onboard the vessel and weighed to determine
an average weight for each taxon. That average weight
is then applied to all specimens identified during the
tally period, and the resulting proportional species com-
position is extrapolated to the total gear set to obtain
a total catch weight for each species for each set. The
basic data unit used for this study is the extrapolated

catch weight for each taxon from each observed haul
or set (hereafter trawl hauls and longline sets will be
collectively referred to as “hauls”). The total observed
skate catch (OSC) was calculated by summing extrapo-
lated catch weights for all skate taxa (including the
following unidentified skate groups: “skate unidenti-
fied,” “Bathyraja sp.,” and “ Raja sp.”) across all hauls
in which skates were identified. Scientific and common
names for skate taxa follow Stevenson et al. (2007).

From the inception of the NPGOP through the sam-
pling year 2002, observers were not trained to identify
skates and were therefore not required to identify them
beyond the family level. During 2002 and 2003, a field
identification key was developed (Stevenson, 2004) and
experienced observers began receiving training in skate
identification during annual briefings. Feedback from
experienced observers was used to refine the identifica-
tion materials and classroom training, and beginning
with the 2004 sampling year, all new and returning
observers were provided with skate identification train-
ing and materials for identification of skate in the field.
Since 1 January 2004 all observers have been required
to identify skates to the species level when possible.
Because of these changes in observer identification
training and policies, two separate but overlapping
time frames were used in this study. To investigate the
trends in observed skate catch and the history of skate
identification by observers an 11-year time frame was
chosen and queries were restricted to data collected
from 1 January 1998 through 31 December 2008. For
investigations of species-level trends in observer data,
queries were restricted to data collected from 1 Janu-
ary 2004 through 31 December 2008 — a period that
corresponds with the time period in which all new and
returning observers have been trained to identify skates
to the species level. Regions were defined on the basis
of NMFS management areas: Bering Sea comprises the
Bering Sea NMFS management areas 509-524; the
Aleutian Islands region comprises NMFS management
areas 541-543; and the Gulf of Alaska comprises NMFS
management areas 610-650 (Fig. 1). All catch propor-
tions are presented as a percentage of total observer
reported catch weight.

The targeted resource was not directly recorded in ob-
server catch data, so that for the purposes of this study,
the term “target species” is defined as the predominant
species in the catch. “Predominant species” was defined
as the species accounting for the highest percentage
of the extrapolated weight in the species composition
sample and was determined on a haul-by-haul basis.
Percent retained data are subjective estimates made
by observers using visual approximations, along with
information provided by the vessel’s captain or factory
manager. Mean retention rates used here are weighted
averages calculated annually for each species with the
following equation:

xaa
Xa '

210

Fishery Bulletin 108(2)

150°E 160°E 170°E 180° 170°W 160°W 150°W 140°W 130°W 120°W

0 250 500 1,000 km

1 l L l I

, 1 1 1 1

180° 170°W 160°W 150°W 140°W

Figure 1

Map showing NMFS management areas in which observed skate catch was examined from 1998 through 2008.
Stippled areas = Bering Sea, shaded areas=Aleutian Islands, diagonal hatching=Gulf of Alaska.

where i?;/ = the observer reported retention rate of spe-
cies i in haul j\ and

Ci; = the extrapolated catch weight of species i
in haul j.

Historical skate price information was derived from
Alaska state fish-ticket data, and was compiled for the
study period by Terry Hiatt (unpubl. data1). An annual
mean price was determined for each taxon by 1) calcu-
lating the exvessel price paid per pound round weight
at each delivery to all processors where the purchase
of raw skates from Alaska waters was recorded, and
then 2) calculating the simple average of those delivery
price points over the calendar year. Round weight refers
to intact whole specimens. For deliveries consisting of
nonwhole specimens, round weight (in pounds) was
calculated from net delivery weight by using a product

1 Hiatt, T. 2009. NMFS Alaska Fisheries Science Center,
Seattle, WA 98115.

recovery rate (PRR) of 0.32 for “wings” and 0.9 for
gutted animals (National Marine Fisheries Service,
http://www.fakr.noaa.gov/rr/tables/tabl3.pdf, accessed
November 2009). Each annual mean represents at least
334 (range: 334-2247) data points.

Results

Skate species composition reported by observers over
the past decade has changed considerably. Up to and
including 2002, over 98% of OSC was reported as “skate
unidentified” (Table 1). In 2003, less than 90% of OSC
was unidentified, and the proportion of unidentified
skates has continued to drop through 2008, a year in
which only 2% of OSC was unidentified. Because the
proportion of unidentified skates has dropped, the pro-
portions of skates identified to the genus level ( Bathy -
raja) and to the species level (Bathyraja parmifera. Raja
binoculata, etc.) have continued to rise. In 2008, 46% of
OSC was identified to the genus level and approximately

Stevenson and Lewis: Skate bycatch in the commercial groundfish fisheries of Alaska

211

Table 1

Species composition (% by weight) of observed skate catch by year reported in Alaska’s groundfish fisheries for 1998-
less than 0.1%.

-2008. * =

Taxon

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008 2004-2008

Skate unidentified

99.7

99.6

99.5

98.6

88.7

61.3

25.2

21.4

7.1

2.4

23.2

Bathyraja sp.

0.1

1.1

0.2

0.4

0.5

39.3

34.2

42.6

47.4

33.1

Bathyraja parmifera
(Alaska skate)

0.1

0.2

0.7

7.9

30.2

27.2

36.6

40.0

40.1

34.8

Bathyraja aleutica
(Aleutian skate)

0.1

0.7

2.2

2.6

1.9

2.5

2.7

2.4

Bathyraja interrupta
(Bering skate)

0.3

1.6

1.5

1.1

1.3

2.7

1.7

Bathyraja maculata
(whiteblotched skate)

0.1

1.1

0.4

0.7

0.5

1.0

0.7

Bathyraja lindbergi
(Commander skate)

0.1

0.2

0.1

0.2

Bathyraja taranetzi
(mud skate)

0.2

0.1

0.3

0.2

Bathyraja trachura
(roughtail skate)

0.1

Bathryaja minispinosa
(whitebrow skate)

Raja sp.

0.1

0.5

0.1

Raja binoculata
(big skate)

0.3

0.4

0.3

0.2

1.7

2.3

2.4

3.7

2.1

2.5

Raja rhina

(longnose skate)

0.2

0.5

1.0

1.9

1.0

1.1

52% was identified to species (i.e., Bathyraja parmifera
and other species).

The portion of the OSC that was identified to the spe-
cies level was dominated by Bathyraja parmifera, Raja
binoculata, and Bathyraja aleutica (Aleutian skate),
which accounted for 40.1%, 2.1%, and 2.7%, respectively,
of OSC in 2008 (Table 1). These proportions have re-
mained relatively stable since observers began identi-
fying skates in 2004, with B. parmifera, R. binoculata,
and B. aleutica averaging 34.8%, 2.5%, and 2.4%, re-
spectively, of the annual OSC from 2004 through 2008.
Seven other species of skates, including R. rhina and
six species of Bathyraja ( B . interrupta, B. maculata, B.
lindbergi, B. taranetzi, B. trachura, B. minispinosa ),
have been regularly reported in smaller proportions
by observers since 2004. Although unidentified skates
now constitute less than 5% of OSC, a large propor-
tion of skates are still identified only to the genus level
(“ Bathyraja sp.” and “ Raja sp.”).

The species composition of OSC varied by region and
by gear type within each region. During the 2004-08
period, Bathyraja parmifera was the most commonly
observed species in both the Bering Sea and Aleutian
Islands region (Table 2). In the Bering Sea, no other
single species made up more than 1.7% of OSC, and a
large percentage of skates were identified only to the
genus level. Species composition profiles were similar

for both types of trawl, but for fisheries using longline
gear a much higher percentage of skates were not iden-
tified to the species level.

In the Aleutian Islands, B. parmifera again accounted
for a higher proportion of OSC than any other species
(Table 2). However, notable proportions of B. maculata
and B. aleutica were reported in this region as well.
As in the Bering Sea, a large proportion of the skates
were not identified to the species level, and most of
the unidentified skates and skates identified to genus
were encountered in fisheries using longline gear. The
species composition profile for pelagic trawl gear in the
Aleutian Islands, with only two species reported and B.
interrupta accounting for over 80% of OSC, was mark-
edly different from any of the other region-gear combi-
nations reported in our study. However, that profile was
based on only two species composition samples in which
skates were reported.

The species composition of OSC was quite different
in the Gulf of Alaska, where the two species of Raja (R.
binoculata and R. rhina) are more common, accounting
for over half of OSC in the region (Table 2). Among
species of Bathyraja, B. aleutica accounted for the high-
est proportion in the Gulf of Alaska. The proportion of
skates not identified to the species level was consider-
ably lower in the Gulf of Alaska than in either the Ber-
ing Sea or Aleutian Islands, and the species composition

212

Fishery Bulletin 108(2)

Table 2

Species composition (% by weight) of observed skate catch by region and by gear type within each region of Alaska for 2004-2008.
Regions: BS=Bering Sea, AI=Aleutian Islands, GOA=Gulf of Alaska. Gear types: l=Nonpelagic trawl, 2=Pelagic trawl, 3=Long-
line. * = less than 0.1%.

Taxon

GOA

All

Skate unidentified

0.6

1.4

33.5

24.2

3.9

25.1

17.6

2.4

3.5

16.7

9.6

Bathyraja sp.

1.9

1.5

47.5

34.5

4.7

40.7

28.0

2.2

3.3

16.5

9.4

Bathyraja parmifera

90.0

92.8

14.9

36.6

50.9

18.0

29.6

3.6

9.4

2.8

3.2

Bathyraja aleutica

3.0

2.4

1.2

1.7

16.1

3.7

8.0

9.5

9.8

13.7

11.6

Bathyraja interrupta

1.3

1.7

1.6

1.1

80.6

0.4

0.7

5.9

6.3

3.0

4.4

Bathyraja minispinosa

0.1

Bathyraja maculata

0.4

0.2

17.2

9.3

12.1

0.2

0.1

Bathyraja lindbergi

0.1

0.2

1.7

1.1

0.2

0.1

Bathyraja taranetzi

0.2

0.1

5.2

19.4

0.6

2.2

Bathyraja trachura

0.1

0.2

1.4

0.7

Raja sp.

0.6

0.3

4.6

2.6

Raja binoculata

2.4

0.4

0.7

1.0

0.4

0.2

0.3

52.3

24.8

19.6

35.7

Raja rhina

0.1

23.4

42.6

21.3

22.4

Total

100

profiles varied more by gear type than in the other two
regions. All three gear types were dominated by species
of Raja, but R. binoculata accounted for over 50% of
OSC from nonpelagic trawl gear, whereas R. rhina was
the dominant species in pelagic trawl and longline gear.
As in the other two regions, proportions of unidentified
skates were much higher on longliners than on vessels
with other gear types, although a much higher percent-
age of skates were identified to the species level even
with longline gear in the Gulf of Alaska.

Significant amounts of skate bycatch were encoun-
tered by observers in fisheries targeting a variety of
commercial groundfish species, including Pacific cod,
walleye pollock, Atka mackerel, shallow-water flatfishes
(primarily yellowfin and rock soles), and others. During
the 1998-2008 study period, nearly 72% of OSC was re-
ported in longline fisheries, and over 65% was reported
in longline hauls targeting Pacific cod (Table 3). Non-
pelagic trawl fisheries accounted for only 22% of OSC,
most of which was reported in hauls targeting miscel-
laneous flatfishes. Pelagic trawl fisheries, essentially all
of which target walleye pollock, accounted for very little
of OSC (6%). These results reflect the percentages for
the Bering Sea, a region in which over 90% of OSC was
reported. In the Aleutian Islands significant numbers
of skates were also encountered on trawlers targeting
Atka mackerel, and in the Gulf of Alaska on trawlers
targeting deepwater flatfishes (arrowtooth flounder and
Greenland turbot).

The percentage of OSC retained by commercial fish-
ermen has increased over the past decade (Fig. 2). In
1998, overall mean skate retention was just over 12%,

and that figure steadily increased to a peak of nearly
40% in 2003. For the most recent 4 years (2005-08)
overall skate retention has remained relatively consis-
tent at around 30-35%. Species-level retention data
were erratic from 1998 through 2003. They have be-
come more stable since 2004 when observers began
consistently identifying skates to the species level, but
the annual mean retention for some of the species,
particularly the genus Raja, still appears relatively
inconsistent from year to year. Since 2004, the largest
species of skates ( Raja binoculata, R. rhina, Bathyraja
parmifera, B. aleutica, and B. maculata) have generally
been retained at 30% of OSC or above, and smaller spe-
cies, such as B. interrupta, B. lindbergi, B. taranetzi,
and B. minispinosa, have been retained at lower levels
(5-15%).

Discussion

From the inception of the NPGOP through 2003, field
identification tools for the skates of Alaska were limited,
and skate bycatch data were collected at a very basic
level. Almost all skates were reported by observers as
“skate unidentified.” However, from 2004 through 2008
this situation changed rapidly. With the development
and deployment of a field guide and the implementation
of an observer training protocol (Stevenson, 2004), the
proportion of skates identified to the species level has
increased dramatically. For the last year included in this
study, over 95% of OSC was identified at least to genus,
and that proportion may continue to rise in future years

Stevenson and Lewis: Skate bycatch in the commercial groundfish fisheries of Alaska

213

Table 3

Observed skate catch (in tons) by region, gear type, and target species reported in Alaska’s groundfish fisheries for 1998-2008.
Target species is defined as the predominant species (by % weight) in the catch. * = less than 100 tons.

Region
Gear type

Target species

>*/

\-y

cjT „

A/ /

.21

sfi

<?'>"

0°

<t%o"

o & cy

■ %•* %y'

Other

Total

1367

26,675

8967

2714

94,766

2067

793

5597

448

3445

286

2566

1823

32,187

9061

3794

102,929

5630

144,177

Bering Sea

Nonpelagic trawl

2606

18,400

2476

1677

Pelagic trawl

8912

Longline

90,314

206

967

461

Aleutian Islands

Nonpelagic trawl

491

Pelagic trawl

Longline

3850

256

497

Gulf of Alaska

Nonpelagic trawl

437

756

1393

Pelagic trawl

Longline

1486

443

All areas

Nonpelagic trawl

3534

19,177

2563

3123

101

Pelagic trawl

8967

Longline

95,650

208

1229

1401

Total

99,222

19,253

11,738

4366

1504

1021

1072

1077

443

262

744

843

140

319

493

544

as training methods and identification tools
are further refined.

Patterns of species composition in OSC
generally parallel recent biomass estimates
for regional skate populations derived from
bottom trawl surveys. Bathyraja parmifera
accounts for the large majority of OSC,
which is not surprising given that B. par-
mifera is the most abundant species of
skate encountered in bottom trawl surveys
conducted in Alaska waters (Stevenson et
al., 2008). In fact, B. parmifera is particu-
larly common on the Bering Sea continental
shelf, where its populations make up about
95% of the total skate biomass (Acuna and
Lauth, 2008; Lauth and Acuna, 2009) and
where commercial fishing effort for wall-
eye pollock, Pacific cod, and flatfishes is
concentrated. Many of the other species
encountered by observers in the Bering Sea
are recorded from fishing activity on the
upper continental slope, where B. aleutica,
B. maculata, and B. interrupts populations

100

All skates
q R. binoculata
o- B. rhina
• — B. parmifera
« — B. aleutica
+ — B. maculata

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Year

Figure 2

Overall mean percent retention of skate catch in commercial fisheries
for each year from 1998 through 2008 (gray bars), as well as mean
percent retention for Raja binoculata , R. rhina , Bathyraja parmifera,
B. aleutica, and B. maculata.

214

Fishery Bulletin 108(2)

are concentrated (Hoff and Britt, 2003, 2005, 2009;
Stevenson et al., 2008).

In the Aleutian Islands, over 50% of OSC consists
of B. parmifera, B. maculata, and B. aleutica (Table
2), which are the top three species in terms of recent
biomass estimates for the region (Zenger, 2004; Rooper,
2008; Rooper and Wilkins, 2008). However, the pro-
portion of B. parmifera is higher (29.6% of observed
skate catch) and that of B. maculata considerably lower
(12.1%) in commercial catches in the Aleutian Islands
than their biomass estimates in the region (20-25%
and 48% of total skate biomass, respectively) would
indicate. The reasons for these differences in relative
catch weight are unclear, but may be due to geographi-
cally and bathymetrically concentrated commercial fish-
ing effort. Skate populations in Alaska are primarily
segregated by depth, and B. maculata tends to be found
in deeper waters than those inhabited by B. parmifera
(Rooper, 2008; Stevenson et al., 2008). Therefore, shal-
low-water fisheries are more likely to catch B. par-
mifera, and although observers reported skates in the
Aleutian Islands from depths to 2000 m, the majority of
OSC came from 200 m or less. Thus, Aleutian popula-
tions of B. parmifera may be disproportionately affected
by fishing activity because of the shallow depth distri-
bution of this species.

The two species of Raja (and unidentified Raja — “ Raja
sp.”) account for over 60% of OSC in the Gulf of Alaska.
These results are also consistent with fishery-indepen-
dent survey data, which indicate that Raja binoculata
and R. rhina are the most abundant species in the
Gulf of Alaska, making up about 37% and 33%, respec-
tively, of the skate biomass in the region (Stevenson et
al., 2008; von Szalay et al., 2009). Among species of
Bathyraja in the Gulf of Alaska, survey-derived biomass
estimates indicate that B. aleutica is the most common,
and indeed B. aleutica accounts for a greater proportion
of OSC in this region than all other species of Bathyraja
combined.

Deepwater skate species, such as B. lindbergi, B.
minispinosa, and B. trachura, are rarely reported by
observers in any of the three regions, probably due to
the relatively small amount of fishing effort targeting
deepwater species. Other species known to be rare in
Alaska waters, such as B. abyssicola, B. mariposa, and
Amblyraja badia, have been only rarely reported by
observers, and only B. mariposa has been confirmed by
photographs and collected specimens.

Although the percentage of unidentified skates in
observer species composition data has declined to a very
low level, a large percentage of OSC is still identified
only to genus. These less specific skate identifications
are largely the result of uncertainty with identification
in the field. Because observers encounter a relatively
high diversity of skates, particularly of the genus Bathy-
raja, and must often interpret subtle characteristics to
identify skates to the species level, they are encour-
aged to identify a skate only to the genus level if the
specimen is not brought to hand for inspection or if
the identification of the specimen is questionable. As a

result, species composition of OSC is clearly affected by
fish-handling practices and observer sampling methods
on vessels with different gear types. Observers in trawl
fisheries select their species composition samples at
random from the catch after it is onboard the vessel.
Therefore, the entire composition sample is weighed,
and all specimens in the composition sample are iden-
tified in hand. In contrast, on longline vessels species
composition data are collected as the gear is being re-
trieved, and not all of the specimens in the composi-
tion sample are brought on board and weighed. Some
specimens counted during the tally period, particularly
larger species such as many of the skates common in
Alaska waters, become “drop-offs.” These specimens are
retrieved to the surface on the line but either fall off
before they can be brought onboard or are intentionally
released to save strain on the gear, the personnel, and
the fishes. Therefore, many of the skates in the compo-
sition sample from longline vessels are not brought to
hand for identification, and are recorded at the genus
level. Thus, the way the catch is handled and sampled
in longline fisheries largely explain the influence of
gear type on the species composition profiles reported
here (Table 2).

The influence of longline data is significant because
the majority of OSC in Alaska waters comes from long-
liners. In fact, the data presented here (Table 3) indi-
cate that the longline fishery for Pacific cod in the Ber-
ing Sea accounts for more skate bycatch than all other
federally managed groundfish fisheries combined. This
result must be interpreted with some caution because
differences in observer coverage for different fisheries
and regions may have influenced these figures, and
a predominant species is not a precise indicator of a
target fishery. But it is clear that longliners targeting
Pacific cod catch a lot of skates. Moreover, longline gear
is often fished deeper than trawl gear, and therefore
may affect a greater diversity of skate species than gear
fished in more shallow water because skate diversity in
Alaska waters tends to be highest on the continental
slope (Stevenson et al., 2008). Therefore, as long as a
high proportion of skates encountered on longliners are
identified only to genus, a potentially important seg-
ment of species-specific catch data is still not available
for analysis.

The presence of skates in the catch of pelagic trawls
may seem counterintuitive because skates are generally
benthic, substrate-oriented fishes unlikely to be found in
the path of midwater nets. Indeed, the amount of skate
catch reported in pelagic trawls (about 6% of OSC) is
much lower than in the other two gear types. There
are two general explanations for the skates that are
collected in pelagic nets: either the skates were swim-
ming up in the water column or the net contacted the
seafloor. The target of most pelagic trawling in Alaska
is walleye pollock, a species that is often found very
close to the bottom, and catch data from pelagic trawl-
ers often include a variety of benthic species, such as
flatfishes and sculpins, in addition to skates. Therefore,
it is likely that at least a large proportion of the skate

Stevenson and Lewis: Skate bycatch in the commercial groundfish fisheries of Alaska

215

Figure 3

Annual mean exvessel price paid by processors in Alaska for big skate
( Raja binoculata), longnose skate ( Raja rhina), and miscellaneous skates
from 1998 through 2007.

catch in pelagic trawls is the result of the net contact-
ing, or at least coming very close to, the seafloor.

Historically, skates have not been considered valuable
by Alaska’s commercial fishermen. Even though skates
are large fishes that represent a significant potential
source of protein, retention of skates in the commercial
fisheries of Alaska has been low. However, groundfish
observer data, coupled with exvessel pricing informa-
tion, may indicate that this situation is beginning to
change. Overall mean retention was less than 15% in
the late 1990s, and presumably before that time as well;
however, it has increased to 30-35% in recent years.
Species-level catch data collected since 2004 indicate
that the large species (such as both species of Raja,
Bathyraja parmifera , and B. aleutica) are retained at
a higher rate than smaller species, and that retention
rates for the large species are not necessarily consistent
from year to year. The general increase in retention
rates may reflect changes in the market value for skate
products. Although the mean exvessel price for general
skate catch has remained fairly stable over the past
decade (Fig. 3), the price paid to Alaskan fishermen
for big skates and longnose skates has risen sharply.
Since 2004, when processors began reporting landings
data by species owing to changes in the Fishery Man-
agement Plan for groundfish of the Gulf of Alaska, the
mean annual price paid for big and longnose skates has
nearly tripled.

Although the data presented here signify a dramatic
improvement in the information available to fishery
managers, some noteworthy gaps persist. The data pre-
sented here represent only sampled hauls on vessels
requiring observer coverage in federally managed fish-
eries, and therefore other sources of skate bycatch are
not represented. Commercial fishing activity in the Ber-
ing Sea and Aleutian Islands is conducted primarily on
large vessels, which are required to have 100% observer
coverage, and therefore observer data should provide a

good representation of skate bycatch in those regions.
In contrast, many of the commercial vessels operating
in the Gulf of Alaska are small enough that observer
coverage is only required on 30% of fishing days or is
not required at all. Therefore, observer data for this
region may provide much less reliable estimates of skate
bycatch. Because the two species of the genus Raja are
common in the Gulf of Alaska, and are among the larg-
est skate species in the region, the unobserved catch of
those species is of particular concern. Disproportionate
retention of larger skates is prevalent in many fisheries
worldwide, and as larger, more vulnerable species are
removed, smaller species may become more abundant
(Russ, 1991; Agnew et al., 2000; Cedrola et al., 2005;
Swain et al., 2005). In the North Atlantic, severe reduc-
tion in biomass for some larger, less resilient skate spe-
cies has been accompanied by an increased biomass for
smaller, more resilient species (Casey and Myers, 1998;
Walker and Hislop, 1998; Dulvy et al., 2000). Species-
specific observer data on skate bycatch can document
this phenomenon, but only if the data are representa-
tive of total fishing effort. Therefore, undocumented
sources of skate bycatch, as well as nonspecific data
from observed longline fisheries (see above comments on
longline species composition data), present significant
remaining challenges to fishery managers.

Observer data on skate bycatch in the groundfish fish-
eries of Alaska represent a rich source of information
for managers charged with protecting skate populations
from future overexploitation. The species-level catch
data now being collected by observers have facilitated
the development of an age-structured stock assessment
model for B. parmifera (B. Matta, personal commun.2),
which is a critical aid in setting appropriate catch lim-
its for the species, and similar models for other species

2 Matta, Beth. 2009. NMFS Alaska Fisheries Science Center,
Seattle, WA 98115.

216

Fishery Bulletin 108(2)

are on the horizon. These fishery-dependent data can
now be compared directly with fishery-independent sur-
vey data, creating two independent lines of evidence for
management strategies. Specific catch data may also
be used to identify areas in which the most vulnerable
species may be most heavily impacted and thus can
help identify areas in which restrictions or closures
are necessary. Although observer data do not give a
complete account of skate bycatch in the fisheries of
Alaska, the information currently provided allows this
diverse assemblage of species to be managed in a more
biologically appropriate way than was possible in the
past. As fishing pressure on Alaska’s skate populations
increases, the consequences of data deficiencies will be
magnified, and observer data will play an increasingly
important role in protecting skates from the declines in
biomass and shifts in community structure that have
befallen these fishes in other parts of the world.

Acknowledgments

We thank the multitude of staff and observers of the
North Pacific Groundfish Observer Program that have
helped to collect the data used here. We also thank S.
Gaichas, O. Ormseth, and B. Matta for discussions about
skate stock assessments, R. Narita for assistance with
data retrieval, and T. Hiatt for providing skate price
information. For comments on an early draft of the
manuscript, we thank M. Loefflad, B. Mason, B. Matta,
P. Nelson, O. Ormseth, and J. Orr.

Literature cited

Acuna, E., and R. R. Lauth.

2008. Results of the 2007 eastern Bering Sea conti-
nental shelf bottom trawl survey of groundfish and
invertebrate resources. NOAA Tech. Memo. NMFS-
AFSC-181, 195 p.

Agnew, D. J., C. P. Nolan, J. R. Beddington, and R. Baranovski.
2000. Approaches to the assessment and management of
multispecies skate and ray fisheries using the Falkland
Islands fishery as an example. Can. J. Fish. Aquat.
Sci. 57:429-440.

Brander, K.

1981. Disappearance of common skate Raia batis from
Irish Sea. Nature 290:48-49.

Casey, J. M., and R. A. Myers.

1998. Near extinction of a large, widely distributed
fish. Science 281:690-691.

Cedrola, P. V., A. M. Gonzalez, and A. D. Pettovello.

2005. Bycatch of skates (Elasmobranchii: Arhyncho-
batidae, Rajidae) in the Patagonian red shrimp
fishery. Fish. Res. 75:141-150.

Dulvy, N. K., J. D. Metcalfe, J. Glanville, M. G. Pawson, and
J. D. Reynolds.

2000. Fishery stability, local extinctions, and shifts in com-
munity structure in skates. Conserv. Biol. 14:283—293.
Ebert, D. A.

2005. Reproductive biology of skates, Bathyraja (Ishi-
yama), along the eastern Bering Sea continental
slope. J. Fish Biol. 66:618-649.

Ebert, D. A., W. D. Smith, and G. M. Cailliet.

2008. Reproductive biology of two commercially ex-
ploited skates, Raja binoculata and Raja binoculata, in
the western Gulf of Alaska. Fish. Res. 94:48-57.

Hoff, G. R., and L. L. Britt.

2003. The 2002 eastern Bering Sea upper continental slope
survey of groundfish and invertebrate resources. NOAA
Tech. Memo. NMFS-AFSC-141, 261 p.

2005. Results of the 2004 eastern Bering Sea upper
continental slope survey of groundfish and inverte-
brate resources. NOAA Tech. Memo. NMFS-AFSC-156,
276 p.

2009. Results of the 2008 eastern Bering Sea upper
continental slope survey of groundfish and invertebrate
resources. NOAA Tech. Memo. NMFS-AFSC-197,
294 p.

Ishihara, H., and R. Ishiyama.

1985. Two new North Pacific skates (Rajidae) and a
revised key to Bathyraja in the area. Jpn. J. Ichthyol.
32:143-179.

1986. Systematics and distribution of the skates of the
North Pacific (Chondrichthyes, Rajoidei). In Indo-Pacific
fish biology: proceedings of the second international
conference on Indo-Pacific fishes (T. Uyeno, R. Aria, T.
Taniuchi, and K. Matsuura, eds.), p. 269-280. Ichthyol.
Soc. Japan, Tokyo.

Ishiyama, R., and H. Ishihara.

1977. Five new species of skates in the genus Bathy-
raja from the western North Pacific, with reference
to their interspecific relationships. Jpn. J. Ichthyol.
24:71-90.

Laptikhovsky, V. V.

2004. Survival rates for rays discarded by the bottom
trawl squid fishery off the Falkland Islands. Fish.
Bull. 102:757-759.

Lauth, R. R., and E. Acuna.

2009. Results of the 2008 eastern Bering Sea conti-
nental shelf bottom trawl survey of groundfish and
invertebrate resources. NOAA Tech. Memo. NMFS-
AFSC-195, 229 p.

Matta, M. E.

2006. Aspects of the life history of the Alaska skate,
Bathyraja parmifera, in the eastern Bering Sea. M.S.
thesis, 92 p. Univ. Washington, Seattle, WA.

Ormseth, O., B. Matta, and J. Hoff.

2009. Bering Sea and Aleutian Islands skates. In Stock
assessment and fishery evaluation report for the ground-
fish resources of the Bering Sea/Aleutian Islands region,
chapter 18a, p. 1087-1178. [Available from North Pacific
Fishery Management Council, 605 West 4th Ave, Suite
306, Anchorage, AK 99501.]

Rooper, C. N.

2008. Data report: 2006 Aleutian Islands bottom trawl
survey. NOAA Tech. Memo. NMFS-AFSC-179, 237 p.

Rooper, C. N., and M. E. Wilkins.

2008. Data report: 2004 Aleutian Islands bottom
trawl survey. NOAA Tech. Memo. NMFS-AFSC-185,
207 p.

Russ, G. R.

1991. Coral reef fisheries: effects and yields. In The
ecology of fishes on coral reefs, p. 601-635. Academic
Press, San Diego, California.

Stevens, J. D., R. Bonfil, N. K. Dulvy, and P. A. Walker.

2000. The effects of fishing on sharks, rays, and chimae-
ras (chondrichthyans), and the implications for marine
ecosystems. ICES J. Mar. Sci. 57:476-494.

Stevenson and Lewis: Skate bycatch in the commercial groundfish fisheries of Alaska

217

Stevenson, D. E.

2004. Identification of skates, sculpins, and smelts by
observers in North Pacific groundfish fisheries (2002-
2003). NOAA Tech. Memo. NMFS-AFSC-142, 67 p.

Stevenson, D. E., J. W. Orr, G. R. Hoff, and J. D. McEachran.

2004. Bathyraja mariposa : a new species of skate (Rajidae:
Arhynchobatinae) from the Aleutian Islands. Copeia
2004:305-314.

2007. Field guide to sharks, skates, and ratfish of Alaska,
77 p. Alaska Sea Grant College Program, Fairbanks,
AK.

2008. Emerging patterns of species richness, diversity,
population density, and distribution in the skates (Raji-
dae) of Alaska. Fish. Bull. 106:24-39.

Stobutzki, I. C., M. J. Miller, D. S. Heales, and D. T. Brewer.

2002. Sustainability of elasmobranchs caught as bycatch
in a tropical prawn (shrimp) trawl fishery. Fish. Bull.
100:800-821.

Swain, D. P., T. Hurlburt, and H. P. Benoit.

2005. Changes in the abundance and size of skates in the
Southern Gulf of St. Lawrence, 1971-2002. J. Northw.
Atl. Fish. Sci. 36:19-30.

von Szalay, P. G., M. E. Wilkins, and M. M. Martin.

2009. Data report: 2007 Gulf of Alaska bottom trawl
survey. NOAA Tech. Memo. NMFS-AFSC-189,
247 p.

Walker, P. A., and J. R. G. Hislop.

1998. Sensitive skates or resilient rays? Spatial and
temporal shifts in ray species composition in the central
and north-western North Sea between 1930 and the
present day. ICES J. Mar. Sci. 55:392-402.

Zenger, H. H., Jr.

2004. Data report: 2002 Aleutian Islands bottom
trawl survey. NOAA Tech. Memo. NMFS-AFSC-143,
247 p.

218

Effects of starvation on energy density
of juvenile chum salmon ( Oncorhynchus keta )
captured in marine waters
of Southeastern Alaska

Emily A. Fergusson (contact author)

Molly V. Sturdevant
Joseph A. Orsi

E-mail address for contact author: emily.fergusson@noaa.gov

Auke Bay Laboratories
Alaska Fisheries Science Center
National Marine Fisheries Service
17109 Point Lena Loop Road
Juneau, Alaska 99801

Abstract — We conducted laboratory
starvation experiments on juvenile
chum salmon ( Oncorhynchus keta)
captured in the neritic marine waters
of northern Southeast Alaska in June
and July 2003. Temporal changes in
fish energy density (whole body energy
content [WBEC], cal/g dry weight),
percent moisture content, wet weight
(g), length (mm), and size-related con-
dition residuals were measured in the
laboratory and were then compared
to long-term field data. Laboratory
water temperatures and salinities
averaged 9°C and 32 psu in both
months. Trends in response variables
were similar for both experimental
groups, although sampling intervals
were limited in July because fewer
fish were available (n = 54) than in
June (« = 101). Overall, for June (45-
d experimental period, 9 intervals),
WBEC, wet weight, and condition
residuals decreased and percent
moisture content increased, whereas
fork length did not change. For July
(20-d experimental period, 5 inter-
vals), WBEC and condition residuals
decreased, percent moisture content
and fork length increased, and wet
weight did not change. WBEC, per-
cent moisture content, and condition
residuals fell outside the norm of long-
term data ranges within 10-15 days
of starvation, and may be more useful
than fork length and wet weight for
detecting fish condition responses to
suboptimal environments.

Manuscript submitted 19 May 2009.
Manuscript accepted 20 January 2010.
Fish. Bull. 218-225(3020).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

Energy density is an important mea-
sure of fish nutritional condition and
is used to assess growth, construct
energy budgets, and measure energy
flow in ecosystems (Brett et al., 1969;
Jobling, 1994; Ban et al., 1996; Edsall
et al., 1999). Energy density is also
a critical parameter for bioenergetic
models (Orsi et al., 2004; Trudel et al.,
2005; Wuenschel et al., 2006; Breck,
2008). Along with other measures of
fish condition, such as body composi-
tion, growth, and length-weight condi-
tion indices, energy density integrates
and reflects the history of fish feed-
ing environments before the time of
sampling (LeBrasseur, 1969; Edsall et
al., 1999; Breck, 2008). During good
feeding periods, fish condition will be
high, whereas the reverse is expected
during poor feeding periods as energy
reserves are depleted to maintain
standard metabolic needs (Jobling,

1994) . However, an examination of
how quickly energy density responds
during periods of poor feeding that are
usually associated with low growth
has been limited to a few studies. In
general, a balanced energy budget is
expressed as the equation: ingestion
= metabolism + growth + excretion,
which outlines how an energy source
is used by an organism and what pro-
portion is allocated to each component
of the equation (Jobling, 1994; Brett,

1995) . These allocations depend on the
initial amount of energy, as well as the
environmental conditions that affect

physiological rates, such as tempera-
ture and salinity (Brett et al., 1969;
Hoar, 1988; Jobling, 1994). When fish
are starved, growth typically ceases
and energy density declines; when
energy stores are used, the percent-
ages of fat and protein in the fish
decrease as the relative water content
increases (Brett, 1995; Breck, 2008).
Changes in fish energy density may be
more detectable on small scales than
other fish parameters, such as growth,
during periods of poor feeding condi-
tions in marginal habitats.

Juvenile Pacific salmon (Oncorhyn-
chus spp.) use transitional habitats
along their seaward migration from
near shore to the open ocean and
can experience rapid environmen-
tal changes that may affect growth
and energy allocation (Orsi et al.,
2000; Cross et al., 2008). Fish tran-
sit these demanding environments at
the same time that they are experi-
encing increasing energy demands
while undergoing ontogenetic changes
in metabolic rate related to salinity
and smoltification (Hoar, 1998). These
transitional habitats are presumed to
be critical feeding areas because prey
fields also change dramatically, and
juvenile salmon are often found in
the presence of planktivorous forage
fish species that potentially impact
carrying capacity (Purcell and Stur-
devant, 2001; Park et al., 2004; Orsi
et al., 2004). Therefore, understand-
ing how changes in juvenile salmon

Fergusson et al.: Effects of starvation on energy density of Oncorhynchus keta

219

energy density reflect habitat quality may give insight
into factors that affect their growth and survival, par-
ticularly if food resources may be limited during this
critical time in their life history (Paul and Willette,
1997; Boldt and Haldorson, 2004; Cross et ah, 2008).

We initiated a study to measure changes in condition
of juvenile chum salmon ( O . keta) captured at sea and
later denied food resources in the laboratory. In previ-
ous studies on fish starvation, juvenile chum salmon
were reared entirely in the laboratory (LeBrasseur,
1969; Akiyama and Nose, 1980; Murai et al., 1983;
Ban et al., 1996); however, in our study they experi-
enced variable environmental conditions at sea before
being captured and transported back to the labora-
tory. Thus, these salmon from field collections represent
natural variation of fish in marine waters better than
fish reared in controlled laboratory environments. Our
primary objective was to measure changes in energy
density, moisture content, weight, length, and a size-
related condition residual index for field-caught juvenile
chum salmon in response to starvation in the labora-
tory over time. We also compared the condition of these
experimentally starved fish to that determined from a
long-term data series on field-caught fish 1) to assess
the range of normally occurring condition values and 2)
to identify the length of time before experimental values
fell outside the observed range.

Methods

Juvenile chum salmon for the experiments were cap-
tured in the vicinity of Icy Strait (58°N latitude, 135°W
longitude) about 50 km west of Juneau, Alaska, in June
and July 2003. Fish were obtained during the South-
east Alaska Coastal Monitoring (SECM) Project long-
term annual survey of juvenile salmon by the Auke Bay
Laboratories (ABL) aboard the NOAA ship John N. Cobb
(Orsi et al., 2004). Juvenile chum salmon were collected
from the neritic waters of Icy Strait and Upper Chatham
Strait, along the primary seaward migration corridor
in the northern region of Southeast Alaska (Orsi et al.,
2000, 2004). Preliminary observations along this corridor
showed that juvenile chum salmon exhibit approximately
a five-fold increase in body length, 100-fold increase in
weight, 25% increase in energy density, and more than
6% decline in body moisture content between May and
September. We used fish from this locality in June and
July, the periods of highest abundance and greatest
interaction with other juvenile salmon species. In June,
fish were captured with a Kodiak pair-trawl fished at 1
m/sec for 10 min (Mortensen et al., 2000). In July, fish
were captured with a Nordic 264 rope trawl fished at
1.5 m/sec for 20 min (Orsi et al., 2000). All fish caught
were immediately transferred from the trawl codend to
static live tanks containing sea water. Juvenile chum
salmon were then identified and sorted into flow-through
“live” tanks. The sea water for the tanks was pumped
from a depth of 3 m and then filtered to prevent feeding
on zooplankton prey. Before transfer to the laboratory,

the juvenile chum salmon were held onboard for one
day in June and four days in July while the surveys
were completed. To establish a baseline for the start
of the starvation experiments, on the day of capture a
subsample of fish were measured (fork length, FL, mm)
and frozen (-5°C) for later laboratory analysis. Daily
temperature and salinity measurements were recorded
and averaged 11.4°C and 26.1 psu in June and 12.7°C
and 23.2 psu in July.

In the laboratory, the juvenile chum salmon were
placed in two living-stream tanks (Frigid Units, Inc.,
Toledo, OH) (200x50x48 cm) with screened baffles sepa-
rating the inflow and outflow pipes. One unit was allo-
cated the salmon captured in June; the other unit — the
salmon captured in July. Ambient sea water from a
25-m depth in Auke Bay was supplied to the tanks at
a rate of 3 L/min. Daily temperature and salinity mea-
surements were recorded in the laboratory tanks and
averaged 8.6°C and 31.7 psu for June and 8.6°C and
32.1 psu for July. Sea water was filtered to prevent feed-
ing on zooplankton prey. The fish were not subjected to
any strong currents that would increase activity costs.
To best mimic the photoperiod in the natural environ-
ment at the time of capture, light conditions in the labo-
ratory were set at a standard eight hours of darkness,
one hour of dusk, one hour of dawn, and 14 hours of
daylight. Subsamples of 10-15 fish were removed from
the tank at predetermined intervals and sacrificed with
an overdose of tricaine methanesulfonate (MS-222),
then frozen (-5°C) individually for later size and calo-
rimetric analyses. Fish that had died between sacrifice
intervals were not included in the experiments.

Frozen juvenile chum salmon were processed for data,
including energy density in terms of whole body energy
content (WBEC, cal/g wet weight [WW]), dry weight
(DW, mg), percent moisture content (%<MC), FL, and wet
weight (mg). After excising each stomach and removing
and weighing its contents, we dried the fish to obtain
DW (full gut minus empty gut, nearest mg) so that un-
digested prey would not bias the final values. Stomachs
examined from fish sacrificed after the first time inter-
val were devoid of prey and therefore stomachs were not
excised in subsequent time intervals. All viscera were
replaced in the body cavity before the fish were dried to
a stable weight (<5 mg change), requiring a minimum
of 48 hours at 55°C. The DW was recorded and %MC
of each fish was calculated as ([1 -DW/WW] x 100).
Each fish was homogenized with a Waring pulverizer,
then finely ground with a mortar and pestle to yield
a uniform powder. Susamples of 15 mg were formed
into pellets with a pellet press and stored in a desic-
cator to prevent rehydration. A 1425 Parr micro-bomb
calorimeter was used to obtain cal/g DW for each fish;
this measure was converted to WBEC by multiplying by
DW/WW. Estimates of WBEC from replicate subsamples
were consistent (<2% coefficient of variation). To ac-
count for potential effects of size variation on WBEC
and %MC, size-related condition residuals (CR) were
calculated by using the ln-transformed experimental
FL and WW measures for each fish. We first derived

220

Fishery Bulletin 108(2)

a regression equation from all paired ln-weights and
ln-lengths (rc = 8475; -700 per year) of field-caught ju-
venile chum salmon collected during June-August for
the SECM project from 1997 to 2008. We then used
this regression equation to predict ln(WW) for each
experimental ln(FL). Finally, we obtained the CRs by
subtracting the predicted ln(WW) from the observed
ln(WW) (Jakob et al., 1996; Brodeur et al., 2004).

To account for potential stock-related differences
in condition of the experimental chum salmon (of un-
known stocks), WBEC was determined for additional
field-caught fish of known stocks. Historically, between
70% and 90% of fish caught in June originated from
Macaulay Hatchery (MH), whereas mixed hatchery
stocks were present during July (Orsi et al., 2004).
Otoliths were not retained from the fish used in the
experimental groups; however, stock of origin was de-
termined from thermal marks present on the otoliths
of juvenile chum salmon captured in the study area in
July and these marks indicated that the fish were from
unmarked stocks (UM, presumably wild) and MH and

Hidden Falls Hatchery (HF) stocks. Both hatcheries
mark 100% of chum salmon released. Energy densities
were determined (as described above) for these three
stock groups.

One-way analyses of variance (ANOVA) were used for
initial statistical analyses to compare WBEC, %MC,
FL, and WW of fish across sampling intervals for each
experimental group and for July stock groups. If sig-
nificant differences were detected, Tukey’s paired com-
parison tests were performed to identify the interval
in which they were found. We used graphical analyses
to compare the WBEC and %MC for each experimen-
tal group to the norms (one standard deviation about
the mean) derived from the entire SECM field data
set (1997-2008) from June and July (n = 1257; WBEC:
993.4 ±72.3 and %MC: 79.4 ±1.2). The temporal data
from the experiments were compared to these norms
to identify the duration of starvation before the ex-
perimental measures fell outside the long-term range
of field values.

Days starved

Figure t

Average whole-body energy content (WBEC, cal/g wet weight) and one
standard error about the mean for juvenile chum salmon ( Oncorhyn -
chus keta) starved over time in the laboratory after capture in
the marine waters of Icy Strait and Upper Chatham Strait in the
northern region of southeastern Alaska, June and July 2003. The
grey band indicates one standard deviation about the mean for
all field-caught juvenile chum salmon examined for WBEC during
the Southeast Coastal Monitoring project, June-July (n = 1257),
1997-2008. Significant differences (Tukey’s paired comparisons;
P<0.05) and percent change between sample intervals are shown
in inset boxes.

Results

The numbers of juvenile chum salmon
obtained for the two starvation trials included
101 fish for June and 54 fish for July. The
higher number of juvenile chum salmon
available in June allowed nine experimental
time intervals to be tested, spanning 45 days
(mean of five days per interval, range of 1-16
days between intervals). The smaller number
of juvenile chum salmon available in July
allowed only five experimental time intervals
to be tested, spanning 20 days (mean of five
days per interval, range of 1-10 days between
intervals). Both experimental groups had
common intervals at about 10 and 20 days.
Mortality between sampling intervals was
minimal in both groups: 13 fish died in June
(70% during the first 10 days of the experi-
ment) and two died in July (both during the
first 2 days).

The energy content of juvenile chum salm-
on declined over time in both experimental
groups (Fig. 1). Initial WBEC was significant-
ly higher in June than in July (1081.2 cal/g
WW compared to 960.5 cal/g WW; P<0.001).
For the June sample group, WBEC decreased
significantly (P<0.001) by 19% between days
zero and 19 and by 40% between days zero
and 45; see table insets in figures for signifi-
cant differences (Tukey’s paired comparisons)
between intervals. For the July sample group,
WBEC decreased significantly (P<0.001) by
11% between days zero and 20. Overall, the
relative loss of energy content was almost
twice as great in June as in July at day 20.

In contrast to WBEC, %MC of juvenile
chum salmon increased over time in both

Fergusson et al.: Effects of starvation on energy density of Oncorhynchus keta

221

Table 1

Average fork length (FL, mm), wet weight (WW, g), percent moisture content (%MC, [(1 -dry weight/WW)x 100]), and whole
body energy content (WBEC, cal/g WW), for unmarked (presumably wild) and hatchery stock groups of juvenile chum salmon
( Oncorhynchus keta ) captured in the marine waters of Icy Strait and Upper Chatham Strait in the northern region of southeast-
ern Alaska, July 2003. Standard errors are given in parentheses.

Stock group

%MC

WBEC

Unmarked

120 (1.7)

17.5 (0.8)

80.4 (0.1)

954.0(5.7)

Macaulay Hatchery

137 (3.0)

29.0(1.5)

80.3 (0.2)

957.5 (14.3)

Hidden Falls Hatchery

127 (2.9)

22.1 (1.6)

80.4 (0.1)

959.5 (9.6)

experimental groups (Fig. 2). Initial %MC was
significantly lower (P<0.001) in June than in
July (77.8% compared to 80.1%). For the June
sample group, %MC increased significantly
(P<0.001) by 4% between days zero and 19 and
by 9% between days zero and 45. For the July
sample group, %MC increased significantly
(P<0.001) by 1% between days zero and 20.
Overall, the increase in %MC was four times
as great in June as in July at day 20.

Changes in the WW and FL of juvenile chum
salmon over time were not consistent between
the experimental groups (Fig. 3). For WW,
initial values did not differ (P>0.05) between
June and July (14.2 compared to 13.6 g). For
the June sample group, WW decreased signifi-
cantly (PcO.Ol) by 39% between days zero and
45. For the July sample group, no significant
(P>0.05) differences in WW were observed.
Similarly, initial FL values did not differ
(P>0.05) between June and July (112 com-
pared to 110 mm). For the June sample group,
FL did not change significantly (P>0.05) be-
tween days zero and 45. For the July sample
group, FL increased significantly (PcO.001) by
19% between days zero and 20.

The CR of juvenile chum salmon became
increasingly negative over time in both ex-
perimental groups (Fig. 4). Initial CRs were
positive in both months, but June CRs were
lower than those for July. For the June sample
group, CR declined significantly (P<0.001) be-
tween days zero and 19 and between days
zero and 45. For the July sample group, CR
declined significantly (P<0.001) between days
zero and 20. Mean CRs shifted from positive
to negative after approximately 10 days of

Days starved

Figure 2

Average percent moisture content (%MC, [(1-dry weight / wet
weight) x 100]) and one standard error about the mean for juve-
nile chum salmon ( Oncorhynchus keta) starved over time in the
laboratory after capture in the marine waters of Icy Strait and
Upper Chatham Strait in the northern region of southeastern
Alaska, June and July 2003. The grey band indicates one stan-
dard deviation about the mean for all field-caught juvenile chum
salmon examined for %MC during the Southeast Coastal Monitoring
project, June-July (n = 1257), 1997-2008. Significant differences
(Tukey’s paired comparisons; P<0.05) and percent change between
sample intervals are shown in inset boxes.

starvation in each sample group and continued to de-
cline, indicating increasingly poor condition for a given
size fish.

Hatchery stock group did not affect the WBEC or
%MC of the July-caught juvenile chum salmon. A total
of 33 fish were examined: UM (n- 13), MH (n=10), and
HF (n = 10) (Table 1). Stock had no effect on WBEC or
%MC (P>0.05). However, WW and FL did differ signifi-

cantly (P<0.001) among stocks and were highest for the
MH stock and lowest for the UM stock (Table 1).

Discussion

To our knowledge, this is the first published study of
the change in energy density and %MC of field-captured

222

Fishery Bulletin 108(2)

Days starved

Figure 3

Average fork length (mm, top panels) and wet weight (g, bottom panels) for juvenile chum
salmon ( Oncorhynclius keta) starved over time in the laboratory after capture in the
marine waters of Icy Strait and Upper Chatham Strait in the northern region of south-
eastern Alaska, during June (left panels) and July (right panels) 2003. Error bars are
one standard error about the mean. Significant differences (Tukey’s paired comparisons;
P<0.05) and percent change between sample intervals are shown in inset boxes.

juvenile chum salmon during starvation. Limited infor-
mation has been published on the changes in condition of
laboratory-reared chum salmon due to starvation. Such
studies typically show depletion of stored nutrients and
declines in condition and size over time, despite differ-
ences in methods (LeBrasseur, 1969; Akiyama and Nose,
1980; Murai et ah, 1983; Ban et al., 1996). For nutrient
responses, lipid and serum protein levels of laboratory-
reared juvenile chum salmon were lowest after 10 and
20 days of starvation, respectively (Ban et ah, 1996);
unfortunately, however, energy content was not deter-
mined. We did not directly measure lipid and protein,
but the decline in WBEC that we observed between days
zero and 10 and between days 20 and 45 in June could
reflect similar declines in these nutrient measures. For
condition responses, two studies showed that %MC of
small starved juvenile chum salmon increased by 4.3%
(41 mm and 0.45 g initial size; 42-d starvation; LeBras-
seur, 1969) to 5.4% (0.26 g initial size; 28-d starvation;
Murai et ah, 1983) at ~15°C; another study showed that
% MC of larger starved juvenile chum salmon increased
by 12% (94.5 mm and 7.9 g initial size; 91-d starvation;
Akiyama and Nose, 1980) at 17°C. Trends in %MC of
our juvenile chum salmon were comparable despite the
differences in fish size, duration of starvation, and water

temperature. For size responses, weight decreased for
five size-groups of juvenile chum salmon (0.46-7.95 g
initial size; 5-13 wk starvation); however, the percentage
weight loss decreased as fish size increased (Akiyama
and Nose, 1980). These differences in weight loss among
fish sizes indicate that physiological responses to starva-
tion may vary with ontogeny.

Our results are also comparable to information avail-
able for other salmonid species and stages. For starved
juvenile sockeye salmon (O. nerka), energy density de-
clined more rapidly and %MC increased more rapidly
with increasing temperatures (Brett et ah, 1969). In our
study, chum salmon in June exhibited a 40% decline in
WBEC and a 9% increase in %MC after 45 days of star-
vation at an average temperature of ~9°C. By compari-
son, at similar temperatures (10°C), laboratory-reared
juvenile sockeye salmon lost 37% of initial WBEC and
gained 9% MC during 99 days of starvation (Table 3 in
Brett et al., 1969). Such inverse relationships between
fraction water and fraction lipid or energy content are
often reported during starvation (Miglavs and Jobling,
1989; Simpkins et al., 2004; Breck, 2008). In a few
studies, size changes similar to those that we observed
have also been reported among other starved salmo-
nids. Weight decreased for starved juvenile Arctic charr

Fergusson et al.: Effects of starvation on energy density of Oncorhynchus keta

223

0.6-1

0.4-

°'2 1
0.0

-0.2-

-0.4-

June

! i

1 i

ra -0.6J

Figure 4

Condition residuals (CR) for individual juvenile chum salmon (Oncorhyn-
chus keta) starved over time in the laboratory after capture in the marine
waters of Icy Strait and Upper Chatham Strait in the northern region
of southeastern Alaska, June and July 2003. The CRs were calculated
by using the In-transformed experimental fork length and wet weight
measures for each fish in a regression equation derived from all paired
ln-weights and ln-lengths of field-caught juvenile chum salmon col-
lected during the Southeast Coastal Monitoring project, June-August
(n = 8476) from 1997 to 2008. The 0.0-line represents the expected CR
of an average fish; therefore, positive values indicate above average
condition and negative values indicate below average condition.

( Salvelinus alpinus; Miglavs and Jobling,

1989), rainbow trout ( O . mykiss; Simp-
kins et al., 2004), and Atlantic salmon
(Salmo salar\ Stefansson et al., 2009) for
starvation periods of 4-6 weeks. Length
and weight of small (30.1-mm and 0.14-g)
sockeye salmon decreased significantly
after 14-49 days of starvation in colder
water (7.9°C; Bilton and Robins, 1973)
than that used in our experiment. Like
the salmonids in the above studies,
weight of our juvenile chum salmon de-
creased for the June experimental group,
but similar conclusions about the July
fish could not be reported because of the
shorter experimental period.

The chum salmon caught in June ini-
tially had approximately 11% higher
WBEC and approximately 3% lower %MC
than fish caught in July — differences that
could be accounted for by both environ-
mental and biological variables. In both
the June and July experimental groups,
a measurable increase in WBEC and de-
crease in %MC occurred between days
zero and one. These changes may have
been attributed to a physiological stress
response that caused the fish to lose wa-
ter and therefore increased the relative
WBEC and decreased the %MC (Breck,

2008). Temperature and salinity both af-
fect fish physiological rates and influence
ingestion, metabolism, and growth (Brett
et al., 1969; Mason, 1974; Sheridan et al.,

1983; Jobling, 1994; Weatherley and Gill,

1995). In our study, field temperature
was cooler and salinity was higher in
June (11°C; 26 psu) than in July (13°C;

23 psu), but fish captured in both months
were transferred into identical, colder
(9°C) and more saline (32 psu) environments in the
laboratory. Monthly differences in temperature and
salinity were therefore eliminated as variables in the
experiments. However, the fish captured in June had
probably smolted more recently (Zaporozhec and Za-
porozhec, 1993; Hoar, 1998) and spent less time in the
marine environment, and probably had lower growth
rates (Orsi et al., 2000) and energy requirements than
fish captured in July, when it was warmer.

We accounted for potential size-related effects on
WBEC and %MC by using length-weight regression
analysis, which corrected for natural variation in fish
size; however, the results may still be misleading be-
cause this regression did not account for differences in
actual nutritional status or body composition, such as
protein, lipid, and water content (Miglavs and Jobling,
1989; Edsall et al., 1999; Kotiaho, 1999; Trudel et al.,
2005; Congleton and Wagner, 2006). Length-weight re-
gression analysis is useful for initially identifying con-
dition in relation to a long-term index and to anticipate

trends in energy density, but to account for changes in
nutritional status or body composition WBEC, %MC,
or proximate composition, should be used to verify the
CR results.

In our study, stocks of juvenile chum salmon sampled
from the same habitat did not differ in WBEC or %MC,
but size did differ significantly. By comparison, for ju-
venile pink salmon (O. gorbuscha ) captured together
in marine habitats of Prince William Sound, Alaska,
differences in length and WBEC between stock groups
have not been consistent (Paul and Willette, 1997; Boldt
and Haldorson, 2004; Cross et al., 2008). For fish ~80
mm in length, the occurrence of length differences be-
tween juvenile pink salmon stocks depended on the size
of hatchery fish at time of release (Cross et al., 2008).
In a concurrent study, juvenile pink salmon length dif-
fered between stock groups, but WBEC did not (Boldt
and Haldorson, 2004). Conversely, energy content (so-
matic) of smaller juvenile pink salmon (~35 mm) did
differ between stock groups (Paul and Willette, 1997).

224

Fishery Bulletin 108(2)

These studies, along with ours, support the idea that
different stock groups of juvenile salmon may have
similar WBEC in common habitats despite stock-specific
size differences, and thus emphasize the importance of
habitat quality on fish condition. These different results
could also be related to ontogenetic changes in physiol-
ogy (Hoar, 1998; Wuenschel et al., 2006).

Because so little mortality occurred within each ex-
perimental group, we conclude that juvenile salmon
can survive for prolonged periods without food during
the summer months, as has also been reported by Ste-
fansson et al. (2009). Most of the mortalities occurred
within the first eight days of the June experiment. As
discussed previously, the June fish were younger and
less robust (lower CR) and could have been more sus-
ceptible to environmental stresses because of scale loss
(Bouck and Smith, 1979) from net abrasion during cap-
ture, for example. However, even though juvenile chum
salmon were still alive after 45 days of starvation, many
salmonids cannot recover physiologically after extended
periods of starvation because of compromised seawater
tolerance or impaired compensatory growth (Bilton and
Robins, 1973; Ban et al., 1996; Stefansson et al., 2009);
such recovery capabilities in juvenile chum salmon re-
main unclear.

The experimental WBEC, %MC, and CR differed from
the long-term average of the SECM data sets during
both months. After about 10 days of starvation, WBEC
was below the normal range, %MC was above the nor-
mal range, and CR shifted from positive to negative,
in both months. More specifically, by day 20, the June
fish had lost twice their WBEC and CR, and had gained
four times %MC as the July fish. The WBEC of the
June fish required only 3-7 days of starvation before
dropping to the lower initial level of the July fish.

Our study on the effects of starvation on field-caught
juvenile chum salmon indicates that WBEC, %MC, and
CR are more responsive measures than WW and FL to
prolonged food deprivation in a controlled laboratory
environment. Although starvation is an extreme case
of limited food resources, clearly juvenile chum salmon
can survive these conditions for extended periods, but
may consequently be less tolerant of variable environ-
mental conditions and more susceptible to other sources
of mortality, such as predation. Future studies will
focus on monitoring the seasonal response of juvenile
salmon condition measures, such as WBEC, %MC, and
CR, in different habitats at sea.

Acknowledgments

We thank the command and crew of the NOAA ship
John N. Cobb for help in collecting samples. We thank D.
Tersteeg and the staff at the Macaulay Hatchery otolith
laboratory for decoding all of the otoliths used in this
study. This manuscript was improved with suggestions
from three anonymous reviewers. Finally, we thank A.
Wertheimer and A. Moles for statistical and editorial
help with this manuscript.

Literature cited

Akiyama, T., and T. Nose.

1980. Changes in body weight, condition factor and body
composition of fingerling chum salmon with various
sizes during starvation. Bull. Natl. Res. Inst. Aquae.
1:71-78.

Ban, M., H. Hasegawa, and M. Ezure.

1996. Effects of starvation and refeeding on physiologi-
cal condition of juvenile chum salmon, Oncorhynchus
keta. Sci. Rep. Hokkaido Salmon Hatch. 50:117-123.

Bilton, H. T., and G. L. Robins.

1973. The effects of starvation and subsequent feed-
ing on survival and growth of Fulton channel sockeye
salmon fry (Oncorhynchus nerka). J. Fish. Res. Board
Can. 30:1-5.

Boldt, J. L., and L. J. Haldorson.

2004. Size and condition of wild and hatchery pink salmon
juveniles in Prince William Sound, Alaska. Trans.
Am. Fish. Soc. 133:173-184.

Bouck, G. R., and S. D. Smith.

1979. Mortality of experimentally descaled smolts of
coho salmon (Oncorhynchus kisutch) in fresh and salt
water. Trans. Am. Fish. Soc. 108:67-69.

Breck, J. E.

2008. Enhancing bioenergetics models to account for
dynamic changes in fish body composition and energy
density. Trans. Am. Fish. Soc. 137:340-356.

Brett, J. R.

1995. Energetics. In Physiological ecology of Pacific
salmon (C. Groot, L. Margolis, and W. C. Clarke, eds.),
p. 3-68. UBC Press, Vancouver, B.C., Canada.

Brett, J. R., J. E. Shelbourn, and C. T. Shoop.

1969. Growth rate and body composition of fingerling
sockeye salmon, Oncorhynchus nerka, in relation to tem-
perature and ration size. J. Fish. Res. Board Can.
26:2363-2394.

Brodeur, R. D., J. P. Fisher, D. J. Teel, R. L. Emmett, E. Casillas,
and T. W. Miller.

2004. Juvenile salmonid distribution, growth, condition,
origin, and environmental and species associations in the
Northern California Current. Fish. Bull. 102:25-46.

Congleton, J. L., and T. Wagner.

2006. Blood-chemistry indicators of nutritional status in
juvenile salmonids. J. Fish Biol. 69:473-490.

Cross, A. D., D. A. Beauchamp, K. W. Myers, and J. H. Moss.

2008. Early marine growth of pink salmon in Prince
William Sound and the Coastal Gulf of Alaska during
years of low and high survival. Trans. Am. Fish. Soc.
137:927-939.

Edsall, T. A., A. M. Frank, D. V. Rottiers, and J. V. Adams.

1999. The effect of temperature and ration size on the
growth, body composition, and energy content of juvenile
coho salmon. J. Great Lakes Res. 25:355-362.

Hoar, W. S.

1988. The physiology of smolting salmonids. In Fish
physiology, vol 11 (Hoar, W. S., and D. J. Randall, eds.),
p. 375-343. Academic Press, New York, NY.

Jakob, E. M., S. D. Marshall, and G. W. Uetz.

1996. Estimating fitness: a comparison of body condition
indices. Oikos 77:61-67.

Jobling, M.

1994. Fish bioenergetics, 309 p. Chapman and Hill,
London, U.K.

Fergusson et al.: Effects of starvation on energy density of Oncorhynchus keta

225

Kotiaho, J. S.

1999. Estimating fitness: comparison of body condition
indices revisited. Oikos 87:399-400.

LeBrasseur, R. J.

1969. Growth of juvenile chum salmon (Oncorhynchus
keta) under different feeding regimes. J. Fish. Res.
Board Can. 26:1631-1645.

Mason, J. C.

1974. Behavioral ecology of chum salmon fry ( Oncorhyn-
chus keta) in a small estuary. J. Fish. Res. Board Can.
31:83-92.

Miglavs, I., and M. Jobling

1989. The effects of feeding regime on proximate body
composition and patterns of energy deposition in juve-
nile Arctic charr, Salvelinus alpinus. J. Fish Biol.
35:1-11.

Mortensen, D., A. Wertheimer, S. Taylor, and J. Landingham.

2000. The relation between early marine growth of pink
salmon, Oncorhynchus gorbuscha, and marine water
temperature, secondary production, and survival to
adulthood. Fish. Bull. 98:319-335.

Murai, T., Y. Hirasawa, T. Akiyama, and T. Nose.

1983. Effects of previous dietary history on the mortality
and changes in body compositions of chum salmon fry
during starvation in seawater. Bull. Natl. Res. Inst.
Aquae. 4:79-86.

Orsi, J. A., M. V. Sturdevant, J. M. Murphy, D. G. Mortensen,
and B. L. Wing.

2000. Seasonal habitat use and early marine ecology
of juvenile Pacific salmon in southeastern Alaska. N.
Pac. Anadr. Fish Comm. Bull. No. 2:111-122.

Orsi, J. A., A. C. Wertheimer, M. V. Sturdevant, E. A. Fergusson,
D. G. Mortensen, and B. L. Wing.

2004. Juvenile chum salmon consumption of zooplank-
ton in marine waters of southeastern Alaska: a bio-
energetics approach to implications of hatchery stock
interactions. Rev. Fish Biol. Fish. 14:335-359.

Park, W., M. Sturdevant, J. Orsi, A. Wertheimer, E. Fergusson,
W. Heard, and T. Shirley.

2004. Interannual abundance patterns of copepods
during an ENSO event in Icy Strait, southeastern
Alaska. ICES J. Mar. Sci. 61:464—477.

Paul, A. J., and M. Willette.

1997. Geographical variation in somatic energy content of
migrating pink salmon fry from Prince William Sound:
A tool to measure nutritional status. In Forage fishes
in marine ecosystems: proceedings of the international
symposium on the role of forage fishes in marine eco-
systems (C. W. Mecklenburg, ed.), p. 707-720. Alaska
Sea Grant College Program Report 97-01, Univ. Alaska,
Fairbanks, AK.

Purcell, J. E., and M. V. Sturdevant.

2001. Prey selection and dietary overlap among zooplank-
tivorous jellyfish and juvenile fishes in Prince William
Sound, Alaska. Mar. Ecol. Prog. Ser. 210:67-83.

Sheridan, M. A., W. V. Allen, and T. H. Kerstetter.

1983. Seasonal variations in the lipid composition of the
steelhead trout, Sal/no gairdneri Richardson, associ-
ated with the parr-smolt transformation. J. Fish Biol.
23:125-134.

Simpkins, D. G„ W. A. Hubert, C. M. Del Rio, and D. C. Rule.

2004. Constraints of body size and swimming velocity on
the ability of juvenile rainbow trout to endure periods
without food. J. Fish Biol. 65:530-544.

Stefansson, S. O., A. K. Imsland, and S. O. Handeland.

2009. Food-deprivation, compensatory growth and hydro-
mineral balance in Atlantic salmon ( Salmo salar) post-
smolts in sea water. Aquaculture 290:243-249.

Trudel, M., S. Tucker, J. F. T. Morris, D. A. Higgs, and D. W.

Welch.

2005. Indicators of energetic status in juvenile coho
salmon and Chinook salmon. N. Am. J. Fish. Manag.
25:374-390.

Weatherley, A. H„ and H. S. Gill.

1995. Growth. In Physiological ecology of Pacific salmon
(C. Groot, L. Margolis, and W. C. Clarke, eds.), p. 103-
158. UBC Press, Vancouver, B.C., Canada.

Wuenschel, M. J., A. R. Jugovich, and J. A. Hare.

2006. Estimating the energy density of fish: the impor-
tance of ontogeny. Trans. Am. Fish. Soc. 135:379-385.

Zaporozhec, O. M., and G. V. Zaporozhec.

1993. Preparation of hatchery-reared chum fry for life
at sea: osmoregulation dynamics. Fish. Oceanog.
2:91-96.

226

Accuracy of sex determination

for northeastern Pacific Ocean thornyheads

(Sebastolobus altivelis and S. alascanus)

Erica L. Fruh (contact author)1
Aimee Keller2
Jessica Trantham3
Victor Simon2

Email address for contact author: Erica.Fruh@noaa.gov

' National Oceanographic and Atmospheric Administration
National Marine Fisheries Service
Northwest Fisheries Science Center
Fishery Resource Analysis and Monitoring Division
2032 SE OSU Drive
Newport, Oregon 97365

2 National Oceanographic and Atmospheric Administration
National Marine Fisheries Service

Northwest Fisheries Science Center
Fishery Resource Analysis and Monitoring Division
2725 Montlake Blvd. East
Seattle, Washington 98112

3 Husbandry Department
Underwater World

1245 Pate San Vitores RD Ste 400
Turnon, Guam 96913

Abstract — Determining the sex of
thornyheads ( Sebastolobus alasca-
nus and S. altivelis ) can be difficult
under field conditions. We assessed
our ability to correctly assign sex in
the field by comparing results from
field observations to results obtained
in the laboratory through both mac-
roscopic and microscopic examination
of gonads. Sex of longspine thorny-
heads was more difficult to determine
than that of shortspine thornyheads
and correct determination of sex
was significantly related to size.
By restricting the minimum size of
thornyheads to 18 cm for macroscopic
determination of sex we reduced the
number of fish with misidentified sex
by approximately 65%.

Manuscript submitted 25 June 2009.
Manuscript accepted 11 February 2010.
Fish. Bull. 108:226-232 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National
Marine Fisheries Service, NOAA.

Accurate sex-specific data are essen-
tial for fitting age-structured popula-
tion dynamic models and estimating
spawning biomass (Methot, 2000).
Assessing sex ratio is of added
importance if sex-based selectiv-
ity occurs within a fishery; because
separate management measures may
be required for male and female fish
(Cochrane, 2009).

Thornyheads are a common conti-
nental slope species and support a
large commercial fishery (Gunder-
son, 1997). Longspine thornyheads
( Sebastolobus altivelis) are found
from the Gulf of Alaska to southern
Baja California, whereas shortspine
thornyheads (Sebastolobus alasca-
nus) are distributed from the Ber-
ing Sea to northern Baja (Orr et al.,
2000). Longspine thornyheads gener-
ally inhabit depths greater than 400
m, have a distribution range to about
1400 m depth (Jacobson and Vetter,
1996), and a peak in abundance and
spawning biomass at about 1000 m
depth (Wakefield, 1990; Jacobson
and Vetter, 1996). Shortspine thorny-
heads are found from 20 m to over

1500 m in depth, are most abundant
in the range of 180 to 450 m, and
the majority of the spawning bio-
mass occurs between 600 and 1400
m, where longspine thornyheads
are most abundant (Jacobson and
Vetter, 1996). The maximum size of
shortspine thornyheads (>70 cm) is
larger than that of longspine thorny-
heads (—38 cm). Shortspine thorny-
heads migrate to deeper water as
their body size increases, whereas
longspine thornyheads do not mi-
grate to deeper water with increas-
ing size.

Identifying the sex of mature long-
spine thornyheads and shortspine
thornyheads by gross visual exami-
nation is difficult when gonads re-
gress to a resting state (Pearson and
Gunderson, 2003) because male and
female gonads are small, not fully
developed, and are morphologically
similar. Determining the sex of in-
dividual thornyheads collected dur-
ing the annual Northwest Fisheries
Science Center (NWFSC) West Coast
Groundfish Bottom Trawl Survey is
difficult because the survey occurs

Fruh et al.: Accuracy of sex determination for Sebastolobus altivelis and S. aloscanus

227

from May to October when thornyheads are not re-
productively active and gonads are in a resting state
(Moser, 1974; Wakefield, 1990).

The addition of sex identification for both thornyhead
species to survey sampling protocols will improve the
information available for management of the resource.
To address concerns about the ability of field person-
nel to correctly determine sex of thornyheads while at
sea, we examined the sex of longspine and shortspine
thornyheads in the laboratory using macroscopic ex-
amination of gonads (as a correlate for field work) in
contrast to microscopic techniques (for confirmation of
results). An additional goal was to determine a mini-
mum size below which the error rate for classification
of sex of thornyheads in the field was judged to be too
high by investigating the relationship between sex mis-
identification and length, geographic area, and month
captured. Because assessment scientists are interested
in the actual proportion of males to females, we also
evaluated absolute percent error after accounting for
the portion of the error that was cancelled out by bal-
ancing the number of misidentified males reported as
females against the number of misidentified females
reported as males.

Materials and methods

The 2003 NWFSC West Coast Groundfish Bottom Trawl
Survey was conducted between 24 June and 23 October,
from the area off Cape Flattery, Washington (48°10'N
lat.) to the U.S. -Mexico border (32°30'N lat.) at water
depths of 55-1280 m. The survey area was covered twice
by chartered commercial fishing vessels (20 to 28 m
length). The first sampling period was from 24 June to
13 August and the second from 31 August to 23 October.
A stratified random sampling design was used and the
survey area was subdivided into adjacent cells of equal
area (1.5 nmi long, by 2.0 nmi lat., Albers equal area
projection). A total of 620 primary sites were randomly
selected from cells stratified by geographic location and
depth. The geographic allocation was based on assign-
ing 15-25% of the cells to each of five International
North Pacific Fisheries Commission (INPFC) statis-
tical areas: U.S. -Vancouver (47°30'N to U.S. -Canada
border), Columbia (43°00' to 47°30'N), Eureka (40°30' to
43°00'N), Monterey (36°00' to 40°30'N), and Conception
(U.S. -Mexico border to 36°00'N). The survey area was
further stratified into depth zones with 45% of the cells
allocated to the shallow depth zone (55-183 m), 30% to
mid-depth (184-549 m) and 25% to the deep stratum
(550-1280 m). Each of four chartered fishing vessels
was assigned 155 stations to sample.

The bottom trawl survey is a standardized fishery in-
dependent survey and all fishing operations are conduct-
ed in strict compliance to national protocols (Stauffer,
2004). Vessels were equipped with standard Aberdeen-
style nets with small mesh (1.5-inch stretched measure)
liner in the codend. All thornyheads randomly selected
for biological sampling were assigned a unique identi-

fication number, individually weighed (kg), measured
(fork length, cm), and frozen while at sea. All frozen
specimens were brought back to the laboratory where
fish were thawed, dissected, and examined macroscopi-
cally to identify sex. For macroscopic examination of
gonads, an incision was made with a scalpel on the
ventral surface of each thornyhead from the vent to the
base of the pectoral fin. The lateral side of the fish was
opened to expose the gonads, and a visual identification
of sex was based on the physical structure of the gonad-
al tissue as described by Lagler et al. (1962). Sex was
recorded as male, female, or unknown. For microscopic
identification of sex, a section of gonad tissue from each
fish was placed on a glass microscope slide, stained with
acetocarmine solution and compressed with a cover slip.
The stain acted on the gonad tissue by readily staining
oocytes dark pink (Guerrero, 1974). The slides were
viewed under a lOx power microscope (Leica DM LS2,
Bannockburn, IL), and females were distinguished from
males by the presence of dark pink stained oocytes.

Accuracy of sex determination was examined in rela-
tion to length by species, geographic region, and month
of capture (June-October). To determine a size thresh-
old below which sex determination should not be at-
tempted in the field, we examined both the total and
absolute percentage of incorrectly sexed thornyheads
in relation to length. To avoid biasing results, we did
not consider our ability to correctly identify female
thornyheads at smaller sizes, as opposed to our ability
to correctly identify males at smaller sizes. Absolute er-
ror was calculated as the absolute value of misidentified
males minus misidentified females divided by the total
number examined at each 1-cm size interval, and this
value was then expressed as a percentage. Size data
were transformed (natural logarithm) to reduce hetero-
geneity of variance before statistical analysis. Data were
statistically compared by analysis of variance (ANOVA)
by using SAS for Windows (SAS Institute, Inc., Cary,
NC). Significant ANOVAs were followed by a nonpara-
metric comparison of means test (Tukey’s test). Fish in
which the gonad could not be found, stained, or micro-
scopically identified were not included in the analyses.

Results

A total of 574 successful tows were completed. Figure 1
shows the distribution and relative abundance (kg/ha)
of thornyheads from the 2003 survey. Both species
were concentrated in the mid- and deep depth strata
(183-1280 m) and exhibited higher relative abundance
north of Pt. Conception, CA (34°30'N lat.). Longspine
thornyheads were collected in 214 tows at depths of
328—1280 m (mean depth 802 m) and shortspine thorny-
heads were collected in 311 tows at depths of 88-1280 m
(mean depth 605 m). A total of 2325 thornyheads were
collected for later processing in the laboratory. Sex was
determined for 852 longspine thornyheads and 1148
shortspine thornyheads. Sex was indeterminable for
189 longspine and 136 shortspine thornyheads (average

228

Fishery Bulletin 108(2)

Figure 1

Distribution and relative abundance (kg/ha) of (A) longspine thornyhead ( Sebastolobus altivelis) and (B) shortspine
thornyhead ( Sebastolobus alascanus) determined from the 2003 Northwest Fisheries Science Center west coast ground-
fish trawl survey. SD = standard deviation.

length 14.3 cm). Longspine thornyhead sex was misiden-
tified by visual examination in 23.1% of males and 22.4%
of females, and for shortspine thornyheads, in 9.4% of
males and 9.3% of females.

Average lengths of longspine and shortspine thorny-
heads (females, males, and total) for which sex was
misidentified were significantly lower than the lengths
for fish whose sex was correctly assigned (Table 1).
For shortspine thornyheads, the average length of sex-
misidentified females was significantly smaller than
that of males (ANOVA: df=6, F- 5.5, P= 0.02). Similar
tendencies were seen for longspine thornyhead lengths
but the results were not significant (Table 1).

Determining sex for longspine thornyheads greater
than 22 cm would eliminate approximately 80% of the
overall error rate, but would also eliminate 50% of the
fish whose sex was correctly determined. By proposing
18 cm as the minimum size for examining longspine
thornyheads in the field we eliminated approximately
65% of the incorrectly sexed fish, while retaining >70%
of those correctly sexed (Fig. 2A). On average, the sex
of 50.5% of longspine thornyheads ranging in size from
11 to 17 cm was incorrectly determined. This average
dropped to approximately 10% for longspine thorny-

heads at lengths from 18 to 34 cm. A similar result was
seen for shortspine thornyheads (Fig. 2A). The average
percentage of shortspine thornyheads with misidentified
sex was 53.7% at lengths from 11 to 17 cm. This value
decreased to 5.9% for larger fish (18-71 cm) (Fig. 2A).

With a single exception, more males were misiden-
tified as females in every size category for both spe-
cies, and the absolute percentage of sex-misidentified
fish decreased at fork lengths greater than 17 cm (Fig.
2B). For longspine thornyheads the average decreased
from 15.8% for fish 11-17 cm to 2.2% for fish 18-34
cm length and the average percentage for shortspine
thornyheads dropped from 24.5% to 3.0% in the larger
size category (Fig. 2B).

Sex misidentification in longspine thornyheads did not
vary significantly by month from June through October
(ANOVA: df=7, F=1.74, P=0.34; Fig. 3A). However, sex
misidentification for shortspine thornyheads was signifi-
cantly higher in August, with an increasing trend from
June through August followed by a decline (ANOVA:
df=7, F=15.5, P=0.02; Fig. 3A).

The accuracy of sex determination varied by geo-
graphic area for both species (Fig. 3B). The sex of long-
spine thornyheads was more frequently misidentified

Fruh et al.: Accuracy of sex determination for Sebastolobus altivelis and 5. alascanus

229

Table 1

Number ( n ), mean fork length (cm, ±standard error [SE] ), and analyses of variance (ANOVAs) for sizes for female, male, and
total longspine ( Sebastolobus altivelis) and for female, male, and total shortspine thornyheads ( S . alascanus) captured during
the 2003 Northwest Fisheries Science Center west coast groundfish trawl survey, correctly and incorrectly assigned sex based
on visual examination.

Correct

Incorrect

ANOVAs

Species

Mean length ( ± SE )

Longspine thornyhead

female

396

21.6 (0.21)

114

18.5 (0.37)

509

48.9

0.0001

male

259

23.4 (0.20)

19.0 (0.47)

341

96.6

0.0001

total

655

22.3 (0.15)

197

18.7 (0.29)

851

52.7

0.0001

Shortspine thornyhead

female

560

35.5 (0.52)

23.6 (1.16)

617

50.8

0.0001

male

481

34.9 (0.45)

28.1 (1.52)

529

21.5

0.0001

total

1041

35.2 (0.35)

107

25.7 (0.96)

1147

36.2

0.0001

above 43°N latitude, and the U.S. -Vancouver and Co-
lumbia areas had a significantly higher average per-
centage of misidentification than the Eureka, Monterey,
and Conception areas (ANOVA: df=4, F=44.1, P-0.007).
The sex of shortspine thornyheads became more diffi-

cult to correctly identify below 40°N latitude, and both
the Monterey and Conception areas had a significantly
higher average percentage of misidentification com-
pared to the Eureka, Columbia, and U.S. -Vancouver
areas (ANOVA: df=4, P=13.9, P=0.03). There were no

230

Fishery Bulletin 108(2)

Fork length (cm)

Figure 2

(A) Total percentage of sex-misidentified longspine ( Sebastolobus altivelis)
and shortspine thornyheads (Sebastolobus alascanus) determined by compar-
ing the gross morphological features of gonads to a section of each gonad
subsequently stained and viewed microscopically, by size (fork length, cm);
and (B) the absolute percent error in identifying the sex of thornyheads
after accounting for the portion of the total error that is cancelled out by
balancing the number of sex-misidentified males against the number of
sex-misidentified females.

significant differences in mean fork length for longspine
thornyheads between the different areas (ANOVA: df=
858, F=0.3, P=0.9), but for shortspine thornyheads,
size was significantly larger in the Monterey and U.S.-
Vancouver areas (ANOVA: df=1140, F= 4.7, P=0.0009),
and large fish in the Monterey area had a higher rate
of individuals for which sex was incorrectly determined
than similar size shortspine thornyheads in the U.S.-
Vancouver area.

Discussion

This study provides guidance for a minimum size limit
below which sex of thornyheads should not be deter-
mined at-sea because of high error rates. High quality
biological information is important for management and
modeling of thornyhead populations along the U.S. west
coast (Fay, 2005). Fishery scientists need estimates of

sex ratio for fish populations because shifts in these
values can indicate overfishing on one sex or the other
due to selective gear, differential growth rates, segrega-
tion by sex or any combination of these (Cochrane, 2009).

In previous studies of the reproductive biology of
thornyheads, the longspine thornyhead spawning was
determined to begin in January, peak in February and
March, and continue at least through April (Wakefield,
1990; Pearson and Gunderson, 2003; Cooper et al.,
2005). Shortspine thornyheads spawn between Decem-
ber and May along the U.S. west coast. The onset of
sexual maturity occurs at 17-19 cm total length (10%
mature females) in both species and 90% are mature
at 25-27 cm (Pearson and Gunderson, 2003). Sex of
smaller thornyheads is difficult to determine, particu-
larly during the summer, because of the small size of
the gonads — size being a function of the annual spawn-
ing cycle. Pearson and Gunderson (2003) noted that
of 36 longspine thornyheads designated as immature

Fruh et at: Accuracy of sex determination for Sebastolobus altivelis and S. alascanus

231

females in the field on the basis of gross morphological
features, nine were actually males.

Correct visual identification of sex for both shortspine
and longspine thornyheads increased in fish longer than
17 cm. Overall accuracy is greater for shortspine than
for longspine thornyheads, and greater for females than
for males, and this accuracy is related to size in both
instances. For both species, 18 cm was selected as the
lower limit for determining the sex of thornyheads in
the field because the majority of sex-misidentified fish
fell below this value. In 2003, 66% of the longspine
thornyheads and 90% of the shortspine thornyheads
measured in the field throughout the survey period were
greater than 17 cm. The selected size falls within the
range of lengths noted for the onset of sexual maturity
in both species.

Because the survey is conducted after the completion
of the spawning season for longspine thornyheads (Janu-
ary-April), the samples are collected exclusively during
the reproductive resting stage. Sex misidentification was
relatively constant for longspine thornyheads through-
out the sample period and there were no significant
differences among months. Sex misidentification was
greater for longspine than for shortspine thornyheads
for each time period. The lower rate of sex misidentifica-
tion for shortspine thornyheads may be related to their
longer spawning season (December— May). Differences
in the reproductive cycles of the two species resulted
in the cessation of spawning coinciding with the start
of the survey sampling for shortspine thornyheads and
may partially explain the observed overall lower rate of
sex misidentification for this species. The middle of the
reproductive resting-stage period correlated with high
levels of sex misidentification for both species, although
only for shortspine thornyheads was the difference sig-
nificant (in August).

The differences in sex misidentification among geo-
graphic areas are more difficult to explain. Sex of long-
spine thornyhead was more frequently misidentified in
the U.S. -Vancouver and Columbia areas. Samples in
these areas were collected primarily in June and Sep-
tember, the periods with the highest rates of sex mis-
identification. The lack of any significant differences in
mean length for longspine thornyheads between INPFC
areas indicates that the higher rates of misidentification
of sex farther north were not a function of size, but were
related to the timing of the annual spawning cycle at
differing latitudes.

Shortspine thornyhead samples collected in the Eure-
ka, Columbia, and U.S. -Vancouver areas (i.e., those with
significantly lower rates of sex misidentification) were
primarily taken in June, July, and September when the
rate of sex misidentification for shortspine thornyheads
was lowest. Additionally, there were significant differ-
ences in the lengths of shortspine thornyheads among
areas, indicating that the lower rates of sex misidentifi-
cation in the U.S. -Vancouver area may also be partially
related to size (although similar size differences were
not observed in the Eureka and Columbia areas). Be-
cause differences in geographic area were related to size

q3 35

30
25
20
15
10
5
0

Conception Monterey Eureka Columbia US-Vancouver

Figure 3

Percentage of sex-misidentified longspine ( Sebastolo-
bus altivelis) and shortspine (Sebastolobus alascanus )
thornyheads determined by comparing the gross mor-
phological features of gonads to a section of each gonad
subsequently stained and viewed microscopically (A)
by month, and (B) by geographic area as defined by
the International North Pacific Fisheries Commission
regions: Conception (U.S. -Mexico border to 36°00'N lat.),
Monterey (36°00' to 40°30'N lat.), Eureka (40°30' to
43°00'N lat.), Columbia (43°00' to 47°30'N lat.), and
U.S. -Vancouver (47°30'N lat. to U.S. -Canada border).

for at least one thornyhead species and the differences
in seasonal determination of sex were variable, we rec-
ommend that sex determination of thornyheads <18 cm
not be attempted in the field. This is likely a conserva-
tive estimate because identifying sex in fresh specimens
at sea is somewhat more reliable than examining frozen
and thawed specimens in the laboratory. The approach
described here establishes a protocol for determining
a minimum size for at-sea sex identification of thorny-
heads, but may be applicable for use with any species
where ambiguity may exist in correctly identifying the
sex of fish at smaller sizes, within different regions, or
across spawning or other seasonal cycles.

Acknowledgments

We thank the captains and crew of the fishing vessels
Ms. Julie , Excalibur, Captain Jack, and Blue Horizon for
their effort during the 2003 NWFSC West Coast Ground-
fish Bottom Trawl Survey. We also thank the biologists
who participated in this study, including K. Bosley, J.
Buchanan, D. Kamikawa, and V. Tuttle.

232

Fishery Bulletin 108(2)

Literature cited

Cochrane, K.

2009. Current paradigms and forms of advice. In Fish
reproductive biology implications for assessment and
management (Jakobsen, T., M. J. Fogerty, B. A. Megrey,
and E. Moksness, eds.), p. 335-354. Wiley-Blackwell,
Ames, IA.

Cooper, D. W., K. E. Pearson, and D. R. Gunderson.

2005. Fecundity of shortspine thornyhead ( Sebastolo -
bus alascanus) and longspine thornyhead (S. altive-
lis) ( Scorpaenidae) from the northeastern Pacific
Ocean, determined by stereological and gravimetric
techniques. Fish. Bull. 103:15-22.

Fay, G.

2005. Stock assessment and status of longspine thorny-
head (Sebastolobus altivelis) off California, Oregon, and
Washington in 2005, 90 p. Pacific Fishery Management
Council, Portland, OR.

Guerrero III, R. D.

1974. An aceto-carmine squash method for sexing juve-
nile fishes. Prog. Fish Cult. 36:56.

Gunderson, D. R.

1997. Spatial patterns in the dynamics of slope rockfish
stocks and their implications for management. Fish.
Bull. 95:219-230.

Jacobson, L. D., and R. D. Vetter.

1996. Bathymetric demography and niche separa-
tion of thornyhead rockfish: Sebastolobus alascanus
and Sebastolobus altivelis. Can. J. Fish. Aquat. Sci.
53:600-609.

Lagler, K. F., J. E. Bardach and R. R. Miller.

1962. Ichthyology, 545 p. John Wiley and Sons, Inc.,
New York.

Methot, R.

2000. Technical description of the stock synthesis assess-
ment program. U.S. Dep. Commer., NOAA. Tech. Memo.
NMFS-NWFSC-43, 46 p.

Moser, H. G.

1974. Development and distribution of larvae and juveniles
of Sebastolobus (Pisces: Family Scorpaenidae). Fish.
Bull. 72:865-884.

Orr, J. W., M. A. Brown, and D. C. Baker.

2000. Guide to rockfishes (Scorpaenidae) of the genera
Sebastes , Sebastolobus, and Adelosebastes of the North-
east Pacific Ocean, 2nd ed. NOAA Tech. Memo. NMFS-
AFSC-117, 48 p.

Pearson, K. E., and D. R. Gunderson.

2003. Reproductive biology and ecology of shortspine
thornyhead rockfish, Sebastolobus alascanus and long-
spine thornyhead rockfish, S. altivelis, from the north-
easter Pacific Ocean. Environ. Biol. Fish. 62:117-136.

Stauffer, G.

2004. NOAA protocols for groundfish bottom trawl sur-
veys of the nation’s fishery resources. U.S. Dep. Com-
merce, NOAA Tech. Memo. NMFS-F/SPO-65, 205 p.

Wakefield, W. W.

1990. Patterns in the distribution of demersal fishes
on the upper continental shelf off central California
with studies of ontogenetic vertical migration in par-
ticle flux. Ph.D. thesis, 281 p. Scripps Institution of
Oceanography, Univ. California, San Diego, CA.

233

Abstract — Body-size measurement
errors are usually ignored in stock
assessments, but may be important
when body-size data (e.g., from visual
surveys) are imprecise. We used
experiments and models to quantify
measurement errors and their effects
on assessment models for sea scallops
( Placopecten magellanieus). Errors in
size data obscured modes from strong
year classes and increased frequency
and size of the largest and smallest
sizes, potentially biasing growth, mor-
tality, and biomass estimates. Model-
ing techniques for errors in age data
proved useful for errors in size data.
In terms of a goodness of model fit
to the assessment data, it was more
important to accommodate variance
than bias. Models that accommodated
size errors fitted size data substan-
tially better. We recommend experi-
mental quantification of errors along
with a modeling approach that accom-
modates measurement errors because
a direct algebraic approach was not
robust and because error parameters
were difficult to estimate in our
assessment model. The importance
of measurement errors depends on
many factors and should be evaluated
on a case by case basis.

Manuscript submitted 22 June 2009.
Manuscript accepted 25 January 2010.
Fish. Bull. 108:233-247 (2010).

The views and opinions expressed or
implied in this article are those of the
author (or authors) and do not necessarily
reflect the position of the National Marine
Fisheries Service, NOAA.

Measurement errors in body size

of sea scallops ( Placopecten magellanieus )

and their effect on stock assessment models

Larry D. Jacobson (contact author)1 Deborah Hart1

Kevin D. E. Stokesbury2
Melissa A. Allard2
Antonie Chute1
Bradley P. Harris2

Tom Jaffarian2
Michael C. Marino II2
Jacob I. Nogueira2
Paul Rago1

Email address for contact author: Larry.Jacobson@noaa.gov

1 NOAA Fisheries

Northeast Fisheries Science Center
166 Water Street

Woods Hole, Massachusetts 02543-1026

2 Department of Fisheries Oceanography
School for Marine Science and Technology
University of Massachusetts School of Marine Sciences
838 South Rodney French Boulevard

New Bedford, Massachusetts 02744-1221

Two fishery-independent surveys are
important for monitoring Atlantic
sea scallop (Placopecten magellani-
cus) abundance and biomass levels
off the northeastern coast of the
United States because they provide
abundance, body size,1 meat weight
(weight of marketable adductor mus-
cles), and other data (NEFSC2’3). The
National Marine Fisheries Service,
Northeast Fisheries Science Center
(NEFSC) sea scallop dredge survey
has been conducted annually since
1977 (Serchuk et ah, 1979; Serchuk
and Wigley, 1986). In addition, an
underwater video survey for sea scal-
lops and other benthic organisms has
been conducted annually since 2003
(Stokesbury, 2002; Stokesbury et al.,
2004) by the University of Massachu-
setts Dartmouth, School for Marine
Science and Technology (SMAST). The
dredge and video surveys are carried
out across the range of sea scallops in
U.S. waters.

In this analysis, we used sea scal-
lops to draw attention to errors in
body-size data when the data are
used in a length-structured stock
assessment model. The topic of mea-
surement errors in body-size data
has received relatively little atten-
tion, although Heery and Berkson

(2009) evaluated effects of systematic
errors (biased sampling) in fishery
size-composition data used in an age-
structured model. Our work was mo-
tivated by questions that arose from
examining video survey shell-height
data in sea scallop stock assessments
(NEFSC2’3). Our experimental and
analytical results may be important
and useful in other situations where
body-size data are imprecise. Body-

1 Shell height (SH, the distance in mm
between the umbo and shell margin)
is the body size measurement for sea
scallops.

2 NEFSC (Northeast Fisheries Science
Center). 2004. Stock assessment for
Atlantic sea scallops. In 39th north-
east regional stock assessment workshop
(39th SAW) assessment summary report
and assessment report. Northeast Fish-
eries Science Center, National Marine
Fisheries Service, Woods Hole Labora-
tory, 166 Water St., Woods Hole, MA
02543. Ref. Doc. 04-10, p. 87-211.

3 NEFSC (Northeast Fisheries Science
Center). 2007. Stock assessment for
Atlantic sea scallops. In 45th northeast
regional stock assessment workshop (45th
SAW) assessment summary report and
assessment report. Northeast Fisheries
Science Center, National Marine Fisher-
ies Service, Woods Hole Laboratory, 166
Water St., Woods Hole, MA 02543. Ref.
Doc. 07-16, p. 139-370.

234

Fishery Bulletin 108(2)

size data may be imprecise, for example, when col-
lected by scuba (St. John et al., 1990; Edgar et al.,
2004), remotely operated underwater vehicles (ROV;
Butler et al., 2006), camera sleds (Rosenkranz and
Byersdorfer, 2004), or in other optical surveys where
body-size measurements are obtained without handling
individual specimens.

In fishery stock assessment modeling, body-size mea-
surements are almost always assumed to be without
error. In contrast, statistical sampling errors that arise
from too few are often considered in modeling (Fournier
and Archibald, 1982; Pennington et al., 2001). Measure-
ment errors in fishery age data have received substan-
tial attention and are often addressed in stock assess-
ment modeling (Methot, 1989, 1990). Approaches to
dealing with measurement error in body-size data have
not been explored.

Shell-height composition data for sea scallops are
of two types: 1) distributions of shell-height measure-
ments, which include measurement errors and true
variability among individuals in size; and 2) distri-
butions of shell-height measurements, which include
measurement errors only. It is important to distinguish
between these two types of data. In particular, shell-
height compositions are sample specific and depend on
the underlying distribution of true sizes. In our study
measurement errors are the difference between the
video or board measurements and the true shell height
of individual specimens (i.e., after removing differences
in true shell height among individuals). Shell-height
composition data are important because they are in-
terpreted in stock assessments to estimate year-class
strength, mortality, and other biological characteris-
tics. In our study measurement errors are important
because they can be used to quantify the accuracy of
the measurement process itself and because they affect
shell-height data from all samples.

Two types of measurement errors are considered in
this study. The first type is bias that causes individual
shell-height measurements and estimated sample means
to differ, on average, from their true values (Cochran,
1977). The second type is random errors, which cause
variability in shell-height measurements and affect the
precision of measurements and estimated mean values
(Cochran, 1977).

Figure 1 shows how hypothetical errors in sea scal-
lop shell-height measurements tend to smooth the true
underlying distribution of the data. Measurement errors
tend to smooth modes in the data (which usually cor-
respond to recruitment events) by moving individuals
from size bins with relatively high numbers into adja-
cent bins with lower numbers. Random measurement
errors also tend to expand the range of observed sizes
by decreasing the smallest observed size and increas-
ing the largest (Fig. 1). Bias degrades body-size data
by making measurements consistently larger or smaller
than the true value. Methot (1989, 1990) highlighted
these issues in the context of age data from survey and
fishery samples. We use Methot’s modeling methods in
our analysis for shell-height data.

In principal, body-size measurement errors can cause
errors in a wide range of important fishery estimates
but biomass estimates are of particular importance. In
the absence of bias, imprecise body-size data tend to
cause positive bias in mean weight and biomass esti-
mates because of the nonlinear relationship between
size and biomass and Jensen’s inequality (Feller, 1966).
For example, according to Jensen’s inequality, if body
weight is a cubic function of body size, then a -10%
error in body size will cause a 0.93-l = -27% error in
estimated body weight for one individual. In contrast,
a +10% error in body size will cause a 1.13-1 = +33%
error in body weight. The combined effect of the two
errors for two scallops of the same size would be a posi-
tive bias of +6%.

The length-based Beverton-Holt mortality estimator
involves equilibrium and other assumptions that may
make it inappropriate to use in some cases (Gedamke
and Hoenig, 2006), but it clearly demonstrates the po-
tential effects of errors in body-size measurements on
stock assessment model mortality estimates:

L-L

(1)

where Z =
=

K =
L =

Lc =

the instantaneous rate of mortality from all
sources;

asymptotic length;

rate parameter from the von Bertalanffy
growth equation;

average length of individuals in a sample
from the fishery; and

the “critical” length at which individuals
are fully vulnerable to the fishery (Quinn
and Deriso, 1999).

With all other factors held constant, a positive bias in
L will make the numerator in Equation 1 too small, the
denominator too large, and the mortality estimate will
be biased low. Conversely, a negative bias in L will bias
the mortality estimate high.

In this article, we characterize measurement
errors in shell-height data for sea scallops in two
types of surveys, using experimental data. The
experimental results are used to evaluate effects
on mean body weight and swept-area biomass es-
timates, and on biomass and mortality estimates
from a modern size-structured stock assessment
model. The assessment model demonstrates a
promising approach (used originally for age data)
for accommodating measurement errors in body-
size data. In the appendices, we use numerical and
bootstrap techniques to evaluate robustness of the
assessment model approach in comparison to an
algebraic one. Our purpose is not to evaluate the
merits of any particular survey, rather, we use sea
scallops as an example for dealing with general
problems arising from body-size measurement er-
rors in survey and fishery-dependent data, and for
suggesting possible approaches to using such data.

Jacobson et al.: Measurement errors in body size of Placopecten magellanicus

235

Simulated shell heights Simulated shell heights

with and without errors with out errors and residuals

Bias only

Figure 1

Rootograms (Tukey, 1977) showing hypothetical distributions of Atlantic sea scallop ( Placopecten
magellanicus) shell-height (SH) measurements with and without simulated measurement errors. The
black line in each panel shows the distribution of measurements with no errors (5-mm size bins).
In the left column, bars show distributions of shell heights with measurement errors. In the right
column, bars show residuals (measurement with no errors minus measurements with errors). For
the “bias only” scenario (A and B), precise measurement errors were assumed with a bias of -4.1
mm. For the “imprecision only” scenario (C and D) unbiased measurement errors were assumed
with a standard deviation of 6.1 mm. For the “imprecision and bias” scenario (E and F), measure-
ment errors were assumed with a bias of -4.1 mm and standard deviation of 6.1 mm.

Materials and methods

The SMAST sea scallop survey is conducted with video
cameras mounted on a steel pyramid frame to provide
a 3.24-m2 view of the sea floor and associated macro-
benthos (Stokesbury, 2002; Stokesbury et al., 2004).
Video images are recorded at sea on high-resolution

S-VHS videotape and then replayed in the laboratory
where digitized images are created. All sea scallops
are counted, and all clearly visible sea scallops (with
the hinge and opposite edge visible) within the digi-
tized images are measured to the nearest mm by using
Image Pro Plus® software (Media Cybernetics, Inc.,
Bethesda, MD).

236

Fishery Bulletin 108(2)

In previous analyses, correction factors were applied
to the raw video shell-height measurements to account
for distance from the origin (DFO), which is the dis-
tance of a specimen from the “origin” (center) of the
sampling frame (Stokesbury et ah, 2004). Subsequent
work during routine stock assessments (unpublished)
indicated that adjustments were unnecessary because
the distributions of measurement errors were simpler
and easier to describe statistically, and data were easier
to model without adjustments. Moreover, adjusted data
were sometimes less accurate than the unadjusted data.
Additional research may result in more accurate adjust-
ments or transformations of body-size data. However,
unadjusted video data from the “large” camera on the
sampling frame are used in current stock assessments
and in this analysis.

NEFSC sea scallop surveys are conducted with a
2.44-m New Bedford sea scallop dredge with a 38-mm
liner. The catch is sorted, counted, and measured on the
deck of the research vessel. In most cases, the entire
catch is counted and measured, but a few large catches
were subsampled. During the early 1980s through 2003,
sea scallops in the catch were measured to the nearest
5-mm shell-height interval with a standard NEFSC sea
scallop measuring board.

Experiments

Two experiments were conducted during 20 and 23 Feb-
ruary 2003 when the SMAST video pyramid was placed
in a 341,000-L tank filled with seawater in the SMAST
laboratory. NEFSC sea scallop measuring boards and
SMAST video equipment in the experiments were con-
figured and used in a realistic manner that was similar
to use during actual surveys at sea. Accurate measure-
ments used as true shell heights in this analysis were
made to the nearest mm by using scientific calipers
under laboratory conditions with adequate lighting.

We used the experimental data to evaluate statisti-
cal characteristics of shell-height composition data and
shell-height measurement errors.

Accuracy, bias, and precision of measurements were
quantified by comparing data obtained from the mea-
suring board and video camera with data from the
caliper. Accuracy is the closeness to the true underlying
value and is measured by mean square error (MSE). For
shell-height composition data,

MSE = (h-H)2, (2)

where h = the mean of the measurements; and

H = the mean of the true values for the sample
(Cochran, 1977).

For measurement errors in our analysis,

MSE = — — , (3)

where e- - h—Hj = the error for the jth observation (where
hj is the measurement and is the
true value).

Bias and variance both contribute to MSE. In fact,
MSE = s2 + b2, where s2 is the variance and b is bias
(Cochran, 1977). In our study, b-h-H where h is the
mean of shell-height measurements and H is the mean
of the true shell heights in the sample. Bias is the same
for shell-height composition data and measurement er-
rors as shown below:

^ ](hj-Hj)/n = h-H . (4)

7=1

Variance (s2) was computed from shell-height composi-
tion data or measurement errors by using the standard
formula. Variance of shell-height composition data and
measurement errors will generally be different because
true shell heights usually differ among specimens in a
sample.

It is convenient to express accuracy, bias, and preci-
sion in terms of the square root of the MSE (RMSE),
bias (b), and standard deviation (s) because all three
are absolute measures with the same units (mm for sea
scallop shell-height data). Percent RMSE (RMSE /htrue),
percent bias ( blhtrue ), and the CV (s/h) are useful for
making comparisons on a relative basis.

The third and fourth moment statistics, gl and g2
were used to measure skewness (asymmetry) and kur-
tosis (peakedness) of shell-height composition data and
measurement errors, in relation to what would be ex-
pected from a normal distribution (Sokal and Rohlf,
1995). Skewness and kurtosis statistics for shell-height
composition data and measurement errors from the
same sample differ if there is variability in size among
specimens. For normally distributed random variables
with no skewness, gr - 0. Negative g1 values indicate
skewness to the left (a distribution with a long left
tail and more small values than expected in a normal
distribution). Positive g1 values indicate skewness to
the right (long right tail with more large values than
expected). Similarly, positive g2 values indicate dis-
tributions more peaked than expected for a normal
distribution, and negative g2 values indicate distribu-
tions that are less peaked (flatter) than expected. The
two statistics convey information about the shape of
any distribution in relation to a normal distribution,
but care is required in interpreting g 1 and g2, particu-
larly for data that are far from normally distributed.
The skewness and kurtosis statistics were easier to
interpret for measurement errors than for shell-height
measurements because the latter were not normally
distributed.

We used a test for normally distributed statistics
(Sokal and Rohlf, 1995) to evaluate the statistical sig-
nificance of skewness and kurtosis for distributions of
measurement errors that might be otherwise assumed
normally distributed. Statistical tests were carried out

Jacobson et at: Measurement errors in body size of Placopecten magellanicus

237

for distributions of measurement errors because
they were closer to normally distributed.

Multiple shell height-measurements were usu-
ally made from single specimens in our experi-
ments. We made allowance for repeated sampling
when testing skewness and kurtosis by using the
number of unique specimens in the experiment as
the degrees of freedom instead of the number of
measurements (i.e., if n measurements were made
on each of k specimens, we used k as the degrees
of freedom in statistical tests). The effect of this
adjustment was to make the statistical tests more
conservative (less likely to reject the null hypoth-
esis of no difference). The number of specimens
is a reasonable lower bound estimate of the true
effective sample size.

Body weights for sea scallops and other marine
organisms are often computed from body size. For
sea scallops in this analysis,

w = ea+pUh)^ (5)

where W = sea scallop meat weight ( g , the weight
of the marketable adductor muscle);
h = shell height (mm); and the parameter
values a=-12.01 and 3.22.

Bland-Altman plots (1986, 1995) were used to
characterize shell-height measurement errors. In
the case of measuring boards, for example, the dif-
ference between the measuring board and caliper
shell-height measurements for each sea scallop was
plotted on the y-axis against the average of the
two measures for the same individual on the x-
axis. Bland-Altman plots are typically presented
as scatter plots with a point for each difference
(pair of measurements); however, boxplots may be
more useful in some circumstances (see below).
Bland-Altman plots are useful because they elimi-
nate spurious correlations when the difference of
y-x is plotted against the more precise measure (x)
and because patterns are easier to discern along a
horizontal line (the x-axis) than along a diagonal
line. Spurious correlations occur because the mea-
surement error in x affects the variables plotted on
both the x- and y-axes.

Experiment 1 was designed to measure the accuracy
of video measurements for objects of known size (square
ceramic tiles) as a function of position in the video
frame as measured by DFO (Fig. 2). Scuba divers in
experiment 1 placed black and white ceramic floor tiles
(all were 48.5x48.5 mm) in a closely packed square grid
on the bottom of the tank, starting at the center of the
video pyramid and covering the entire range of view
in actual surveys (Fig. 2). The width and height of 91
tiles across the field of view and at various distances
and positions from the center of the sampling frame
(Fig. 2) were estimated from video images by using the
standard video survey procedures described above. Data
were recorded in such a way that the length and height

measurements from the same tile could be associated
with each other and with the particular position of the
tile in the video image. The tiles used in experiment 1
(48.5x48.5 mm) corresponded roughly with the size of
the smallest scallops fully recruited to the dredge and
video surveys (about 40 mm SH) and included in stock
assessment analyses. Sea scallops, according to actual
survey data, cover a much wider range of shell heights
(to about 190 mm SH in experiment 2, see Discussion
section).

Experiment 2 was designed to measure the accu-
racy of video shell-height measurements for sea scal-
lop shells of varying sizes (39 to 192 mm SH) placed
randomly on a sand-granule-pebble substrate, similar

238

Fishery Bulletin 108(2)

to the random aggregations observed on Georges Bank.
All shell-height measurements could be linked with
each individual sea scallop in experiment 2 because
the right valve of 172 individual sea scallop shells was
numbered uniquely. The identification numbers were
large and written under the valve with dark indelible
ink and clearly visible with video equipment when the
sea scallops were turned over so that the labels faced
the camera. The numbered sea scallops were assigned
randomly to fifteen groups. All members of the same
group were stored together in a bag with a unique label
for group identification.

In each experimental replicate, a group of shell
valves was placed randomly on the bottom of the tank.
Two video images were made for each group. The first
image (with the valve turned towards the sediment
and identification numbers hidden) was used by four
technicians to independently measure shell heights.
The second image was taken with identification num-
bers visible after divers turned the shells over and
replaced them in their original positions. After video
images were recorded, the shell valves were measured
with measuring boards by two technicians who could
not see the identification numbers and once by a third
technician with calipers.

A stock assessment model that incorporates
errors from shell-height measurements

Following NEFSC2’3 procedures, we used results from
experiment 2 and a modified version of the CASA (catch-
at-size-analysis, Sullivan et al., 1990) stock assessment
model (Appendix 1) to investigate potential effects of
shell-height measurement errors on model-based bio-
mass and fishing mortality estimates for two sea scallop
stocks. Assessment model results in this article should
not be used by managers because model runs were
tailored to investigate potential effects of shell-height
measurement errors and because some types of data
were omitted.

As described in Appendix 1, the CASA model that
is routinely used for sea scallop stock assessments ac-
commodates both bias and imprecision in shell-height
measurements. CASA models were run for sea scal-
lops in the Mid-Atlantic Bight during 1982-2006. In
contrast to NEFSC2, measurement error parameters
were obtained from experiments and not estimated in
the CASA model itself. The data used in modeling in-
cluded commercial landings in metric tons (t), survey
trend data (numbers per unit of sampling effort) from
the camera video and dredge surveys, and shell-height
composition data from the commercial fishery, video,
and dredge surveys. Survey selectivity patterns were
not estimated because the video and dredge surveys
have flat selectivity patterns (catch sea scallops equally
well) at shell height >40 mm, and goodness-of-fit calcu-
lations were restricted to this size range (Appendices
B7-B8 in NEFSC3). Measurement errors in commercial
shell-height data were assumed to be the same as those
in the dredge survey for lack of better information and

because procedures for measuring sea scallops on land
in port samples and at-sea in fishery observer samples
are similar to procedures followed in surveys.

As described in Appendix 1, bias and precision of
shell-height measurements are represented in the CASA
model by an error matrix ( E ) that gives the probability
that a sea scallop in each true shell-height bin is as-
signed to a range of observed shell-height bins (a range
that accommodates measurement errors). As described
by Methot (1989, 1990) for age data, the error matrix
E can be set up to deal with a wide range of situations
for bias and variance (e.g., both can vary among shell-
height bins or over time).

For the calculation of E for sea scallops in this analy-
sis, shell-height measurement error distributions were
assumed to be normally distributed with means and
standard deviations from experiment 2. The normal
distributions for measurement errors were truncated
three standard deviations above and below the mean.
In calculating distributions of measurement errors, true
shell heights were assumed with or without bias to be
uniformly distributed within each true 5-mm SH bin
so that, for example, the frequency of sea scallops with
true shell heights of 70, 71, 72, 73, and 74 mm (in the
70-74.9 mm SH bin with midpoint 72.5) was the same.
Distributions for measurement errors were normalized
to sum to one before use in the CASA model.

Results

Height and width measurements from the same tiles
in experiment 1 were not significantly different by a
paired t-test (£=-0.23, P=0.30, 91 df). Therefore, height
and width measurements from 91 tiles in experiment 1
were combined to form a single set of video data (a total
of 182 measurements) (Table 1).

The RMSE statistic for video tile-size composition
and measurement errors in experiment 1 (Table 1) was
3.5 mm (%RMSE = 7%, Table 1). Bias (-2.2 mm) and
imprecision (standard deviation 2.7 mm) of video tile
measurements were similar. In comparison to the true
size of the tiles (48.5 mm), the smallest measurement
was 38 mm, and the largest measurement was 50 mm.
The video size-composition data and measurement er-
rors were left skewed (g1=-0.28) and flatter (g2=-0.53)
than expected for a normal distribution. There were
gaps in the distribution of the video tile measurements
(Fig. 3) due to the resolution of the video images used
in digitizing (each pixef=3x3 mm).

Measurement error increased with DFO for the video
tile measurements (Fig. 3). Bias was positive for DFO
<400 mm and negative at larger DFO levels.

RMSE for shell-height composition data in experi-
ment 2 was 33 mm (%RMSE 30%) for video and 34 mm
(%RMSE = 31%) for measuring board data (Table 2).
Mean shell height was 106 mm for video and 109 mm
for measuring boards, compared to 110 mm for calipers.
Minimum shell height was 34 mm for video, 38 mm for
measuring boards, and 39 mm for calipers. Maximum

Jacobson et al.: Measurement errors in body size of Placopecten mage/lanicus

239

shell height was 201 mm for video, 193 mm for measur-
ing boards, and 192 mm for calipers.

Bland-Altman plots for experiment 2 show that mea-
suring board shell heights were more accurate than
video measurements, and that bias in video and mea-
suring board data was relatively constant across the
range of shell heights in experiment 2 (Fig. 4). However,
relatively large outliers sometimes occurred in video
measurements at 80-130 mm SH (Fig. 4).

Video and measuring-board shell-height compositions
in experiment 2 were similar in terms of skewness with
gj=-0.41 for video measurements and -0.47 for measur-
ing boards compared to -0.46 for calipers (Table 2). The
video shell-height distribution was more peaked with
g2~- 0.65 compared to g2=- 0.85 for measuring boards,
and g2=-0.84 for calipers (Table 2). Video measurement
errors were skewed to the left (g1=-0.60) compared to
measuring-board errors which were nearly symmetrical
(^1=-0.05). The distribution of errors for measuring
boards was flatter (g2~- 0.85) and video measurement
errors were more peaked (g2- 1-84) than would be ex-
pected for normal distribution. The error distribution
for measuring boards had a nearly flat mode about
5-mm wide because shell heights are automatically
truncated by measuring boards to the next lowest 5-mm
shell-height bin.

On a proportional basis, meat weights calculated from
shell heights in experiment 2 were much less accu-
rate than the original shell-height measurements. In

Table 1

Summary of size-composition data and measurement
errors for 182 tile measurements (height and width from
91 tiles, each 48.5x48.5 mm) by video equipment in exper-
iment 1.

Statistic

Video

Measurements and measurement errors

Bias

-2.2

Standard deviation

2.7

Square root of the mean squared error

3.5

Skewness (gq)

-0.28

Kurtosis (g2)

-0.53

Measurements

Minimum

38.3

5% quantile

41.2

95% quantile

50.1

Maximum

50.1

Average

46.3

Percent bias

-5%

Coefficient of variation

Percent square root of the mean squared

error 7%

particular, %RMSE values for meat weights were 71%
and 74% for video and measuring boards, respectively
(Table 3), compared to 30% and 31% for the original

Table 2

Summary statistics for shell-height composition data and measurement errors (in mm) from 172 uniquely identified Atlantic sea
scallop ( Placopecten magellanicus) shell valves in experiment 2. “NA” means that a statistic is not applicable.

Statistic

True shell height (calipers)

Video

Measuring boards

Shell heights and measurement errors

n measurements used

172

670

344

n omitted

Bias

-4.5

-0.6

Shell heights

Minimum

38.5

34.3

37.5

5% quantile

54.8

48.8

52.5

95% quantile

149.6

147.3

147.5

Maximum

192.0

200.6

192.5

Average

109.9

106.5

109.3

Percent bias

-4%

-1%

Standard deviation

33.5

33.1

33.6

Coefficient of variation

30%

31%

Square root of the mean squared error

33.4

33.6

Percent square root of the mean squared error

30%

31%

Skewness (gx)

-0.46

-0.41

-0.47

Kurtosis (g2)

-0.84

-0.65

-0.85

Measurement errors

Standard deviation

6.1

1.7

Square root of the mean squared error

7.6

1.8

Skewness (gj)

-0.60

-0.044

Kurtosis (g2)

1.84

-0.85

240

Fishery Bulletin 108(2)

shell heights (Table 2). The nonlinear shell-height to
meat-weight relationship showed exaggerated extremes
of the distributions so that the ratio of maximum to
mean meat weight was 158/27=5.9 for video data and
138/29 = 4.8 for measuring boards (Table 3) compared to
201/106=1.9 and 193/109=1.8 for shell heights (Table 2).
Variance in meat-weight measurements increases as
true meat-weight increases for video data and, to a
lesser extent, for measuring boards (Fig. 5).

The meat-weight composition data were more right
skewed (gj=1.53) and flatter (g2= 6.22) than the meat-
weight composition data from measuring boards
(^1=0.92 and ^2=2.61) or calipers (^=0.99 and g2=3.00).
Errors in meat-weight data were left skewed and not as
peaked for video (^1=-0.80 and g2=2.48) than measur-
ing board data (^1=-1.06 andg\2 = 4.68).

801 A

Measurement (mm)

i 1 1 1 1 1

0 200 400 600 800 1000

Distance from origin (mm)

Figure 3

(A) Video measurements for tiles in experiment 1. The verti-
cal line shows the true value at 48.5 mm. ( B) Measurement
errors (video measurement minus caliper measurement)
for tiles in experiment 1 as a function of distance from the
origin (DFO). The nonlinear LOESS regression line shows
the overall trend in measurement errors as a function of
DFO.

Results from the assessment models

Based on results from experiment 2 (Table 2) and
assumptions listed above, video shell-height measure-
ments for sea scallops with true sizes evenly distributed
over 100-104.99 mm SH (i.e., the 100-mm bin with
midpoint 102.5 mm) would fall into nine observed shell-
height bins with midpoints from 77.5 to 117.5 mm (Table
4). Measuring board shell-height measurements would
fall into four observed shell-height bins with midpoints
ranging from 92.5 to 107.5 mm (Table 4).

Four model configurations were used. The “no mea-
surement error” model configuration was fitted by as-
suming no errors in shell-height data. The “bias only”
model was fitted by assuming that shell-height data
were biased (to the extent measured in experiment 2),
but precise (with zero variance). The “imprecision
only” model was fitted by assuming that shell-height
measurements were imprecise (standard deviations
from experiment 2), but not biased. The “impreci-
sion and bias” model was fitted by assuming both
types of shell-height measurement errors.

Models which accommodated measurement errors
fitted better, with substantially lower negative log
likelihoods for both stocks, than models that ig-
nored measurement errors. Differences in negative
log likelihood were mostly for shell-height compo-
sition data. Mean 2004-06 biomass and fishing
mortality rates and coefficients of variation (CV)
for biomass and fishing mortality estimates were
similar for all model configurations (Table 5).

Discussion

The importance of body-size measurement errors
and the need to accommodate them in modeling
probably depends on the situation. Biological factors
(growth rate, recruitment variability), assessment
model type, quality and quantity of fishery and fish-
ery-independent data may be important. Sea scallops
may be an atypical case because they are a data-rich
species. We suggest that the potential importance of
body size measurement errors should be evaluated
on a case by case basis, particularly if body-size data
may be imprecise or biased. Simulation studies may
be useful in determining the importance of experi-
mentally derived body-size measurement errors on
stock assessment results.

In the sea scallop case, models that accommodat-
ed measurement errors fitted substantially better,
but there was little effect on point estimates and
variances for recent biomass and fishing mortality.
We hypothesize that effects on biomass and mortal-
ity estimates would be larger in cases with positive
biases in body-size measurements. For both video
and measuring boards, the positive bias in meat
weights due to the nonlinear relationship between
body size and meat weight was mitigated to some
extent by the negative bias in shell-height mea-

Jacobson et al.: Measurement errors in body size of Placopecten magellanicus

241

Measurement

Figure 4

Modified Bland-Altman plots for Atlantic sea scallop ( Placopecten
magellanicus) shell-height (SH) measurements in experiment 2. The
y-axis shows the difference between the experimental measurement
(measuring boards in A or video in B) and the caliper measure-
ment. The jc-axis shows the average of the experimental and caliper
measurement. Boxplots and 30-mm shell-height bins were used
instead of traditional scatter plots for shell height measurements
in experiment 2 because the large number of samples between 120
and 150 mm SH gave the impression that variance was higher for
those sizes. Boxplots show the interquartile range (a robust vari-
ance measure) and are not sensitive to sample size. The width of
the boxplots is proportional to the number of observations for the
shell-height bin.

surements. In contrast, Heery and Berkson
(2009) used simulations to evaluate effects of
systematic sampling errors (too many small
or too many large individuals) in size-com-
position data from commercial catches and
three simulated stocks. The simulated data
were used in a forward-projecting age-struc-
tured stock assessment and in projection
models to estimate stock size and fishing
mortality in relation to threshold values, and
rebuilding trajectories. Body-size data with
too many large individuals biased stock size
high and fishing mortality low and tended to
support management measures that did not
meet management goals, particularly for lon-
ger lived and depleted stocks. Body-size data
with too many small individuals were less
problematic, but tended to support overly
restrictive management actions in extreme
cases. Heery and Berkson’s (2009) results
indicate that systematic errors in sampling
may be more important than errors in indi-
vidual measurements of body size.

Variance in calculated meat weights in-
creased rapidly with shell height with both
video and measuring board techniques, in
contrast to the variance in shell heights
(Figs. 4 and 5). This additional source of
variability likely increases variance in bio-
mass estimates, particularly for relatively
large fishable sea scallops.

In our analysis, assessment models that
accommodated shell-height measurement er-
rors fitted better, even though no additional
parameters were estimated. The Mid-Atlan-
tic Bight model that accommodated impre-
cise (but not biased) shell-height measure-
ment errors had a negative log likelihood
that was 15 units smaller than the negative
log likelihood for the no measurement er-
ror model (Table 5). Results for the Georges
Bank stock (not shown to conserve space) were similar.
In contrast and based on likelihood theory, a difference
in negative log likelihoods of just 1.92 units is sufficient
to justify an additional parameter in a statistical model
at the P= 0.05 level (Venzon and Moolgavkar, 1988).
Comparing results of the “bias only” scenario to results
from the “imprecision only” and “imprecision and bias”
scenarios, we found that improvements in goodness of
fit were mostly due to accommodating imprecision; bias
was less important (Table 5).

Experiments

Our results highlight the value and information that
may be gained from evaluating body size measurement
errors experimentally. Body-size measurement error
experiments should be conducted when survey equip-
ment is changed, particularly if body-size measurements
are imprecise. In some cases, frequent “mini-experi-

ments” may be required if the accuracy of the equip-
ment tends to drift over time or change in response to
environmental conditions.

Our results indicate the importance of designing mea-
surement error experiments so that individual speci-
mens can be identified and associated with individual
measurements; otherwise measurement errors can not
be estimated individually and evaluated directly. Data
from experiment 2 were most useful because individual
sea scallops were numbered and replicate measure-
ments of different types could be linked and analyzed
in detail. In addition, the full range of variability for all
important factors (i.e., distance from the origin (DFO),
shell height, and identity of individual technicians)
should be included in the experimental design.

We ignored skewness and kurtosis in measurement
errors in calculating measurement error matrices for
use in the CASA stock assessment model. In future
modeling, it may be better to use the experimental dis-

242

Fishery Bulletin 108(2)

Table 3

Summary statistics of meat weights and meat weight measurement errors (g) for Atlantic sea scallop (Placopecten magellanicus)
shell-height measurements in experiment 2 (sample sizes are the same as those for shell-height measurements in Table 2). The
original shell heights were obtained with calipers, video camera, and measure boards. “NA” means that a statistic is not applicable.

Statistic

True (calipers)

Video

Measuring boards

Meat weights and measurement errors

Bias

-3.2

-0.4

Meat weights

Minimum

0.8

0.5

0.7

5% quantile

2.4

1.7

2.1

95% quantile

61.3

58.3

58.6

Maximum

136.9

157.7

138.0

Average

29.8

27.3

29.4

Percent bias

-10%

-1%

Standard deviation

22.2

21.4

21.8

Coefficient of deviation

74%

78%

74%

Square root of the mean squared error

21.6

21.8

Percent square root of the mean squared error

71%

74%

Skewness (gx)

0.99

1.53

0.92

Kurtosis (g2)

3.00

6.22

2.61

Measurement errors

Standard deviation

5.1

1.5

Square root of the mean squared error

6.0

1.6

Skewness (g^)

-0.80

-1.06

Kurtosis ( g2 )

2.48

4.68

Table 4

Estimated probability distributions for Atlantic sea scallop ( Placopecten magellanicus) shell-height (SH) measurements based on
bias and standard deviations from experiment 2. Condition factors for error matrices used in the catch-at-size-analysis (CASA)
stock assessment model scenarios are given also. The shell-height bins are 5-mm wide and identified by their midpoint. For
example, sea scallops 80-84.9 mm SH fall into a bin whose midpoint is 82.5 mm.

Video scenario

Measuring board scenario

Calipers

Imprecision

Statistic (true shell height)

Bias only

only

and bias

Bias only

only

and bias

Condition factor (k)

3xl015

5457

2638

1.6

2.1

2.3

Bias (mm)

-4.5

-0.6

Standard deviation (mm)

6.1

1.7

Shell height bin (mm)

Probability of observed bins

72.5

77.5

0.0009

82.5

0.0014

0.0167

87.5

0.0203

0.0820

92.5

0.0929

0.2158

0.0001

97.5

0.8000

0.2300

0.3101

0.2000

0.1325

0.2008

102.5

1.0000

0.2000

0.3110

0.2436

0.8000

0.7349

0.7181

107.5

0.2300

0.1045

0.1325

0.0810

112.5

0.0929

0.0243

117.5

0.0203

0.0020

122.5

0.0014

127.5

Jacobson et al.: Measurement errors in body size of Placopecten magellanicus

243

50 100

Mean shell height (mm)

Figure 5

Bland-Altman plots for Atlantic sea scallop ( Placopecten mag-
ellanicus) meat weights calculated from experimental shell-
height measurements in experiment 2 (measuring boards in
panel A and video in panel B). The y-axis shows the difference
between the meat weights calculated from the experimental
(video or measuring board) shell height measurements and the
meat weights calculated from caliper measurements. The x-axis
shows the average of the experimental and caliper-derived
measurements.

tributions of measurement errors directly in er-
ror matrices, particularly if experimental sample
sizes are large.

Drouineau et al. (2008) used simulation analy-
sis to show the importance of alternative as-
sumptions about the distribution of individuals
within size groups and the statistical distribu-
tion of growth increments in length-structured
models like the CASA (catch-at-size-analysis)
model. Our experience indicates that the same
types of assumptions are important in calcu-
lating body-size measurement-error matrices.

In particular, it was important to assume that
individuals were uniformly distributed within
size groups, to make realistic assumptions about
the distributions of measurement errors, and to
be careful in programming to ensure consistent
calculations at the boundaries of length bins for
calculating error matrices and for the stock as-
sessment model.

Statistical methods for repeated measurements
or random effects may be suitable for analysis of
our experimental data. We made allowances for
repeated measures in bootstrap calculations (Ap-
pendix 2) and in calculating P-values for skew-
ness and kurtosis tests, but not in calculating
other statistics (Tables 1-3).

Our experiments were conducted under ideal
conditions with tiles and shell valves, rather
than live sea scallops. Our results may under-
estimate the magnitude of errors under more
realistic field conditions.

Model results may depend on shell-height bin
width such that larger shell height bins would
cause measurement errors to have a greater im-
pact on biomass and mortality estimates. We
used 5-mm SH bins for sea scallops because 5-
mm is the resolution and approximate accuracy
for the survey shell-height data. In general, it
may be important to consider the magnitude of
measurement errors in making decisions about size bins
used in stock assessment modeling.

Body-size measurement errors

Random measurement errors are unavoidable. One may
conclude that it is incumbent on the researcher to search
out and correct sources of bias, whatever the source. We
suggest that it may be more cost effective to quantify
measurement errors experimentally and to accommodate
them in modeling. Time series with consistent body-size
measurement errors are probably easiest to interpret.
Models may become overly complex if multiple sets of
assumptions about measurement errors are required
to interpret one survey time series. Resources required
to quantify measurement errors after each adjustment
to survey procedures or equipment may be better spent
on more accurately characterizing the measurement
errors for survey gear that remains the same for longer
periods of time.

Bootstrap results also showed that an algebraic ap-
proach to removing errors from the data by using the
inverse error matrix E _1 gave negative proportions for
both video and measuring board data in at least some
size groups (Appendix 2). The sampling distribution for
algebraically adjusted shell-height data may be difficult
to characterize. These results indicate that it may be
difficult to remove measurement errors directly from
body-size data and we hypothesize that approaches like
the one used in the CASA model will generally perform
better. Bootstrap results showed that estimates of pre-
dicted shell-height composition data with measurement
errors as carried out in the CASA model were robust
to uncertainties in the measurement-error matrix E
(Appendix 2). Models can be designed to be robust to
measurement errors. For example, the last size bin in
the CASA model is a plus-group that absorbs data for
large scallops that may have been strongly affected
by measurement errors. Other data in the model may
have also contributed to the robustness of biomass and

244

Fishery Bulletin 108(2)

Table 5

Results from the catch-at-size-analysis (CASA) model for Mid-Atlantic Bight sea scallops (Placopecten magellanicus) and four
model configurations. The “no measurement error” model configuration does not accommodate shell-height measurement errors.
Other model configurations accommodate bias and imprecise measurement errors in various combinations as shown in the table.
Lower negative log likelihood (NLL) values indicate better model fit. Coefficients of variation (CV) shown in parenthesis are
asymptotic variances calculated by the delta method. For ease of comparison, the “no measurement error” configuration NLL
values were subtracted from corresponding NLL statistics for all three configurations. The lowest NLL, biomass or fishing mor-
tality estimates in each row are printed in boldface.

Variable or estimate

measurement

error

Bias only

Imprecision

only

Imprecision

and

bias

Bias and precision (mm) assumed in modeling

Standard deviation — video survey

0.0

6.1

Bias — video survey

0.0

-4.5

0.0

-4.5

Standard deviation — dredge survey

0.0

1.7

Bias — survey

0.0

-0.6

0.0

-0.6

Negative log likelihood (NLL)

Total

0.00

20.92

-14.62

-1.16

Commercial fishery shell-height data

0.00

4.99

-0.34

2.06

Dredge survey shell-height data

0.00

-4.14

-10.66

-6.97

Video survey shell-height data

0.00

19.45

-3.00

4.59

Mean biomass and fishing mortality during 2004-06

Fishing mortality (y-1)

0.45

0.41

0.46

0.42

(8%)

(7%)

(8%)

Biomass (t meats)

81,211

84,650

80,844

83,602

(5%)

fishing mortality estimates to assumptions about shell-
height measurement errors.

In principal, measurement-error parameters could be
estimated directly in stock assessment models without
resorting to experiments. Measurement-error param-
eters in the CASA model were estimated in the NEFSC
study,2 but the estimates proved to be unstable (NEF-
SC3). Without at least one source of accurate body-size
data, there may be too little information about mea-
surement errors to estimate parameters. In addition,
there may be strong correlations between estimated
measurement errors and estimates of other factors that
affect interpretation of body-size data, such as survey
and fishery selectivity, natural mortality, and recruit-
ment variability.

Acknowledgements

We thank F. Serchuk (Northeast Fisheries Science
Center, Woods Hole, MA), S. Correia (Massachusetts
Division of Marine Fisheries, New Bedford, MA), C.
O’Keefe and C. Adams (SMAST, New Bedford, MA),
and five anonymous reviewers for useful technical and
editorial suggestions. We are grateful for support from
the School of Marine Science and Technology, the Mas-
sachusetts Division of Marine Fisheries, and NOAA
awards: NA04NMF4720332, NA05NMF4721131, and

NA06NMF4720097. We are grateful to the crews and
scientific staff who collected and measured sea scallops
in NEFSC and SMAST surveys. Live sea scallops used in
the experiments were provided by commercial sea scallop
vessels from New Bedford and Fairhaven, MA.

Literature cited

Bland, J. M., and D. G. Altman.

1986. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet
1:307-310.

1995. Comparing methods of measurement: why plotting
difference against standard method is misleading. Lan-
cet 346:1085-1087.

Butler, J., M. Neuman, D. Pinkard, R. Kvitek, and G. Cochrane.

2006. The use of sonar mapping techniques to refine
population estimates of the endangered white abalone
( Haliotis sorenseni). Fish. Bull. 104:521-532.
Cochran, W. G.

1977. Sampling techniques, 428 p. John Wiley and Sons,
Inc., New York.

Drouineau, H., S. Mahevas, M. Bertignac, and A. Fertin.

2008. Assessing the impact of discretisation assumptions
in a length-structured population growth model. Fish.
Res. 91:160-167.

Edgar, G. J., N. S. Barrett, and A. J. Morton.

2004. Bias associated with the use of underwater
visual census techniques to quantify the density and

Jacobson et al.: Measurement errors in body size of Placopecten magellanicus

245

size-structure of fish populations. J. Exp. Biol. Ecol.
308:269-290.

Feller. W.

1966. An introduction to probability theory and its appli-
cations. Volume II. 626 p. John Wiley and Sons, Inc.,
New York.

Fournier, D., and C .P. Archibald.

1982. General theory for analyzing catch at age
data. Can. J. Fish. Aquat. Sci. 39:1195-1207.

Gedamke, T., and J. M. Hoenig.

2006. Estimating mortality from mean length data
in nonequilibrium situations, with application to
the assessment of goosefish. Trans. Am. Fish. Soc.
135:476-487.

Heery, E. C., and J. Berkson.

2009. Systematic errors in length frequency data and their
effect on age-structured stock assessment models and
management. Trans. Am. Fish. Soc. 138:218-232.

Horn, R. A., and C. R. Johnson.

1985. Matrix analysis, 561 p. Cambridge Univ. Press,
New York.

Methot, R. D.

1989. Synthetic estimates of historical abundance and
mortality for northern anchovy. Am. Fish. Soc. Symp.
6:66-82.

1990. Synthesis model: an adaptive framework for analy-
sis of diverse stock assessment data. Int. N. Pacific
Fish. Comm. Bull. 50:259-277.

Pennington, M., L.-M. Burmeister, and V. Hjellvik.

2001. Assessing the precision of frequency distribu-
tions estimated from trawl-survey samples. Fish.
Bull. 100:74-80.

Quinn, T., and R. B. Deriso.

1999. Quantitative fish dynamics, 570 p. Oxford Univ.
Press, Oxford, U K.

Ricker, W. E.

1975. Comparison and interpretation of biological sta-
tistics of fish populations. Fish. Res. Board Can. Bull.
191:1-382.

Rosenkranz, G. E., and S. C. Byersdorfer.

2004. Video scallop survey in the eastern Gulf of Alaska,
USA. Fish. Res. 69:131-140.

Serchuk, F. M, and S. E. Wigley.

1986. Evaluation of USA and Canadian research vessel
surveys for sea scallops, (Placopecten magellanicus) on
Georges Bank. J. Northw. Atl. Fish. Sci. 7:1-13.

Serchuk, F. M., P. W. Wood Jr., J. A. Posgay, and B. E. Brown.

1979. Assessment and status of sea scallop (Placopec-
ten magellanicus), populations off the Northeast coast
of the United States. Proc. Natl. Shellfish. Assoc.
69:161-191.

Sokal, R. R., and F. J. Rohlf.

1995. Biometry, 859 p. Freeman, New York.

St. John, J., G. R. Russ, and W. Gladstone.

1990. Accuracy and bias of visual estimates of numbers,
size structure, and biomass of a coral reef fish. Mar.
Ecol. Prog. Ser. 64:253-262.

Stokesbury, K. D. E.

2002. Estimation of sea scallop abundance in closed
areas of Georges Bank, USA. Trans. Am. Fish. Soc.
131:1081-1092.

Stokesbury, K. D. E., B. P. Harris, M. C. Marino II, and J. I.

Nogueira.

2004. Estimation of sea scallop abundance using a
video survey in off-shore US waters. J. Shellfish Res.
23:33-44.

Sullivan, P. J., H.-L. Lai, and V. F. Gallucci.

1990. A catch-at-length analysis that incorporates a
stochastic model of growth. Can. J. Fish. Aquat. Sci.
47:184-198.

Venzon, D. J., and S. H. Moolgavkar.

1988. A method for computing profile-likelihood based
confidence intervals. Appl. Stat. 37:87-94.

Tukey, J. W.

1977. Exploratory data analysis, 688 p. Addison-Wesley
Pub. Co., Reading, MA.

Appendix 1

Following the approach of the Northeast Fisheries Sci-
ence Center (NEFSC, 2>3) we used a likelihood approach
to fitting the CASA model to sea scallop stock assessment
data. The best estimates from the model minimized the
combined negative log likelihood of all the data. Relevant
details are described below. Appendix B10 in the NEFSC
report (NEFSC3) is a complete technical description of
the CASA model for sea scallops. Appendix B12 in that
same report (NEFSC3) describes CASA model perfor-
mance with simulated stock assessment data.

Estimates of population abundance and survey size
selectivity are available for each shell height and year
as the CASA model is fitted. In a single year, for ex-
ample, we calculated the number of sea scallops in the
population that were available or selected by the video
gear with the following equation:

nh=QhNh > (A1>

where Nh = the predicted number of sea scallops in the
population for shell height bin h;
qh = the size-specific probability of detection
(selectivity) in the video survey (on a scale
of 0 to 1 and relative to the bin with maxi-
mum probability of detection); and
nh = the estimated number of sea scallops in the
population that are available to the video
survey gear.

In the absence of measurement error, the predicted shell-
height composition jih for the survey is

1=1

where L = the number of shell-height bins in the model.

If if is a row vector of length L containing the predicted
proportions (before measurement errors) for each length
group in the survey, then

p = nE, (A3)

where p the row vector of predicted proportions (includ-
ing measurement errors).

246

Fishery Bulletin 108(2)

In Equation A3, E is a square measurement error ma-
trix with L rows and columns that distributes numbers
at true shell height into observed shell heights bins that
are larger and smaller than the true shell height. For
example, the first row of E sums to one and gives the
probability of observed shell heights for sea scallops in
the first true shell height bin. The last row of E sums
to one and gives the probabilities that sea scallops in
each shell height bin would be assigned to the “plus
group” because of measurement error. As described in
the text, we estimated E for sea scallops using results
from experiment 2.

Appendix 2

Equation A3 in Appendix 1 indicates the possibility of
correcting shell-height data measurement algebraically,

without resorting to an approach like the CASA model.
In particular, if the matrix E is invertible, then it may
be possible to estimate the true sample proportions n
by multiplying both sides of Equation A3 by the inverse
matrix E~u.

n - pE~x. (A4)

However, the inverse calculation in Equation A4 will be
unreliable if the estimated error matrix E is poorly con-
ditioned. If the error matrix is poorly conditioned, then
small inaccuracies in the estimate of E will propagate
into larger errors in the inverse E~l and the predicted
proportions k.

As described by Horn and Johnson (1985), the condi-
tion factor for an invertible matrix E is

k=\e ||||e'1||, (A5)

where ||Z?|| = the matrix norm of E.

The condition factor k is always at least one and
is an upper bound measure of the extent to which
errors in the original error matrix E (ignoring
errors in p ) will propagate to its inverse. If k is
slightly larger than one, then uncertainty in E 1
and n from Equation A4 will be at most slightly
greater then uncertainty in E. If k is large, then
uncertainty in E -1 and n may be much larger
than uncertainty in E.

The measurement-error matrices that included
both bias and imprecision are the most realistic
according to results from experiment 2. The con-
dition factors for these error matrices were 2638
for video and 2.3 for measuring boards (Table 4).
These condition factors indicate that uncertainty
in E -1 and “corrected” shell-height composition
data could be much higher than uncertainty in
the original error matrix E for video and at most
2.3 times higher for measuring boards.

Bootstrap analyses show the practical signifi-
cance of condition factors for video and measur-
ing board data in our study. For example, for the
video shell-height measurements in experiment
2, the first step was to resample n data records
(including one video measurement and the corre-
sponding caliper measurement) with replacement
from the data in experiment 2.

Sample sizes (77 = 670 for video and 77 = 344 for
measuring boards) were the same as the number
of experimental measurements and constituted
an upper bound on the true effective sample size
because they ignore repeated measurements on
the same specimens (Table 2). The effect of us-
ing an upper bound estimate for effective sample
size was to understate effects of uncertainty in
error matrices. Our interest was, however, in a
“best case” scenario with relatively large sample
sizes. Next, the measurement errors (e.g. video
or measuring board minus caliper measure-

0.10-

0.15

0.00

Measurement

boards

S-

' VA

Video

100

50 -

-50 -

-100

B— — —

1 1 1 1 1 1
30 50

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
70 90 110 130 150

1 1 1 1
185

• .

o § 8

§ 1 ?
i i T : i
I — : R_l|—

if s 8 r i i * °

Rath ■ ^ „ o a . - „ » „ .«

1 - : V

8 tig
“ 8 §

Borers . _ t ; * 8

lift s ° T T T o

? § s

o §

° ° o

•

i i i i i i i

30 50

150 185

90 110 130

Shell height (mm)

Appendix Figure 1

Boxplots showing bootstrap distributions (1000 iterations) of
estimated true shell-height (SH) composition for Atlantic sea
scallops ( Placopecten magellanicus ) in experiment 2, based on
measurement boards (A) and video (B) shell-height data. True
shell-height compositions were estimated by using bootstrap
estimates of the inverse of the measurement error matrix E
and Equation A4. The solid line in (A) shows the actual caliper-
derived shell-height data in the experiment. The solid line is
not visible in (B) because of the scale of the y-axis.

Jacobson et al.: Measurement errors in body size of Placopecten magellanicus

247

Shell height (mm)

Appendix Figure 2

Bootstrap distributions (1000 iterations) for Atlantic sea scal-
lop ( Placopecten magellanicus ) shell-height data obtained from
measurement boards (A) and video (B), with measurement
errors. The solid line shows the actual caliper-derived shell-
height data in experiment 2.

ments), their mean (bias), and variance were
used to calculate the bootstrap measurement er-
ror matrix and its inverse. Finally, the original
video shell-height composition data used in ex-
periment 2 (expressed as proportions) were then
multiplied by the bootstrap inverse matrix (Eq.

A4) to remove measurement errors and obtain a
bootstrap estimate of the true shell-height com-
position. There were 1000 bootstrap iterations
for both the video and measurement board data.

The variability among bootstrap estimates of the
true shell-height composition was due entirely to
errors in the measurement error matrix E and
its inverse .

As expected, based on condition factors (see
above) and measurement error statistics (Table
2), bootstrap estimates of true caliper shell-
height composition data from video data were
highly variable and predicted proportions ranged
from -188 to 195 (i.e., outside the feasible range
for proportions). Bootstrap estimates from mea-
surement board data resembled the correspond-
ing true caliper measurements. However, the
estimated proportions for both measurement
methods were often negative and infeasible (Ap-
pdx. Fig. 1).

We used a similar bootstrap procedure to eval-
uate effects of uncertainty in predicted length
compositions with measurement errors (Eq. A3
in Appdx. 1), which is the approach used in the
CASA model. In this bootstrap analysis, the
caliper shell height composition data from ex-
periment 2 were assumed to be true and error
matrices were generated by bootstrapping the
experimental and video and measuring board
data as described above. The sample size was
n=172 for both video and measuring boards and
the same as the number of individual specimens
in experiment 2. This lower bound estimate of
the effective sample size was used in order to
overstate effects of uncertainty in error matrices. Re-
sults indicated that the calculations used in the CASA
model for measurement errors were robust to uncer-

tainty about the error matrices and the magnitude of
the errors because variability in predicted shell height
compositions was relatively minor (Appdx. Fig. 2).

248

Fishery Bulletin

Guidelines for authors

Manuscript Preparation

Contributions published in Fishery Bulletin describe
original research in marine fishery science, fishery
engineering and economics, as well as the areas of
marine environmental and ecological sciences (including
modeling). Preference will be given to manuscripts that
examine processes and underlying patterns. Descriptive
reports, surveys, and observational papers may occa-
sionally be published but should appeal to an audience
outside the locale in which the study was conducted.
Although all contributions are subject to peer review,
responsibility for the contents of papers rests upon the
authors and not on the editor or publisher. Submission
of an article implies that the article is original and is not
being considered for publication elsewhere. Articles may
range from relatively short contributions (10-15 typed,
double-spaced pages, tables and figures not included)
to extensive contributions (20-30 typed pages). Manu-
scripts must be written in English; authors whose native
language is not English are strongly advised to have
their manuscripts checked by English-speaking col-
leagues before submission.

Title page should include authors’ full names and
mailing addresses and the senior author’s telephone,
fax number, and e-mail address, and a list of key words
to describe the contents of the manuscript. Abstract
should be limited to 200 words (one-half typed page),
state the main scope of the research, and emphasize
the author’s conclusions and relevant findings. Do not
review the methods of the study or list the contents of
the paper. Because abstracts are circulated by abstract-
ing agencies, it is important that they represent the
research clearly and concisely. Text must be typed in
12 point Times New Roman font throughout. A brief
introduction should convey the broad significance of
the paper; the remainder of the paper should be divided
into the following sections: Materials and methods,
Results, Discussion (or Conclusions), and Acknowl-
edgments. Headings within each section must be short,
reflect a logical sequence, and follow the rules of multi-
ple subdivision (i.e., there can be no subdivision without
at least two items). The entire text should be intelligible
to interdisciplinary readers; therefore, all acronyms,
abbreviations, and technical terms should be written
out in full the first time they are mentioned. Include
FAO common names for species in the list of keywords
and in the introduction. Regional common names may
be used throughout the rest of the text if they are dif-
ferent from FAO common names which can be found at
http://www.fishbase.org/search.html. Follow the U.S.
Government Printing Office Style Manual (2000 ed.)
and Scientific Style and Format: the CSE Manual for
Authors, Editors, and Publishers (7th ed.) for editorial

style; for fish nomenclature follow the most current issue
of the American Fisheries Society’s Common and Scien-
tific Names of Fishes from the United States, Canada,
and Mexico, 6th ed. Dates should be written as follows:
11 November 2000. Measurements should be expressed
in metric units, e.g., 58 metric tons (t); if other units of
measurement are used, please make this fact explicit
to the reader. Write out the numbers zero through
nine unless they form part of measurement units (e.g.,
nine fish but 9 mm). Refrain from using the shorthand
slash (/), an ambiguous symbol, in the general text
Literature cited comprises published works and
those accepted for publication in peer-reviewed literature
(in press). Follow the name and year system for citation
format in the “Literature cited” section (that is say,
citations should be listed alphabetically by the authors’
last names, and then by year if there is more than one
citation with the same authorship). If there is a sequence
of citations in the text, list chronologically: (Smith,
1932; Green, 1947; Smith and Jones, 1985). Abbrevia-
tions of serials should conform to abbreviations given in
the Serial Sources for the BIOSIS Previews Database.
Authors are responsible for the accuracy and complete-
ness of all citations. Literature citation format: Author
(last name, followed by first-name initials). Year. Title
of report or manuscript. Abbreviated title of the series
to which it belongs. Always include number of pages.
Cite all software and special equipment or chemical
solutions used in the study, not in a footnote but within
parentheses in the text (e.g., SAS, vers. 6.03, SAS Inst.,
Inc., Cary, NC).

Tables are often overused in scientific papers; it
is seldom necessary or even desirable to present all
the data associated with a study. Tables should not
be excessive in size and must be cited in numerical
order in the text. Headings should be short but ample
enough to allow the table to be intelligible on its own.
All unusual symbols must be explained in the table
legend. Other incidental comments may be footnoted
with italic numeral footnote markers. Use asterisks to
indicate probability in statistical data. Do not type table
legends on a separate page; place them above the table
data. Do not submit tables in photo mode.

• Zeros should precede all decimal points for values
less than one.

• Sample size, n, should be italicized.

• Capitalize the first letter of the first word in all labels
within figures.

• Do not use overly large font sizes in maps and for units
of measurements along axes in figures.

• Do not use bold fonts or bold lines in figures.

• Do not place outline rules around graphs.

• Do not use horizontal lines through graphs to indicate
measurement units.

• Use a comma in numbers of five digits or more (e.g.
13,000 but 3000).

• Maps require a North arrow and degrees latitude-
longitude (e.g., 170°E).

Fishery Bulletin 108(2)

249

Figures include line illustrations, photographs (or
slides), and computer-generated graphs and must be
cited in numerical order in the text. Graphics should
aid in the comprehension of the text, but they should
be limited to presenting patterns rather than raw data.
Figures should not exceed one figure for every four
pages of text. Figures must be labeled with author’s
name and number of the figure. Avoid placing labels
vertically (except of y axis). Figure legends should
explain all symbols and abbreviations and should be
double-spaced on a separate page at the end of the
manuscript. Color is allowed in figures to show morpho-
logical differences among species (for species identifica-
tion), to show stain reactions, and to show gradations in
temperature contours within maps. Color is discouraged
in graphs, and for the few instances where color may
be allowed, the use of color will be determined by the
Managing Editor.

Failure to follow these guidelines
and failure to correspond with editors
in a timely manner will delay
publication of a manuscript.

Copyright law does not apply to Fishery Bulletin,
which falls within the public domain. However, if an
author reproduces any part of an article from Fishery
Bulletin in his or her work, reference to source is consid-
ered correct form (e.g.. Source: Fish. Bull 97:105).

Submission

The Scientific Editorial Office encourages authors
to submit their manuscripts as a single PDF (pre-
ferred) or Word (zipped) document by e-mail to

Fishery.Bulletin@noaa.gov. Please use the subject
heading, “Fishery Bulletin manuscript submission”.
Do not send encrypted files. Please provide names
and contact information for 3-4 suggested reviewers.
Commerce Department personnel should submit papers
under a completed NOAA Form 25-700. Or you may
send your manuscript on a compact disc in one of the
above formats. For further details on electronic sub-
mission, please contact the Scientific Editorial Office
directly (see address below).

Richard D. Brodeur, Ph.D.

Scientific Editor, Fishery Bulletin
Northwest Fisheries Science Center

2030 S. Marine Science Dr.

Newport, Oregon 97365-5296

Once the manuscript has been accepted for publication,
you will be asked to submit a final electronic copy of
your manuscript. When requested, the text and tables
should be submitted in Word or Word Rich Text Format.
Figures should be sent as PDF files, Windows metafiles,
tiff files, or EPS files. Send a copy of figures in the origi-
nal software if conversion to any of these formats yields
a degraded version.

Questions? If you have questions regarding these
guidelines, please contact the Managing Editor, Sharyn
Matriotti, at

Sharyn.Matriotti@noaa.gov

Questions regarding manuscripts under review should
be addressed to Richard Brodeur, Scientific Editor, at
Rick.Brodeur@noaa.gov.

Fishery Bulletin

Subscription form

Superintendent of Documents Publications Order Form
*5178

I I YES, please send me the following publications:

Subscriptions to Fishery Bulletin

for $36.00 per year ($50.40 foreign)

The total cost of my order is $ . Prices include regular domestic

postage and handling and are subject to change.

(Company or Personal Name) (Please type or print)

(Additional address/attention line)

(Street address)

(City, State, ZIP Code)

(Daytime phone including area code)

(Purchase Order No.)

Charge

your

order.

IT’S

EASY!

Please Choose Method of Payment:

| | Check Payable to the Superintendent of Documents

| | GPO Deposit Account

] VISA or MasterCard Account your orders

| | | | | p| | | | | | | | | | | | | | | (202) 512-2250

(Credit card expiration date)

(Authorizing Signature)

Mail To: Superintendent of Documents

P.O. Box 371954, Pittsburgh, PA 15250-7954

Thank you for
your order!

Also available online at http://bookstore.gpo.gov/collections/fishery-bulletin

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Fishery bulletin"

See other formats