Fisheries Research 165 (2015) 100–114
Contents lists available at ScienceDirect
Fisheries Research
journal homepage: www.elsevier.com/locate/fishres
Variability in trophic positions of four commercially important
groundfish species in the Gulf of Alaska
Jennifer M. Marsh a,∗ , Robert J. Foy b , Nicola Hillgruber a,1 , Gordon H. Kruse a
a
b
University of Alaska Fairbanks, School of Fisheries and Ocean Sciences, 17101 Point Lena Loop Road, Juneau, AK 99801, USA
Kodiak Laboratory, Alaska Fisheries Science Center, National Marine Fisheries Service, NOAA, 301 Research Court, Kodiak, AK 99615, USA
a r t i c l e
i n f o
Article history:
Received 7 May 2014
Received in revised form 2 January 2015
Accepted 4 January 2015
Handling Editor Prof. George A. Rose
Keywords:
Mean trophic level of catch
Stable isotope analysis
Ecosystem indicators
Gulf of Alaska
Groundfish
a b s t r a c t
We examined trends in nitrogen stable isotope data as a proxy for trophic position (mean trophic level,
TL) of commercial and survey catches as an ecosystem-based indicator of sustainability of four groundfish species in the Gulf of Alaska. From 2000 to 2004, walleye pollock (Gadus chalcogrammus), Pacific
cod (Gadus macrocephalus), arrowtooth flounder (Atheresthes stomias), and Pacific halibut (Hippoglossus
stenolepis) were collected from the waters surrounding Kodiak Island, Alaska. Several analyses of covariance (ANCOVA) models were tested to detect variations in mean TL among years with fish length as
a covariate. Best-fit models were selected using the Akaiki Information Criterion to estimate trends in
mean TL of commercial catch using length-frequency data from onboard fishery observers for each target species. Then, linear regression models were used to estimate mean TL of commercial catch over
1990–2009 and the mean TL of population biomass over 1984–2007 based on length-frequency data and
biomass estimates from trawl surveys conducted by National Marine Fisheries Service and from historical
catch data. The TL of catch for each species except walleye pollock remained stable over the time frame
of the study. Walleye pollock TLs became increasingly variable after 1999. Similar trends in mean TL
were observed for the survey biomass of walleye pollock. Additionally, there was an observed decrease
of the occurrence of higher TL Pacific halibut over time. While the decline had no impact on overall TL
estimates during 1990–2009, a continued decline may affect mean TL in the future. Overall, length seems
to be the most important factor in estimating a species’ TL. Therefore, including relationships between
length of catch and TL estimates could lead to an early detection of TL declines that may be associated
with unsustainable fishing mortality.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
Since the 1990s, fisheries managers have been advised to
broaden awareness beyond a single-species focus by taking an
ecosystem approach to fisheries, also termed ecosystem-based
fisheries management (EBFM; US National Research Council, 1998).
The broader set of considerations in EBFM include system sustainability (or productivity), maintenance of biodiversity, and
protection of habitat (Sinclair et al., 2002; Zhang et al., 2009). This
more holistic approach has led to the development of a host of
ecosystem indicators as metrics to evaluate whether EBFM objectives are being achieved (FAO, 2003; Livingston et al., 2005). Some
∗ Corresponding author. Tel.: +1 011 907 796 5462; fax: +1 011 907 796 5447.
E-mail address: [email protected] (J.M. Marsh).
1
Current address: Thünen Institute of Fisheries Ecology, Wulfsdorfer Weg 204,
D-22926 Ahrensburg, Germany.
http://dx.doi.org/10.1016/j.fishres.2015.01.003
0165-7836/© 2015 Elsevier B.V. All rights reserved.
of these ecosystem indicators involve considerations of trophic
position or trophic level (TL). Following the practice in the marine
fisheries literature, here we use TL and trophic position interchangeably.
The mean TL of fisheries landings has been advanced as a
particularly important indicator in exploited marine ecosystems
(Pauly et al., 1998, 2001); trends in this index have been adopted
as indicators of sustainability and/or biodiversity (Gislason et al.,
2000; Livingston et al., 2005; Zhang et al., 2009, 2011). Using
mass-balance models and United Nations Food and Agriculture
Organization (FAO) data, Pauly et al. (1998) documented a worldwide average decrease of 0.1–0.3 TL per decade in commercial
catches over 1950–1994. The proposed mechanism, “fishing down”
marine food webs, is a process that commences with depletion of
high TL predators, followed by the sequential depletion of successively lower TL fisheries (Pauly et al., 1998), resulting in a decline
in mean TL of the catch and simplification of the food web (Pauly
et al., 2002). An alternative mechanism, “fishing through” marine
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
food webs, also leads to declines in mean TL of commercial catches,
but through addition of lower TL fisheries while maintaining higher
TL fisheries (Essington et al., 2006). Declines in mean TL have mostly
been observed in the North Atlantic Ocean (Pauly et al., 1998, 2001).
In the North Pacific Ocean (western and eastern regions combined),
the mean TL of commercial landings also declined between the
1970s and 1990s (Pauly et al., 1998); however, in the northeast
Pacific Ocean (i.e., Gulf of Alaska (GOA), Bering Sea, and Aleutian
Islands), this index remained stable over a similar time period
(Livingston, 2005).
This use of mean TL of the catch as an ecosystem indicator
of unsustainable fisheries and biodiversity degradation has been
controversial (Gislason et al., 2000). There are concerns about
the interpretation of catch-based estimates of mean TL, as results
often diverge from trends in mean TL estimated from fisheryindependent surveys (Branch et al., 2010). This divergence may
arise from factors such as changes in fishery economics (e.g., price,
market demand), fishery management, fishing technology (e.g.,
gear changes) and other factors (Branch et al., 2010). Moreover,
studies using catch- and assessment-based estimates of mean TL
typically utilize a single fixed estimate of TL for each species. Apparent trends in mean TL of global landings are sensitive to changes
in species-specific TL estimates; for instance, an updated estimate
for the most heavily exploited marine fish species in the world,
Peruvian anchoveta (Engraulis ringens), from TL of 2.2 to 2.7 had a
particularly marked effect on perceived trends in global mean TL
(Branch et al., 2010). Divergent results from such analyses using
fixed values of species-specific TLs emphasize the importance of
considering temporal variability in species’ trophic position.
Trophic position may vary in relation to changes in size and community composition. At the population level, size-selective fishing
and entry of a large year class into the exploited stock lead to
a shift toward smaller body sizes, while propagation of a strong
year class to older ages or reductions in fishing mortality result
in a shift to larger individuals. Such changes in size composition
may alter mean TL, as trophic position increases with size for most
fish species (see Jennings et al., 2002a for opposite trend) likely
attributed to an increase in gape size, burst swimming speed, capture efficiency, energetic requirements, and visual acuity (Scharf
et al., 2000; Jennings et al., 2002b; Sherwood and Rose, 2005). At
the ecosystem level, an example of a change in community composition resulting in a change in mean TL occurred following a climate
shift to a warm regime in the late 1970s in the GOA. The ecosystem
shifted from a community dominated by lower TL pelagic forage
fish and benthic invertebrates (e.g., crab and shrimp) to one dominated by higher TL piscivorous gadoids and flatfish (Anderson and
Piatt, 1999; Litzow, 2006). This shift in community structure led
to an increase in the overall TL of the commercial catch from the
1970s through the early 1990s (Urban and Vining, 2008).
Mean TL of catch is often evaluated using mass-balance models, such as Ecopath, which combines diet data from gut content
analysis (GCA), production, and biomass estimates (Pauly et al.,
1998, 2000). These models typically assign a single trophic position to each species or species group; at most, these models assign
a separate trophic position for adults and juveniles of the same
species, respectively. Gut content analysis provides good resolution for trophic linkages between identifiable organisms, but it
only reflects recent predation and often overlooks gelatinous and
detrital matter and ignores specimens with empty stomachs. In
contrast, stable isotope analysis (SIA) utilizes the principle of a
consistent enrichment of ␦15 N from prey to consumer (Minagawa
and Wada, 1984; Post, 2002). Moreover, SIA incorporates only prey
items assimilated by consumers, thus more accurately representing
the transfer of energy between trophic levels. While both methods
have their own caveats and assumptions, SIA is attractive because it
integrates prey selection over a longer time scale, depending on the
101
biological turnover rate of the tissue (Hesslein et al., 1993; Miller,
2000); thus, it may provide more accurate information about the
mean trophic position of a consumer than GCA. In addition, SIA
allows for easier assessment of relationships between TL and body
size for a given species.
In a previous study using SIA, we assessed seasonal, interannual,
and ontogenetic variations in the trophic roles of walleye pollock
(Gadus chalcogrammus, hereafter pollock), Pacific cod (G. macrocephalus, hereafter cod), arrowtooth flounder (Atheresthes stomias;
hereafter ATF), and Pacific halibut (Hippoglossus stenolepis; hereafter halibut) (Marsh et al., 2012). These four groundfish species
have dominated the demersal fish biomass in the GOA since the
early 1980s (Mueter and Norcross, 2002) and comprise about 75%
of the total groundfish catch (NPFMC, 2009). For each species,
trophic position, as indicated by nitrogen stable isotope signature (␦15 N), increased with total length. Both trophic position and
relative contribution of benthic versus pelagic diet, indicated by
lipid-normalized carbon stable isotope signature (␦13 C ), varied
interannually, coinciding with a more pelagic diet in summer and
a more benthic diet in fall. Such findings on temporal and ontogenetic variability in trophic position enable an evaluation of some of
the methods and assumptions in the application of mean TL of the
catch as an ecosystem indicator.
Two goals motivated our study of pollock, cod, ATF and halibut
in the GOA. First, we were interested to compare patterns in mean
TL of the catch based on Ecopath-derived estimates using fixed
values of trophic position for each species from GCA with those
derived from SIA (␦15 N) with annually estimated trophic positions.
If use of mean TL of the catch depends heavily on method (Ecopath vs. SIA) or an assumption of fixed species trophic position
despite evidence of annual variability, then the utility of mean TL
as an ecosystem indicator would be degraded. Second, although
no groundfish species in the GOA are currently subject to overfishing (NOAA, 2014), development of a high-resolution baseline
on trophic position of these species at the commercial catch or
stock abundance scale may provide a useful basis against which
to judge future effects of fishing or climate regime shifts. As these
four groundfish are highly connected within the GOA food web,
a significant decline in the abundance of one or more of these
species could potentially signal another major restructuring of the
marine ecosystem (Gaichas and Francis, 2008), which could have
substantial impacts on fishing-dependent communities. To accomplish these goals, our specific objectives were to: (1) estimate
mean annual TL and TL range of the commercial catch for the four
species; (2) estimate mean annual TL and TL range of population
biomass estimates for each of the four species; and (3) compare
TL estimates based on SIA with published Ecopath-derived TL
estimates.
2. Methods
2.1. Data collection and processing
Fish samples were obtained during field collections of the Gulf
Apex Predator-Prey Project (GAP; http://seagrant.uaf.edu/map/
gap/) at the University of Alaska Fairbanks (UAF). Fish were collected from midwater and bottom trawls in the waters surrounding
Kodiak Island in the central GOA from 2000 to 2004 in NMFS reporting area 630 (Fig. 1). Details on trawl gear and tow specifications
were described by Marsh et al. (2012). All samples were identified to species and measured (wet weight to the nearest 0.1 g and
total length to the nearest 1.0 mm). Stomachs were excised and
removed from larger fish. The remaining fish tissue was ground up
using a meat grinder, frozen at −30 ◦ C, freeze-dried, and ground
into a powder (see Marsh et al., 2012 for details). A total of 229
102
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
Fig. 1. Map of National Marine Fisheries Service (NMFS) reporting areas in the Gulf of Alaska (figure credit: NOAA). All samples for this study were collected in NMFS reporting
area 630.
pollock, 146 cod, 100 ATF, and 65 halibut samples were collected
and subsequently analyzed for stable isotopes (Table 1).
1984; Post, 2002). To estimate the TL based on ␦15 N data, the following equation was used:
TLi =
2.2. Sample processing
Between 0.2 and 0.5 mg of each powdered whole-fish sample
was weighed on a Sartorius CP2P microbalance and enclosed in
3.5 mm × 3.5 mm tin capsules for SIA. Stable nitrogen and carbon
isotopes were analyzed at the Alaska Stable Isotope Facility, UAF,
using a Costech ECS4010 elemental analyzer interfaced through a
CONFLO III to a Finnigan Deltaplus XP isotope ratio mass spectrometer (IRMS).
Results are presented in delta (␦) notation per mil (‰) and calculated using:
␦15 N =
R
SAMPLE
RSTANDARD
− 1 × 1000,
(1)
where R is the ratio of heavy to light isotope concentration
(15 N:14 N) for Nitrogen (N). The standard material for N was atmospheric air. The precision of the IRMS analysis was 0.23‰ for ␦15 N,
based on the average standard deviation of replicates of peptone
run every 10th sample (n = 319). Any samples with an average
standard deviation greater than 0.40 were re-run.
Delta 15 N values were analyzed to estimate the TLs of the demersal fish caught in this study. We assumed that ␦15 N has a constant
enrichment of 3.4‰ from diet to consumer (Minagawa and Wada,
(ı15 Ni − ı15 Nref )
3.4
+ 3.5,
(2)
where TLi is the trophic level of organism i, ␦15 Ni is the measured
␦15 N value for organism i, and ␦15 Nref is the measured ␦15 N value
for the baseline organism.
Eulachon (Thaleichthys pacificus), was used as the baseline
organism in this study and was assigned a trophic position of 3.5
based on an Ecopath model for the GOA (Aydin et al., 2007). Obtaining a baseline from which to quantify estimates of trophic positions
is one of the most difficult problems facing SIA applications (Post,
2002). Ideally, a ubiquitous organism is chosen with a well-defined
trophic position, such as a long-lived primary consumer, e.g., snails,
mussels, or other bivalves (Post, 2002). In our particular case,
an additional constraint is the requirement for samples from all
sampling periods as estimation of temporal variability in ␦15 N values was essential to meet our research objectives. Unfortunately,
no primary consumer was consistently available to us. Although
possessing a higher trophic position than desired, eulachon was
chosen as the best baseline organism, because: (1) specimens were
caught in the same nets as target species and were available during
each sampling period; (2) during their marine phase, eulachon are
demersal (Hay and McCarter, 2000) and therefore occupy a habitat
similar to that of the four groundfish species in this study, (3) they
are an occasional prey item for the groundfish species (Yang and
Table 1
Sample size (N), mean total length and total length range (cm, in parentheses) of walleye pollock, Pacific cod, arrowtooth flounder, and Pacific halibut collected around Kodiak
Island during 2000–2004.
Species
N
Walleye pollock
229
Pacific cod
146
Arrowtooth flounder
100
Pacific halibut
65
Year
2000
Total length
2001
Total length
2002
Total length
2003
Total length
2004
Total length
46
(12–64)
38
(16–76)
25
(17–66)
24
(39–77)
54
(12–69)
28
(21–76)
6
(24–48)
6
(35–98)
34
(10–68)
26
(10–79)
18
(25–74)
3
(39, 44, 70)
60
(14–78)
46
(21–87)
43
(24–82)
29
(38–97)
35
(7–70)
8
(30–64)
8
(29–37)
3
(53, 77, 81)
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
Nelson, 2000), (4) the diet of eulachon in the GOA is highly consistent across a large eulachon size range and throughout a diurnal
cycle (Wilson et al., 2009), (5) the ␦15 N of eulachon was found to
be stable over the study period (Marsh et al., 2012) and (6) as a
longer-lived consumer, they integrate the isotopic signature of the
food web on a large time scale, minimizing the effects of short term
variation (Post, 2002).
2.3. Model selection
Analysis of covariance (ANCOVA) models were used to test for
variation in the relationship between length and TL among years.
The full model for cod, ATF, and halibut included main effects for
years and separate slopes for each year:
Yij = + Ai + ˇi (Xij − X̄i ) + εij ,
(3)
where is the mean TL, Ai is the year effect (i = 1,. . .,5 year), Xij is the
covariate (length) measured for observation Yij (TL), X̄i is the average value of the covariate for treatment group i, and ˇi is the slope
term for length (covariate) in each year. For pollock, we elected
to use piecewise regression ANCOVA models with a breakpoint at
52 cm to better represent pollock feeding patterns following Marsh
et al. (2012).
The full ANCOVA models were compared to simpler models,
including an ANCOVA model with a single fixed slope across years
(ˇC ) and a different intercept for each treatment, and a simple
regression model in which TL varies with length only, and with
no year effect (Marsh et al., 2012; Table 3). The best-fit models
were selected using the Akaike Information Criterion (AIC) (Akaike,
1974). The model with the lowest AIC value was selected as the
best fit. If there were multiple models with a AIC of ≤2, the simplest model was selected. Prior to modeling, the TL values for both
pollock and halibut were loge transformed to meet the normality
assumption; normality was assessed using the Shapiro–Wilks test
for normality. Additionally, the data were fitted by linear regression models to estimate the TL of population biomass data and
commercial catch data for years (1984–2009) outside of the study.
For all statistical tests, a p-value < 0.05 was considered statistically
significant.
Size compositions of the GOA commercial catch from 2000
through 2004 were provided by the NMFS Fisheries Monitoring
and Analysis Division of the Alaska Fisheries Science Center based
on onboard observer data for pollock, cod, ATF, and halibut for
NMFS reporting area 630 (Fig. 2). The total commercial catch weight
(excluding discards and bycatch) and number for each of the species
in area 630 were also provided for the same time frame. It was
assumed that the length frequency of the total catch was the same
as that of the observed catch. Best-fit models from the previous
analyses of TL as a function of length for each species were applied
to the observed commercial catch lengths for each year to estimate
TL. These TL estimates were multiplied by the corresponding fraction of observed catch to total catch at each length to provide an
estimated mean TL of commercial catch for each species for each
year. Box plots were used to show TL ranges of commercial catch for
each species by year. No formal statistical tests were performed on
TL differences among years due to an unknown and much smaller
effective sample size than the number of fish observed. The overall
weighted mean TL by year of commercial catch in area 630 (TLall,j )
for the four groundfish species was estimated as:
4
TLi,j × Ci,j
i=1
CT,j
,
where TLi,j is the mean TL based on best-fit models, linear regression models and a GOA Ecopath model (Aydin et al., 2007) of a given
species i (i = pollock, cod, ATF, halibut) over year j (j = 2000–2004),
Ci,j is the weight of commercial catch (t) or number caught for
species i in area 630 during year j, and CT,j is the weight (t) or abundance of total commercial catch for the four species combined over
year j.
Additionally, abundance and biomass estimates based on raw
area swept and length-frequency data for pollock, cod, ATF, and
halibut were provided by NOAA Fisheries from bottom trawl
surveys over 1984–2007 for area 630 (Fig. 3). Surveys were
conducted in the GOA triennially during 1984–1999 and biennially during 1999–2007. For a complete description of NMFS
bottom trawl survey sampling design and procedure, see Britt
and Martin (2001). Model parameters from the linear-regression
models were applied to the survey data to estimate a mean
TL by survey year for each species, as well as an estimated
TL range for each year based on length-frequency distributions.
For comparison with published Ecopath TL estimates, an overall weighted mean TL for the target groundfish species combined
was estimated using Eq. (4) with estimated population abundance and biomass data for area 630 instead of commercial catch
data.
Even though all fish samples selected for SIA were collected in a relatively small area in the waters surrounding
Kodiak Island each year, their feeding in prior months may
have occurred in different locations. To visually assess interannual differences in the spatial distribution of each target
species, we mapped the scaled catch-per-unit-effort (CPUE)
by weight (kg) from NMFS bottom trawl survey stations that
caught pollock, cod, ATF, halibut and eulachon in NMFS statistical
reporting areas 620 and 630 during 2001 and 2003 (data found at:
http://www.afsc.noaa.gov/RACE/groundfish/survey data/data.htm).
The years 2001 and 2003 were selected because for all target fish
species the mean ␦13 C values were the most depleted in 2003,
the highest in 2001 and regardless of length the ␦15 N values were
significantly lower for cod, halibut and ATF in 2003 compared to
2001 (Marsh et al., 2012).
3. Results
2.4. Trophic level estimates
TLall,j =
103
(4)
3.1. Length-frequency and spatial distributions
Pollock, cod, ATF, and halibut all exhibited to varying degrees
interannual variability in length-frequency distributions of the
observed commercial catch (Fig. 2) and trawl surveys (Fig. 3). Pollock was the most variable, with higher proportions of smaller
fish encountered in bottom trawl surveys in 1999, 2001, 2005,
and 2007 (Fig. 3) and in the commercial catch in 2002 and 2003
(Fig. 2). Prior to 1999, there were higher proportions of larger pollock in the trawl survey and in the commercial catch (Fig. 2). Cod
had a higher proportion of smaller-sized individuals in the 1996
and 2007 trawl surveys and a slightly higher proportion of smaller
individuals in the 2003 catch (Fig. 3). In contrast, the length distributions of ATF were relatively stable throughout both time series
(Figs. 2 and 3), with the exception of proportionally more small
individuals (<40 cm) in 2007 and large individuals (≥50 cm) in
2008 captured by the fishery. In 1984, 1987, and 2002, there were
larger proportions of small halibut collected in the NMFS trawl survey than in other years (Fig. 3). Numbers of halibut in the largest
size classes appeared to have declined through time (Fig. 3). The
spatial distributions of pollock, cod, ATF, halibut and eulachon in
statistical reporting areas 620 and 630 caught in the NMFS trawl
survey in 2001 and 2003 were largely similar throughout the region
(Fig. 4).
104
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
Fig. 2. Cumulative densities by 10 cm length-bin of commercial catch by year (2000–2004) for walleye pollock, Pacific cod, arrowtooth flounder, and Pacific halibut in NMFS
reporting area 630 as provided by the NMFS observer program.
3.2. Model selection
The best-fit models for estimating TL varied by year for all
species. For pollock < 52 cm, the best-fit piecewise model had a
different average TL by year (intercept) and a varying TL-length
relationship by year (slope). For larger pollock (>52 cm) there was
a constant change in slope for each year (Tables 2 and 3; Fig. 5).
Trophic levels of 3.0 and 4.7 corresponded to ␦15 N values of 11.38‰
and 17.16‰. The best-fit models for cod, ATF, and halibut had an
increasing TL with length (constant positive slope) and the overall average TL fluctuated interannaully (different intercept for each
year; Tables 2 and 3; Fig. 5).
3.3. Trophic level estimates of commercial catch
For the years examined, cod had the highest estimated mean
TL of commercial catch every year, followed by halibut, ATF, and
pollock, respectively (Table 5; Fig. 6). Cod, halibut, and ATF all displayed a dip in mean TL in 2003 (Fig. 6). Corresponding with this dip
in mean TL, mean length of the commercial catch was also lower
for cod and ATF in 2003, but not for pollock and halibut. While the
mean TL fluctuated for all species, the annual interquartile range of
TLs remained relatively consistent for cod, ATF, and halibut during
2000–2004 (Fig. 6). However, the TL interquartile range for pollock
varied interannually, with 2001 and 2004 having the most expansive TL range and 2000 and 2002 being the least variable (Fig. 6).
The estimated mean TLs of catch for pollock, ATF, and halibut were
lower than Ecopath estimates of each adult population (Aydin et al.,
2007), while the mean TL catch estimate for cod was higher (Fig. 6).
The relative proportion of pollock in the catch in area 630
declined in the years 2000–2004, while the proportion of cod rose
and that of ATF and halibut remained relatively stable (Table 5;
Fig. 7a). Cod surpassed pollock as the dominant species caught in
2001 by weight (Fig. 7a) and therefore had the greatest influence on
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
105
Fig. 3. Cumulative densities by 10 cm length-bin of biomass from NMFS triennial (1984–1996) and biennial (1999–2007) trawl surveys for walleye pollock, Pacific cod,
arrowtooth flounder and Pacific halibut in NMFS reporting area 630 as provided by NOAA Fisheries.
the weighted mean TL of catch biomass after 2000 (Fig. 8a). Both
ATF and halibut contributed the least to catch biomass (Fig. 7a),
thus had the smallest influence on the weighted mean TL (Fig. 8a).
The overall catch biomass-weighted mean TL of commercial catch
estimated from the best-fit models and linear regression models remained fairly stable at ∼4 (Fig. 8a). There was a slight dip
in 2003, followed by an increase in 2004, in the weighted mean
TL based on the best-fit models. The weighted mean TL of catch
based on the linear regression models increased slightly over time
(Fig. 8a), likely due to the increase in proportion of cod caught in
area 630 (Fig. 7a). In contrast, the catch biomass-weighted mean
TLs based on Ecopath values peaked in 2002 when halibut and
ATF comprised the largest proportion of the catch during the time
series (Fig. 8a). Patterns differ somewhat when considering catch
abundance instead of catch biomass. Pollock contribute a greater
fraction of catch numbers owing to their relatively small body size,
though a clear declining trend is still evident (Fig. 7b). Cod and ATF
contribute more equally to catch abundance, and the contribution
of halibut is smaller than when considering catch biomass owing
to their relatively large body size. All three methods of computing weighted mean TL of the catch based on abundance suggest an
increasing trend, although values dip in 2002 and 2003 for both
linear regression and best-fit models (Fig. 8b).
When our linear-regression models were applied to catch
length-frequency data over 1990–2009, cod had the highest estimated mean TL, followed by halibut, ATF, and pollock (Table 4;
Fig. 9). The TL of pollock appeared to decrease slightly after the
late 1990s. Additionally, after the late 1990s, the distribution of TL
estimates for pollock were more variable than in previous years,
while the distribution of TL estimates for cod and halibut remained
relatively stable over the same time frame (Fig. 9). However, the
TL distributions for ATF varied in 2008 and 2009, with lower and
higher than average values, respectively (Fig. 9). This is likely due
to smaller sample sizes in those years.
106
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
Fig. 4. Spatial distribution of 2001 and 2003 NMFS bottom trawl survey biomass for walleye pollock, Pacific cod, arrowtooth flounder, Pacific halibut and eulachon in NMFS
reporting areas 620 and 630. Open circles represent CPUE (kg) values scaled to the observed maximum CPUE (kg) for each species year combination. The 200 m bathymetric
line is shown.
3.4. Trophic level estimates of surveyed population
Estimated TLs of the surveyed population were highest for cod,
followed by halibut, ATF, and pollock (Fig. 10). This rank order
Table 2
Results of ANCOVA model fitting using AIC. Pacific cod and arrowtooth flounder TL
covaried with total length (cm) for each year, whereas walleye pollock and Pacific
halibut ln(TL) covaried with total length for each year.
Species
Treatment
Model
df
AIC
Walleye pollock
Year
MP1
MP2*
MP3
MP4
MP5
16
12
12
8
4
−578.1
−577.1
−571.8
−565.6
−570.3
Pacific cod
Year
M1
M2*
M3
11
7
3
−76.5
−79.8
−68.6
Arrowtooth flounder
Year
M1
M2*
M3
11
7
3
−31.96
−37.12
−25.32
Pacific halibut
Year
M1
M2*
M3
7
5
3
*
Indicates the best-fit model.
−189.0
−192.3
−187.5
corresponded with the TL estimates of commercial catch. After
the mid-late 1990s, the TL estimate for the two gadid species
became more variable, while the TL estimates for the two flatfish species remained stable (Fig. 10). The TL estimates of pollock
slightly decreased after the late 1990s.
3.5. Weighted mean trophic positions
From 1984 to 2007, ATF dominated the survey biomass in area
630, followed by pollock, cod, and halibut (Fig. 11b). On the other
hand, pollock dominated the commercial catch (fish retained plus
discards) in the early 1990s, with cod and ATF comprising larger
proportions of the catch in recent years through 2008 (Fig. 11a).
The overall weighted mean TL of survey biomass for the four
species remained relatively stable over time (Fig. 12). The weighted
mean TL of the commercial catch based on the regression models started at 3.9 in 1990, slightly decreased until 1994, peaked
in 1996 at 4.1, declined and leveled out around 4.0 in 2001, and
slightly declined from 2004 to 2008 (Fig. 12). The weighted mean
TL based on Ecopath values followed a slightly different trend, in
which TL increased during 2000–2002 to above 4.1, dipped in 2005,
and increased again until 2008 to a TL of above 4.1 (Fig. 12). Overall,
the TL estimates of the commercial catch were more variable than
TL estimates of the survey biomass (Fig. 12).
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
107
Fig. 5. Plots of TL (1◦ y-axis) and ␦15 N (2◦ y-axis) versus total length (cm) and predicted relationships (based on parameters of the best fit ANCOVA models in Table 3 and
regression models in Table 4) between total length and TL (␦15 N) by year and all years combined for Pacific cod, arrowtooth flounder, walleye pollock and Pacific halibut.
4. Discussion
Application of TL models to nitrogen stable isotopes and lengthfrequency data revealed interesting interannual variability in mean
TL of pollock, cod, ATF, and halibut in commercial and survey
catches in the central GOA that could not be evidenced by Ecopathbased estimates with fixed values of trophic position of component
species. Best-fit models indicate that TL varied interannually as
Table 3
Coefficients and significance for the best fit ANCOVA models by year. Pacific cod and arrowtooth flounder TL covaried with total length (cm), whereas walleye pollock and
Pacific halibut ln(TL) covaried with total length for each year. The trophic level of pollock and halibut were log transformed to meet normality assumptions.
Species
Best fit model
Year
a
b
b2
r2
F
p
Walleye pollock
MP2
2000
2001
2002
2003
2004
1.221
1.161
1.243
1.180
1.097
0.00126
0.00261
0.00042
0.00243
0.00417
0.00322
0.41
15.26
<0.001
Pacific cod
M2
2000
2001
2002
2003
2004
3.778
3.775
3.679
3.637
3.809
0.00999
0.00999
0.00999
0.00999
0.00999
0.49
27.23
<0.001
Arrowtooth flounder
M2
2000
2001
2002
2003
2004
3.616
3.654
3.590
3.423
3.610
0.00848
0.00848
0.00848
0.00848
0.00848
0.34
9.62
<0.001
Pacific halibut
M2
2000
2001
2003
1.303
1.309
1.270
0.00160
0.00160
0.00160
0.33
9.172
<0.001
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
4.6
4.4
4.2
3.8
4.0
3.6
3.8
3.4
2001
2002
2003
2004
C
2000
2001
2002
2003
2004
2001
2002
2003
2004
D
5.5
4.4
2000
3.6
4.0
3.8
4.5
4.0
5.0
4.2
Trophic Level
B
4.8
A
4.0
4.2
108
2000
2001
2002
2003
2004
2000
Year
Fig. 6. Estimated trophic level ranges of commercial catch by year for (A) walleye pollock, (B) Pacific cod, (C) arrowtooth flounder, and (D) Pacific halibut based on best-fit
ANCOVA models (Table 3). The middle line represents the median TL, the ends of the boxes (hinges) represent the 1st and 3rd quartiles, the whiskers extend to data points
within a factor of 1.5 of the hinges, and the dots are the remaining data points. Overlaid on each boxplot is the Ecopath based trophic level estimates (horizontal dotted lines):
3.7 for walleye pollock, 4.1 for Pacific cod, 4.3 for arrowtooth flounder, and 4.5 for Pacific halibut (Aydin et al., 2007).
4.3
A
4.2
4.1
4.0
Weight mean - Best fit
3.9
3.8
Weighted mean Linear regression
Trophic Level
3.7
Weighted mean Ecopath values
3.6
3.5
1999
4.3
2000
2001
2002
2003
2004
2005
2000
2001
2002
Year
2003
2004
2005
B
4.2
4.1
4.0
3.9
3.8
3.7
3.6
3.5
1999
Fig. 7. Percent of commercial catch by (A) weight and (B) abundance in NMFS
reporting area 630 for walleye pollock, Pacific cod, arrowtooth flounder, and Pacific
halibut for 2000–2004.
Fig. 8. Plots of the weighted mean TL of commercial catch by (A) catch biomass
and (B) catch abundance for walleye pollock, Pacific cod, arrowtooth flounder, and
Pacific halibut in NMFS reporting area 630 by year using best-fit ANCOVA models,
linear regression models based on stable isotope analysis, and fixed Ecopath values
from Aydin et al. (2007).
109
A
3.5
4.0
4.5
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
1992
1994
1996
1998
2000
2002
2004
2006
2008
B
3.8
4.3
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
C
3.7
3.9
4.1
Trophic Level
4.2
4.6
5.0
1990
1992
1995
1997
1999
2001
2003
2005
2007
2009
D
4.0
4.5
5.0
5.5
1990
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
Fig. 9. Estimated annual trophic level ranges of commercially caught (A) walleye pollock, (B) Pacific cod, (C) arrowtooth flounder, and (D) Pacific halibut in NMFS reporting
area 630 from 1990 to 2008 based on SIA-derived linear regression parameter estimates. The middle line represents the median TL, the ends of the boxes (hinges) represent
the 1st and 3rd quartiles, the whiskers extend to data points within a factor of 1.5 of the hinges, and the dots are the remaining data points.
well as with total length. Including “year” as a variable, effectively
doubled the range of TLs that would have been considered without a temporal component. While the annual SIA-based regression
modeled estimate of mean TL of each species remained fairly consistent during 1984–2009, the annual distribution of TL values for
each species revealed trends that were not apparent by looking at
the mean TL alone. Selected models revealed species-specific ontogenetic and temporal shifts in SIA-based TL-length relationships
that should be considered when estimating the mean TL of catch or
population biomass.
Table 4
Parameters and test statistics for linear relationships between total length (cm) and
trophic level or ln(TL). The form of the linear relationship is TL = a + b(length) for
Pacific cod and arrowtooth and ln(TL) = a + b(length) for walleye pollock and Pacific
halibut.
Species
Walleye pollock – ln(TL)
Pacific cod – TL
Arrowtooth flounder – TL
Pacific halibut – ln(TL)
Length versus TL or ln(TL)
a
b
b2
r2
F
p
1.18
3.73
3.63
1.29
0.00213
0.00978
0.00634
0.00155
0.00327
0.35
0.42
0.19
0.23
60.9
105.0
23.6
16.7
<0.001
<0.001
<0.001
<0.001
Year effects in best-fit models included the lowest predicted
TL in 2003, regardless of length, for ATF, cod, and halibut, resulting in the lowest estimated mean TL of catch based on observer
length-frequency data among 2000–2004. These three species are
considered opportunistic feeders, which prey on the most available prey items (Yang et al., 2006). In 2003 euphausiids comprised
a larger proportion of the diet of the four target groundfish than
in 2001 (Yang personal communication; Yang et al., 2006). Additionally, all four species had relatively depleted ␦13 C signatures
indicative of a more pelagic diet in 2003 (Marsh et al., 2012). However, temporal diet shifts may only partially explain the observed
interannual differences in the mean TL. For instance, among the
years considered, mean lengths of both cod and ATF were smallest
in 2003, contributing to the decline in mean TL. Likewise, interannual variability in pollock mean TL were influenced by variability in
total length. For each species, the intraspecific annual TL differences
were minimal (<0.3 TL), while TL range of the commercially caught
species spanned up to 1.75 TL for halibut, 1.1 TL for pollock, 0.95 TL
for cod, and 0.8 TL for ATF indicating that TL was heavily influenced
by body length.
Generally, pollock, cod, ATF, and halibut become increasingly
more piscivorous as they grow (Yang and Nelson, 2000; Yang et al.,
2006), thus increasing TL with body length (Marsh et al., 2012).
110
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
B
3.8
1984
1990
1996
2001
2005
1984
1990
1996
2001
2005
D
1984
1990
1996
2001
2005
1984
1990
1996
2001
2005
5.0
4.2
C
3.7
3.8
3.8
4.2
3.9
4.0
4.6
4.1
Trophic Level
3.4
4.0
3.6
4.2
3.8
4.4
4.0
4.6
4.2
A
Year
Fig. 10. Estimated trophic level ranges of survey caught (A) walleye pollock, (B) Pacific cod, (C) arrowtooth flounder, and (D) Pacific halibut populations in NMFS reporting
area 630 based SIA-derived linear regression parameter estimates. The middle line represents the median TL, the ends of the boxes (hinges) represent the 1st and 3rd quartiles,
the whiskers extend to data points within a factor of 1.5 of the hinges, and the dots are the remaining data points.
However, the increase in TL (or ␦15 N) with body size, varies by
species (Yang et al., 2006; Marsh et al., 2012). Patterns of increasing
piscivory have led many isotope studies to assume a linear relationship between TL and length (Jennings et al., 2002a; Badalamenti
et al., 2002); in a prior study, however, we found that pollock
changed their diet at around 52 cm by feeding at an increasingly
higher TL (␦15 N) and on a larger range of prey items (Marsh et al.,
2012). A similar observation had been made for pollock in Prince
William Sound, which demonstrated a sharp increase in TL that
occurred at a length of 50–60 cm, probably due to an increase in
piscivory (Kline, 2008). Our model for pollock accounted for a rapid
increase in TL after 52 cm total length; this ontogenetic diet shift
means that selective fishery removal of large pollock will disproportionately lower the mean TL estimate for this species.
Table 5
Total annual commercial catch in NMFS reporting area 630 by weight and number of individuals for walleye pollock, Pacific cod, arrowtooth flounder, and Pacific halibuts.
Annual mean length of catch is provided based on observer length frequency data. Corresponding estimated mean trophic level of catch ± SD is calculated from best-fit
models (Table 3; Fig. 5) and regression models (Table 4; Fig. 5).
Species
Year
Catch (mt)
Catch (#)
% Observed
Length (±SD)
TL best-fit (±SD)
Walleye pollock
2000
2001
2002
2003
2004
35,933
20,273
10,902
12,315
14,093
86,039,619
58,920,504
48,857,601
42,240,276
35,178,920
13.1%
9.8%
12.3%
13.6%
15.5%
50.1
47.6
42.8
42.8
47.8
±
±
±
±
±
7.9
12.3
10.2
6.4
6.2
3.64
3.65
3.54
3.62
3.67
±
±
±
±
±
0.07
0.16
0.06
0.08
0.12
3.66
3.65
3.59
3.58
3.62
±
±
±
±
±
0.09
0.13
0.12
0.07
0.08
Pacific cod
2000
2001
2002
2003
2004
26,235
22,894
18,627
22,524
26,565
19,685,661
19,623,076
16,355,061
18,858,842
19,725,845
5.8%
8.1%
6.6%
8.3%
6.4%
64.2
61.1
62.3
58.6
62.5
±
±
±
±
±
8.7
10.8
11.6
10.5
9.5
4.42
4.39
4.30
4.22
4.43
±
±
±
±
±
0.09
0.11
0.12
0.10
0.09
4.36
4.33
4.34
4.30
4.34
±
±
±
±
±
0.09
0.11
0.11
0.10
0.09
Arrowtooth flounder
2000
2001
2002
2003
2004
13,049
9562
9753
8683
8383
23,133,421
18,561,231
18,372,697
21,269,697
18,867,972
17.8%
11.7%
19.1%
16.9%
9.8%
50.1
48.8
49.5
47.5
48.1
±
±
±
±
±
10.3
11.7
10.0
10.1
9.9
4.04
4.07
4.01
3.83
4.02
±
±
±
±
±
0.09
0.10
0.08
0.09
0.08
3.94
3.94
3.94
3.93
3.93
±
±
±
±
±
0.07
0.07
0.06
0.06
0.06
Pacific halibut
2000
2001
2002
2003
2004
7317
8174
8198
8075
8672
2,784,422
3,517,516
3,368,517
3,795,383
4,336,181
7.1%
6.6%
5.7%
5.2%
4.9%
72.2
64.9
63.8
67.2
68.4
±
±
±
±
±
22.5
24.6
24.8
21.8
21.1
4.13 ± 0.15
4.10 ± 0.17
4.07
4.02
4.01
4.03
4.04
±
±
±
±
±
0.14
0.16
0.16
0.14
0.13
3.97 ± 0.14
TL regression (±SD)
A
1.0
Commercial catch composition
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
0.8
111
120
A
100
90
0.6
80
0.4
70
60
0.2
Commercial removals (1000 mt)
110
50
0.0
Walleye pollock
Pacific cod
Arrowtooth flounder
2400 Pacific halibut
40
1990
1992
1994
1996
1998
2000
2002
2004
2006
B 1.0
2008
2000
1800
0.6
1600
1400
0.4
1200
1000
0.2
Population biomass (1000 mt)
Survey Composition (biomass)
2200
0.8
800
0.0
600
1985
1990
1995
2000
2005
Year
Fig. 11. (A) Commercial catch composition by weight (shaded bars) and total commercial removals (t) (line) during 1991–2008 and (B) NMFS trawl survey composition and
combined population biomass estimates (t) during 1984–2007. Both panels consider only walleye pollock, Pacific cod, arrowtooth flounder, and Pacific halibut in federal
reporting area 630.
When estimated over two decades (1990–2009), the annual
mean estimated TL of catch for each of the target species based
on SIA-derived linear regression models and length frequency data
remained remarkably stable. It should be noted, that during this
time period, there were no major climate regime shifts resulting in ecosystem restructuring, which could have revealed how
SIA-derived TL measures would have detected the response by
studying these four focal fish. Despite the overall stability, the
4.2
4.1
Trophic Level
4.0
3.9
3.8
Biomass
Catch - best fit
Catch - regression
Catch - Ecopath
3.7
3.6
1985
1990
1995
2000
2005
Year
Fig. 12. Plot of the weighted mean TL of survey biomass and commercial for walleye pollock, Pacific cod, arrowtooth flounder, and Pacific halibut by year in federal reporting
area 630 using SIA-derived best-fit ANCOVA models, SIA-derived linear regression models and Ecopath values from Aydin et al. (2007).
112
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
interannual minimum and maximum TL estimates varied for each
species. Most notable were the expansion and contraction of the
range in pollock TL values after 1999, the lower pollock TL values
in 2008, the higher ATF TL values in 2009, and the disappearance
of higher TL halibut with time. Variations in pollock TLs can be
partially explained by two strong year classes (1999 and 2000)
recruiting into the fishery in 2002 and 2003 as age-2 fish (Dorn
et al., 2009), which resulted in a lower mean TL. In recent years,
the market for ATF has expanded, resulting in a gradual shift from
being discarded as bycatch to moving toward a targeted fishery
(Turnock and Wilderbuer, 2009). Retention of ATF in trawl fisheries has increased from <10% in the early 1990s to 54–69% during
2005–2009 (Turnock and Wilderbuer, 2009). Despite the decline
in mean weight-at-age due to decreased growth rates of halibut
since the mid-1980s (Clark and Hare, 2002), the fishery seems to
be selecting for the same size fish (although now older than before).
A continued loss of larger halibut resulting in a reduction of higher
TL could potentially cause future declines in the mean TL of the
catch.
In general, our population TL estimates based on biomass for
pollock, cod, ATF, and halibut were lower than the TL estimates
of the commercial catch. These results are not surprising, because
the NMFS bottom trawl survey is intended to census a broad size
range of fish for stock assessment, whereas commercial fisheries
target age 3+ pollock, cod, and ATF, and age 6+ halibut. Additionally, some gear types select for larger sized fish more than
others. For example, cod longline and pot fisheries select for a
larger size composition than the cod trawl fishery (Thompson et al.,
2009). The majority of cod landings were taken by trawl gear during 1991–2002 and by pot gear from 2003 onward (Thompson
et al., 2009). Similarly, halibut are primarily caught by longline
in the commercial fishery and the corresponding mean TL estimates of the surveyed population biomass were lower than those
of the fishery removals. The majority (90%) of the pollock catch
is taken by pelagic trawls (Dorn et al., 2009). Estimated mean TL
for the pollock population and fishery were within 0.03 through
1996, after which the mean TL of the population declined more
rapidly than that of the fishery. One additional important factor is
that the weighted mean TL of the population biomass of the four
species was mainly driven by the proportionately large biomass
of ATF, whereas the weighted mean trophic level of the catch was
driven by pollock and cod, which command much higher prices
than ATF.
Compared to static Ecopath-based estimates of mean TL for the
GOA (Aydin et al., 2007), our estimated mean TL values for pollock, ATF, and halibut were lower, while our estimated values for
cod were higher. Despite the differences in species-specific TL values, the weighted mean TL values based on Ecopath TL values and
those based on our estimated TL values were almost identical in
2000, 2001, and 2004. In 2002 and 2003, however, the weighted
mean Ecopath values were higher than our estimated TL values. The
higher Ecopath TL values in 2002 and 2003 resulted from a higher
proportion of halibut and ATF in the catch composition, which were
assigned lower TL values in our study. Interannual differences in
the mean estimated TL based on survey biomass were more apparent for the gadids, with mean TL values for pollock decreasing and
becoming more variable after 1996. Both the mean TL for halibut
and for ATF remained stable over the years of the survey. Despite
incorporating ontogenetic diet shifts, the weighted mean TL of the
four groundfish combined remained fairly stable for both the population biomass (heavily influenced by ATF) and the commercial
catch in NMFS reporting area 630.
Both nitrogen stable isotope-derived TL estimates and Ecopathderived TL estimates based on gut content analysis have advantages
and disadvantages. Gut content analysis provides detailed trophic
linkages, but over a short timeframe. Stable isotope analysis
integrates diet over a longer time scale, but it is difficult to infer
detailed trophic linkages. In the current study, past diet and stable
isotope studies demonstrate that these target fish species exhibit
ontogenetic and temporal shifts in their diets and trophic levels
(Yang et al., 2006; Marsh et al., 2012) that are not accounted for in
Ecopath-derived TL estimates. However, Ecopath models estimate
TLs for the entire ecosystem at once, albeit using fixed values for
functional groups or species. Branch et al. (2010) demonstrated that
changes in fixed values of trophic positions can change the perceptions of mean TL trends using Ecopath. Similarly, assuming that the
interannual variability revealed by the current study is more representative of nature, then failure to include such variability can
lead to errors in TL estimates. Annually varying nitrogen-derived
TL estimates may be the most accurate, but are potentially more
costly. The Ecopath approach may be more practical, but can miss
important ecosystem dynamics. Perhaps the best approach is a
combination of the two.
Lastly, precaution should be exercised when interpreting these
results. For instance, results may have differed if we were able
to use a different baseline species, such as a sessile primary consumer (e.g., bivalve such as clams or scallops). Estimating trophic
position from ␦15 N values requires estimates of trophic fractionation of ␦15 N and knowledge of the variability in ␦15 N at the
base of the food chain. Fractionation factors can vary between
species (Hussey et al., 2013 and citations therein) and within
species (Sherwood and Rose, 2005; Ankjærø et al., 2012). Due to
biogeochemical processes, the baseline of benthic food webs is
often more enriched in 15 N relative to the pelagic baseline (e.g.,
Sherwood and Rose, 2005), though not always, as microbial degradation can enrich, deplete or have no effect on 15 N depending on
the chemical makeup of the substrate (McTigue and Dunton, 2014
and citations therein). However, we elected to interpret changes in
␦15 N as changes in TL based on eulachon, as other lower trophic
level alternatives were not broadly available throughout area 630
for the years of the study and due to its stable feeding ecology.
Additionally, samples were taken only from a limited area surrounding Kodiak, while our models were applied to the larger
NMFS reporting area 630, which contains quite disparate habitats with likely differences in prey availability and distribution.
We observed variations in the intraspecific TL at each size range,
indicating that a fish at a given size could potentially feed on
a range of prey items, depending on their availability. Unfortunately, we had few samples of halibut > 80 cm, which most likely
resulted in an underestimation of TL for this species. The size
composition of halibut came from the NMFS onboard observer
program, which only included halibut bycatch, largely (> 80%)
from the groundfish trawl fishery that tends to catch smaller halibut than longline fisheries (Sullivan et al., 1994). Also, fishery
discards were not included in any of the length compositions;
inclusion of small discarded fish would likely have lowered the
average TL of the catch. Finally, both ATF and halibut are sexually dimorphic, with females being larger and maturing later
than males (Zimmermann, 1997; IPHC, 1998). Because we did not
have sex data for all samples, we were not able to fit sex-specific
models, which may have improved model fits. Accuracy of the SIAderived TL estimates could be improved through laboratory feeding
experiments providing direct estimates species-specific of ␦15 N
trophic fractionation rates, continued sampling from a larger area,
and additional sampling of benthic baseline organisms. Estimates
of mean TL could be improved by inclusion of all catch data; fishery,
survey and discards (Hornborg et al., 2013). Despite the uncertainties in SIA-derived TLs, monitoring of stable isotope signatures from
species (caught during routine trawl surveys) shows promise in
providing accurate TL estimates that account for ontogenetic and
temporal variation that could compliment and potentially validate
Ecopath or ecosystem model derived-TLs.
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
5. Conclusions
We found that SIA-based estimates of mean TLs of the catches
and populations of four commercially important groundfish species
vary interannually. Fish length is a statistically significant covariate
of mean TL for each of these species. Variability in recruitment of
small fish and size-selective removals of large fish are likely to be
the proximate causes for observed changes in TL. However, while
TL is mostly influenced by fish length, all focal species also showed
some temporal co-variation in their TL, which may be due to other
factors. Consequently, continued monitoring using stable isotope
signatures of these fishes and of other species, used in conjunction with mass-balance models, could aid in detecting long-term
trends of trophic status, especially in species that are harvested
by size-selective fishing techniques. Consistent with previous findings (Livingston, 2005), our results indicate no “fishing down the
food web” in the GOA. Average SIA-based estimates of trophic
position were significantly higher than Ecopath-based estimated
for cod, but significantly lower for pollock, ATF, and halibut. Our
results also demonstrate that the use of catch length-frequency
data could further improve estimates of the trophic status as an
ecosystem-based indicator of sustainability and may lead to an earlier detection of potential declines in TL of a given fish species in the
GOA.
Acknowledgements
We thank all those who aided in the collection and processing
of the fish samples, including Lei Guo, Mike Trussell, and the
captain and crew of the F/V Laura; Matthew Wooller, and Alexander Andrews for their helpful suggestions on earlier drafts; Franz
Mueter for statistical advice; and Norma Haubenstock and Tim
Howe (Alaska Stable Isotope Facility) for processing the samples.
This work was funded through the Gulf Apex Predator–Prey Study
(NOAA, National Marine Fisheries Service) and by the Rasmuson
Fisheries Research Center (University of Alaska Fairbanks). The
findings and conclusions in the paper are those of the authors
and do not necessarily represent the views of the National Marine
Fisheries Service. Reference to trade names does not imply endorsement by the National Marine Fisheries Service.
References
Akaike, H., 1974. A new look at statistical model identification. IEEE Trans. Autom.
Control 19, 716–722.
Anderson, P.J., Piatt, J.E., 1999. Community reorganization in the Gulf of Alaska following ocean climate regime shift. Mar. Ecol. Prog. Ser. 189, 117–123.
Ankjærø, T., Christensen, J.T., Grønkjær, P., 2012. Tissue-specific turnover rates and
trophic enrichment of stable N and C isotopes in juvenile Atlantic cod Gadus
morhua fed three different diets. Mar. Ecol. Prog. Ser. 461, 197–209.
Aydin, K., Gaichas, S., Ortiz, I., Kinzey, D., Friday, N., 2007. A Comparison of the Bering
Sea, Gulf of Alaska, and Aleutian Islands Large Marine Ecosystems Through Food
Web Modeling. U.S. Department of Commerce NOAA Technical Memorandum
NMFS-AFSC-178, Seattle, Washington.
Badalamenti, F., D’Anna, G., Pinnegar, J.K., Polunin, N.V.C., 2002. Size-related trophodynamic changes in the three target fish species recovering from intensive
trawling. Mar. Biol. 141, 561–570.
Branch, T.A., Watson, R., Fulton, E.A., Jennings, S., McGilliard, C.R., Pablico, G.T., Ricard,
D., Tracey, S.R., 2010. The trophic fingerprint of marine fisheries. Nature 468
(7322), 431–435.
Britt, L.L., Martin, M.H., 2001. Data Report: 1999 Gulf of Alaska Bottom Trawl Survey.
U.S. Department of Commerce NOAA Technical Memorandum NMFS-AFSC-121,
Seattle, Washington.
Clark, C.G., Hare, S.R., 2002. Effects of climate and stock size on recruitment and
growth of Pacific halibut. N. Am. J. Fish. Manag. 22, 852–862.
Dorn, M., Aydin, K., Barbeaux, S., Guttormsen, M., Megrey, B., Spalinger, K., Wilkins,
M., 2009. Assessment of the Walleye Pollock Stock in the Gulf of Alaska. In: Stock
Assessment and Fishery Evaluation Report for the Groundfish Resources on the
Bering Sea and Aleutian Islands for 2009. North Pacific Fisheries Management
Council, 605W, 4th Ave., Suite 306, Anchorage, Alaska 99501.
Essington, T.E., Beaudreau, A.H., Wiedenmann, J., 2006. Fishing through marine food
webs. Proc. Natl. Acad. Sci. U.S.A. 103, 3171–3175.
113
Food and Agriculture Organization (FAO), 2003. The Ecosystem Approach to Fisheries. FAO Technical Guidelines for Responsible Fisheries No. 4 (Suppl. 2). FAO,
Rome.
Gaichas, S.K., Francis, R.C., 2008. Network models for ecosystem-based fishery analysis: a review of concepts and application to the Gulf of Alaska marine food web.
Can. J. Fish. Aquat. Sci. 65, 1965–1982.
Gislason, H., Sinclair, M., Sainsbury, K., O’Boyle, R., 2000. Symposium overview:
incorporating ecosystem objectives within fisheries management. ICES J. Mar.
Sci. 57, 468–475.
Hay, D.E., McCarter, P.B., 2000. Status of the Eulachon Thaleichthys pacificus in
Canada. Department of Fisheries and Oceans Canada, Canadian Stock Assessment Secretariat, Research Document 2000-145, Ottawa, Canada.
Hesslein, R.H., Hallard, K.A., Ramlal, R., 1993. Replacement of sulfur, carbon, and
nitrogen in tissue of growing broad whitefish (Coregonus nasus) in response
to a change in diet traced by ␦34 S, ␦13 C, and ␦15 N. Can. J. Fish. Aquat. Sci. 50,
2071–2076.
Hornborg, S., Belgranon, A., Bartolino, V., Valentinsson, D., Ziegler, F., 2013.
Trophic indicators in fisheries: a call for re-evaluation. Biol. Lett. 9,
http://dx.doi.org/10.1098/rsbl.2012.1050.
Hussey, N.E., MacNeil, M.A., McMeans, B.C., Olin, J.A., Dudley, S.F.J., Cliff, G., Wintner,
S.P., Fennessy, S.T., Fisk, A.T., 2013. Rescaling the trophic structure of marine
food webs. Ecol. Lett., 239–250, http://dx.doi.org/10.1111/ele.12226.
International Pacific Halibut Commission (IPHC), 1998. The Pacific Halibut: Biology,
Fishery, and Management. International Pacific Halibut Commission Technical
Report No. 40. IPHC, Seattle, Washington, USA.
Jennings, S., Greenstreet, S.P.R., Hill, L., Piet, G.J., Pinnegar, J.K., Warr, K.J., 2002a. Longterm trends in the trophic structure of the North Sea fish community: evidence
from stable-isotope analysis, size-spectra and community metrics. Mar. Biol.
141, 1085–1097.
Jennings, S., Pinnegar, J.K., Polunin, N.V.C., Warr, K.J., 2002b. Linking size-based
and trophic analyses of benthic community structure. Mar. Ecol. Prog. Ser. 226,
77–85.
Kline Jr., T.C., 2008. Ontogenetic, temporal, and spatial variation of feeding niche in an
unexploited population of walleye pollock (Theragra chalcogramma). In: Kruse,
G.H., Drinkwater, K., Ianelli, J.N., Link, J.S., Stram, D.L., Wespestad, V., Woodby,
D. (Eds.), Resiliency of Gadid Stocks to Fishing and Climate Change. Alaska
Sea Grant College Program AK-SG-08-01. University of Alaska, Fairbanks, USA,
pp. 251–269.
Litzow, M.A., 2006. Climate regime shifts and community reorganization in the Gulf
of Alaska: how do recent shifts compare with 1976/1977? ICES J. Mar. Sci. 63,
1386–1396.
Livingston, P.A., 2005. Ecosystem-based management indices and information:
trophic level of catch. Ecosystem considerations. In: Stock Assessment and Fishery Evaluation Report for the Groundfish Resources on the Bering Sea and
Aleutian Islands for 2002. North Pacific Fishery Management Council, 605W,
4th Ave., Suite 306, Anchorage, Alaska 99501, USA.
Livingston, P.A., Aydin, K., Boldt, J., Ianelli, J., Jurado-Molina, J., 2005. A framework
for ecosystem impacts assessment using an indicator approach. ICES J. Mar. Sci.
62 (3), 592–597.
Marsh, J.M., Hillgruber, N., Foy, R.J., 2012. Temporal and ontogenetic variability in
trophic role of four groundfish species – walleye pollock, Pacific cod, arrowtooth
flounder, and Pacific halibut – around Kodiak Island in the Gulf of Alaska. Trans.
Am. Fish. Soc. 141, 468–486.
McTigue, N.D., Dunton, K.H., 2014. Trophodynamics and organic matter assimilation pathways in the northeast Chukchi Sea, Alaska. Deep-Sea Res. II 102,
84–96.
Miller, T.W., 2000. Tissue-specific response of ␦15 N in adult Pacific herring (Clupea
pallasi) following an isotopic shift in diet. Environ. Biol. Fish. 76, 177–189.
Minagawa, M., Wada, E., 1984. Stepwise enrichment of 15 N along food chains: further
evidence and the relation between ␦15 N and animal age. Geochim. Cosmochim.
Acta 48, 1135–1140.
Mueter, F.J., Norcross, B.L., 2002. Spatial and temporal patterns in the demersal fish
community on the shelf and upper slope regions of the Gulf of Alaska. Fish. Bull.
100, 559–581.
National Oceanic and Atmospheric Administration (NOAA), 2014. Status of Stocks
2013. Annual Report to Congress on the Status of U.S. Fisheries. NOAA, Silver
Spring, MD.
North Pacific Fishery Management Council (NPFMC), 2009. GOA Introduction. In:
Stock Assessment and Fishery Evaluation Report for the Groundfish Resources
on the Bering Sea and Aleutian Islands for 2002. North Pacific Fisheries Management Council, 605W, 4th Ave., Suite 306, Anchorage, Alaska 99501, USA.
Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., Torres Jr., F.C., 1998. Fishing down
marine food webs. Science 279, 860–863.
Pauly, D., Christensen, V., Walters, C., 2000. Ecopath, Ecosim, and Ecospace as tools
for evaluating ecosystem impact on fisheries. ICES J. Mar. Sci. 57, 697–706.
Pauly, D., Palomares, M.L., Froese, R., Sa-a, P., Vakily, M., Preikshot, D., Wallace, S.,
2001. Fishing down Canadian aquatic food webs. Can. J. Fish. Aquat. Sci. 58,
51–62.
Pauly, D., Christensen, V., Guénette, S., Pitcher, T.J., Sumaila, U.R., Walters, C.J., Watson, R., Zeller, D., 2002. Towards sustainability in world fisheries. Nature 418,
689–695.
Post, D.M., 2002. Using stable isotopes to estimate trophic position: models, methods, and assumptions. Ecology 83, 703–718.
Scharf, F.S., Juanes, F., Roundtree, R.A., 2000. Predator size–prey size relationships of
marine fish predators: interspecific variation and effects of ontogeny and body
size on trophic-niche breadth. Mar. Ecol. Prog. Ser. 208, 229–248.
114
J.M. Marsh et al. / Fisheries Research 165 (2015) 100–114
Sherwood, G.D., Rose, G.A., 2005. Stable isotope analysis of some representative fish
and invertebrates of the Newfoundland and Labrador continental shelf food web.
Estuar. Coastal Shelf Sci. 63, 537–549.
Sinclair, M., Arnason, R., Csirke, J., Karnicki, Z., Sigurjonsson, J., Rune Skjoldal, H.,
Valdimarsson, G., 2002. Responsible fisheries in the marine ecosystem. Fish.
Res. 58, 255–265.
Sullivan, P.J., Trumble, R.J., Adlerstein, S.A., 1994. Pacific Halibut Bycatch in the
Groundfish Fisheries: Effects on Management Implications for the Halibut Fishery. International Pacific Halibut Commission Scientific Report No. 78, Seattle,
Washington, USA.
Thompson, G.G., Ianelli, J.N., Wilkins, M.E., 2009. Assessment of the Pacific cod stock
in the Gulf of Alaska. In: Stock Assessment and Fishery Evaluation Report for the
Groundfish Resources on the Bering Sea and Aleutian Islands for 2009. North
Pacific Fishery Management Council, 605W, 4th Ave., Suite 306, Anchorage,
Alaska 99501, USA (Chapter 2).
Turnock, B.J., Wilderbuer, T.K., 2009. Gulf of Alaska arrowtooth flounder stock assessment. In: Stock Assessment and Fishery Evaluation Report for the Groundfish
Resources on the Bering Sea and Aleutian Islands for 2009. North Pacific Fishery Management Council, 605W, 4th Ave., Suite 306, Anchorage, Alaska 99501,
USA.
Urban, D., Vining, I., 2008. Variation in the trophic level of Pacific cod with changes
in size and season. In: Kruse, G.H., Drinkwater, K., Ianelli, J.N., Link, J.S., Stram,
D.L., Wespestad, V., Woodby, D. (Eds.), Resiliency of Gadid Stocks to Fishing and
Climate Change. Alaska Sea Grant College Program AK-SG-08-01. University of
Alaska, Fairbanks, USA, pp. 305–315.
US National Research Council, 1998. Sustaining Marine Fisheries. A Report of the
Committee on Ecosystem Management for Sustainable Fisheries. Ocean Studies Board, Commission on Geosciences, Environment and Resources, National
Research Council. National Academy Press, Washington, DC, pp. 167.
Wilson, M.T., Jump, C.M., Buchheister, A., 2009. Ecology of small neritic fishes in the
western Gulf of Alaska: II. Consumption of krill in relation to krill standing stock
and the physical environment. Mar. Ecol. Prog. Ser. 392, 239–251.
Yang, M.S., Dodd, K., Hibpshman, R., Whitehouse, A., 2006. Food Habits of Groundfishes in the Gulf of Alaska in 1999 and 2001. U.S. Department of Commerce,
NOAA Technical Memorandum NMFS-AFSC-164, Seattle, Washington, USA.
Yang, M.S., Nelson, M.W., 2000. Food Habits of the Commercially Important Groundfishes in the Gulf of Alaska in 1990, 1993, and 1996. U.S. Department of
Commerce, NOAA Technical Memorandum NMFS-AFSC-112, Seattle, Washington, USA.
Zhang, C.I., Hollowed, A.B., Lee, J.-B., Kim, D.-H., 2011. An IFRAME approach for
assessing impacts of climate change on fisheries. ICES J. Mar. Sci. 68, 1318–1328.
Zhang, C.I., Kim, S., Gunderson, D., Marasco, R., Lee, J.B., Park, H.W., Lee, J.H., 2009. An
ecosystem-based fisheries assessment approach for Korean fisheries. Fish. Res.
100, 26–41.
Zimmermann, M., 1997. Maturity and fecundity of arrowtooth flounder, Atheresthes
stomias, from the Gulf of Alaska. Fish. Bull. 95, 598–611.