- Email: [email protected]
Seasonal changes in pelagic fish biomass around the Chiswell Island Steller sea lion rookery in 2003
Fisheries Research 93 (2008) 179–185
Contents lists available at ScienceDirect
Fisheries Research journal homepage: www.elsevier.com/locate/fishres
Seasonal changes in pelagic fish biomass around the Chiswell Island Steller sea lion rookery in 2003 Charles F. Adams a,b,∗ , Robert J. Foy c,1 , Devin S. Johnson d , Kenneth O. Coyle e a
Seward Marine Center, School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, P.O. Box 730, Seward, AK 99664-0730, USA Alaska SeaLife Center, P.O. Box 1329, Seward, AK 99664-1329, USA c School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, 118 Trident Way, Kodiak, AK 99615-7401, USA d National Marine Mammal Laboratory, Alaska Fisheries Science Center, National Oceanographic and Atmospheric Administration, 7600 Sand Point Way, Seattle, WA 98115-6349, USA e Institute of Marine Science, School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, P.O. Box 757220, Fairbanks, AK 99775-7220, USA b
a r t i c l e
i n f o
Article history: Received 5 March 2008 Received in revised form 1 April 2008 Accepted 10 April 2008 Keywords: Steller sea lion Eumetopias jubatus Walleye pollock Theragra chalcogramma Pacific herring Clupea pallasii
a b s t r a c t Fisheries acoustics surveys were conducted around the Chiswell Island rookery in the northern Gulf of Alaska at night in April and August 2003 to assess seasonal changes in prey available to Steller sea lions (Eumetopias jubatus) foraging around the rookery. Adult walleye pollock (Theragra chalcogramma) ≥28 cm fork length was the dominant biomass in the upper 50 m of the water column in both months, increasing from 122.8 kg/nmi2 in April to 457.9 kg/nmi2 in August. A similar pattern was observed for Pacific herring (Clupea pallasii), which averaged 2.8 and 65.6 kg/nmi2 in April and August, respectively. Incidental trawl catch suggested the appearance of age-0 pollock and juvenile salmonids (Oncorhynchus spp.) around the rookery in August as well. The increased biomass of these key prey species is linked to increased foraging trip durations by lactating sea lions from Chiswell Island, and supports the general view that sea lions in the northern Gulf of Alaska are not food limited during summer months. © 2008 Elsevier B.V. All rights reserved.
1. Introduction The western stock (i.e., west of 144◦ W) of Steller sea lions (SSLs), Eumetopias jubatus, decreased almost 80% between the late 1970s and late 1990s (Loughlin et al., 2003). One of the early hypotheses to explain this decline was a change in the availability or quality of their prey (Merrick et al., 1987). Possible causes were thought to be commercial fisheries (Braham et al., 1980); environmental changes resulting in a switch from feeding on high energy herring and osmerids to low energy gadids and flatfish (Alverson, 1992); and interspecific competition with groundfish for small forage fishes (Merrick, 1997). The population dynamics of SSLs are largely determined by the reproductive success of adult females and the survival of their young (York, 1994). Thus, the foraging behavior of these two groups
∗ Corresponding author at: School for Marine Science and Technology, University of Massachusetts Dartmouth, 706 South Rodney French Boulevard, New Bedford, MA 02744-1221, USA. Tel.: +1 508 910 6386; fax: +1 508 910 6396. E-mail address: [email protected] (C.F. Adams). 1 Kodiak Laboratory, Alaska Fisheries Science Center, National Oceanographic and Atmospheric Administration, Kodiak Fisheries Research Center, 301 Research Court, Kodiak, AK 99615-7400, USA. 0165-7836/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2008.04.003
is often used to make inferences regarding prey availability. In the central Gulf of Alaska (GOA) and Aleutian Islands, female SSLs with dependent young increase their foraging effort over the course of the year by spending more time at sea, diving deeper and having greater home ranges in winter as compared with summer (Merrick and Loughlin, 1997). In the northern GOA there is a sharp increase in the duration of foraging trips by lactating females in August (Maniscalco et al., 2006). Both of these observations are thought to be linked to seasonal changes in prey availability (Merrick and Loughlin, 1997; Maniscalco et al., 2006). Chiswell Island is a SSL rookery in the northern GOA that is used by about 90 breeding individuals, producing about 60 pups per year (Maniscalco et al., 2002). SSL diet in the Chiswell region during the mid-1970s consisted of 67% walleye pollock (Theragra chalcogramma) by frequency of occurrence (Pitcher, 1981). Current analysis of Chiswell SSL scats suggests that salmonids (Oncorhynchus spp.) are now the primary prey item, comprising 71% of the diet by frequency of occurrence, while pollock rank 2nd at 45%, and Pacific herring (Clupea pallasii) rank third at 37% (Jason Waite, Alaska SeaLife Center, unpublished data). Our objective was to provide a seasonal assessment of pelagic fish biomass available to SSLs foraging around the rookery in 2003. We conducted our surveys at night because it is generally thought that SSLs feed
180
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
Fig. 1. General map of Alaska (above), with a close up of the study area (below). Large circle depicts the 18.5-km (10-nmi) study radius, centered around the Chiswell Island rookery. Parallel lines represent acoustic transects. April 2003 trawl stations are denoted with a solid circle (䊉). August 2003 trawl stations are denoted with a solid square (), with the “H” further depicting the location of the herring school targeted with trawl 2, and empty squares () showing the locations of trawls 1 and 5, which were omitted from echogram analysis (see Section 2.3 for details).
primarily at this time (Higgins et al., 1988; Thomas and Thorne, 2001; Trites and Porter, 2002). 2. Materials and methods 2.1. Field sampling Echo-integration midwater trawl surveys were conducted from 11 to 17 April 2003, and 22 to 28 August 2003, within an 18.5km (10 nmi) radius of Chiswell Island (59◦ 36 N, 149◦ 34 W; Fig. 1). The April cruise was conducted aboard the 20-m R/V Pandalus, while the August cruise was conducted aboard the 23-m F/V Nightwatch. Surveys were done at night, between evening and morning civil twilights, when the sun is 6◦ below the horizon (http://aa.usno.navy.mil/). Acoustic data were collected with a Hydroacoustic Technology, Inc. (HTI), model 244 split-beam echosounder, equipped with a 38-kHz transducer. The transducer was towed beside the vessel at about 3 m/s in a dead-weight tow body about 4 m from the hull and 2 m below the surface. Sampling was restricted to calm conditions when noise due to surface bubbles was not observed in the data. The system col-
lected 20-log-R data, which were integrated at 15-s time intervals and 1-m depth intervals, to a maximum depth of 200 m. Data within 3 m of the transducer were excluded to avoid near-field effects. Differential global positioning system (GPS) position and Greenwich Meridian Time (GMT) from the ship’s navigation system were appended to each record before writing the data to disk. The echosounder was calibrated by HTI at their calibration barge before the field season according to the methods described in Foote et al. (1987). Briefly, the beam pattern was measured by mounting the transducer on a rotator whose angular position was coupled to a special purpose instrumentation system. The echosounder was used to generate pulses which were measured by a U.S. Navy reference standard transducer (ST). The two transducers were aligned for maximum on-axis response, and the plotter was normalized for 0◦ and 0 dB. The beam pattern was started with the transducer rotated off-axis by 90◦ . As the echosounder transmitter pulsed, the transducer was rotated back toward, and then past the ST. The echo level detected at the ST was converted to dB and plotted on the beam pattern plotter as a function of the off-axis angle. The system acoustic level was measured with the transducer and the ST aligned. Next, gain was determined with on-axis alignment of the transducer and the ST, with the latter used as the transmitting transducer. The known acoustic level was corrected for one-way transmission loss. Finally, the electrical/mechanical ratio was calculated as the relationship between the electrical phase angle and the physical angle of the target relative to the transducer axis (Ransom et al., 1995). In addition to the factory calibration, on-axis measurements were done in the field using a 38-mm tungsten carbide calibration sphere to verify that echosounder settings (Table 1) optimized by HTI were not altered during the cruises. Noise levels were determined at the start of each cruise. The signal threshold was set at 6 dB above the measured noise level for each depth interval and time-varied gain correction. Temperature effects were not observed. Parallel acoustics transects were spaced at 3-nmi intervals, perpendicular to the 200-m isobath, which is close to the mean depth of pollock distribution in the region (Hughes and Hirschhorn, 1979). Opportunistic midwater trawls were also done at night. Tows were done at a constant depth, aimed at layers of high backscatter, and averaged 25 min in duration. A Gourock, Inc., midwater trawl with vertical and horizontal openings of 12 and 22 m, respectively, was fitted with a 1.9 cm mesh cod end liner. Trawl stations were distributed throughout the study area in April, while they were concentrated in Aialik Bay and Dora Passage in August (Fig. 1). Trawl depths ranged between 22 and 87 m in April; and between 16 and 124 m in August (Table 2). Up to 200 random specimens of each species per tow were retained for shipboard measurements of FL, while additional fish were counted as they were released. Of the 200 retained fish, a sub-sample of 50 specimens was frozen for length–weight measurements in the laboratory.
Table 1 38-kHz settings for the HTI model 244 split-beam echosounder used within a 10-nmi radius around Chiswell Island in 2003 Absorption coefficient Beam angle Ping rate Equivalent beam angle Pulse length Source level Transmit power TVG Gain Note that 33 dBW = 1995.3 W. Calculation: W = 10(33/10) .
10.2 dB/km 7◦ 0.8 pps −29.05 dB re 1 sr 5 ms 217.78 dB re 1 Pa 33 dBW −2.4 dB
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
181
Table 2 Trawl catch data within a 10-nmi radius around Chiswell Island in 2003 Min HR (m)
Max FR (m)
Capelin (n)
Eulachon (n)
Herring (n)
Pollock (n)
Other (n)
April Trawl 1 Trawl 2 Trawl 3 Trawl 4 Trawl 5 Trawl 6 Trawl 7
29.3 25.6 22.0 22.0 25.6 38.4 25.6
46.8 38.4 78.6 62.2 86.6 58.5 76.8
– – 2 2 1 – –
– – 3 – – 6 –
– – – – 6 – –
63 79 378 23 92 183 1107
– – 1a – – – –
August Trawl 1 Trawl 2 Trawl 3 Trawl 4 Trawl 5
100.6 18.2 18.3 23.8 15.5
124.4 49.4 43.9 47.5 36.6
16 1 – – –
1 – – – –
– 441 127 4 3
11 13 315 85 1
2b 11c 2d – 4e
Min HR: minimum headrope depth during tow. Max FR: maximum footrope depth during tow. Note that Min HR and Max HR were recorded at the start and end of acoustic data collection and do not include deployment and retrieval of the net. n: number of fish. a Arrowtooth flounder (Atheresthes stomias). b Juvenile salmonids (Oncorhynchus spp.). c Includes six arrowtooth flounder, four juvenile salmonids and one spiny dogfish (Squalus acanthias). d Prowfish (Zaprora silenus). e Sablefish (Anoplopoma fimbria).
2.2. Laboratory procedures FL for all frozen specimens was measured to the nearest mm and weighed to the nearest 0.1 g. As total length (TL) measurements are required for the use of the target strength (TS) models of Gauthier and Horne (2004), the FL measurements for pollock, capelin (Mallotus villosus) and eulachon (Thaleichthys pacificus) were converted to TL with the equations provided in Buchheister and Wilson (2005). Similarly, FL measurements for herring were converted with the equations provided in Karpov and Kwiecien (1988). Shrinkage due to freezing (ca. 1–3%) was also corrected for pollock and eulachon length measurements (Buchheister and Wilson, 2005). A similar correction for herring is currently not available. Length–weight regressions were calculated for pollock, herring and eulachon. No capelin were collected for laboratory measurements and so the equation provided in Brown (2002) was used to generate a length–weight regression.
Table 3 Target strength–length equations from Gauthier and Horne (2004) used to partition echograms within a 10-nmi radius around Chiswell Island in 2003 Capelin Eulachon Pacific herring Walleye pollock L: total length in cm.
transect. Furthermore, we note that trawls 1 and 5 were excluded from the August species mix because the former was the only tow in the 100–150-m depth bin, and because of the low number of fish (n = 8) in the latter tow (Table 2). Once the species mix for each cruise was calculated, echointegrals Ei were partitioned with:
Ei =
wi i Em
2.3. Echogram processing
M wi =
q /qk k=1 ik M
I
i=1
Echo-integration data were processed with Echoview software (SonarData Pty. Ltd., Hobart, Tasmania, Australia). Echo-integrals contaminated by the intrusion of bottom echoes into the water column were defined as bad data regions in Echoview and omitted from further analysis (Simmonds and MacLennan, 2005). The April catch data were combined into a species mix to partition all the echograms for that cruise using the equation: (1)
where wi is the proportion w of species i, qik is the quantity of species i caught at station k, qk is the total catch at station k, and M is the total number of trawls for the respective cruise (Simmonds and MacLennan, 2005). August trawls 3 and 4 were combined into a species mix to partition all the echograms for that cruise with Eq. (1). The only exception to this procedure was n = 3 small schools (<5% of transects) that could be identified as herring based on school structure. These schools were partitioned with the catch data for trawl 2 (Table 2), which was aimed at a herring school. The biomass for each school was added to the species mix biomass for the respective
24.9 log L–75 27.3 log L–94 13.3 log L–55.9 19.2 log L–66
(2)
wi i
where I is the total number of species, i is the mean backscattering cross-section of species i, and Em is the volume backscattering coefficient (Simmonds and MacLennan, 2005). The mean backscattering cross-section of species i was calculated as
i =
wij 10[ai +bi log(Lij )/10]
(3)
j
where Lij is the midpoint of the jth size class for species i, and ai and bi are constants for the ith species (Table 3). Note that only pollock were split into different size classes because it is known that juvenile SSLs feed on smaller pollock (Merrick and Calkins, 1996). This splitting was done by visual inspection of peaks and troughs in length–frequency histograms (Anderson and Neumann, 1996). Backscatter not attributed to capelin, eulachon, herring or pollock was considered “other.” The abundance of species i, expressed as the number/nmi2 , was calculated using: i =
(1852)2
z2 z1
i
/V
dz (4)
182
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
Table 4 Summary of length–weight regressions used to convert abundance estimates to biomass within a 10-nmi radius around Chiswell Island in 2003 n Eulachon Herring Pollock
40 211 379
Min (cm) 16.4 7.0 9.7
Max (cm) 23.3 27.0 52.6
Regression 3.0697
0.0046L 0.008L3.0555 0.0095L2.9095
r2 0.84 0.94 0.97
Weight is expressed in g. n: number of fish. L: length of fish in cm.
where V is the volume insonified and z is depth. This abundance estimate was then converted to kg/nmi2 using the length–weight regressions described above (Simmonds and MacLennan, 2005). Biomass data were exported from Echoview in two depth bins: 20–50 m, and 50–100 m. This was done for two reasons. One, to facilitate comparison with previous studies of SSL diving behavior that used the same depth bins (e.g., Merrick and Loughlin, 1997; Loughlin et al., 2003). Two, to minimize the vertical aspect of the error associated with partitioning echograms based on trawl catch data, because only two trawls were used to calculate wi for August. This was done by rounding the minimum/maximum headrope/footrope depths up or down to the nearest bin cutoff. For example, the minimum headrope depth for trawls 3 and 4 in August was 18.3 m, while the maximum footrope depth was 47.5 m; the closest bin cutoffs were 20 and 50 m; thus, August echo-integration data were partitioned to biomass between 20 and 50 m.
Fig. 2. Length–frequency histograms for walleye pollock sampled by midwater trawl within a 10-nmi radius around Chiswell Island in 2003.
2.4. Biomass estimates A retrospective analysis was done on the elementary distance sampling unit (EDSU), which is defined as the distance along a transect over which acoustic measurements are averaged to give one sample. If the EDSU is too large, information about the geographical distribution of the stock will be lost, whereas if it is too small, successive samples will be dominated by local variability (Simmonds and MacLennan, 2005, p. 324). Thus, data were binned at 0.1, 0.5 and 1.0 nmi, and experimental semivariograms (Rivoirard et al., 2000) were generated for each bin size in each month with geoR (http://www.est.ufpr.br/geoR/) to determine the most appropriate EDSU. It was found that binning at 0.5 nmi was more appropriate than 0.1 or 1.0 nmi. Adjacent acoustic measurements exhibit autocorrelation (Simmonds and MacLennan, 2005, p. 324), so calculating a mean biomass and 95% confidence limit for each species would be a form of pseudoreplication (Hurlbert, 1984; Millar and Anderson, 2004). To account for the spatial correlation in the data we estimated the mean biomass and 95% confidence limit for each species with a generalized linear mixed model (Venables and Dichmont, 2004) in PROC MIXED (SAS Institute, Cary, NC, U.S.A.), which allows the use of spatially correlated error terms in the standard linear model Y = X + . Error terms were modeled with a geostatistical spherical covariance structure (Rivoirard et al., 2000). One of the assumptions of a generalized linear mixed model is that the data are normally distributed. Even after logtransformation all biomass data still failed a Shapiro–Wilk test for normality (Zar, 1999). Thus, we present 95% confidence limits with the caveat that they are only approximate intervals. 3. Results 3.1. Laboratory procedures Length–weight regressions for pollock, herring and eulachon are presented in Table 4. The length–frequency histogram for April pollock showed clear peaks at 14.5 and 36.0 cm (Fig. 2), which corresponds to age-1
juveniles and age-3 adults, respectively (Hughes and Hirschhorn, 1979; Shima et al., 2002; Duffy-Anderson et al., 2003). August peaks occurred at 9.0, 19.0 and 32.5 cm, which correspond to age-0 juveniles, age-1 juveniles and age-3 adults, respectively. For brevity, the age-3 fish will be referred to as adults for the remainder of this paper, even though the left hand tail of this distribution clearly contains some age-2 juveniles. 3.2. Echogram processing Values used to partition echograms into the species mix for each cruise are presented in Table 5. 3.3. Biomass estimates Adult pollock was the dominant biomass in the 20–50 m depth bin in April, averaging 122.8 kg/nmi2 (Table 6). This was two orders of magnitude greater than the mean biomass of herring and age-1 pollock, both of which averaged 2.8 kg/nmi2 . Eulachon and capelin Table 5 Data used to partition echograms into species mix within a 10-nmi radius around Chiswell Island in 2003 TS (dB)
TL (cm)
Weight (g)
% mix
April Capelin Eulachon Herring Pollock, age-1 Pollock, adult Other
−49.72 −57.89 −37.36 −43.98 −36.28 NA
10.4 21.0 24.8 14.0 35.3 NA
6.7 52.9 145.7 20.6 303.3 NA
1.36 0.57 0.87 6.12 91.05 0.04
August Herring Pollock, age-0 Pollock, age-1 Pollock, adult Other
−37.63 −48.33 −41.41 −36.44 NA
23.7 8.3 19.1 34.6 NA
126.2 4.5 50.6 286.0 NA
18.11 2.57 0.34 78.76 0.22
TS: target strength. TL: average total length.
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
183
Table 6 Predicted mean biomass (Yˆ ), arithmetic mean biomass (¯x), and respective 95% confidence intervals within a 10-nmi radius of Chiswell Island, 2003 Confidence limits Yˆ
Upper
x¯
0.12 0.28 0.93 0.92 9.52
0.86 2.20 6.51 6.48 1455.13
0.43 1.02 2.75 2.74 110.83
0.32 0.77 2.08 2.07 66.47
0.56 1.30 3.58 3.57 184.35
0.84 1.91 5.15 5.13 271.40
0.00a 0.27 1.02 1.02 33.51
2.86 5.69 17.69 17.62 2148.31
0.81 1.87 4.93 4.92 229.80
0.63 1.43 3.66 3.64 134.42
1.00 2.38 6.56 6.54 392.35
65.56 0.83 1.20 457.88
18.36 0.46 0.65 94.61
227.88 1.30 1.93 2201.42
62.56 0.74 1.07 453.29
45.98 0.59 0.86 315.54
85.00 0.90 1.31 650.99
April 20–50 m Capelin Eulachon Herring Pollock, age-1 Pollock, adult
0.44 1.02 2.81 2.79 122.77
50–100 m Capelin Eulachon Herring Pollock, age-1 Pollock, adult August 20–50 m Herring Pollock, age-0 Pollock, age-1 Pollock, adult a
Confidence limits
Lower
Lower
Upper
Impossible negative confidence limit replaced with 0.
were the least abundant, averaging 1.02 and 0.44 kg/nmi2 , respectively. Trends for the 50–100 m depth bin in April were similar, except that mean biomass was greater than that of the shallower depth bin (Table 6). For example, average biomass of adult pollock was 271.4 kg/nmi2 in the 50–100 m depth bin, as compared with 122.8 kg/nmi2 in the 20–50 m depth bin. Adult pollock was the dominant biomass in the 20–50-m depth bin in August, averaging 457.9 kg/nmi2 (Table 6). Herring was an order of magnitude lower at 65.6 kg/nmi2 . Age-1 and age-0 pollock were the third and fourth most abundant, with a mean biomass of 1.2 and 0.8 kg/nmi2 , respectively. The biomass of adult pollock in the 20–50 m depth bin increased from 122.8 to 457.9 kg/nmi2 between April and August, respectively (Table 6). Similarly, there was a 23-fold increase in the mean biomass of herring, from 2.8 to 65.6 kg/nmi2 . The biomass of age1 pollock decreased slightly from 2.8 to 1.2 kg/nmi2 . Age-0 pollock appeared in August, while capelin and eulachon were not observed. A comparison of the standard confidence limits in Table 6 with the approximate intervals derived from accounting for the spatial correlation in the data illustrate that any inferences with respect to potentially significant differences between different categories should be made with caution. For example, the standard confidence interval for August herring (46.0–85.0 kg/nmi2 ) does not overlap with the interval for adult pollock (315.5–651.0 kg/nmi2 ); but once they are adjusted for spatial correlation the interval for herring (18.4–227.9 kg/nmi2 ) does indeed overlap with the interval for adult pollock (94.6–2201.4 kg/nmi2 ).
4. Discussion 4.1. Error budget Acoustically derived biomass estimates contain both sampling error, caused by the measurements being stochastic samples of the true mean density, and systematic error, which affects all the observations equally. Examples of the latter include equipment sensitivity, transducer motion, the surface bubble layer, etc. (Simmonds and MacLennan, 2005).
The primary source of sampling error results from survey design (Simmonds and MacLennan, 2005). Other sources of sampling error are variation in the mean TS over the surveyed region; species discrimination through echogram partitioning; diurnal variability affecting fish distribution and TS; sampling echogram marks by fishing to provide species proportions; and sudden changes in fish distribution and behavior due to storm events, etc. (Simmonds and MacLennan, 2005). Given the small size of the study area, it was assumed that we were dealing with a single fish population, and so it follows that variation in mean TS would be negligible. All echograms were categorized as a species mix based on trawl catch, so any error resulting from species discrimination based on echogram characteristics would have been restricted to the three small herring schools in August. Transects were run between evening and morning civil twilights, during the period when vertical migrators are near the surface, so it was assumed that there was no diurnal variability in fish distribution or TS during data collection. Finally, there was no major storm event during any of the cruises so it is unlikely there were major changes in fish distribution over the course of each cruise. This leaves partitioning echogram marks based on species proportions in trawl catch as the largest source of other sampling error in this study. A minimum of five to seven trawl stations are needed per 100 nmi2 in order to minimize the variance in abundance estimates (Godø et al., 1998). Although we met these minimum criteria in both cruises, not all catch data were used to calculate wi in August due to the concerns described in Section 2.3. Our method of partitioning echograms based on as little as two tows undoubtedly introduced the largest source of error in this study. However, we attempted to minimize the vertical aspect of this error by not extrapolating species composition and biomass estimates outside trawl depths. 4.2. Pelagic fish biomass available to Chiswell SSLs Our objective in this study was to provide a seasonal assessment of pelagic fish biomass available to SSLs foraging around the Chiswell Island rookery in 2003. We presented our data in depth bins comparable with previous studies of diving behavior
184
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
in SSLs (e.g., Merrick and Loughlin, 1997; Loughlin et al., 2003). Our results offer several potential insights into the foraging behavior of Chiswell SSLs. In April there was more fish biomass in the 50–100 m depth bin as compared with the 20–50 m depth bin. This is consistent with the observation that female SSLs dive more in depth bins >50 m in late winter as compared with summer (Merrick and Loughlin, 1997). There were two seasonal differences in the 20–50 m depth bin between April and August. One was an increase in the biomass of adult pollock. This is important because the historical mean size of pollock consumed by SSLs in the Chiswell region was 29.8 cm (Pitcher, 1981), and contemporary analysis indicates that 75% of consumed pollock are ≥27.0 cm (Jason Waite, Alaska SeaLife Center, unpublished data). As we noted in Section 3.1, these sizes are considered adults in the GOA (Hughes and Hirschhorn, 1979; Shima et al., 2002; Duffy-Anderson et al., 2003). The other seasonal trend in the 20–50 m depth bin was an increase in the biomass of forage fishes. The clearest evidence for this was the ca. 23-fold increase in the mean biomass of herring between April and August. Furthermore, trawls 1 and 2 suggest the appearance of juvenile salmonids around the rookery at this time. The latter observation is consistent with other reports that juvenile pink salmon (Oncorhynchus gorbuscha) move into the eastern side of our study area in August (Armstrong et al., 2005). These findings are important because current analysis of Chiswell SSL scats suggests that salmonids are the primary prey item, comprising 71% of the diet by frequency of occurrence, while herring rank third at 37% (Jason Waite, Alaska SeaLife Center, unpublished data). The increased availability of adult pollock, herring and salmon in August is coincident with an increase in the duration of foraging trips by lactating Chiswell SSLs (Maniscalco et al., 2006). This suggests that female SSLs increase their foraging trip durations in response to increased prey availability. However, this conclusion should be viewed with caution, given that the observations by Maniscalco et al. (2006) were done during daylight hours, while our surveys were done at night. Increased foraging trip durations in the presence of increased prey availability can be explained by optimal foraging theory (Stephens and Krebs, 1986). Laboratory study of shallow dives by female SSLs found that increased prey encounter rates resulted in increased dive durations and foraging times (Cornick and Horning, 2003). In contrast, individual dive duration and foraging time were reduced in cost-increased dives that were comparable with reduced prey encounter rates (Cornick et al., 2006). Although we did not do any tows less than 15 m, our results offer some tentative insights into prey availability at the shallower depths utilized by juvenile SSLs (e.g., Merrick and Loughlin, 1997). There was a ca. 50% reduction in the mean biomass of age-1 pollock between April and August. This is important because this is the age-class of pollock that is comparable to the peak size of pollock that juvenile SSLs feed most heavily on (Fig. 3 in Merrick and Calkins, 1996). On the other hand, our trawl catch data indicate the appearance of at least small numbers of age-0 pollock around the rookery in August. This study supports the general view that SSLs in the northern GOA have not been food-limited during recent years, at least during summer months (Maniscalco et al., 2006). One of the possible explanations for the change in foraging trip durations of lactating Chiswell SSLs is thought to be a seasonal change in prey availability (Maniscalco et al., 2006). Our finding of an increased biomass of adult pollock, herring and salmonids in August 2003 supports this observation. Obviously, additional surveys are needed to confirm whether these trends are representative of other years. Nevertheless, our study illustrates some of the unique insights that
can be gained by linking fisheries acoustics surveys with studies of the foraging behavior and diet of endangered Steller sea lions. Acknowledgements We thank the crews of the F/V Nightwatch and R/V Pandalus, ´ as well as Stephane Gauthier, Elizabeth Moundalexis and Sarah Norberg for assistance with field collections. We also thank Leslie Cornick for helpful discussions about SSL diving behavior and Susan Inglis for suggestions regarding an earlier draft of this manuscript. Alexei Pinchuk provided Visual Basic code to fix damaged HTI files. Margaret Spahn provided the bathymetry shapefile. Support for this project was provided by the Alaska SeaLife Center Steller Sea Lion Research Program with funds from National Marine Fisheries Service. References Alverson, D.L., 1992. A review of commercial fisheries and the Steller sea lion (Eumetopias jubatus): the conflict arena. Rev. Aquat. Sci. 6, 203–256. Anderson, R.O., Neumann, R.M., 1996. Length, weight, and associated structural indices. In: Murphy, B.R., Willis, D.W. (Eds.), Fisheries Techniques. American Fisheries Society, Bethesda, MD, pp. 447–481. Armstrong, J.L., Boldt, J.L., Cross, A.D., Moss, J.H., Davis, N.D., Myers, K.W., Walker, R.V., Beauchamp, D.A., Haldorson, L.J., 2005. Distribution, size, and interannual, seasonal and diel food habits of northern Gulf of Alaska juvenile pink salmon, Oncorhynchus gorbuscha. Deep-Sea Res. Part II 52, 247–265. Braham, H.W., Everitt, R.D., Rugh, D.J., 1980. Northern sea lion population declines in the eastern Aleutian Islands. J. Wildl. Manage. 44, 25–33. Brown, E.D., 2002. Life history, distribution, and size structure of Pacific capelin in Prince William Sound and the Northern Gulf of Alaska. ICES J. Mar. Sci. 59, 983–996. Buchheister, A., Wilson, M.T., 2005. Shrinkage correction and length conversion equations for Theragra chalcogramma, Mallotus villosus and Thaleichthys pacificus. J. Fish Biol. 67, 541–548. Cornick, L.A., Horning, M., 2003. A test of hypotheses based on optimal foraging considerations for a diving mammal using a novel experimental approach. Can. J. Zool. 81, 1799–1807. Cornick, L.A., Inglis, S.D., Willis, K., Horning, M., 2006. Effects of increased swimming costs on foraging behavior and efficiency of captive Steller sea lions: evidence for behavioral plasticity in the recovery phase of dives. J. Exp. Mar. Biol. Ecol. 333, 306–314. Duffy-Anderson, J.T., Ciannelli, L., Honkalehto, T., Bailey, K.M., Sogard, S.M., Springer, A.M., Buckley, T., 2003. Distribution of age-1 and age-2 walleye pollock in the Gulf of Alaska and eastern Bering Sea: sources of variation and implications for higher trophic levels. In: Browman, H.I., Skiftesvik, A.B. (Eds.), The Big Fish Bang. Proceedings of the 26th Annual Larval Fish Conference, Bergen 2002. Institute of Marine Research, Bergen, pp. 381–394. Foote, K.G., Knudsen, H.P., Vestnes, G., MacLennan, D.N., Simmonds, E.J., 1987. Calibration of acoustic instruments for fish density estimation: a practical guide. ICES Coop. Res. Rep. 144. International Council for the Exploration of the Sea, Copenhagen. Gauthier, S., Horne, J.K., 2004. Acoustic characteristics of forage fish species in the Gulf of Alaska and Bering Sea based on Kirchoff-approximation models. Can. J. Fish. Aquat. Sci. 61, 1839–1850. Godø, O.R., Karp, W.A., Totland, A., 1998. Effects of trawl sampling variability on precision of acoustic abundance estimates of gadoids from the Barents Sea and the Gulf of Alaska. ICES J. Mar. Sci. 55, 86–94. Higgins, L.V., Costa, D.P., Huntley, A.C., Le Boeuf, B.J., 1988. Behavioral and physiological measurements of maternal investment in the Steller sea lion, Eumetopias jubatus. Mar. Mamm. Sci. 4, 44–58. Hurlbert, S.H., 1984. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54, 187–211. Hughes, S.E., Hirschhorn, G., 1979. Biology of walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska 1973–75. Fish. Bull. 77, 263–274. Karpov, K.A., Kwiecien, G.S., 1988. Conversions between total, fork, and standard lengths for 41 species in 15 families of fish from California using preserved and fresh specimens. California Department of Fish and Game Marine Resources Administrative Report No. 88-9, 17 pp. Loughlin, T.R., Sterling, J.T., Merrick, R.L., Sease, J.L., York, A.E., 2003. Diving behavior of immature Steller sea lions (Eumetopias jubatus). Fish. Bull. 101, 566–582. Maniscalco, J., Atkinson, S., Armato, P., 2002. Early maternal care and pup survival in Steller sea lions. A remote video monitoring project in the northern Gulf of Alaska. Arctic Res. U.S. 16, 36–41. Maniscalco, J.M., Parker, P., Atkinson, S., 2006. Interseasonal and interannual measures of maternal care among individual Steller sea lions (Eumetopias jubatus). J. Mamm. 87, 304–311.
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185 Merrick, R.L., 1997. Current and historical roles of apex predators in the Bering Sea ecosystem. J. Northwest Atl. Fish. Sci. 22, 343–355. Merrick, R.L., Calkins, D.G., 1996. Importance of juvenile walleye pollock, Theragra chalcogramma, in the diet of Gulf of Alaska Steller sea lions, Eumetopias jubatus. In: Brodeur, R.D., Livingston, P.A., Loughlin, T.R., Hollowed, A.B. (Eds.), Ecology of Juvenile Walleye Pollock (Theragra chalcogramma). NOAA Tech. Rep. NMFSAFSC-126, pp. 153–166. Merrick, R.L., Loughlin, T.R., 1997. Foraging behavior of adult female and young-ofthe-year Steller sea lions in Alaskan waters. Can. J. Zool. 75, 776–786. Merrick, R.L., Loughlin, T.R., Calkins, D.G., 1987. Decline in abundance of the northern sea lion, Eumetopias jubatus, in Alaska 1956–86. Fish. Bull. 85, 351–365. Millar, R.B., Anderson, M.J., 2004. Remedies for pseudoreplication. Fish. Res. 70, 397–407. Pitcher, K.W., 1981. Prey of the Steller sea lion, Eumetopias jubatus, in the Gulf of Alaska. Fish. Bull. 79, 467–472. Ransom, B.H., Steig, T.W., Johnston, S.V., Ehrenberg, J.E., Torkelson, T.C., Nealson, P.A., 1995. Using Hydroacoustics for Fisheries Assessment. Hydroacoustic Technology, Inc., Seattle.
185
Rivoirard, J., Simmonds, J., Foote, K.G., Fernandes, P., Bez, N., 2000. Geostatistics for Estimating Fish Abundance. Blackwell Science Ltd., Oxford. Shima, M., Hollowed, A.B., VanBlaricom, G.R., 2002. Changes over time in the spatial distribution of walleye pollock (Theragra chalcogramma) in the Gulf of Alaska 1984–1996. Fish. Bull. 100, 307–323. Simmonds, J., MacLennan, D., 2005. Fisheries Acoustics. Blackwell Science Ltd., Oxford. Stephens, D.W., Krebs, J.R., 1986. Foraging Theory. Princeton University Press, New Jersey. Thomas, G.L., Thorne, R.E., 2001. Night-time predation by Steller sea lions. Nature 411, 1013. Trites, A.W., Porter, B.W., 2002. Attendance patterns of Steller sea lions (Eumetopias jubatus) and their young during winter. J. Zool. Lond. 256, 547–556. Venables, W.N., Dichmont, C.M., 2004. GLMs, GAMs and GLMMs: an overview of theory for application in fisheries research. Fish. Res. 70, 319–337. York, A.E., 1994. The population dynamics of northern sea lions 1975–1985. Mar. Mamm. Sci. 10, 38–51. Zar, J.H., 1999. Biostatistical Analysis. Prentice-Hall, Upper Saddle River, New Jersey.
Contents lists available at ScienceDirect
Fisheries Research journal homepage: www.elsevier.com/locate/fishres
Seasonal changes in pelagic fish biomass around the Chiswell Island Steller sea lion rookery in 2003 Charles F. Adams a,b,∗ , Robert J. Foy c,1 , Devin S. Johnson d , Kenneth O. Coyle e a
Seward Marine Center, School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, P.O. Box 730, Seward, AK 99664-0730, USA Alaska SeaLife Center, P.O. Box 1329, Seward, AK 99664-1329, USA c School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, 118 Trident Way, Kodiak, AK 99615-7401, USA d National Marine Mammal Laboratory, Alaska Fisheries Science Center, National Oceanographic and Atmospheric Administration, 7600 Sand Point Way, Seattle, WA 98115-6349, USA e Institute of Marine Science, School of Fisheries and Ocean Sciences, University of Alaska Fairbanks, P.O. Box 757220, Fairbanks, AK 99775-7220, USA b
a r t i c l e
i n f o
Article history: Received 5 March 2008 Received in revised form 1 April 2008 Accepted 10 April 2008 Keywords: Steller sea lion Eumetopias jubatus Walleye pollock Theragra chalcogramma Pacific herring Clupea pallasii
a b s t r a c t Fisheries acoustics surveys were conducted around the Chiswell Island rookery in the northern Gulf of Alaska at night in April and August 2003 to assess seasonal changes in prey available to Steller sea lions (Eumetopias jubatus) foraging around the rookery. Adult walleye pollock (Theragra chalcogramma) ≥28 cm fork length was the dominant biomass in the upper 50 m of the water column in both months, increasing from 122.8 kg/nmi2 in April to 457.9 kg/nmi2 in August. A similar pattern was observed for Pacific herring (Clupea pallasii), which averaged 2.8 and 65.6 kg/nmi2 in April and August, respectively. Incidental trawl catch suggested the appearance of age-0 pollock and juvenile salmonids (Oncorhynchus spp.) around the rookery in August as well. The increased biomass of these key prey species is linked to increased foraging trip durations by lactating sea lions from Chiswell Island, and supports the general view that sea lions in the northern Gulf of Alaska are not food limited during summer months. © 2008 Elsevier B.V. All rights reserved.
1. Introduction The western stock (i.e., west of 144◦ W) of Steller sea lions (SSLs), Eumetopias jubatus, decreased almost 80% between the late 1970s and late 1990s (Loughlin et al., 2003). One of the early hypotheses to explain this decline was a change in the availability or quality of their prey (Merrick et al., 1987). Possible causes were thought to be commercial fisheries (Braham et al., 1980); environmental changes resulting in a switch from feeding on high energy herring and osmerids to low energy gadids and flatfish (Alverson, 1992); and interspecific competition with groundfish for small forage fishes (Merrick, 1997). The population dynamics of SSLs are largely determined by the reproductive success of adult females and the survival of their young (York, 1994). Thus, the foraging behavior of these two groups
∗ Corresponding author at: School for Marine Science and Technology, University of Massachusetts Dartmouth, 706 South Rodney French Boulevard, New Bedford, MA 02744-1221, USA. Tel.: +1 508 910 6386; fax: +1 508 910 6396. E-mail address: [email protected] (C.F. Adams). 1 Kodiak Laboratory, Alaska Fisheries Science Center, National Oceanographic and Atmospheric Administration, Kodiak Fisheries Research Center, 301 Research Court, Kodiak, AK 99615-7400, USA. 0165-7836/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2008.04.003
is often used to make inferences regarding prey availability. In the central Gulf of Alaska (GOA) and Aleutian Islands, female SSLs with dependent young increase their foraging effort over the course of the year by spending more time at sea, diving deeper and having greater home ranges in winter as compared with summer (Merrick and Loughlin, 1997). In the northern GOA there is a sharp increase in the duration of foraging trips by lactating females in August (Maniscalco et al., 2006). Both of these observations are thought to be linked to seasonal changes in prey availability (Merrick and Loughlin, 1997; Maniscalco et al., 2006). Chiswell Island is a SSL rookery in the northern GOA that is used by about 90 breeding individuals, producing about 60 pups per year (Maniscalco et al., 2002). SSL diet in the Chiswell region during the mid-1970s consisted of 67% walleye pollock (Theragra chalcogramma) by frequency of occurrence (Pitcher, 1981). Current analysis of Chiswell SSL scats suggests that salmonids (Oncorhynchus spp.) are now the primary prey item, comprising 71% of the diet by frequency of occurrence, while pollock rank 2nd at 45%, and Pacific herring (Clupea pallasii) rank third at 37% (Jason Waite, Alaska SeaLife Center, unpublished data). Our objective was to provide a seasonal assessment of pelagic fish biomass available to SSLs foraging around the rookery in 2003. We conducted our surveys at night because it is generally thought that SSLs feed
180
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
Fig. 1. General map of Alaska (above), with a close up of the study area (below). Large circle depicts the 18.5-km (10-nmi) study radius, centered around the Chiswell Island rookery. Parallel lines represent acoustic transects. April 2003 trawl stations are denoted with a solid circle (䊉). August 2003 trawl stations are denoted with a solid square (), with the “H” further depicting the location of the herring school targeted with trawl 2, and empty squares () showing the locations of trawls 1 and 5, which were omitted from echogram analysis (see Section 2.3 for details).
primarily at this time (Higgins et al., 1988; Thomas and Thorne, 2001; Trites and Porter, 2002). 2. Materials and methods 2.1. Field sampling Echo-integration midwater trawl surveys were conducted from 11 to 17 April 2003, and 22 to 28 August 2003, within an 18.5km (10 nmi) radius of Chiswell Island (59◦ 36 N, 149◦ 34 W; Fig. 1). The April cruise was conducted aboard the 20-m R/V Pandalus, while the August cruise was conducted aboard the 23-m F/V Nightwatch. Surveys were done at night, between evening and morning civil twilights, when the sun is 6◦ below the horizon (http://aa.usno.navy.mil/). Acoustic data were collected with a Hydroacoustic Technology, Inc. (HTI), model 244 split-beam echosounder, equipped with a 38-kHz transducer. The transducer was towed beside the vessel at about 3 m/s in a dead-weight tow body about 4 m from the hull and 2 m below the surface. Sampling was restricted to calm conditions when noise due to surface bubbles was not observed in the data. The system col-
lected 20-log-R data, which were integrated at 15-s time intervals and 1-m depth intervals, to a maximum depth of 200 m. Data within 3 m of the transducer were excluded to avoid near-field effects. Differential global positioning system (GPS) position and Greenwich Meridian Time (GMT) from the ship’s navigation system were appended to each record before writing the data to disk. The echosounder was calibrated by HTI at their calibration barge before the field season according to the methods described in Foote et al. (1987). Briefly, the beam pattern was measured by mounting the transducer on a rotator whose angular position was coupled to a special purpose instrumentation system. The echosounder was used to generate pulses which were measured by a U.S. Navy reference standard transducer (ST). The two transducers were aligned for maximum on-axis response, and the plotter was normalized for 0◦ and 0 dB. The beam pattern was started with the transducer rotated off-axis by 90◦ . As the echosounder transmitter pulsed, the transducer was rotated back toward, and then past the ST. The echo level detected at the ST was converted to dB and plotted on the beam pattern plotter as a function of the off-axis angle. The system acoustic level was measured with the transducer and the ST aligned. Next, gain was determined with on-axis alignment of the transducer and the ST, with the latter used as the transmitting transducer. The known acoustic level was corrected for one-way transmission loss. Finally, the electrical/mechanical ratio was calculated as the relationship between the electrical phase angle and the physical angle of the target relative to the transducer axis (Ransom et al., 1995). In addition to the factory calibration, on-axis measurements were done in the field using a 38-mm tungsten carbide calibration sphere to verify that echosounder settings (Table 1) optimized by HTI were not altered during the cruises. Noise levels were determined at the start of each cruise. The signal threshold was set at 6 dB above the measured noise level for each depth interval and time-varied gain correction. Temperature effects were not observed. Parallel acoustics transects were spaced at 3-nmi intervals, perpendicular to the 200-m isobath, which is close to the mean depth of pollock distribution in the region (Hughes and Hirschhorn, 1979). Opportunistic midwater trawls were also done at night. Tows were done at a constant depth, aimed at layers of high backscatter, and averaged 25 min in duration. A Gourock, Inc., midwater trawl with vertical and horizontal openings of 12 and 22 m, respectively, was fitted with a 1.9 cm mesh cod end liner. Trawl stations were distributed throughout the study area in April, while they were concentrated in Aialik Bay and Dora Passage in August (Fig. 1). Trawl depths ranged between 22 and 87 m in April; and between 16 and 124 m in August (Table 2). Up to 200 random specimens of each species per tow were retained for shipboard measurements of FL, while additional fish were counted as they were released. Of the 200 retained fish, a sub-sample of 50 specimens was frozen for length–weight measurements in the laboratory.
Table 1 38-kHz settings for the HTI model 244 split-beam echosounder used within a 10-nmi radius around Chiswell Island in 2003 Absorption coefficient Beam angle Ping rate Equivalent beam angle Pulse length Source level Transmit power TVG Gain Note that 33 dBW = 1995.3 W. Calculation: W = 10(33/10) .
10.2 dB/km 7◦ 0.8 pps −29.05 dB re 1 sr 5 ms 217.78 dB re 1 Pa 33 dBW −2.4 dB
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
181
Table 2 Trawl catch data within a 10-nmi radius around Chiswell Island in 2003 Min HR (m)
Max FR (m)
Capelin (n)
Eulachon (n)
Herring (n)
Pollock (n)
Other (n)
April Trawl 1 Trawl 2 Trawl 3 Trawl 4 Trawl 5 Trawl 6 Trawl 7
29.3 25.6 22.0 22.0 25.6 38.4 25.6
46.8 38.4 78.6 62.2 86.6 58.5 76.8
– – 2 2 1 – –
– – 3 – – 6 –
– – – – 6 – –
63 79 378 23 92 183 1107
– – 1a – – – –
August Trawl 1 Trawl 2 Trawl 3 Trawl 4 Trawl 5
100.6 18.2 18.3 23.8 15.5
124.4 49.4 43.9 47.5 36.6
16 1 – – –
1 – – – –
– 441 127 4 3
11 13 315 85 1
2b 11c 2d – 4e
Min HR: minimum headrope depth during tow. Max FR: maximum footrope depth during tow. Note that Min HR and Max HR were recorded at the start and end of acoustic data collection and do not include deployment and retrieval of the net. n: number of fish. a Arrowtooth flounder (Atheresthes stomias). b Juvenile salmonids (Oncorhynchus spp.). c Includes six arrowtooth flounder, four juvenile salmonids and one spiny dogfish (Squalus acanthias). d Prowfish (Zaprora silenus). e Sablefish (Anoplopoma fimbria).
2.2. Laboratory procedures FL for all frozen specimens was measured to the nearest mm and weighed to the nearest 0.1 g. As total length (TL) measurements are required for the use of the target strength (TS) models of Gauthier and Horne (2004), the FL measurements for pollock, capelin (Mallotus villosus) and eulachon (Thaleichthys pacificus) were converted to TL with the equations provided in Buchheister and Wilson (2005). Similarly, FL measurements for herring were converted with the equations provided in Karpov and Kwiecien (1988). Shrinkage due to freezing (ca. 1–3%) was also corrected for pollock and eulachon length measurements (Buchheister and Wilson, 2005). A similar correction for herring is currently not available. Length–weight regressions were calculated for pollock, herring and eulachon. No capelin were collected for laboratory measurements and so the equation provided in Brown (2002) was used to generate a length–weight regression.
Table 3 Target strength–length equations from Gauthier and Horne (2004) used to partition echograms within a 10-nmi radius around Chiswell Island in 2003 Capelin Eulachon Pacific herring Walleye pollock L: total length in cm.
transect. Furthermore, we note that trawls 1 and 5 were excluded from the August species mix because the former was the only tow in the 100–150-m depth bin, and because of the low number of fish (n = 8) in the latter tow (Table 2). Once the species mix for each cruise was calculated, echointegrals Ei were partitioned with:
Ei =
wi i Em
2.3. Echogram processing
M wi =
q /qk k=1 ik M
I
i=1
Echo-integration data were processed with Echoview software (SonarData Pty. Ltd., Hobart, Tasmania, Australia). Echo-integrals contaminated by the intrusion of bottom echoes into the water column were defined as bad data regions in Echoview and omitted from further analysis (Simmonds and MacLennan, 2005). The April catch data were combined into a species mix to partition all the echograms for that cruise using the equation: (1)
where wi is the proportion w of species i, qik is the quantity of species i caught at station k, qk is the total catch at station k, and M is the total number of trawls for the respective cruise (Simmonds and MacLennan, 2005). August trawls 3 and 4 were combined into a species mix to partition all the echograms for that cruise with Eq. (1). The only exception to this procedure was n = 3 small schools (<5% of transects) that could be identified as herring based on school structure. These schools were partitioned with the catch data for trawl 2 (Table 2), which was aimed at a herring school. The biomass for each school was added to the species mix biomass for the respective
24.9 log L–75 27.3 log L–94 13.3 log L–55.9 19.2 log L–66
(2)
wi i
where I is the total number of species, i is the mean backscattering cross-section of species i, and Em is the volume backscattering coefficient (Simmonds and MacLennan, 2005). The mean backscattering cross-section of species i was calculated as
i =
wij 10[ai +bi log(Lij )/10]
(3)
j
where Lij is the midpoint of the jth size class for species i, and ai and bi are constants for the ith species (Table 3). Note that only pollock were split into different size classes because it is known that juvenile SSLs feed on smaller pollock (Merrick and Calkins, 1996). This splitting was done by visual inspection of peaks and troughs in length–frequency histograms (Anderson and Neumann, 1996). Backscatter not attributed to capelin, eulachon, herring or pollock was considered “other.” The abundance of species i, expressed as the number/nmi2 , was calculated using: i =
(1852)2
z2 z1
i
/V
dz (4)
182
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
Table 4 Summary of length–weight regressions used to convert abundance estimates to biomass within a 10-nmi radius around Chiswell Island in 2003 n Eulachon Herring Pollock
40 211 379
Min (cm) 16.4 7.0 9.7
Max (cm) 23.3 27.0 52.6
Regression 3.0697
0.0046L 0.008L3.0555 0.0095L2.9095
r2 0.84 0.94 0.97
Weight is expressed in g. n: number of fish. L: length of fish in cm.
where V is the volume insonified and z is depth. This abundance estimate was then converted to kg/nmi2 using the length–weight regressions described above (Simmonds and MacLennan, 2005). Biomass data were exported from Echoview in two depth bins: 20–50 m, and 50–100 m. This was done for two reasons. One, to facilitate comparison with previous studies of SSL diving behavior that used the same depth bins (e.g., Merrick and Loughlin, 1997; Loughlin et al., 2003). Two, to minimize the vertical aspect of the error associated with partitioning echograms based on trawl catch data, because only two trawls were used to calculate wi for August. This was done by rounding the minimum/maximum headrope/footrope depths up or down to the nearest bin cutoff. For example, the minimum headrope depth for trawls 3 and 4 in August was 18.3 m, while the maximum footrope depth was 47.5 m; the closest bin cutoffs were 20 and 50 m; thus, August echo-integration data were partitioned to biomass between 20 and 50 m.
Fig. 2. Length–frequency histograms for walleye pollock sampled by midwater trawl within a 10-nmi radius around Chiswell Island in 2003.
2.4. Biomass estimates A retrospective analysis was done on the elementary distance sampling unit (EDSU), which is defined as the distance along a transect over which acoustic measurements are averaged to give one sample. If the EDSU is too large, information about the geographical distribution of the stock will be lost, whereas if it is too small, successive samples will be dominated by local variability (Simmonds and MacLennan, 2005, p. 324). Thus, data were binned at 0.1, 0.5 and 1.0 nmi, and experimental semivariograms (Rivoirard et al., 2000) were generated for each bin size in each month with geoR (http://www.est.ufpr.br/geoR/) to determine the most appropriate EDSU. It was found that binning at 0.5 nmi was more appropriate than 0.1 or 1.0 nmi. Adjacent acoustic measurements exhibit autocorrelation (Simmonds and MacLennan, 2005, p. 324), so calculating a mean biomass and 95% confidence limit for each species would be a form of pseudoreplication (Hurlbert, 1984; Millar and Anderson, 2004). To account for the spatial correlation in the data we estimated the mean biomass and 95% confidence limit for each species with a generalized linear mixed model (Venables and Dichmont, 2004) in PROC MIXED (SAS Institute, Cary, NC, U.S.A.), which allows the use of spatially correlated error terms in the standard linear model Y = X + . Error terms were modeled with a geostatistical spherical covariance structure (Rivoirard et al., 2000). One of the assumptions of a generalized linear mixed model is that the data are normally distributed. Even after logtransformation all biomass data still failed a Shapiro–Wilk test for normality (Zar, 1999). Thus, we present 95% confidence limits with the caveat that they are only approximate intervals. 3. Results 3.1. Laboratory procedures Length–weight regressions for pollock, herring and eulachon are presented in Table 4. The length–frequency histogram for April pollock showed clear peaks at 14.5 and 36.0 cm (Fig. 2), which corresponds to age-1
juveniles and age-3 adults, respectively (Hughes and Hirschhorn, 1979; Shima et al., 2002; Duffy-Anderson et al., 2003). August peaks occurred at 9.0, 19.0 and 32.5 cm, which correspond to age-0 juveniles, age-1 juveniles and age-3 adults, respectively. For brevity, the age-3 fish will be referred to as adults for the remainder of this paper, even though the left hand tail of this distribution clearly contains some age-2 juveniles. 3.2. Echogram processing Values used to partition echograms into the species mix for each cruise are presented in Table 5. 3.3. Biomass estimates Adult pollock was the dominant biomass in the 20–50 m depth bin in April, averaging 122.8 kg/nmi2 (Table 6). This was two orders of magnitude greater than the mean biomass of herring and age-1 pollock, both of which averaged 2.8 kg/nmi2 . Eulachon and capelin Table 5 Data used to partition echograms into species mix within a 10-nmi radius around Chiswell Island in 2003 TS (dB)
TL (cm)
Weight (g)
% mix
April Capelin Eulachon Herring Pollock, age-1 Pollock, adult Other
−49.72 −57.89 −37.36 −43.98 −36.28 NA
10.4 21.0 24.8 14.0 35.3 NA
6.7 52.9 145.7 20.6 303.3 NA
1.36 0.57 0.87 6.12 91.05 0.04
August Herring Pollock, age-0 Pollock, age-1 Pollock, adult Other
−37.63 −48.33 −41.41 −36.44 NA
23.7 8.3 19.1 34.6 NA
126.2 4.5 50.6 286.0 NA
18.11 2.57 0.34 78.76 0.22
TS: target strength. TL: average total length.
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
183
Table 6 Predicted mean biomass (Yˆ ), arithmetic mean biomass (¯x), and respective 95% confidence intervals within a 10-nmi radius of Chiswell Island, 2003 Confidence limits Yˆ
Upper
x¯
0.12 0.28 0.93 0.92 9.52
0.86 2.20 6.51 6.48 1455.13
0.43 1.02 2.75 2.74 110.83
0.32 0.77 2.08 2.07 66.47
0.56 1.30 3.58 3.57 184.35
0.84 1.91 5.15 5.13 271.40
0.00a 0.27 1.02 1.02 33.51
2.86 5.69 17.69 17.62 2148.31
0.81 1.87 4.93 4.92 229.80
0.63 1.43 3.66 3.64 134.42
1.00 2.38 6.56 6.54 392.35
65.56 0.83 1.20 457.88
18.36 0.46 0.65 94.61
227.88 1.30 1.93 2201.42
62.56 0.74 1.07 453.29
45.98 0.59 0.86 315.54
85.00 0.90 1.31 650.99
April 20–50 m Capelin Eulachon Herring Pollock, age-1 Pollock, adult
0.44 1.02 2.81 2.79 122.77
50–100 m Capelin Eulachon Herring Pollock, age-1 Pollock, adult August 20–50 m Herring Pollock, age-0 Pollock, age-1 Pollock, adult a
Confidence limits
Lower
Lower
Upper
Impossible negative confidence limit replaced with 0.
were the least abundant, averaging 1.02 and 0.44 kg/nmi2 , respectively. Trends for the 50–100 m depth bin in April were similar, except that mean biomass was greater than that of the shallower depth bin (Table 6). For example, average biomass of adult pollock was 271.4 kg/nmi2 in the 50–100 m depth bin, as compared with 122.8 kg/nmi2 in the 20–50 m depth bin. Adult pollock was the dominant biomass in the 20–50-m depth bin in August, averaging 457.9 kg/nmi2 (Table 6). Herring was an order of magnitude lower at 65.6 kg/nmi2 . Age-1 and age-0 pollock were the third and fourth most abundant, with a mean biomass of 1.2 and 0.8 kg/nmi2 , respectively. The biomass of adult pollock in the 20–50 m depth bin increased from 122.8 to 457.9 kg/nmi2 between April and August, respectively (Table 6). Similarly, there was a 23-fold increase in the mean biomass of herring, from 2.8 to 65.6 kg/nmi2 . The biomass of age1 pollock decreased slightly from 2.8 to 1.2 kg/nmi2 . Age-0 pollock appeared in August, while capelin and eulachon were not observed. A comparison of the standard confidence limits in Table 6 with the approximate intervals derived from accounting for the spatial correlation in the data illustrate that any inferences with respect to potentially significant differences between different categories should be made with caution. For example, the standard confidence interval for August herring (46.0–85.0 kg/nmi2 ) does not overlap with the interval for adult pollock (315.5–651.0 kg/nmi2 ); but once they are adjusted for spatial correlation the interval for herring (18.4–227.9 kg/nmi2 ) does indeed overlap with the interval for adult pollock (94.6–2201.4 kg/nmi2 ).
4. Discussion 4.1. Error budget Acoustically derived biomass estimates contain both sampling error, caused by the measurements being stochastic samples of the true mean density, and systematic error, which affects all the observations equally. Examples of the latter include equipment sensitivity, transducer motion, the surface bubble layer, etc. (Simmonds and MacLennan, 2005).
The primary source of sampling error results from survey design (Simmonds and MacLennan, 2005). Other sources of sampling error are variation in the mean TS over the surveyed region; species discrimination through echogram partitioning; diurnal variability affecting fish distribution and TS; sampling echogram marks by fishing to provide species proportions; and sudden changes in fish distribution and behavior due to storm events, etc. (Simmonds and MacLennan, 2005). Given the small size of the study area, it was assumed that we were dealing with a single fish population, and so it follows that variation in mean TS would be negligible. All echograms were categorized as a species mix based on trawl catch, so any error resulting from species discrimination based on echogram characteristics would have been restricted to the three small herring schools in August. Transects were run between evening and morning civil twilights, during the period when vertical migrators are near the surface, so it was assumed that there was no diurnal variability in fish distribution or TS during data collection. Finally, there was no major storm event during any of the cruises so it is unlikely there were major changes in fish distribution over the course of each cruise. This leaves partitioning echogram marks based on species proportions in trawl catch as the largest source of other sampling error in this study. A minimum of five to seven trawl stations are needed per 100 nmi2 in order to minimize the variance in abundance estimates (Godø et al., 1998). Although we met these minimum criteria in both cruises, not all catch data were used to calculate wi in August due to the concerns described in Section 2.3. Our method of partitioning echograms based on as little as two tows undoubtedly introduced the largest source of error in this study. However, we attempted to minimize the vertical aspect of this error by not extrapolating species composition and biomass estimates outside trawl depths. 4.2. Pelagic fish biomass available to Chiswell SSLs Our objective in this study was to provide a seasonal assessment of pelagic fish biomass available to SSLs foraging around the Chiswell Island rookery in 2003. We presented our data in depth bins comparable with previous studies of diving behavior
184
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185
in SSLs (e.g., Merrick and Loughlin, 1997; Loughlin et al., 2003). Our results offer several potential insights into the foraging behavior of Chiswell SSLs. In April there was more fish biomass in the 50–100 m depth bin as compared with the 20–50 m depth bin. This is consistent with the observation that female SSLs dive more in depth bins >50 m in late winter as compared with summer (Merrick and Loughlin, 1997). There were two seasonal differences in the 20–50 m depth bin between April and August. One was an increase in the biomass of adult pollock. This is important because the historical mean size of pollock consumed by SSLs in the Chiswell region was 29.8 cm (Pitcher, 1981), and contemporary analysis indicates that 75% of consumed pollock are ≥27.0 cm (Jason Waite, Alaska SeaLife Center, unpublished data). As we noted in Section 3.1, these sizes are considered adults in the GOA (Hughes and Hirschhorn, 1979; Shima et al., 2002; Duffy-Anderson et al., 2003). The other seasonal trend in the 20–50 m depth bin was an increase in the biomass of forage fishes. The clearest evidence for this was the ca. 23-fold increase in the mean biomass of herring between April and August. Furthermore, trawls 1 and 2 suggest the appearance of juvenile salmonids around the rookery at this time. The latter observation is consistent with other reports that juvenile pink salmon (Oncorhynchus gorbuscha) move into the eastern side of our study area in August (Armstrong et al., 2005). These findings are important because current analysis of Chiswell SSL scats suggests that salmonids are the primary prey item, comprising 71% of the diet by frequency of occurrence, while herring rank third at 37% (Jason Waite, Alaska SeaLife Center, unpublished data). The increased availability of adult pollock, herring and salmon in August is coincident with an increase in the duration of foraging trips by lactating Chiswell SSLs (Maniscalco et al., 2006). This suggests that female SSLs increase their foraging trip durations in response to increased prey availability. However, this conclusion should be viewed with caution, given that the observations by Maniscalco et al. (2006) were done during daylight hours, while our surveys were done at night. Increased foraging trip durations in the presence of increased prey availability can be explained by optimal foraging theory (Stephens and Krebs, 1986). Laboratory study of shallow dives by female SSLs found that increased prey encounter rates resulted in increased dive durations and foraging times (Cornick and Horning, 2003). In contrast, individual dive duration and foraging time were reduced in cost-increased dives that were comparable with reduced prey encounter rates (Cornick et al., 2006). Although we did not do any tows less than 15 m, our results offer some tentative insights into prey availability at the shallower depths utilized by juvenile SSLs (e.g., Merrick and Loughlin, 1997). There was a ca. 50% reduction in the mean biomass of age-1 pollock between April and August. This is important because this is the age-class of pollock that is comparable to the peak size of pollock that juvenile SSLs feed most heavily on (Fig. 3 in Merrick and Calkins, 1996). On the other hand, our trawl catch data indicate the appearance of at least small numbers of age-0 pollock around the rookery in August. This study supports the general view that SSLs in the northern GOA have not been food-limited during recent years, at least during summer months (Maniscalco et al., 2006). One of the possible explanations for the change in foraging trip durations of lactating Chiswell SSLs is thought to be a seasonal change in prey availability (Maniscalco et al., 2006). Our finding of an increased biomass of adult pollock, herring and salmonids in August 2003 supports this observation. Obviously, additional surveys are needed to confirm whether these trends are representative of other years. Nevertheless, our study illustrates some of the unique insights that
can be gained by linking fisheries acoustics surveys with studies of the foraging behavior and diet of endangered Steller sea lions. Acknowledgements We thank the crews of the F/V Nightwatch and R/V Pandalus, ´ as well as Stephane Gauthier, Elizabeth Moundalexis and Sarah Norberg for assistance with field collections. We also thank Leslie Cornick for helpful discussions about SSL diving behavior and Susan Inglis for suggestions regarding an earlier draft of this manuscript. Alexei Pinchuk provided Visual Basic code to fix damaged HTI files. Margaret Spahn provided the bathymetry shapefile. Support for this project was provided by the Alaska SeaLife Center Steller Sea Lion Research Program with funds from National Marine Fisheries Service. References Alverson, D.L., 1992. A review of commercial fisheries and the Steller sea lion (Eumetopias jubatus): the conflict arena. Rev. Aquat. Sci. 6, 203–256. Anderson, R.O., Neumann, R.M., 1996. Length, weight, and associated structural indices. In: Murphy, B.R., Willis, D.W. (Eds.), Fisheries Techniques. American Fisheries Society, Bethesda, MD, pp. 447–481. Armstrong, J.L., Boldt, J.L., Cross, A.D., Moss, J.H., Davis, N.D., Myers, K.W., Walker, R.V., Beauchamp, D.A., Haldorson, L.J., 2005. Distribution, size, and interannual, seasonal and diel food habits of northern Gulf of Alaska juvenile pink salmon, Oncorhynchus gorbuscha. Deep-Sea Res. Part II 52, 247–265. Braham, H.W., Everitt, R.D., Rugh, D.J., 1980. Northern sea lion population declines in the eastern Aleutian Islands. J. Wildl. Manage. 44, 25–33. Brown, E.D., 2002. Life history, distribution, and size structure of Pacific capelin in Prince William Sound and the Northern Gulf of Alaska. ICES J. Mar. Sci. 59, 983–996. Buchheister, A., Wilson, M.T., 2005. Shrinkage correction and length conversion equations for Theragra chalcogramma, Mallotus villosus and Thaleichthys pacificus. J. Fish Biol. 67, 541–548. Cornick, L.A., Horning, M., 2003. A test of hypotheses based on optimal foraging considerations for a diving mammal using a novel experimental approach. Can. J. Zool. 81, 1799–1807. Cornick, L.A., Inglis, S.D., Willis, K., Horning, M., 2006. Effects of increased swimming costs on foraging behavior and efficiency of captive Steller sea lions: evidence for behavioral plasticity in the recovery phase of dives. J. Exp. Mar. Biol. Ecol. 333, 306–314. Duffy-Anderson, J.T., Ciannelli, L., Honkalehto, T., Bailey, K.M., Sogard, S.M., Springer, A.M., Buckley, T., 2003. Distribution of age-1 and age-2 walleye pollock in the Gulf of Alaska and eastern Bering Sea: sources of variation and implications for higher trophic levels. In: Browman, H.I., Skiftesvik, A.B. (Eds.), The Big Fish Bang. Proceedings of the 26th Annual Larval Fish Conference, Bergen 2002. Institute of Marine Research, Bergen, pp. 381–394. Foote, K.G., Knudsen, H.P., Vestnes, G., MacLennan, D.N., Simmonds, E.J., 1987. Calibration of acoustic instruments for fish density estimation: a practical guide. ICES Coop. Res. Rep. 144. International Council for the Exploration of the Sea, Copenhagen. Gauthier, S., Horne, J.K., 2004. Acoustic characteristics of forage fish species in the Gulf of Alaska and Bering Sea based on Kirchoff-approximation models. Can. J. Fish. Aquat. Sci. 61, 1839–1850. Godø, O.R., Karp, W.A., Totland, A., 1998. Effects of trawl sampling variability on precision of acoustic abundance estimates of gadoids from the Barents Sea and the Gulf of Alaska. ICES J. Mar. Sci. 55, 86–94. Higgins, L.V., Costa, D.P., Huntley, A.C., Le Boeuf, B.J., 1988. Behavioral and physiological measurements of maternal investment in the Steller sea lion, Eumetopias jubatus. Mar. Mamm. Sci. 4, 44–58. Hurlbert, S.H., 1984. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54, 187–211. Hughes, S.E., Hirschhorn, G., 1979. Biology of walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska 1973–75. Fish. Bull. 77, 263–274. Karpov, K.A., Kwiecien, G.S., 1988. Conversions between total, fork, and standard lengths for 41 species in 15 families of fish from California using preserved and fresh specimens. California Department of Fish and Game Marine Resources Administrative Report No. 88-9, 17 pp. Loughlin, T.R., Sterling, J.T., Merrick, R.L., Sease, J.L., York, A.E., 2003. Diving behavior of immature Steller sea lions (Eumetopias jubatus). Fish. Bull. 101, 566–582. Maniscalco, J., Atkinson, S., Armato, P., 2002. Early maternal care and pup survival in Steller sea lions. A remote video monitoring project in the northern Gulf of Alaska. Arctic Res. U.S. 16, 36–41. Maniscalco, J.M., Parker, P., Atkinson, S., 2006. Interseasonal and interannual measures of maternal care among individual Steller sea lions (Eumetopias jubatus). J. Mamm. 87, 304–311.
C.F. Adams et al. / Fisheries Research 93 (2008) 179–185 Merrick, R.L., 1997. Current and historical roles of apex predators in the Bering Sea ecosystem. J. Northwest Atl. Fish. Sci. 22, 343–355. Merrick, R.L., Calkins, D.G., 1996. Importance of juvenile walleye pollock, Theragra chalcogramma, in the diet of Gulf of Alaska Steller sea lions, Eumetopias jubatus. In: Brodeur, R.D., Livingston, P.A., Loughlin, T.R., Hollowed, A.B. (Eds.), Ecology of Juvenile Walleye Pollock (Theragra chalcogramma). NOAA Tech. Rep. NMFSAFSC-126, pp. 153–166. Merrick, R.L., Loughlin, T.R., 1997. Foraging behavior of adult female and young-ofthe-year Steller sea lions in Alaskan waters. Can. J. Zool. 75, 776–786. Merrick, R.L., Loughlin, T.R., Calkins, D.G., 1987. Decline in abundance of the northern sea lion, Eumetopias jubatus, in Alaska 1956–86. Fish. Bull. 85, 351–365. Millar, R.B., Anderson, M.J., 2004. Remedies for pseudoreplication. Fish. Res. 70, 397–407. Pitcher, K.W., 1981. Prey of the Steller sea lion, Eumetopias jubatus, in the Gulf of Alaska. Fish. Bull. 79, 467–472. Ransom, B.H., Steig, T.W., Johnston, S.V., Ehrenberg, J.E., Torkelson, T.C., Nealson, P.A., 1995. Using Hydroacoustics for Fisheries Assessment. Hydroacoustic Technology, Inc., Seattle.
185
Rivoirard, J., Simmonds, J., Foote, K.G., Fernandes, P., Bez, N., 2000. Geostatistics for Estimating Fish Abundance. Blackwell Science Ltd., Oxford. Shima, M., Hollowed, A.B., VanBlaricom, G.R., 2002. Changes over time in the spatial distribution of walleye pollock (Theragra chalcogramma) in the Gulf of Alaska 1984–1996. Fish. Bull. 100, 307–323. Simmonds, J., MacLennan, D., 2005. Fisheries Acoustics. Blackwell Science Ltd., Oxford. Stephens, D.W., Krebs, J.R., 1986. Foraging Theory. Princeton University Press, New Jersey. Thomas, G.L., Thorne, R.E., 2001. Night-time predation by Steller sea lions. Nature 411, 1013. Trites, A.W., Porter, B.W., 2002. Attendance patterns of Steller sea lions (Eumetopias jubatus) and their young during winter. J. Zool. Lond. 256, 547–556. Venables, W.N., Dichmont, C.M., 2004. GLMs, GAMs and GLMMs: an overview of theory for application in fisheries research. Fish. Res. 70, 319–337. York, A.E., 1994. The population dynamics of northern sea lions 1975–1985. Mar. Mamm. Sci. 10, 38–51. Zar, J.H., 1999. Biostatistical Analysis. Prentice-Hall, Upper Saddle River, New Jersey.